tonyair
/
silverstripe-fulltextsearch
mirror of https://github.com/silverstripe/silverstripe-fulltextsearch
1
0
Fork 0
Code Issues Projects Releases Wiki Activity

Converted to PSR-2

This commit is contained in:
helpfulrobot 2015-11-21 19:19:20 +13:00
parent c96d4bf749
commit 314feddd48
44 changed files with 6459 additions and 5854 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

210
code/search/FullTextSearch.php
View File

@ -3,124 +3,134 @@
/**
* Base class to manage active search indexes.
*/
class FullTextSearch {
class FullTextSearch
{
protected static $all_indexes = null;
static protected $all_indexes = null;
protected static $indexes_by_subclass = array();
static protected $indexes_by_subclass = array();
/**
* Optional list of index names to limit to. If left empty, all subclasses of SearchIndex
* will be used
*
* @var array
* @config
*/
private static $indexes = array();
/**
* Optional list of index names to limit to. If left empty, all subclasses of SearchIndex
* will be used
*
* @var array
* @config
*/
private static $indexes = array();
/**
* Get all the instantiable search indexes (so all the user created indexes, but not the connector or library level
* abstract indexes). Can optionally be filtered to only return indexes that are subclasses of some class
*
* @static
* @param String $class - Class name to filter indexes by, so that all returned indexes are subclasses of provided class
* @param bool $rebuild - If true, don't use cached values
*/
public static function get_indexes($class = null, $rebuild = false)
{
if ($rebuild) {
self::$all_indexes = null;
self::$indexes_by_subclass = array();
}
/**
* Get all the instantiable search indexes (so all the user created indexes, but not the connector or library level
* abstract indexes). Can optionally be filtered to only return indexes that are subclasses of some class
*
* @static
* @param String $class - Class name to filter indexes by, so that all returned indexes are subclasses of provided class
* @param bool $rebuild - If true, don't use cached values
*/
static function get_indexes($class = null, $rebuild = false) {
if ($rebuild) {
self::$all_indexes = null;
self::$indexes_by_subclass = array();
}
if (!$class) {
if (self::$all_indexes === null) {
// Get declared indexes, or otherwise default to all subclasses of SearchIndex
$classes = Config::inst()->get(__CLASS__, 'indexes')
?: ClassInfo::subclassesFor('SearchIndex');
if (!$class) {
if (self::$all_indexes === null) {
// Get declared indexes, or otherwise default to all subclasses of SearchIndex
$classes = Config::inst()->get(__CLASS__, 'indexes')
?: ClassInfo::subclassesFor('SearchIndex');
$hidden = array();
$candidates = array();
foreach ($classes as $class) {
// Check if this index is disabled
$hides = $class::config()->hide_ancestor;
if ($hides) {
$hidden[] = $hides;
}
$hidden = array();
$candidates = array();
foreach ($classes as $class) {
// Check if this index is disabled
$hides = $class::config()->hide_ancestor;
if($hides) {
$hidden[] = $hides;
}
// Check if this index is abstract
$ref = new ReflectionClass($class);
if (!$ref->isInstantiable()) {
continue;
}
// Check if this index is abstract
$ref = new ReflectionClass($class);
if (!$ref->isInstantiable()) {
continue;
}
$candidates[] = $class;
}
$candidates[] = $class;
}
if ($hidden) {
$candidates = array_diff($candidates, $hidden);
}
if($hidden) {
$candidates = array_diff($candidates, $hidden);
}
// Create all indexes
$concrete = array();
foreach ($candidates as $class) {
$concrete[$class] = singleton($class);
}
// Create all indexes
$concrete = array();
foreach($candidates as $class) {
$concrete[$class] = singleton($class);
}
self::$all_indexes = $concrete;
}
self::$all_indexes = $concrete;
}
return self::$all_indexes;
} else {
if (!isset(self::$indexes_by_subclass[$class])) {
$all = self::get_indexes();
return self::$all_indexes;
}
else {
if (!isset(self::$indexes_by_subclass[$class])) {
$all = self::get_indexes();
$valid = array();
foreach ($all as $indexclass => $instance) {
if (is_subclass_of($indexclass, $class)) {
$valid[$indexclass] = $instance;
}
}
$valid = array();
foreach ($all as $indexclass => $instance) {
if (is_subclass_of($indexclass, $class)) $valid[$indexclass] = $instance;
}
self::$indexes_by_subclass[$class] = $valid;
}
self::$indexes_by_subclass[$class] = $valid;
}
return self::$indexes_by_subclass[$class];
}
}
return self::$indexes_by_subclass[$class];
}
}
/**
* Sometimes, like when in tests, you want to restrain the actual indexes to a subset
*
* Call with one argument - an array of class names, index instances or classname => indexinstance pairs (can be mixed).
* Alternatively call with multiple arguments, each of which is a class name or index instance
*
* From then on, fulltext search system will only see those indexes passed in this most recent call.
*
* Passing in no arguments resets back to automatic index list
*
* Alternatively you can use `FullTextSearch.indexes` to configure a list of indexes via config.
*/
public static function force_index_list()
{
$indexes = func_get_args();
/**
* Sometimes, like when in tests, you want to restrain the actual indexes to a subset
*
* Call with one argument - an array of class names, index instances or classname => indexinstance pairs (can be mixed).
* Alternatively call with multiple arguments, each of which is a class name or index instance
*
* From then on, fulltext search system will only see those indexes passed in this most recent call.
*
* Passing in no arguments resets back to automatic index list
*
* Alternatively you can use `FullTextSearch.indexes` to configure a list of indexes via config.
*/
static function force_index_list() {
$indexes = func_get_args();
// No arguments = back to automatic
if (!$indexes) {
self::get_indexes(null, true);
return;
}
// No arguments = back to automatic
if (!$indexes) {
self::get_indexes(null, true);
return;
}
// Arguments can be a single array
if (is_array($indexes[0])) {
$indexes = $indexes[0];
}
// Arguments can be a single array
if (is_array($indexes[0])) $indexes = $indexes[0];
// Reset to empty first
self::$all_indexes = array();
self::$indexes_by_subclass = array();
// Reset to empty first
self::$all_indexes = array(); self::$indexes_by_subclass = array();
// And parse out alternative type combos for arguments and add to allIndexes
foreach ($indexes as $class => $index) {
if (is_string($index)) { $class = $index; $index = singleton($class); }
if (is_numeric($class)) $class = get_class($index);
self::$all_indexes[$class] = $index;
}
}
// And parse out alternative type combos for arguments and add to allIndexes
foreach ($indexes as $class => $index) {
if (is_string($index)) {
$class = $index;
$index = singleton($class);
}
if (is_numeric($class)) {
$class = get_class($index);
}
self::$all_indexes[$class] = $index;
}
}
}

1169
code/search/SearchIndex.php
View File

@ -27,576 +27,647 @@
* - Specifying update rules that are not extractable from metadata (because the values come from functions for instance)
*
*/
abstract class SearchIndex extends ViewableData {
/**
* Allows this index to hide a parent index. Specifies the name of a parent index to disable
*
* @var string
* @config
*/
private static $hide_ancestor;
public function __construct() {
parent::__construct();
$this->init();
foreach ($this->getClasses() as $class => $options) {
SearchVariant::with($class, $options['include_children'])->call('alterDefinition', $class, $this);
}
$this->buildDependancyList();
}
public function __toString() {
return 'Search Index ' . get_class($this);
}
/**
* Examines the classes this index is built on to try and find defined fields in the class hierarchy for those classes.
* Looks for db and viewable-data fields, although can't nessecarily find type for viewable-data fields.
*/
public function fieldData($field, $forceType = null, $extraOptions = array()) {
$fullfield = str_replace(".", "_", $field);
$sources = $this->getClasses();
foreach ($sources as $source => $options) {
$sources[$source]['base'] = ClassInfo::baseDataClass($source);
$sources[$source]['lookup_chain'] = array();
}
$found = array();
if (strpos($field, '.') !== false) {
$lookups = explode(".", $field);
$field = array_pop($lookups);
foreach ($lookups as $lookup) {
$next = array();
foreach ($sources as $source => $options) {
$class = null;
foreach (SearchIntrospection::hierarchy($source, $options['include_children']) as $dataclass) {
$singleton = singleton($dataclass);
if ($hasOne = $singleton->has_one($lookup)) {
$class = $hasOne;
$options['lookup_chain'][] = array(
'call' => 'method', 'method' => $lookup,
'through' => 'has_one', 'class' => $dataclass, 'otherclass' => $class, 'foreignkey' => "{$lookup}ID"
);
}
else if ($hasMany = $singleton->has_many($lookup)) {
$class = $hasMany;
$options['multi_valued'] = true;
$options['lookup_chain'][] = array(
'call' => 'method', 'method' => $lookup,
'through' => 'has_many', 'class' => $dataclass, 'otherclass' => $class, 'foreignkey' => $singleton->getRemoteJoinField($lookup, 'has_many')
);
}
else if ($manyMany = $singleton->many_many($lookup)) {
$class = $manyMany[1];
$options['multi_valued'] = true;
$options['lookup_chain'][] = array(
'call' => 'method', 'method' => $lookup,
'through' => 'many_many', 'class' => $dataclass, 'otherclass' => $class, 'details' => $manyMany
);
}
if ($class) {
if (!isset($options['origin'])) $options['origin'] = $dataclass;
$next[$class] = $options;
continue 2;
}
}
}
if (!$next) return $next; // Early out to avoid excessive empty looping
$sources = $next;
}
}
foreach ($sources as $class => $options) {
$dataclasses = SearchIntrospection::hierarchy($class, $options['include_children']);
while (count($dataclasses)) {
$dataclass = array_shift($dataclasses);
$type = null; $fieldoptions = $options;
$fields = DataObject::database_fields($dataclass);
if (isset($fields[$field])) {
$type = $fields[$field];
$fieldoptions['lookup_chain'][] = array('call' => 'property', 'property' => $field);
}
else {
$singleton = singleton($dataclass);
if ($singleton->hasMethod("get$field") || $singleton->hasField($field)) {
$type = $singleton->castingClass($field);
if (!$type) $type = 'String';
if ($singleton->hasMethod("get$field")) $fieldoptions['lookup_chain'][] = array('call' => 'method', 'method' => "get$field");
else $fieldoptions['lookup_chain'][] = array('call' => 'property', 'property' => $field);
}
}
if ($type) {
// Don't search through child classes of a class we matched on. TODO: Should we?
$dataclasses = array_diff($dataclasses, array_values(ClassInfo::subclassesFor($dataclass)));
// Trim arguments off the type string
if (preg_match('/^(\w+)\(/', $type, $match)) $type = $match[1];
// Get the origin
$origin = isset($fieldoptions['origin']) ? $fieldoptions['origin'] : $dataclass;
$found["{$origin}_{$fullfield}"] = array(
'name' => "{$origin}_{$fullfield}",
'field' => $field,
'fullfield' => $fullfield,
'base' => $fieldoptions['base'],
'origin' => $origin,
'class' => $dataclass,
'lookup_chain' => $fieldoptions['lookup_chain'],
'type' => $forceType ? $forceType : $type,
'multi_valued' => isset($fieldoptions['multi_valued']) ? true : false,
'extra_options' => $extraOptions
);
}
}
}
return $found;
}
/** Public, but should only be altered by variants */
protected $classes = array();
protected $fulltextFields = array();
public $filterFields = array();
protected $sortFields = array();
protected $excludedVariantStates = array();
/**
* Add a DataObject subclass whose instances should be included in this index
*
* Can only be called when addFulltextField, addFilterField, addSortField and addAllFulltextFields have not
* yet been called for this index instance
*
* @throws Exception
* @param String $class - The class to include
* @param array $options - TODO: Remove
*/
public function addClass($class, $options = array()) {
if ($this->fulltextFields || $this->filterFields || $this->sortFields) {
throw new Exception('Can\'t add class to Index after fields have already been added');
}
if (!DataObject::has_own_table($class)) {
throw new InvalidArgumentException('Can\'t add classes which don\'t have data tables (no $db or $has_one set on the class)');
}
$options = array_merge(array(
'include_children' => true
), $options);
$this->classes[$class] = $options;
}
/**
* Get the classes added by addClass
*/
public function getClasses() { return $this->classes; }
/**
* Add a field that should be fulltext searchable
* @param String $field - The field to add
* @param String $forceType - The type to force this field as (required in some cases, when not detectable from metadata)
* @param String $extraOptions - Dependent on search implementation
*/
public function addFulltextField($field, $forceType = null, $extraOptions = array()) {
$this->fulltextFields = array_merge($this->fulltextFields, $this->fieldData($field, $forceType, $extraOptions));
}
public function getFulltextFields() { return $this->fulltextFields; }
/**
* Add a field that should be filterable
* @param String $field - The field to add
* @param String $forceType - The type to force this field as (required in some cases, when not detectable from metadata)
* @param String $extraOptions - Dependent on search implementation
*/
public function addFilterField($field, $forceType = null, $extraOptions = array()) {
$this->filterFields = array_merge($this->filterFields, $this->fieldData($field, $forceType, $extraOptions));
}
public function getFilterFields() { return $this->filterFields; }
/**
* Add a field that should be sortable
* @param String $field - The field to add
* @param String $forceType - The type to force this field as (required in some cases, when not detectable from metadata)
* @param String $extraOptions - Dependent on search implementation
*/
public function addSortField($field, $forceType = null, $extraOptions = array()) {
$this->sortFields = array_merge($this->sortFields, $this->fieldData($field, $forceType, $extraOptions));
}
public function getSortFields() { return $this->sortFields; }
/**
* Add all database-backed text fields as fulltext searchable fields.
*
* For every class included in the index, examines those classes and all subclasses looking for "Text" database
* fields (Varchar, Text, HTMLText, etc) and adds them all as fulltext searchable fields.
*/
public function addAllFulltextFields($includeSubclasses = true) {
foreach ($this->getClasses() as $class => $options) {
foreach (SearchIntrospection::hierarchy($class, $includeSubclasses, true) as $dataclass) {
$fields = DataObject::database_fields($dataclass);
foreach ($fields as $field => $type) {
if (preg_match('/^(\w+)\(/', $type, $match)) $type = $match[1];
if (is_subclass_of($type, 'StringField')) $this->addFulltextField($field);
}
}
}
}
/**
* Returns an interator that will let you interate through all added fields, regardless of whether they
* were added as fulltext, filter or sort fields.
*
* @return MultipleArrayIterator
*/
public function getFieldsIterator() {
return new MultipleArrayIterator($this->fulltextFields, $this->filterFields, $this->sortFields);
}
public function excludeVariantState($state) {
$this->excludedVariantStates[] = $state;
}
/** Returns true if some variant state should be ignored */
public function variantStateExcluded($state) {
foreach ($this->excludedVariantStates as $excludedstate) {
$matches = true;
foreach ($excludedstate as $variant => $variantstate) {
if (!isset($state[$variant]) || $state[$variant] != $variantstate) { $matches = false; break; }
}
if ($matches) return true;
}
}
public $dependancyList = array();
public function buildDependancyList() {
$this->dependancyList = array_keys($this->getClasses());
foreach ($this->getFieldsIterator() as $name => $field) {
if (!isset($field['class'])) continue;
SearchIntrospection::add_unique_by_ancestor($this->dependancyList, $field['class']);
}
}
public $derivedFields = null;
/**
* Returns an array where each member is all the fields and the classes that are at the end of some
* specific lookup chain from one of the base classes
*/
public function getDerivedFields() {
if ($this->derivedFields === null) {
$this->derivedFields = array();
foreach ($this->getFieldsIterator() as $name => $field) {
if (count($field['lookup_chain']) < 2) continue;
$key = sha1($field['base'].serialize($field['lookup_chain']));
$fieldname = "{$field['class']}:{$field['field']}";
if (isset($this->derivedFields[$key])) {
$this->derivedFields[$key]['fields'][$fieldname] = $fieldname;
SearchIntrospection::add_unique_by_ancestor($this->derivedFields['classes'], $field['class']);
}
else {
$chain = array_reverse($field['lookup_chain']);
array_shift($chain);
$this->derivedFields[$key] = array(
'base' => $field['base'],
'fields' => array($fieldname => $fieldname),
'classes' => array($field['class']),
'chain' => $chain
);
}
}
}
return $this->derivedFields;
}
/**
* Get the "document ID" (a database & variant unique id) given some "Base" class, DataObject ID and state array
*
* @param String $base - The base class of the object
* @param Integer $id - The ID of the object
* @param Array $state - The variant state of the object
* @return string - The document ID as a string
*/
public function getDocumentIDForState($base, $id, $state) {
ksort($state);
$parts = array('id' => $id, 'base' => $base, 'state' => json_encode($state));
return implode('-', array_values($parts));
}
/**
* Get the "document ID" (a database & variant unique id) given some "Base" class and DataObject
*
* @param DataObject $object - The object
* @param String $base - The base class of the object
* @param Boolean $includesubs - TODO: Probably going away
* @return string - The document ID as a string
*/
public function getDocumentID($object, $base, $includesubs) {
return $this->getDocumentIDForState($base, $object->ID, SearchVariant::current_state($base, $includesubs));
}
/**
* Given an object and a field definition (as returned by fieldData) get the current value of that field on that object
*
* @param DataObject $object - The object to get the value from
* @param Array $field - The field definition to use
* @return Mixed - The value of the field, or null if we couldn't look it up for some reason
*/
protected function _getFieldValue($object, $field) {
set_error_handler(create_function('$no, $str', 'throw new Exception("HTML Parse Error: ".$str);'), E_ALL);
try {
foreach ($field['lookup_chain'] as $step) {
// Just fail if we've fallen off the end of the chain
if (!$object) return null;
// If we're looking up this step on an array or SS_List, do the step on every item, merge result
if (is_array($object) || $object instanceof SS_List) {
$next = array();
foreach ($object as $item) {
if ($step['call'] == 'method') {
$method = $step['method'];
$item = $item->$method();
}
else {
$property = $step['property'];
$item = $item->$property;
}
if ($item instanceof SS_List) $next = array_merge($next, $item->toArray());
elseif (is_array($item)) $next = array_merge($next, $item);
else $next[] = $item;
}
$object = $next;
}
// Otherwise, just call
else {
if ($step['call'] == 'method') {
$method = $step['method'];
$object = $object->$method();
}
elseif ($step['call'] == 'variant') {
$variants = SearchVariant::variants($field['base'], true);
$variant = $variants[$step['variant']]; $method = $step['method'];
$object = $variant->$method($object);
}
else {
$property = $step['property'];
$object = $object->$property;
}
}
}
}
catch (Exception $e) {
$object = null;
}
restore_error_handler();
return $object;
}
/**
* Given a class, object id, set of stateful ids and a list of changed fields (in a special format),
* return what statefulids need updating in this index
*
* Internal function used by SearchUpdater.
*
* @param $class
* @param $id
* @param $statefulids
* @param $fields
* @return array
*/
public function getDirtyIDs($class, $id, $statefulids, $fields) {
$dirty = array();
// First, if this object is directly contained in the index, add it
foreach ($this->classes as $searchclass => $options) {
if ($searchclass == $class || ($options['include_children'] && is_subclass_of($class, $searchclass))) {
$base = ClassInfo::baseDataClass($searchclass);
$dirty[$base] = array();
foreach ($statefulids as $statefulid) {
$key = serialize($statefulid);
$dirty[$base][$key] = $statefulid;
}
}
}
$current = SearchVariant::current_state();
// Then, for every derived field
foreach ($this->getDerivedFields() as $derivation) {
// If the this object is a subclass of any of the classes we want a field from
if (!SearchIntrospection::is_subclass_of($class, $derivation['classes'])) continue;
if (!array_intersect_key($fields, $derivation['fields'])) continue;
foreach (SearchVariant::reindex_states($class, false) as $state) {
SearchVariant::activate_state($state);
$ids = array($id);
foreach ($derivation['chain'] as $step) {
if ($step['through'] == 'has_one') {
$sql = new SQLQuery('"ID"', '"'.$step['class'].'"', '"'.$step['foreignkey'].'" IN ('.implode(',', $ids).')');
singleton($step['class'])->extend('augmentSQL', $sql);
$ids = $sql->execute()->column();
}
else if ($step['through'] == 'has_many') {
$sql = new SQLQuery('"'.$step['class'].'"."ID"', '"'.$step['class'].'"', '"'.$step['otherclass'].'"."ID" IN ('.implode(',', $ids).')');
$sql->addInnerJoin($step['otherclass'], '"'.$step['class'].'"."ID" = "'.$step['otherclass'].'"."'.$step['foreignkey'].'"');
singleton($step['class'])->extend('augmentSQL', $sql);
$ids = $sql->execute()->column();
}
}
SearchVariant::activate_state($current);
if ($ids) {
$base = $derivation['base'];
if (!isset($dirty[$base])) $dirty[$base] = array();
foreach ($ids as $id) {
$statefulid = array('id' => $id, 'state' => $state);
$key = serialize($statefulid);
$dirty[$base][$key] = $statefulid;
}
}
}
}
return $dirty;
}
/** !! These should be implemented by the full text search engine */
abstract public function add($object) ;
abstract public function delete($base, $id, $state) ;
abstract public function commit();
/** !! These should be implemented by the specific index */
/**
* Called during construction, this is the method that builds the structure.
* Used instead of overriding __construct as we have specific execution order - code that has
* to be run before _and/or_ after this.
*/
abstract public function init();
abstract class SearchIndex extends ViewableData
{
/**
* Allows this index to hide a parent index. Specifies the name of a parent index to disable
*
* @var string
* @config
*/
private static $hide_ancestor;
public function __construct()
{
parent::__construct();
$this->init();
foreach ($this->getClasses() as $class => $options) {
SearchVariant::with($class, $options['include_children'])->call('alterDefinition', $class, $this);
}
$this->buildDependancyList();
}
public function __toString()
{
return 'Search Index ' . get_class($this);
}
/**
* Examines the classes this index is built on to try and find defined fields in the class hierarchy for those classes.
* Looks for db and viewable-data fields, although can't nessecarily find type for viewable-data fields.
*/
public function fieldData($field, $forceType = null, $extraOptions = array())
{
$fullfield = str_replace(".", "_", $field);
$sources = $this->getClasses();
foreach ($sources as $source => $options) {
$sources[$source]['base'] = ClassInfo::baseDataClass($source);
$sources[$source]['lookup_chain'] = array();
}
$found = array();
if (strpos($field, '.') !== false) {
$lookups = explode(".", $field);
$field = array_pop($lookups);
foreach ($lookups as $lookup) {
$next = array();
foreach ($sources as $source => $options) {
$class = null;
foreach (SearchIntrospection::hierarchy($source, $options['include_children']) as $dataclass) {
$singleton = singleton($dataclass);
if ($hasOne = $singleton->has_one($lookup)) {
$class = $hasOne;
$options['lookup_chain'][] = array(
'call' => 'method', 'method' => $lookup,
'through' => 'has_one', 'class' => $dataclass, 'otherclass' => $class, 'foreignkey' => "{$lookup}ID"
);
} elseif ($hasMany = $singleton->has_many($lookup)) {
$class = $hasMany;
$options['multi_valued'] = true;
$options['lookup_chain'][] = array(
'call' => 'method', 'method' => $lookup,
'through' => 'has_many', 'class' => $dataclass, 'otherclass' => $class, 'foreignkey' => $singleton->getRemoteJoinField($lookup, 'has_many')
);
} elseif ($manyMany = $singleton->many_many($lookup)) {
$class = $manyMany[1];
$options['multi_valued'] = true;
$options['lookup_chain'][] = array(
'call' => 'method', 'method' => $lookup,
'through' => 'many_many', 'class' => $dataclass, 'otherclass' => $class, 'details' => $manyMany
);
}
if ($class) {
if (!isset($options['origin'])) {
$options['origin'] = $dataclass;
}
$next[$class] = $options;
continue 2;
}
}
}
if (!$next) {
return $next;
} // Early out to avoid excessive empty looping
$sources = $next;
}
}
foreach ($sources as $class => $options) {
$dataclasses = SearchIntrospection::hierarchy($class, $options['include_children']);
while (count($dataclasses)) {
$dataclass = array_shift($dataclasses);
$type = null;
$fieldoptions = $options;
$fields = DataObject::database_fields($dataclass);
if (isset($fields[$field])) {
$type = $fields[$field];
$fieldoptions['lookup_chain'][] = array('call' => 'property', 'property' => $field);
} else {
$singleton = singleton($dataclass);
if ($singleton->hasMethod("get$field") || $singleton->hasField($field)) {
$type = $singleton->castingClass($field);
if (!$type) {
$type = 'String';
}
if ($singleton->hasMethod("get$field")) {
$fieldoptions['lookup_chain'][] = array('call' => 'method', 'method' => "get$field");
} else {
$fieldoptions['lookup_chain'][] = array('call' => 'property', 'property' => $field);
}
}
}
if ($type) {
// Don't search through child classes of a class we matched on. TODO: Should we?
$dataclasses = array_diff($dataclasses, array_values(ClassInfo::subclassesFor($dataclass)));
// Trim arguments off the type string
if (preg_match('/^(\w+)\(/', $type, $match)) {
$type = $match[1];
}
// Get the origin
$origin = isset($fieldoptions['origin']) ? $fieldoptions['origin'] : $dataclass;
$found["{$origin}_{$fullfield}"] = array(
'name' => "{$origin}_{$fullfield}",
'field' => $field,
'fullfield' => $fullfield,
'base' => $fieldoptions['base'],
'origin' => $origin,
'class' => $dataclass,
'lookup_chain' => $fieldoptions['lookup_chain'],
'type' => $forceType ? $forceType : $type,
'multi_valued' => isset($fieldoptions['multi_valued']) ? true : false,
'extra_options' => $extraOptions
);
}
}
}
return $found;
}
/** Public, but should only be altered by variants */
protected $classes = array();
protected $fulltextFields = array();
public $filterFields = array();
protected $sortFields = array();
protected $excludedVariantStates = array();
/**
* Add a DataObject subclass whose instances should be included in this index
*
* Can only be called when addFulltextField, addFilterField, addSortField and addAllFulltextFields have not
* yet been called for this index instance
*
* @throws Exception
* @param String $class - The class to include
* @param array $options - TODO: Remove
*/
public function addClass($class, $options = array())
{
if ($this->fulltextFields || $this->filterFields || $this->sortFields) {
throw new Exception('Can\'t add class to Index after fields have already been added');
}
if (!DataObject::has_own_table($class)) {
throw new InvalidArgumentException('Can\'t add classes which don\'t have data tables (no $db or $has_one set on the class)');
}
$options = array_merge(array(
'include_children' => true
), $options);
$this->classes[$class] = $options;
}
/**
* Get the classes added by addClass
*/
public function getClasses()
{
return $this->classes;
}
/**
* Add a field that should be fulltext searchable
* @param String $field - The field to add
* @param String $forceType - The type to force this field as (required in some cases, when not detectable from metadata)
* @param String $extraOptions - Dependent on search implementation
*/
public function addFulltextField($field, $forceType = null, $extraOptions = array())
{
$this->fulltextFields = array_merge($this->fulltextFields, $this->fieldData($field, $forceType, $extraOptions));
}
public function getFulltextFields()
{
return $this->fulltextFields;
}
/**
* Add a field that should be filterable
* @param String $field - The field to add
* @param String $forceType - The type to force this field as (required in some cases, when not detectable from metadata)
* @param String $extraOptions - Dependent on search implementation
*/
public function addFilterField($field, $forceType = null, $extraOptions = array())
{
$this->filterFields = array_merge($this->filterFields, $this->fieldData($field, $forceType, $extraOptions));
}
public function getFilterFields()
{
return $this->filterFields;
}
/**
* Add a field that should be sortable
* @param String $field - The field to add
* @param String $forceType - The type to force this field as (required in some cases, when not detectable from metadata)
* @param String $extraOptions - Dependent on search implementation
*/
public function addSortField($field, $forceType = null, $extraOptions = array())
{
$this->sortFields = array_merge($this->sortFields, $this->fieldData($field, $forceType, $extraOptions));
}
public function getSortFields()
{
return $this->sortFields;
}
/**
* Add all database-backed text fields as fulltext searchable fields.
*
* For every class included in the index, examines those classes and all subclasses looking for "Text" database
* fields (Varchar, Text, HTMLText, etc) and adds them all as fulltext searchable fields.
*/
public function addAllFulltextFields($includeSubclasses = true)
{
foreach ($this->getClasses() as $class => $options) {
foreach (SearchIntrospection::hierarchy($class, $includeSubclasses, true) as $dataclass) {
$fields = DataObject::database_fields($dataclass);
foreach ($fields as $field => $type) {
if (preg_match('/^(\w+)\(/', $type, $match)) {
$type = $match[1];
}
if (is_subclass_of($type, 'StringField')) {
$this->addFulltextField($field);
}
}
}
}
}
/**
* Returns an interator that will let you interate through all added fields, regardless of whether they
* were added as fulltext, filter or sort fields.
*
* @return MultipleArrayIterator
*/
public function getFieldsIterator()
{
return new MultipleArrayIterator($this->fulltextFields, $this->filterFields, $this->sortFields);
}
public function excludeVariantState($state)
{
$this->excludedVariantStates[] = $state;
}
/** Returns true if some variant state should be ignored */
public function variantStateExcluded($state)
{
foreach ($this->excludedVariantStates as $excludedstate) {
$matches = true;
foreach ($excludedstate as $variant => $variantstate) {
if (!isset($state[$variant]) || $state[$variant] != $variantstate) {
$matches = false;
break;
}
}
if ($matches) {
return true;
}
}
}
public $dependancyList = array();
public function buildDependancyList()
{
$this->dependancyList = array_keys($this->getClasses());
foreach ($this->getFieldsIterator() as $name => $field) {
if (!isset($field['class'])) {
continue;
}
SearchIntrospection::add_unique_by_ancestor($this->dependancyList, $field['class']);
}
}
public $derivedFields = null;
/**
* Returns an array where each member is all the fields and the classes that are at the end of some
* specific lookup chain from one of the base classes
*/
public function getDerivedFields()
{
if ($this->derivedFields === null) {
$this->derivedFields = array();
foreach ($this->getFieldsIterator() as $name => $field) {
if (count($field['lookup_chain']) < 2) {
continue;
}
$key = sha1($field['base'].serialize($field['lookup_chain']));
$fieldname = "{$field['class']}:{$field['field']}";
if (isset($this->derivedFields[$key])) {
$this->derivedFields[$key]['fields'][$fieldname] = $fieldname;
SearchIntrospection::add_unique_by_ancestor($this->derivedFields['classes'], $field['class']);
} else {
$chain = array_reverse($field['lookup_chain']);
array_shift($chain);
$this->derivedFields[$key] = array(
'base' => $field['base'],
'fields' => array($fieldname => $fieldname),
'classes' => array($field['class']),
'chain' => $chain
);
}
}
}
return $this->derivedFields;
}
/**
* Get the "document ID" (a database & variant unique id) given some "Base" class, DataObject ID and state array
*
* @param String $base - The base class of the object
* @param Integer $id - The ID of the object
* @param Array $state - The variant state of the object
* @return string - The document ID as a string
*/
public function getDocumentIDForState($base, $id, $state)
{
ksort($state);
$parts = array('id' => $id, 'base' => $base, 'state' => json_encode($state));
return implode('-', array_values($parts));
}
/**
* Get the "document ID" (a database & variant unique id) given some "Base" class and DataObject
*
* @param DataObject $object - The object
* @param String $base - The base class of the object
* @param Boolean $includesubs - TODO: Probably going away
* @return string - The document ID as a string
*/
public function getDocumentID($object, $base, $includesubs)
{
return $this->getDocumentIDForState($base, $object->ID, SearchVariant::current_state($base, $includesubs));
}
/**
* Given an object and a field definition (as returned by fieldData) get the current value of that field on that object
*
* @param DataObject $object - The object to get the value from
* @param Array $field - The field definition to use
* @return Mixed - The value of the field, or null if we couldn't look it up for some reason
*/
protected function _getFieldValue($object, $field)
{
set_error_handler(create_function('$no, $str', 'throw new Exception("HTML Parse Error: ".$str);'), E_ALL);
try {
foreach ($field['lookup_chain'] as $step) {
// Just fail if we've fallen off the end of the chain
if (!$object) {
return null;
}
// If we're looking up this step on an array or SS_List, do the step on every item, merge result
if (is_array($object) || $object instanceof SS_List) {
$next = array();
foreach ($object as $item) {
if ($step['call'] == 'method') {
$method = $step['method'];
$item = $item->$method();
} else {
$property = $step['property'];
$item = $item->$property;
}
if ($item instanceof SS_List) {
$next = array_merge($next, $item->toArray());
} elseif (is_array($item)) {
$next = array_merge($next, $item);
} else {
$next[] = $item;
}
}
$object = $next;
}
// Otherwise, just call
else {
if ($step['call'] == 'method') {
$method = $step['method'];
$object = $object->$method();
} elseif ($step['call'] == 'variant') {
$variants = SearchVariant::variants($field['base'], true);
$variant = $variants[$step['variant']];
$method = $step['method'];
$object = $variant->$method($object);
} else {
$property = $step['property'];
$object = $object->$property;
}
}
}
} catch (Exception $e) {
$object = null;
}
restore_error_handler();
return $object;
}
/**
* Given a class, object id, set of stateful ids and a list of changed fields (in a special format),
* return what statefulids need updating in this index
*
* Internal function used by SearchUpdater.
*
* @param $class
* @param $id
* @param $statefulids
* @param $fields
* @return array
*/
public function getDirtyIDs($class, $id, $statefulids, $fields)
{
$dirty = array();
// First, if this object is directly contained in the index, add it
foreach ($this->classes as $searchclass => $options) {
if ($searchclass == $class || ($options['include_children'] && is_subclass_of($class, $searchclass))) {
$base = ClassInfo::baseDataClass($searchclass);
$dirty[$base] = array();
foreach ($statefulids as $statefulid) {
$key = serialize($statefulid);
$dirty[$base][$key] = $statefulid;
}
}
}
$current = SearchVariant::current_state();
// Then, for every derived field
foreach ($this->getDerivedFields() as $derivation) {
// If the this object is a subclass of any of the classes we want a field from
if (!SearchIntrospection::is_subclass_of($class, $derivation['classes'])) {
continue;
}
if (!array_intersect_key($fields, $derivation['fields'])) {
continue;
}
foreach (SearchVariant::reindex_states($class, false) as $state) {
SearchVariant::activate_state($state);
$ids = array($id);
foreach ($derivation['chain'] as $step) {
if ($step['through'] == 'has_one') {
$sql = new SQLQuery('"ID"', '"'.$step['class'].'"', '"'.$step['foreignkey'].'" IN ('.implode(',', $ids).')');
singleton($step['class'])->extend('augmentSQL', $sql);
$ids = $sql->execute()->column();
} elseif ($step['through'] == 'has_many') {
$sql = new SQLQuery('"'.$step['class'].'"."ID"', '"'.$step['class'].'"', '"'.$step['otherclass'].'"."ID" IN ('.implode(',', $ids).')');
$sql->addInnerJoin($step['otherclass'], '"'.$step['class'].'"."ID" = "'.$step['otherclass'].'"."'.$step['foreignkey'].'"');
singleton($step['class'])->extend('augmentSQL', $sql);
$ids = $sql->execute()->column();
}
}
SearchVariant::activate_state($current);
if ($ids) {
$base = $derivation['base'];
if (!isset($dirty[$base])) {
$dirty[$base] = array();
}
foreach ($ids as $id) {
$statefulid = array('id' => $id, 'state' => $state);
$key = serialize($statefulid);
$dirty[$base][$key] = $statefulid;
}
}
}
}
return $dirty;
}
/** !! These should be implemented by the full text search engine */
abstract public function add($object) ;
abstract public function delete($base, $id, $state) ;
abstract public function commit();
/** !! These should be implemented by the specific index */
/**
* Called during construction, this is the method that builds the structure.
* Used instead of overriding __construct as we have specific execution order - code that has
* to be run before _and/or_ after this.
*/
abstract public function init();
}
/**
* A search index that does nothing. Useful for testing
*/
abstract class SearchIndex_Null extends SearchIndex {
abstract class SearchIndex_Null extends SearchIndex
{
public function add($object)
{
}
public function add($object) { }
public function delete($base, $id, $state) { }
public function commit() { }
public function delete($base, $id, $state)
{
}
public function commit()
{
}
}
/**
* A search index that just records actions. Useful for testing
*/
abstract class SearchIndex_Recording extends SearchIndex {
abstract class SearchIndex_Recording extends SearchIndex
{
public $added = array();
public $deleted = array();
public $committed = false;
public $added = array();
public $deleted = array();
public $committed = false;
public function reset()
{
$this->added = array();
$this->deleted = array();
$this->committed = false;
}
public function reset() {
$this->added = array();
$this->deleted = array();
$this->committed = false;
}
public function add($object)
{
$res = array();
public function add($object) {
$res = array();
$res['ID'] = $object->ID;
foreach ($this->getFieldsIterator() as $name => $field) {
$val = $this->_getFieldValue($object, $field);
$res[$name] = $val;
}
$res['ID'] = $object->ID;
foreach ($this->getFieldsIterator() as $name => $field) {
$val = $this->_getFieldValue($object, $field);
$res[$name] = $val;
}
$this->added[] = $res;
}
$this->added[] = $res;
}
public function getAdded($fields = array())
{
$res = array();
public function getAdded($fields = array()) {
$res = array();
foreach ($this->added as $added) {
$filtered = array();
foreach ($fields as $field) {
if (isset($added[$field])) {
$filtered[$field] = $added[$field];
}
}
$res[] = $filtered;
}
foreach ($this->added as $added) {
$filtered = array();
foreach ($fields as $field) {
if (isset($added[$field])) $filtered[$field] = $added[$field];
}
$res[] = $filtered;
}
return $res;
}
return $res;
}
public function delete($base, $id, $state)
{
$this->deleted[] = array('base' => $base, 'id' => $id, 'state' => $state);
}
public function delete($base, $id, $state) {
$this->deleted[] = array('base' => $base, 'id' => $id, 'state' => $state);
}
public function commit()
{
$this->committed = true;
}
public function getIndexName()
{
return get_class($this);
}
public function commit() {
$this->committed = true;
}
public function getIndexName() {
return get_class($this);
}
public function getIsCommitted()
{
return $this->committed;
}
public function getIsCommitted() {
return $this->committed;
}
public function getService() {
// Causes commits to the service to be redirected back to the same object
return $this;
}
public function getService()
{
// Causes commits to the service to be redirected back to the same object
return $this;
}
}

137
code/search/SearchIntrospection.php
View File

@ -3,76 +3,91 @@
/**
* Some additional introspection tools that are used often by the fulltext search code
*/
class SearchIntrospection {
class SearchIntrospection
{
protected static $ancestry = array();
protected static $ancestry = array();
/**
* Check if class is subclass of (a) the class in $of, or (b) any of the classes in the array $of
* @static
* @param $class
* @param $of
* @return bool
*/
public static function is_subclass_of($class, $of)
{
$ancestry = isset(self::$ancestry[$class]) ? self::$ancestry[$class] : (self::$ancestry[$class] = ClassInfo::ancestry($class));
return is_array($of) ? (bool)array_intersect($of, $ancestry) : array_key_exists($of, $ancestry);
}
/**
* Check if class is subclass of (a) the class in $of, or (b) any of the classes in the array $of
* @static
* @param $class
* @param $of
* @return bool
*/
static function is_subclass_of ($class, $of) {
$ancestry = isset(self::$ancestry[$class]) ? self::$ancestry[$class] : (self::$ancestry[$class] = ClassInfo::ancestry($class));
return is_array($of) ? (bool)array_intersect($of, $ancestry) : array_key_exists($of, $ancestry);
}
protected static $hierarchy = array();
protected static $hierarchy = array();
/**
* Get all the classes involved in a DataObject hierarchy - both super and optionally subclasses
*
* @static
* @param String $class - The class to query
* @param bool $includeSubclasses - True to return subclasses as well as super classes
* @param bool $dataOnly - True to only return classes that have tables
* @return Array - Integer keys, String values as classes sorted by depth (most super first)
*/
public static function hierarchy($class, $includeSubclasses = true, $dataOnly = false)
{
$key = "$class!" . ($includeSubclasses ? 'sc' : 'an') . '!' . ($dataOnly ? 'do' : 'al');
if (!isset(self::$hierarchy[$key])) {
$classes = array_values(ClassInfo::ancestry($class));
if ($includeSubclasses) {
$classes = array_unique(array_merge($classes, array_values(ClassInfo::subclassesFor($class))));
}
/**
* Get all the classes involved in a DataObject hierarchy - both super and optionally subclasses
*
* @static
* @param String $class - The class to query
* @param bool $includeSubclasses - True to return subclasses as well as super classes
* @param bool $dataOnly - True to only return classes that have tables
* @return Array - Integer keys, String values as classes sorted by depth (most super first)
*/
static function hierarchy ($class, $includeSubclasses = true, $dataOnly = false) {
$key = "$class!" . ($includeSubclasses ? 'sc' : 'an') . '!' . ($dataOnly ? 'do' : 'al');
if (!isset(self::$hierarchy[$key])) {
$classes = array_values(ClassInfo::ancestry($class));
if ($includeSubclasses) $classes = array_unique(array_merge($classes, array_values(ClassInfo::subclassesFor($class))));
$idx = array_search('DataObject', $classes);
if ($idx !== false) {
array_splice($classes, 0, $idx+1);
}
$idx = array_search('DataObject', $classes);
if ($idx !== false) array_splice($classes, 0, $idx+1);
if ($dataOnly) {
foreach ($classes as $i => $class) {
if (!DataObject::has_own_table($class)) {
unset($classes[$i]);
}
}
}
if ($dataOnly) foreach($classes as $i => $class) {
if (!DataObject::has_own_table($class)) unset($classes[$i]);
}
self::$hierarchy[$key] = $classes;
}
self::$hierarchy[$key] = $classes;
}
return self::$hierarchy[$key];
}
return self::$hierarchy[$key];
}
/**
* Add classes to list, keeping only the parent when parent & child are both in list after add
*/
public static function add_unique_by_ancestor(&$list, $class)
{
// If class already has parent in list, just ignore
if (self::is_subclass_of($class, $list)) {
return;
}
/**
* Add classes to list, keeping only the parent when parent & child are both in list after add
*/
static function add_unique_by_ancestor(&$list, $class) {
// If class already has parent in list, just ignore
if (self::is_subclass_of($class, $list)) return;
// Strip out any subclasses of $class already in the list
$children = ClassInfo::subclassesFor($class);
$list = array_diff($list, $children);
// Strip out any subclasses of $class already in the list
$children = ClassInfo::subclassesFor($class);
$list = array_diff($list, $children);
// Then add the class in
$list[] = $class;
}
/**
* Does this class, it's parent (or optionally one of it's children) have the passed extension attached?
*/
static function has_extension($class, $extension, $includeSubclasses = true) {
foreach (self::hierarchy($class, $includeSubclasses) as $relatedclass) {
if ($relatedclass::has_extension($extension)) return true;
}
return false;
}
// Then add the class in
$list[] = $class;
}
/**
* Does this class, it's parent (or optionally one of it's children) have the passed extension attached?
*/
public static function has_extension($class, $extension, $includeSubclasses = true)
{
foreach (self::hierarchy($class, $includeSubclasses) as $relatedclass) {
if ($relatedclass::has_extension($extension)) {
return true;
}
}
return false;
}
}

215
code/search/SearchQuery.php
View File

@ -5,129 +5,148 @@
*
* API very much still in flux.
*/
class SearchQuery extends ViewableData {
class SearchQuery extends ViewableData
{
public static $missing = null;
public static $present = null;
static $missing = null;
static $present = null;
public static $default_page_size = 10;
static $default_page_size = 10;
/** These are public, but only for index & variant access - API users should not manually access these */
/** These are public, but only for index & variant access - API users should not manually access these */
public $search = array();
public $search = array();
public $classes = array();
public $classes = array();
public $require = array();
public $exclude = array();
public $require = array();
public $exclude = array();
protected $start = 0;
protected $limit = -1;
protected $start = 0;
protected $limit = -1;
/** These are the API functions */
/** These are the API functions */
public function __construct()
{
if (self::$missing === null) {
self::$missing = new stdClass();
}
if (self::$present === null) {
self::$present = new stdClass();
}
}
function __construct() {
if (self::$missing === null) self::$missing = new stdClass();
if (self::$present === null) self::$present = new stdClass();
}
/**
* @param String $text Search terms. Exact format (grouping, boolean expressions, etc.) depends on the search implementation.
* @param array $fields Limits the search to specific fields (using composite field names)
* @param array $boost Map of composite field names to float values. The higher the value,
* the more important the field gets for relevancy.
*/
public function search($text, $fields = null, $boost = array())
{
$this->search[] = array('text' => $text, 'fields' => $fields ? (array)$fields : null, 'boost' => $boost, 'fuzzy' => false);
}
/**
* @param String $text Search terms. Exact format (grouping, boolean expressions, etc.) depends on the search implementation.
* @param array $fields Limits the search to specific fields (using composite field names)
* @param array $boost Map of composite field names to float values. The higher the value,
* the more important the field gets for relevancy.
*/
function search($text, $fields = null, $boost = array()) {
$this->search[] = array('text' => $text, 'fields' => $fields ? (array)$fields : null, 'boost' => $boost, 'fuzzy' => false);
}
/**
* Similar to {@link search()}, but uses stemming and other similarity algorithms
* to find the searched terms. For example, a term "fishing" would also likely find results
* containing "fish" or "fisher". Depends on search implementation.
*
* @param String $text See {@link search()}
* @param array $fields See {@link search()}
* @param array $boost See {@link search()}
*/
public function fuzzysearch($text, $fields = null, $boost = array())
{
$this->search[] = array('text' => $text, 'fields' => $fields ? (array)$fields : null, 'boost' => $boost, 'fuzzy' => true);
}
/**
* Similar to {@link search()}, but uses stemming and other similarity algorithms
* to find the searched terms. For example, a term "fishing" would also likely find results
* containing "fish" or "fisher". Depends on search implementation.
*
* @param String $text See {@link search()}
* @param array $fields See {@link search()}
* @param array $boost See {@link search()}
*/
function fuzzysearch($text, $fields = null, $boost = array()) {
$this->search[] = array('text' => $text, 'fields' => $fields ? (array)$fields : null, 'boost' => $boost, 'fuzzy' => true);
}
public function inClass($class, $includeSubclasses = true)
{
$this->classes[] = array('class' => $class, 'includeSubclasses' => $includeSubclasses);
}
function inClass($class, $includeSubclasses = true) {
$this->classes[] = array('class' => $class, 'includeSubclasses' => $includeSubclasses);
}
/**
* Similar to {@link search()}, but typically used to further narrow down
* based on other facets which don't influence the field relevancy.
*
* @param String $field Composite name of the field
* @param Mixed $values Scalar value, array of values, or an instance of SearchQuery_Range
*/
public function filter($field, $values)
{
$requires = isset($this->require[$field]) ? $this->require[$field] : array();
$values = is_array($values) ? $values : array($values);
$this->require[$field] = array_merge($requires, $values);
}
/**
* Similar to {@link search()}, but typically used to further narrow down
* based on other facets which don't influence the field relevancy.
*
* @param String $field Composite name of the field
* @param Mixed $values Scalar value, array of values, or an instance of SearchQuery_Range
*/
function filter($field, $values) {
$requires = isset($this->require[$field]) ? $this->require[$field] : array();
$values = is_array($values) ? $values : array($values);
$this->require[$field] = array_merge($requires, $values);
}
/**
* Excludes results which match these criteria, inverse of {@link filter()}.
*
* @param String $field
* @param mixed $values
*/
public function exclude($field, $values)
{
$excludes = isset($this->exclude[$field]) ? $this->exclude[$field] : array();
$values = is_array($values) ? $values : array($values);
$this->exclude[$field] = array_merge($excludes, $values);
}
/**
* Excludes results which match these criteria, inverse of {@link filter()}.
*
* @param String $field
* @param mixed $values
*/
function exclude($field, $values) {
$excludes = isset($this->exclude[$field]) ? $this->exclude[$field] : array();
$values = is_array($values) ? $values : array($values);
$this->exclude[$field] = array_merge($excludes, $values);
}
public function start($start)
{
$this->start = $start;
}
function start($start) {
$this->start = $start;
}
public function limit($limit)
{
$this->limit = $limit;
}
function limit($limit) {
$this->limit = $limit;
}
public function page($page)
{
$this->start = $page * self::$default_page_size;
$this->limit = self::$default_page_size;
}
function page($page) {
$this->start = $page * self::$default_page_size;
$this->limit = self::$default_page_size;
}
public function isfiltered()
{
return $this->search || $this->classes || $this->require || $this->exclude;
}
function isfiltered() {
return $this->search || $this->classes || $this->require || $this->exclude;
}
function __toString() {
return "Search Query\n";
}
public function __toString()
{
return "Search Query\n";
}
}
/**
* Create one of these and pass as one of the values in filter or exclude to filter or exclude by a (possibly
* open ended) range
*/
class SearchQuery_Range {
class SearchQuery_Range
{
public $start = null;
public $end = null;
public $start = null;
public $end = null;
public function __construct($start = null, $end = null)
{
$this->start = $start;
$this->end = $end;
}
function __construct($start = null, $end = null) {
$this->start = $start;
$this->end = $end;
}
public function start($start)
{
$this->start = $start;
}
function start($start) {
$this->start = $start;
}
public function end($end)
{
$this->end = $end;
}
function end($end) {
$this->end = $end;
}
function isfiltered() {
return $this->start !== null || $this->end !== null;
}
}
public function isfiltered()
{
return $this->start !== null || $this->end !== null;
}
}

385
code/search/SearchUpdater.php
View File

@ -12,22 +12,25 @@
*
* TODO: The way we bind in is awful hacky.
*/
class SearchUpdater extends Object {
class SearchUpdater extends Object
{
/**
* Replace the database object with a subclass that captures all manipulations and passes them to us
*/
public static function bind_manipulation_capture()
{
global $databaseConfig;
/**
* Replace the database object with a subclass that captures all manipulations and passes them to us
*/
static function bind_manipulation_capture() {
global $databaseConfig;
$current = DB::getConn();
if (!$current || @$current->isManipulationCapture) {
return;
} // If not yet set, or its already captured, just return
$current = DB::getConn();
if (!$current || @$current->isManipulationCapture) return; // If not yet set, or its already captured, just return
$type = get_class($current);
$file = TEMP_FOLDER."/.cache.SMC.$type";
$type = get_class($current);
$file = TEMP_FOLDER."/.cache.SMC.$type";
if (!is_file($file)) {
file_put_contents($file, "<?php
if (!is_file($file)) {
file_put_contents($file, "<?php
class SearchManipulateCapture_$type extends $type {
public \$isManipulationCapture = true;
@ -38,162 +41,173 @@ class SearchUpdater extends Object {
}
}
");
}
}
require_once($file);
$dbClass = 'SearchManipulateCapture_'.$type;
require_once($file);
$dbClass = 'SearchManipulateCapture_'.$type;
/** @var SS_Database $captured */
$captured = new $dbClass($databaseConfig);
/** @var SS_Database $captured */
$captured = new $dbClass($databaseConfig);
// Framework 3.2+ ORM needs some dependencies set
if (method_exists($captured, "setConnector")) {
$captured->setConnector($current->getConnector());
$captured->setQueryBuilder($current->getQueryBuilder());
$captured->setSchemaManager($current->getSchemaManager());
}
// Framework 3.2+ ORM needs some dependencies set
if (method_exists($captured, "setConnector")) {
$captured->setConnector($current->getConnector());
$captured->setQueryBuilder($current->getQueryBuilder());
$captured->setSchemaManager($current->getSchemaManager());
}
// The connection might have had it's name changed (like if we're currently in a test)
$captured->selectDatabase($current->currentDatabase());
DB::setConn($captured);
}
// The connection might have had it's name changed (like if we're currently in a test)
$captured->selectDatabase($current->currentDatabase());
DB::setConn($captured);
}
static $registered = false;
/** @var SearchUpdateProcessor */
static $processor = null;
public static $registered = false;
/** @var SearchUpdateProcessor */
public static $processor = null;
/**
* Called by the SearchManiplateCapture database adapter with every manipulation made against the database.
*
* Check every index to see what objects need re-inserting into what indexes to keep the index fresh,
* but doesn't actually do it yet.
*
* TODO: This is pretty sensitive to the format of manipulation that DataObject::write produces. Specifically,
* it expects the actual class of the object to be present as a table, regardless of if any fields changed in that table
* (so a class => array( 'fields' => array() ) item), in order to find the actual class for a set of table manipulations
*/
static function handle_manipulation($manipulation) {
// First, extract any state that is in the manipulation itself
foreach ($manipulation as $table => $details) {
$manipulation[$table]['class'] = $table;
$manipulation[$table]['state'] = array();
}
/**
* Called by the SearchManiplateCapture database adapter with every manipulation made against the database.
*
* Check every index to see what objects need re-inserting into what indexes to keep the index fresh,
* but doesn't actually do it yet.
*
* TODO: This is pretty sensitive to the format of manipulation that DataObject::write produces. Specifically,
* it expects the actual class of the object to be present as a table, regardless of if any fields changed in that table
* (so a class => array( 'fields' => array() ) item), in order to find the actual class for a set of table manipulations
*/
public static function handle_manipulation($manipulation)
{
// First, extract any state that is in the manipulation itself
foreach ($manipulation as $table => $details) {
$manipulation[$table]['class'] = $table;
$manipulation[$table]['state'] = array();
}
SearchVariant::call('extractManipulationState', $manipulation);
SearchVariant::call('extractManipulationState', $manipulation);
// Then combine the manipulation back into object field sets
// Then combine the manipulation back into object field sets
$writes = array();
$writes = array();
foreach ($manipulation as $table => $details) {
if (!isset($details['id']) || !isset($details['fields'])) continue;
foreach ($manipulation as $table => $details) {
if (!isset($details['id']) || !isset($details['fields'])) {
continue;
}
$id = $details['id'];
$state = $details['state'];
$class = $details['class'];
$fields = $details['fields'];
$id = $details['id'];
$state = $details['state'];
$class = $details['class'];
$fields = $details['fields'];
$base = ClassInfo::baseDataClass($class);
$key = "$id:$base:".serialize($state);
$base = ClassInfo::baseDataClass($class);
$key = "$id:$base:".serialize($state);
$statefulids = array(array('id' => $id, 'state' => $state));
$statefulids = array(array('id' => $id, 'state' => $state));
// Is this the first table for this particular object? Then add an item to $writes
if (!isset($writes[$key])) {
$writes[$key] = array(
'base' => $base,
'class' => $class,
'id' => $id,
'statefulids' => $statefulids,
'fields' => array()
);
}
// Otherwise update the class label if it's more specific than the currently recorded one
else if (is_subclass_of($class, $writes[$key]['class'])) {
$writes[$key]['class'] = $class;
}
// Is this the first table for this particular object? Then add an item to $writes
if (!isset($writes[$key])) {
$writes[$key] = array(
'base' => $base,
'class' => $class,
'id' => $id,
'statefulids' => $statefulids,
'fields' => array()
);
}
// Otherwise update the class label if it's more specific than the currently recorded one
elseif (is_subclass_of($class, $writes[$key]['class'])) {
$writes[$key]['class'] = $class;
}
// Update the fields
foreach ($fields as $field => $value) {
$writes[$key]['fields']["$class:$field"] = $value;
}
}
// Update the fields
foreach ($fields as $field => $value) {
$writes[$key]['fields']["$class:$field"] = $value;
}
}
// Then extract any state that is needed for the writes
// Then extract any state that is needed for the writes
SearchVariant::call('extractManipulationWriteState', $writes);
SearchVariant::call('extractManipulationWriteState', $writes);
// Submit all of these writes to the search processor
// Submit all of these writes to the search processor
static::process_writes($writes);
}
static::process_writes($writes);
}
/**
* Send updates to the current search processor for execution
*
* @param array $writes
*/
public static function process_writes($writes) {
foreach ($writes as $write) {
// For every index
foreach (FullTextSearch::get_indexes() as $index => $instance) {
// If that index as a field from this class
if (SearchIntrospection::is_subclass_of($write['class'], $instance->dependancyList)) {
// Get the dirty IDs
$dirtyids = $instance->getDirtyIDs($write['class'], $write['id'], $write['statefulids'], $write['fields']);
/**
* Send updates to the current search processor for execution
*
* @param array $writes
*/
public static function process_writes($writes)
{
foreach ($writes as $write) {
// For every index
foreach (FullTextSearch::get_indexes() as $index => $instance) {
// If that index as a field from this class
if (SearchIntrospection::is_subclass_of($write['class'], $instance->dependancyList)) {
// Get the dirty IDs
$dirtyids = $instance->getDirtyIDs($write['class'], $write['id'], $write['statefulids'], $write['fields']);
// Then add then then to the global list to deal with later
foreach ($dirtyids as $dirtyclass => $ids) {
if ($ids) {
if (!self::$processor) {
self::$processor = Injector::inst()->create('SearchUpdateProcessor');
}
self::$processor->addDirtyIDs($dirtyclass, $ids, $index);
}
}
}
}
}
// Then add then then to the global list to deal with later
foreach ($dirtyids as $dirtyclass => $ids) {
if ($ids) {
if (!self::$processor) {
self::$processor = Injector::inst()->create('SearchUpdateProcessor');
}
self::$processor->addDirtyIDs($dirtyclass, $ids, $index);
}
}
}
}
}
// If we do have some work to do register the shutdown function to actually do the work
// If we do have some work to do register the shutdown function to actually do the work
// Don't do it if we're testing - there's no database connection outside the test methods, so we'd
// just get errors
$runningTests = class_exists('SapphireTest', false) && SapphireTest::is_running_test();
// Don't do it if we're testing - there's no database connection outside the test methods, so we'd
// just get errors
$runningTests = class_exists('SapphireTest', false) && SapphireTest::is_running_test();
if (self::$processor && !self::$registered && !$runningTests) {
register_shutdown_function(array("SearchUpdater", "flush_dirty_indexes"));
self::$registered = true;
}
}
if (self::$processor && !self::$registered && !$runningTests) {
register_shutdown_function(array("SearchUpdater", "flush_dirty_indexes"));
self::$registered = true;
}
}
/**
* Throw away the recorded dirty IDs without doing anything with them.
*/
static function clear_dirty_indexes() {
self::$processor = null;
}
/**
* Throw away the recorded dirty IDs without doing anything with them.
*/
public static function clear_dirty_indexes()
{
self::$processor = null;
}
/**
* Do something with the recorded dirty IDs, where that "something" depends on the value of self::$update_method,
* either immediately update the indexes, queue a messsage to update the indexes at some point in the future, or
* just throw the dirty IDs away.
*/
static function flush_dirty_indexes() {
if (!self::$processor) return;
self::$processor->triggerProcessing();
self::$processor = null;
}
/**
* Do something with the recorded dirty IDs, where that "something" depends on the value of self::$update_method,
* either immediately update the indexes, queue a messsage to update the indexes at some point in the future, or
* just throw the dirty IDs away.
*/
public static function flush_dirty_indexes()
{
if (!self::$processor) {
return;
}
self::$processor->triggerProcessing();
self::$processor = null;
}
}
class SearchUpdater_BindManipulationCaptureFilter implements RequestFilter {
public function preRequest(SS_HTTPRequest $request, Session $session, DataModel $model) {
SearchUpdater::bind_manipulation_capture();
}
class SearchUpdater_BindManipulationCaptureFilter implements RequestFilter
{
public function preRequest(SS_HTTPRequest $request, Session $session, DataModel $model)
{
SearchUpdater::bind_manipulation_capture();
}
public function postRequest(SS_HTTPRequest $request, SS_HTTPResponse $response, DataModel $model) {
/* NOP */
}
public function postRequest(SS_HTTPRequest $request, SS_HTTPResponse $response, DataModel $model)
{
/* NOP */
}
}
/**
@ -203,54 +217,57 @@ class SearchUpdater_BindManipulationCaptureFilter implements RequestFilter {
* indexed. This causes the object to be marked for deletion from the index.
*/
class SearchUpdater_ObjectHandler extends DataExtension {
class SearchUpdater_ObjectHandler extends DataExtension
{
public function onAfterDelete()
{
// Calling delete() on empty objects does nothing
if (!$this->owner->ID) {
return;
}
public function onAfterDelete() {
// Calling delete() on empty objects does nothing
if (!$this->owner->ID) return;
// Force SearchUpdater to mark this record as dirty
$manipulation = array(
$this->owner->ClassName => array(
'fields' => array(),
'id' => $this->owner->ID,
'command' => 'update'
)
);
$this->owner->extend('augmentWrite', $manipulation);
SearchUpdater::handle_manipulation($manipulation);
}
// Force SearchUpdater to mark this record as dirty
$manipulation = array(
$this->owner->ClassName => array(
'fields' => array(),
'id' => $this->owner->ID,
'command' => 'update'
)
);
$this->owner->extend('augmentWrite', $manipulation);
SearchUpdater::handle_manipulation($manipulation);
}
/**
* Forces this object to trigger a re-index in the current state
*/
public function triggerReindex()
{
if (!$this->owner->ID) {
return;
}
/**
* Forces this object to trigger a re-index in the current state
*/
public function triggerReindex() {
if (!$this->owner->ID) {
return;
}
$id = $this->owner->ID;
$class = $this->owner->ClassName;
$state = SearchVariant::current_state($class);
$base = ClassInfo::baseDataClass($class);
$key = "$id:$base:".serialize($state);
$id = $this->owner->ID;
$class = $this->owner->ClassName;
$state = SearchVariant::current_state($class);
$base = ClassInfo::baseDataClass($class);
$key = "$id:$base:".serialize($state);
$statefulids = array(array(
'id' => $id,
'state' => $state
));
$statefulids = array(array(
'id' => $id,
'state' => $state
));
$writes = array(
$key => array(
'base' => $base,
'class' => $class,
'id' => $id,
'statefulids' => $statefulids,
'fields' => array()
)
);
SearchUpdater::process_writes($writes);
}
$writes = array(
$key => array(
'base' => $base,
'class' => $class,
'id' => $id,
'statefulids' => $statefulids,
'fields' => array()
)
);
SearchUpdater::process_writes($writes);
}
}

359
code/search/SearchVariant.php
View File

@ -4,205 +4,226 @@
* A Search Variant handles decorators and other situations where the items to reindex or search through are modified
* from the default state - for instance, dealing with Versioned or Subsite
*/
abstract class SearchVariant {
abstract class SearchVariant
{
public function __construct()
{
}
function __construct() {}
/*** OVERRIDES start here */
/*** OVERRIDES start here */
/**
* Variants can provide any functions they want, but they _must_ override these functions
* with specific ones
*/
/**
* Variants can provide any functions they want, but they _must_ override these functions
* with specific ones
*/
/**
* Return false if there is something missing from the environment (probably a
* not installed module) that means this variant can't apply to any class
*/
abstract public function appliesToEnvironment();
/**
* Return false if there is something missing from the environment (probably a
* not installed module) that means this variant can't apply to any class
*/
abstract function appliesToEnvironment();
/**
* Return true if this variant applies to the passed class & subclass
*/
abstract public function appliesTo($class, $includeSubclasses);
/**
* Return true if this variant applies to the passed class & subclass
*/
abstract function appliesTo($class, $includeSubclasses);
/**
* Return the current state
*/
abstract public function currentState();
/**
* Return all states to step through to reindex all items
*/
abstract public function reindexStates();
/**
* Activate the passed state
*/
abstract public function activateState($state);
/**
* Return the current state
*/
abstract function currentState();
/**
* Return all states to step through to reindex all items
*/
abstract function reindexStates();
/**
* Activate the passed state
*/
abstract function activateState($state);
/**
* Apply this variant to a search query
*
* @param SearchQuery $query
* @param SearchIndex $index
*/
abstract public function alterQuery($query, $index);
/**
* Apply this variant to a search query
*
* @param SearchQuery $query
* @param SearchIndex $index
*/
abstract public function alterQuery($query, $index);
/*** OVERRIDES end here*/
/*** OVERRIDES end here*/
/** Holds a cache of all variants */
protected static $variants = null;
/** Holds a cache of the variants keyed by "class!" "1"? (1 = include subclasses) */
protected static $class_variants = array();
/** Holds a cache of all variants */
protected static $variants = null;
/** Holds a cache of the variants keyed by "class!" "1"? (1 = include subclasses) */
protected static $class_variants = array();
/**
* Returns an array of variants.
*
* With no arguments, returns all variants
*
* With a classname as the first argument, returns the variants that apply to that class
* (optionally including subclasses)
*
* @static
* @param string $class - The class name to get variants for
* @param bool $includeSubclasses - True if variants should be included if they apply to at least one subclass of $class
* @return array - An array of (string)$variantClassName => (Object)$variantInstance pairs
*/
public static function variants($class = null, $includeSubclasses = true)
{
if (!$class) {
if (self::$variants === null) {
$classes = ClassInfo::subclassesFor('SearchVariant');
/**
* Returns an array of variants.
*
* With no arguments, returns all variants
*
* With a classname as the first argument, returns the variants that apply to that class
* (optionally including subclasses)
*
* @static
* @param string $class - The class name to get variants for
* @param bool $includeSubclasses - True if variants should be included if they apply to at least one subclass of $class
* @return array - An array of (string)$variantClassName => (Object)$variantInstance pairs
*/
public static function variants($class = null, $includeSubclasses = true) {
if (!$class) {
if (self::$variants === null) {
$classes = ClassInfo::subclassesFor('SearchVariant');
$concrete = array();
foreach ($classes as $variantclass) {
$ref = new ReflectionClass($variantclass);
if ($ref->isInstantiable()) {
$variant = singleton($variantclass);
if ($variant->appliesToEnvironment()) {
$concrete[$variantclass] = $variant;
}
}
}
$concrete = array();
foreach ($classes as $variantclass) {
$ref = new ReflectionClass($variantclass);
if ($ref->isInstantiable()) {
$variant = singleton($variantclass);
if ($variant->appliesToEnvironment()) $concrete[$variantclass] = $variant;
}
}
self::$variants = $concrete;
}
self::$variants = $concrete;
}
return self::$variants;
} else {
$key = $class . '!' . $includeSubclasses;
return self::$variants;
}
else {
$key = $class . '!' . $includeSubclasses;
if (!isset(self::$class_variants[$key])) {
self::$class_variants[$key] = array();
if (!isset(self::$class_variants[$key])) {
self::$class_variants[$key] = array();
foreach (self::variants() as $variantclass => $instance) {
if ($instance->appliesTo($class, $includeSubclasses)) {
self::$class_variants[$key][$variantclass] = $instance;
}
}
}
foreach (self::variants() as $variantclass => $instance) {
if ($instance->appliesTo($class, $includeSubclasses)) self::$class_variants[$key][$variantclass] = $instance;
}
}
return self::$class_variants[$key];
}
}
return self::$class_variants[$key];
}
}
/** Holds a cache of SearchVariant_Caller instances, one for each class/includeSubclasses setting */
protected static $call_instances = array();
/** Holds a cache of SearchVariant_Caller instances, one for each class/includeSubclasses setting */
protected static $call_instances = array();
/**
* Lets you call any function on all variants that support it, in the same manner as "Object#extend" calls
* a method from extensions.
*
* Usage: SearchVariant::with(...)->call($method, $arg1, ...);
*
* @static
*
* @param string $class - (Optional) a classname. If passed, only variants that apply to that class will be checked / called
*
* @param bool $includeSubclasses - (Optional) If false, only variants that apply strictly to the passed class or its super-classes
* will be checked. If true (the default), variants that apply to any sub-class of the passed class with also be checked
*
* @return An object with one method, call()
*/
public static function with($class = null, $includeSubclasses = true)
{
// Make the cache key
$key = $class ? $class . '!' . $includeSubclasses : '!';
// If no SearchVariant_Caller instance yet, create it
if (!isset(self::$call_instances[$key])) {
self::$call_instances[$key] = new SearchVariant_Caller(self::variants($class, $includeSubclasses));
}
// Then return it
return self::$call_instances[$key];
}
/**
* Lets you call any function on all variants that support it, in the same manner as "Object#extend" calls
* a method from extensions.
*
* Usage: SearchVariant::with(...)->call($method, $arg1, ...);
*
* @static
*
* @param string $class - (Optional) a classname. If passed, only variants that apply to that class will be checked / called
*
* @param bool $includeSubclasses - (Optional) If false, only variants that apply strictly to the passed class or its super-classes
* will be checked. If true (the default), variants that apply to any sub-class of the passed class with also be checked
*
* @return An object with one method, call()
*/
static function with($class = null, $includeSubclasses = true) {
// Make the cache key
$key = $class ? $class . '!' . $includeSubclasses : '!';
// If no SearchVariant_Caller instance yet, create it
if (!isset(self::$call_instances[$key])) self::$call_instances[$key] = new SearchVariant_Caller(self::variants($class, $includeSubclasses));
// Then return it
return self::$call_instances[$key];
}
/**
* A shortcut to with when calling without passing in a class,
*
* SearchVariant::call(...) ==== SearchVariant::with()->call(...);
*/
public static function call($method, &$a1=null, &$a2=null, &$a3=null, &$a4=null, &$a5=null, &$a6=null, &$a7=null)
{
return self::with()->call($method, $a1, $a2, $a3, $a4, $a5, $a6, $a7);
}
/**
* A shortcut to with when calling without passing in a class,
*
* SearchVariant::call(...) ==== SearchVariant::with()->call(...);
*/
static function call($method, &$a1=null, &$a2=null, &$a3=null, &$a4=null, &$a5=null, &$a6=null, &$a7=null) {
return self::with()->call($method, $a1, $a2, $a3, $a4, $a5, $a6, $a7);
}
/**
* Get the current state of every variant
* @static
* @return array
*/
public static function current_state($class = null, $includeSubclasses = true)
{
$state = array();
foreach (self::variants($class, $includeSubclasses) as $variant => $instance) {
$state[$variant] = $instance->currentState();
}
return $state;
}
/**
* Get the current state of every variant
* @static
* @return array
*/
static function current_state($class = null, $includeSubclasses = true) {
$state = array();
foreach (self::variants($class, $includeSubclasses) as $variant => $instance) {
$state[$variant] = $instance->currentState();
}
return $state;
}
/**
* Activate all the states in the passed argument
* @static
* @param (array) $state. A set of (string)$variantClass => (any)$state pairs , e.g. as returned by
* SearchVariant::current_state()
* @return void
*/
public static function activate_state($state)
{
foreach (self::variants() as $variant => $instance) {
if (isset($state[$variant])) {
$instance->activateState($state[$variant]);
}
}
}
/**
* Activate all the states in the passed argument
* @static
* @param (array) $state. A set of (string)$variantClass => (any)$state pairs , e.g. as returned by
* SearchVariant::current_state()
* @return void
*/
static function activate_state($state) {
foreach (self::variants() as $variant => $instance) {
if (isset($state[$variant])) $instance->activateState($state[$variant]);
}
}
/**
* Return an iterator that, when used in a for loop, activates one combination of reindex states per loop, and restores
* back to the original state at the end
* @static
* @param string $class - The class name to get variants for
* @param bool $includeSubclasses - True if variants should be included if they apply to at least one subclass of $class
* @return SearchVariant_ReindexStateIteratorRet - The iterator to foreach loop over
*/
public static function reindex_states($class = null, $includeSubclasses = true)
{
$allstates = array();
/**
* Return an iterator that, when used in a for loop, activates one combination of reindex states per loop, and restores
* back to the original state at the end
* @static
* @param string $class - The class name to get variants for
* @param bool $includeSubclasses - True if variants should be included if they apply to at least one subclass of $class
* @return SearchVariant_ReindexStateIteratorRet - The iterator to foreach loop over
*/
static function reindex_states($class = null, $includeSubclasses = true) {
$allstates = array();
foreach (self::variants($class, $includeSubclasses) as $variant => $instance) {
if ($states = $instance->reindexStates()) {
$allstates[$variant] = $states;
}
}
foreach (self::variants($class, $includeSubclasses) as $variant => $instance) {
if ($states = $instance->reindexStates()) $allstates[$variant] = $states;
}
return $allstates ? new CombinationsArrayIterator($allstates) : array(array());
}
return $allstates ? new CombinationsArrayIterator($allstates) : array(array());
}
}
/**
* Internal utility class used to hold the state of the SearchVariant::with call
*/
class SearchVariant_Caller {
protected $variants = null;
class SearchVariant_Caller
{
protected $variants = null;
function __construct($variants) {
$this->variants = $variants;
}
public function __construct($variants)
{
$this->variants = $variants;
}
function call($method, &$a1=null, &$a2=null, &$a3=null, &$a4=null, &$a5=null, &$a6=null, &$a7=null) {
$values = array();
public function call($method, &$a1=null, &$a2=null, &$a3=null, &$a4=null, &$a5=null, &$a6=null, &$a7=null)
{
$values = array();
foreach ($this->variants as $variant) {
if (method_exists($variant, $method)) {
$value = $variant->$method($a1, $a2, $a3, $a4, $a5, $a6, $a7);
if ($value !== null) $values[] = $value;
}
}
foreach ($this->variants as $variant) {
if (method_exists($variant, $method)) {
$value = $variant->$method($a1, $a2, $a3, $a4, $a5, $a6, $a7);
if ($value !== null) {
$values[] = $value;
}
}
}
return $values;
}
return $values;
}
}

148
code/search/SearchVariantSiteTreeSubsitesPolyhome.php
View File

@ -1,85 +1,95 @@
<?php
class SearchVariantSiteTreeSubsitesPolyhome extends SearchVariant {
class SearchVariantSiteTreeSubsitesPolyhome extends SearchVariant
{
public function appliesToEnvironment()
{
return class_exists('Subsite') && class_exists('SubsitePolyhome');
}
function appliesToEnvironment() {
return class_exists('Subsite') && class_exists('SubsitePolyhome');
}
public function appliesTo($class, $includeSubclasses)
{
return SearchIntrospection::has_extension($class, 'SiteTreeSubsitesPolyhome', $includeSubclasses);
}
function appliesTo($class, $includeSubclasses) {
return SearchIntrospection::has_extension($class, 'SiteTreeSubsitesPolyhome', $includeSubclasses);
}
public function currentState()
{
return Subsite::currentSubsiteID();
}
public function reindexStates()
{
static $ids = null;
function currentState() {
return Subsite::currentSubsiteID();
}
function reindexStates() {
static $ids = null;
if ($ids === null) {
$ids = array(0);
foreach (DataObject::get('Subsite') as $subsite) {
$ids[] = $subsite->ID;
}
}
if ($ids === null) {
$ids = array(0);
foreach (DataObject::get('Subsite') as $subsite) $ids[] = $subsite->ID;
}
return $ids;
}
public function activateState($state)
{
if (Controller::has_curr()) {
Subsite::changeSubsite($state);
} else {
// TODO: This is a nasty hack - calling Subsite::changeSubsite after request ends
// throws error because no current controller to access session on
$_REQUEST['SubsiteID'] = $state;
}
}
return $ids;
}
function activateState($state) {
if (Controller::has_curr()) {
Subsite::changeSubsite($state);
}
else {
// TODO: This is a nasty hack - calling Subsite::changeSubsite after request ends
// throws error because no current controller to access session on
$_REQUEST['SubsiteID'] = $state;
}
}
public function alterDefinition($base, $index)
{
$self = get_class($this);
$index->filterFields['_subsite'] = array(
'name' => '_subsite',
'field' => '_subsite',
'fullfield' => '_subsite',
'base' => $base,
'origin' => $base,
'type' => 'Int',
'lookup_chain' => array(array('call' => 'variant', 'variant' => $self, 'method' => 'currentState'))
);
}
function alterDefinition($base, $index) {
$self = get_class($this);
$index->filterFields['_subsite'] = array(
'name' => '_subsite',
'field' => '_subsite',
'fullfield' => '_subsite',
'base' => $base,
'origin' => $base,
'type' => 'Int',
'lookup_chain' => array(array('call' => 'variant', 'variant' => $self, 'method' => 'currentState'))
);
}
public function alterQuery($query, $index)
{
$subsite = Subsite::currentSubsiteID();
$query->filter('_subsite', array($subsite, SearchQuery::$missing));
}
public function alterQuery($query, $index) {
$subsite = Subsite::currentSubsiteID();
$query->filter('_subsite', array($subsite, SearchQuery::$missing));
}
public static $subsites = null;
static $subsites = null;
/**
* We need _really_ complicated logic to find just the changed subsites (because we use versions there's no explicit
* deletes, just new versions with different members) so just always use all of them
*/
public function extractManipulationWriteState(&$writes)
{
$self = get_class($this);
/**
* We need _really_ complicated logic to find just the changed subsites (because we use versions there's no explicit
* deletes, just new versions with different members) so just always use all of them
*/
function extractManipulationWriteState(&$writes) {
$self = get_class($this);
foreach ($writes as $key => $write) {
if (!$this->appliesTo($write['class'], true)) {
continue;
}
foreach ($writes as $key => $write) {
if (!$this->appliesTo($write['class'], true)) continue;
if (self::$subsites === null) {
$query = new SQLQuery('ID', 'Subsite');
self::$subsites = array_merge(array('0'), $query->execute()->column());
}
if (self::$subsites === null) {
$query = new SQLQuery('ID', 'Subsite');
self::$subsites = array_merge(array('0'), $query->execute()->column());
}
$next = array();
$next = array();
foreach ($write['statefulids'] as $i => $statefulid) {
foreach (self::$subsites as $subsiteID) {
$next[] = array('id' => $statefulid['id'], 'state' => array_merge($statefulid['state'], array($self => $subsiteID)));
}
}
$writes[$key]['statefulids'] = $next;
}
}
foreach ($write['statefulids'] as $i => $statefulid) {
foreach (self::$subsites as $subsiteID) {
$next[] = array('id' => $statefulid['id'], 'state' => array_merge($statefulid['state'], array($self => $subsiteID)));
}
}
$writes[$key]['statefulids'] = $next;
}
}
}

151
code/search/SearchVariantSubsites.php
View File

@ -1,88 +1,99 @@
<?php
class SearchVariantSubsites extends SearchVariant {
class SearchVariantSubsites extends SearchVariant
{
public function appliesToEnvironment()
{
return class_exists('Subsite');
}
function appliesToEnvironment() {
return class_exists('Subsite');
}
public function appliesTo($class, $includeSubclasses)
{
// Include all DataExtensions that contain a SubsiteID.
// TODO: refactor subsites to inherit a common interface, so we can run introspection once only.
return SearchIntrospection::has_extension($class, 'SiteTreeSubsites', $includeSubclasses) ||
SearchIntrospection::has_extension($class, 'GroupSubsites', $includeSubclasses) ||
SearchIntrospection::has_extension($class, 'FileSubsites', $includeSubclasses) ||
SearchIntrospection::has_extension($class, 'SiteConfigSubsites', $includeSubclasses);
}
function appliesTo($class, $includeSubclasses) {
// Include all DataExtensions that contain a SubsiteID.
// TODO: refactor subsites to inherit a common interface, so we can run introspection once only.
return SearchIntrospection::has_extension($class, 'SiteTreeSubsites', $includeSubclasses) ||
SearchIntrospection::has_extension($class, 'GroupSubsites', $includeSubclasses) ||
SearchIntrospection::has_extension($class, 'FileSubsites', $includeSubclasses) ||
SearchIntrospection::has_extension($class, 'SiteConfigSubsites', $includeSubclasses);
}
public function currentState()
{
return (string)Subsite::currentSubsiteID();
}
function currentState() {
return (string)Subsite::currentSubsiteID();
}
public function reindexStates()
{
static $ids = null;
function reindexStates() {
static $ids = null;
if ($ids === null) {
$ids = array('0');
foreach (DataObject::get('Subsite') as $subsite) {
$ids[] = (string)$subsite->ID;
}
}
if ($ids === null) {
$ids = array('0');
foreach (DataObject::get('Subsite') as $subsite) $ids[] = (string)$subsite->ID;
}
return $ids;
}
return $ids;
}
public function activateState($state)
{
// We always just set the $_GET variable rather than store in Session - this always works, has highest priority
// in Subsite::currentSubsiteID() and doesn't persist unlike Subsite::changeSubsite
$_GET['SubsiteID'] = $state;
Permission::flush_permission_cache();
}
function activateState($state) {
// We always just set the $_GET variable rather than store in Session - this always works, has highest priority
// in Subsite::currentSubsiteID() and doesn't persist unlike Subsite::changeSubsite
$_GET['SubsiteID'] = $state;
Permission::flush_permission_cache();
}
public function alterDefinition($base, $index)
{
$self = get_class($this);
$index->filterFields['_subsite'] = array(
'name' => '_subsite',
'field' => '_subsite',
'fullfield' => '_subsite',
'base' => $base,
'origin' => $base,
'type' => 'Int',
'lookup_chain' => array(array('call' => 'variant', 'variant' => $self, 'method' => 'currentState'))
);
}
function alterDefinition($base, $index) {
$self = get_class($this);
$index->filterFields['_subsite'] = array(
'name' => '_subsite',
'field' => '_subsite',
'fullfield' => '_subsite',
'base' => $base,
'origin' => $base,
'type' => 'Int',
'lookup_chain' => array(array('call' => 'variant', 'variant' => $self, 'method' => 'currentState'))
);
}
public function alterQuery($query, $index)
{
$subsite = Subsite::currentSubsiteID();
$query->filter('_subsite', array($subsite, SearchQuery::$missing));
}
function alterQuery($query, $index) {
$subsite = Subsite::currentSubsiteID();
$query->filter('_subsite', array($subsite, SearchQuery::$missing));
}
public static $subsites = null;
static $subsites = null;
/**
* We need _really_ complicated logic to find just the changed subsites (because we use versions there's no explicit
* deletes, just new versions with different members) so just always use all of them
*/
public function extractManipulationWriteState(&$writes)
{
$self = get_class($this);
/**
* We need _really_ complicated logic to find just the changed subsites (because we use versions there's no explicit
* deletes, just new versions with different members) so just always use all of them
*/
function extractManipulationWriteState(&$writes) {
$self = get_class($this);
foreach ($writes as $key => $write) {
if (!$this->appliesTo($write['class'], true)) {
continue;
}
foreach ($writes as $key => $write) {
if (!$this->appliesTo($write['class'], true)) continue;
if (self::$subsites === null) {
$query = new SQLQuery('"ID"', '"Subsite"');
self::$subsites = array_merge(array('0'), $query->execute()->column());
}
if (self::$subsites === null) {
$query = new SQLQuery('"ID"', '"Subsite"');
self::$subsites = array_merge(array('0'), $query->execute()->column());
}
$next = array();
$next = array();
foreach ($write['statefulids'] as $i => $statefulid) {
foreach (self::$subsites as $subsiteID) {
$next[] = array('id' => $statefulid['id'], 'state' => array_merge($statefulid['state'], array($self => (string)$subsiteID)));
}
}
$writes[$key]['statefulids'] = $next;
}
}
foreach ($write['statefulids'] as $i => $statefulid) {
foreach (self::$subsites as $subsiteID) {
$next[] = array('id' => $statefulid['id'], 'state' => array_merge($statefulid['state'], array($self => (string)$subsiteID)));
}
}
$writes[$key]['statefulids'] = $next;
}
}
}

126
code/search/SearchVariantVersioned.php
View File

@ -1,70 +1,84 @@
<?php
class SearchVariantVersioned extends SearchVariant {
class SearchVariantVersioned extends SearchVariant
{
public function appliesToEnvironment()
{
return class_exists('Versioned');
}
function appliesToEnvironment() {
return class_exists('Versioned');
}
public function appliesTo($class, $includeSubclasses)
{
return SearchIntrospection::has_extension($class, 'Versioned', $includeSubclasses);
}
function appliesTo($class, $includeSubclasses) {
return SearchIntrospection::has_extension($class, 'Versioned', $includeSubclasses);
}
public function currentState()
{
return Versioned::current_stage();
}
public function reindexStates()
{
return array('Stage', 'Live');
}
public function activateState($state)
{
Versioned::reading_stage($state);
}
function currentState() { return Versioned::current_stage(); }
function reindexStates() { return array('Stage', 'Live'); }
function activateState($state) { Versioned::reading_stage($state); }
public function alterDefinition($base, $index)
{
$self = get_class($this);
function alterDefinition($base, $index) {
$self = get_class($this);
$index->filterFields['_versionedstage'] = array(
'name' => '_versionedstage',
'field' => '_versionedstage',
'fullfield' => '_versionedstage',
'base' => $base,
'origin' => $base,
'type' => 'String',
'lookup_chain' => array(array('call' => 'variant', 'variant' => $self, 'method' => 'currentState'))
);
}
$index->filterFields['_versionedstage'] = array(
'name' => '_versionedstage',
'field' => '_versionedstage',
'fullfield' => '_versionedstage',
'base' => $base,
'origin' => $base,
'type' => 'String',
'lookup_chain' => array(array('call' => 'variant', 'variant' => $self, 'method' => 'currentState'))
);
}
public function alterQuery($query, $index)
{
$stage = Versioned::current_stage();
$query->filter('_versionedstage', array($stage, SearchQuery::$missing));
}
public function extractManipulationState(&$manipulation)
{
$self = get_class($this);
foreach ($manipulation as $table => $details) {
$class = $details['class'];
$stage = 'Stage';
public function alterQuery($query, $index) {
$stage = Versioned::current_stage();
$query->filter('_versionedstage', array($stage, SearchQuery::$missing));
}
function extractManipulationState(&$manipulation) {
$self = get_class($this);
foreach ($manipulation as $table => $details) {
$class = $details['class'];
$stage = 'Stage';
if (preg_match('/^(.*)_Live$/', $table, $matches)) {
$class = $matches[1];
$stage = 'Live';
}
if (preg_match('/^(.*)_Live$/', $table, $matches)) {
$class = $matches[1];
$stage = 'Live';
}
if (ClassInfo::exists($class) && $this->appliesTo($class, false)) {
$manipulation[$table]['class'] = $class;
$manipulation[$table]['state'][$self] = $stage;
}
}
}
if (ClassInfo::exists($class) && $this->appliesTo($class, false)) {
$manipulation[$table]['class'] = $class;
$manipulation[$table]['state'][$self] = $stage;
}
}
}
function extractStates(&$table, &$ids, &$fields) {
$class = $table;
$suffix = null;
public function extractStates(&$table, &$ids, &$fields)
{
$class = $table;
$suffix = null;
if (ClassInfo::exists($class) && $this->appliesTo($class, false)) {
$table = $class;
$self = get_class($this);
foreach ($ids as $i => $statefulid) {
$ids[$i]['state'][$self] = $suffix ? $suffix : 'Stage';
}
}
}
if (ClassInfo::exists($class) && $this->appliesTo($class, false)) {
$table = $class;
$self = get_class($this);
foreach ($ids as $i => $statefulid) {
$ids[$i]['state'][$self] = $suffix ? $suffix : 'Stage';
}
}
}
}

319
code/search/processors/SearchUpdateBatchedProcessor.php
View File

@ -3,155 +3,176 @@
/**
* Provides batching of search updates
*/
abstract class SearchUpdateBatchedProcessor extends SearchUpdateProcessor {
/**
* List of batches to be processed
*
* @var array
*/
protected $batches;
/**
* Pointer to index of $batches assigned to $current.
* Set to 0 (first index) if not started, or count + 1 if completed.
*
* @var int
*/
protected $currentBatch;
/**
* List of indexes successfully comitted in the current batch
*
* @var array
*/
protected $completedIndexes;
/**
* Maximum number of record-states to process in one batch.
* Set to zero to process all records in a single batch
*
* @config
* @var int
*/
private static $batch_size = 100;
/**
* Up to this number of additional ids can be added to any batch in order to reduce the number
* of batches
*
* @config
* @var int
*/
private static $batch_soft_cap = 10;
public function __construct() {
parent::__construct();
$this->batches = array();
$this->setBatch(0);
}
/**
* Set the current batch index
*
* @param int $batch Index of the batch
*/
protected function setBatch($batch) {
$this->currentBatch = $batch;
}
protected function getSource() {
if(isset($this->batches[$this->currentBatch])) {
return $this->batches[$this->currentBatch];
}
}
/**
* Process the current queue
*
* @return boolean
*/
public function process() {
// Skip blank queues
if(empty($this->batches)) return true;
// Don't re-process completed queue
if($this->currentBatch >= count($this->batches)) return true;
// Send current patch to indexes
$this->prepareIndexes();
// Advance to next batch if successful
$this->setBatch($this->currentBatch + 1);
return true;
}
/**
* Segments batches acording to the specified rules
*
* @param array $source Source input
* @return array Batches
*/
protected function segmentBatches($source) {
// Measure batch_size
$batchSize = Config::inst()->get(get_class(), 'batch_size');
if($batchSize === 0) return array($source);
$softCap = Config::inst()->get(get_class(), 'batch_soft_cap');
// Clear batches
$batches = array();
$current = array();
$currentSize = 0;
abstract class SearchUpdateBatchedProcessor extends SearchUpdateProcessor
{
/**
* List of batches to be processed
*
* @var array
*/
protected $batches;
/**
* Pointer to index of $batches assigned to $current.
* Set to 0 (first index) if not started, or count + 1 if completed.
*
* @var int
*/
protected $currentBatch;
/**
* List of indexes successfully comitted in the current batch
*
* @var array
*/
protected $completedIndexes;
/**
* Maximum number of record-states to process in one batch.
* Set to zero to process all records in a single batch
*
* @config
* @var int
*/
private static $batch_size = 100;
/**
* Up to this number of additional ids can be added to any batch in order to reduce the number
* of batches
*
* @config
* @var int
*/
private static $batch_soft_cap = 10;
public function __construct()
{
parent::__construct();
$this->batches = array();
$this->setBatch(0);
}
/**
* Set the current batch index
*
* @param int $batch Index of the batch
*/
protected function setBatch($batch)
{
$this->currentBatch = $batch;
}
protected function getSource()
{
if (isset($this->batches[$this->currentBatch])) {
return $this->batches[$this->currentBatch];
}
}
/**
* Process the current queue
*
* @return boolean
*/
public function process()
{
// Skip blank queues
if (empty($this->batches)) {
return true;
}
// Don't re-process completed queue
if ($this->currentBatch >= count($this->batches)) {
return true;
}
// Send current patch to indexes
$this->prepareIndexes();
// Advance to next batch if successful
$this->setBatch($this->currentBatch + 1);
return true;
}
/**
* Segments batches acording to the specified rules
*
* @param array $source Source input
* @return array Batches
*/
protected function segmentBatches($source)
{
// Measure batch_size
$batchSize = Config::inst()->get(get_class(), 'batch_size');
if ($batchSize === 0) {
return array($source);
}
$softCap = Config::inst()->get(get_class(), 'batch_soft_cap');
// Clear batches
$batches = array();
$current = array();
$currentSize = 0;
// Build batches from data
foreach ($source as $base => $statefulids) {
if (!$statefulids) continue;
// Build batches from data
foreach ($source as $base => $statefulids) {
if (!$statefulids) {
continue;
}
foreach ($statefulids as $stateKey => $statefulid) {
$state = $statefulid['state'];
$ids = $statefulid['ids'];
if(!$ids) continue;
// Extract items from $ids until empty
while($ids) {
// Estimate maximum number of items to take for this iteration, allowing for the soft cap
$take = $batchSize - $currentSize;
if(count($ids) <= $take + $softCap) $take += $softCap;
$items = array_slice($ids, 0, $take, true);
$ids = array_slice($ids, count($items), null, true);
// Update batch
$currentSize += count($items);
$merge = array(
$base => array(
$stateKey => array(
'state' => $state,
'ids' => $items
)
)
);
$current = $current ? array_merge_recursive($current, $merge) : $merge;
if($currentSize >= $batchSize) {
$batches[] = $current;
$current = array();
$currentSize = 0;
}
}
}
}
// Add incomplete batch
if($currentSize) $batches[] = $current;
return $batches;
}
public function batchData() {
$this->batches = $this->segmentBatches($this->dirty);
$this->setBatch(0);
}
public function triggerProcessing() {
$this->batchData();
}
foreach ($statefulids as $stateKey => $statefulid) {
$state = $statefulid['state'];
$ids = $statefulid['ids'];
if (!$ids) {
continue;
}
// Extract items from $ids until empty
while ($ids) {
// Estimate maximum number of items to take for this iteration, allowing for the soft cap
$take = $batchSize - $currentSize;
if (count($ids) <= $take + $softCap) {
$take += $softCap;
}
$items = array_slice($ids, 0, $take, true);
$ids = array_slice($ids, count($items), null, true);
// Update batch
$currentSize += count($items);
$merge = array(
$base => array(
$stateKey => array(
'state' => $state,
'ids' => $items
)
)
);
$current = $current ? array_merge_recursive($current, $merge) : $merge;
if ($currentSize >= $batchSize) {
$batches[] = $current;
$current = array();
$currentSize = 0;
}
}
}
}
// Add incomplete batch
if ($currentSize) {
$batches[] = $current;
}
return $batches;
}
public function batchData()
{
$this->batches = $this->segmentBatches($this->dirty);
$this->setBatch(0);
}
public function triggerProcessing()
{
$this->batchData();
}
}

509
code/search/processors/SearchUpdateCommitJobProcessor.php
View File

@ -1,249 +1,266 @@
<?php
if(!interface_exists('QueuedJob')) return;
class SearchUpdateCommitJobProcessor implements QueuedJob {
/**
* The QueuedJob queue to use when processing commits
*
* @config
* @var int
*/
private static $commit_queue = 2; // QueuedJob::QUEUED;
/**
* List of indexes to commit
*
* @var array
*/
protected $indexes = array();
/**
* True if this job is skipped to be be re-scheduled in the future
*
* @var boolean
*/
protected $skipped = false;
/**
* List of completed indexes
*
* @var array
*/
protected $completed = array();
/**
* List of messages
*
* @var array
*/
protected $messages = array();
/**
* List of dirty indexes to be committed
*
* @var array
*/
public static $dirty_indexes = true;
/**
* If solrindex::commit has already been performed, but additional commits are necessary,
* how long do we wait before attempting to touch the index again?
*
* {@see http://stackoverflow.com/questions/7512945/how-to-fix-exceeded-limit-of-maxwarmingsearchers}
*
* @var int
* @config
*/
private static $cooldown = 300;
/**
* True if any commits have been executed this request. If so, any attempts to run subsequent commits
* should be delayed until next queuedjob to prevent solr reaching maxWarmingSearchers
*
* {@see http://stackoverflow.com/questions/7512945/how-to-fix-exceeded-limit-of-maxwarmingsearchers}
*
* @var boolean
*/
public static $has_run = false;
/**
* This method is invoked once indexes with dirty ids have been updapted and a commit is necessary
*
* @param boolean $dirty Marks all indexes as dirty by default. Set to false if there are known comitted and
* clean indexes
* @param string $startAfter Start date
* @return int The ID of the next queuedjob to run. This could be a new one or an existing one.
*/
public static function queue($dirty = true, $startAfter = null) {
$commit = Injector::inst()->create(__CLASS__);
$id = singleton('QueuedJobService')->queueJob($commit, $startAfter);
if($dirty) {
$indexes = FullTextSearch::get_indexes();
static::$dirty_indexes = array_keys($indexes);
}
return $id;
}
public function getJobType() {
return Config::inst()->get(__CLASS__, 'commit_queue');
}
public function getSignature() {
// There is only ever one commit job on the queue so the signature is consistent
// See QueuedJobService::queueJob() for the code that prevents duplication
return __CLASS__;
}
public function getTitle() {
return "FullTextSearch Commit Job";
}
/**
* Get the list of index names we should process
*
* @return array
*/
public function getAllIndexes() {
if(empty($this->indexes)) {
$indexes = FullTextSearch::get_indexes();
$this->indexes = array_keys($indexes);
}
return $this->indexes;
}
public function jobFinished() {
// If we've indexed exactly as many as we would like, we are done
return $this->skipped
|| (count($this->getAllIndexes()) <= count($this->completed));
}
public function prepareForRestart() {
// NOOP
}
public function afterComplete() {
// NOOP
}
/**
* Abort this job, potentially rescheduling a replacement if there is still work to do
*/
protected function discardJob() {
$this->skipped = true;
// If we do not have dirty records, then assume that these dirty records were committed
// already this request (but probably another job), so we don't need to commit anything else.
// This could occur if we completed multiple searchupdate jobs in a prior request, and
// we only need one commit job to commit all of them in the current request.
if(empty(static::$dirty_indexes)) {
$this->addMessage("Indexing already completed this request: Discarding this job");
return;
}
// If any commit has run, but some (or all) indexes are un-comitted, we must re-schedule this task.
// This could occur if we completed a searchupdate job in a prior request, as well as in
// the current request
$cooldown = Config::inst()->get(__CLASS__, 'cooldown');
$now = new DateTime(SS_Datetime::now()->getValue());
$now->add(new DateInterval('PT'.$cooldown.'S'));
$runat = $now->Format('Y-m-d H:i:s');
$this->addMessage("Indexing already run this request, but incomplete. Re-scheduling for {$runat}");
// Queue after the given cooldown
static::queue(false, $runat);
}
public function process() {
// If we have already run an instance of SearchUpdateCommitJobProcessor this request, immediately
// quit this job to prevent hitting warming search limits in Solr
if(static::$has_run) {
$this->discardJob();
return true;
}
// To prevent other commit jobs from running this request
static::$has_run = true;
// Run all incompleted indexes
$indexNames = $this->getAllIndexes();
foreach ($indexNames as $name) {
$index = singleton($name);
$this->commitIndex($index);
}
$this->addMessage("All indexes committed");
return true;
}
/**
* Commits a specific index
*
* @param SolrIndex $index
* @throws Exception
*/
protected function commitIndex($index) {
// Skip index if this is already complete
$name = get_class($index);
if(in_array($name, $this->completed)) {
$this->addMessage("Skipping already comitted index {$name}");
return;
}
// Bypass SolrIndex::commit exception handling so that queuedjobs can handle the error
$this->addMessage("Committing index {$name}");
$index->getService()->commit(false, false, false);
$this->addMessage("Committing index {$name} was successful");
// If this index is currently marked as dirty, it's now clean
if(in_array($name, static::$dirty_indexes)) {
static::$dirty_indexes = array_diff(static::$dirty_indexes, array($name));
}
// Mark complete
$this->completed[] = $name;
}
public function setup() {
// NOOP
}
public function getJobData() {
$data = new stdClass();
$data->totalSteps = count($this->getAllIndexes());
$data->currentStep = count($this->completed);
$data->isComplete = $this->jobFinished();
$data->messages = $this->messages;
$data->jobData = new stdClass();
$data->jobData->skipped = $this->skipped;
$data->jobData->completed = $this->completed;
$data->jobData->indexes = $this->getAllIndexes();
return $data;
}
public function setJobData($totalSteps, $currentStep, $isComplete, $jobData, $messages) {
$this->isComplete = $isComplete;
$this->messages = $messages;
$this->skipped = $jobData->skipped;
$this->completed = $jobData->completed;
$this->indexes = $jobData->indexes;
}
public function addMessage($message, $severity='INFO') {
$severity = strtoupper($severity);
$this->messages[] = '[' . date('Y-m-d H:i:s') . "][$severity] $message";
}
public function getMessages() {
return $this->messages;
}
if (!interface_exists('QueuedJob')) {
return;
}
class SearchUpdateCommitJobProcessor implements QueuedJob
{
/**
* The QueuedJob queue to use when processing commits
*
* @config
* @var int
*/
private static $commit_queue = 2; // QueuedJob::QUEUED;
/**
* List of indexes to commit
*
* @var array
*/
protected $indexes = array();
/**
* True if this job is skipped to be be re-scheduled in the future
*
* @var boolean
*/
protected $skipped = false;
/**
* List of completed indexes
*
* @var array
*/
protected $completed = array();
/**
* List of messages
*
* @var array
*/
protected $messages = array();
/**
* List of dirty indexes to be committed
*
* @var array
*/
public static $dirty_indexes = true;
/**
* If solrindex::commit has already been performed, but additional commits are necessary,
* how long do we wait before attempting to touch the index again?
*
* {@see http://stackoverflow.com/questions/7512945/how-to-fix-exceeded-limit-of-maxwarmingsearchers}
*
* @var int
* @config
*/
private static $cooldown = 300;
/**
* True if any commits have been executed this request. If so, any attempts to run subsequent commits
* should be delayed until next queuedjob to prevent solr reaching maxWarmingSearchers
*
* {@see http://stackoverflow.com/questions/7512945/how-to-fix-exceeded-limit-of-maxwarmingsearchers}
*
* @var boolean
*/
public static $has_run = false;
/**
* This method is invoked once indexes with dirty ids have been updapted and a commit is necessary
*
* @param boolean $dirty Marks all indexes as dirty by default. Set to false if there are known comitted and
* clean indexes
* @param string $startAfter Start date
* @return int The ID of the next queuedjob to run. This could be a new one or an existing one.
*/
public static function queue($dirty = true, $startAfter = null)
{
$commit = Injector::inst()->create(__CLASS__);
$id = singleton('QueuedJobService')->queueJob($commit, $startAfter);
if ($dirty) {
$indexes = FullTextSearch::get_indexes();
static::$dirty_indexes = array_keys($indexes);
}
return $id;
}
public function getJobType()
{
return Config::inst()->get(__CLASS__, 'commit_queue');
}
public function getSignature()
{
// There is only ever one commit job on the queue so the signature is consistent
// See QueuedJobService::queueJob() for the code that prevents duplication
return __CLASS__;
}
public function getTitle()
{
return "FullTextSearch Commit Job";
}
/**
* Get the list of index names we should process
*
* @return array
*/
public function getAllIndexes()
{
if (empty($this->indexes)) {
$indexes = FullTextSearch::get_indexes();
$this->indexes = array_keys($indexes);
}
return $this->indexes;
}
public function jobFinished()
{
// If we've indexed exactly as many as we would like, we are done
return $this->skipped
|| (count($this->getAllIndexes()) <= count($this->completed));
}
public function prepareForRestart()
{
// NOOP
}
public function afterComplete()
{
// NOOP
}
/**
* Abort this job, potentially rescheduling a replacement if there is still work to do
*/
protected function discardJob()
{
$this->skipped = true;
// If we do not have dirty records, then assume that these dirty records were committed
// already this request (but probably another job), so we don't need to commit anything else.
// This could occur if we completed multiple searchupdate jobs in a prior request, and
// we only need one commit job to commit all of them in the current request.
if (empty(static::$dirty_indexes)) {
$this->addMessage("Indexing already completed this request: Discarding this job");
return;
}
// If any commit has run, but some (or all) indexes are un-comitted, we must re-schedule this task.
// This could occur if we completed a searchupdate job in a prior request, as well as in
// the current request
$cooldown = Config::inst()->get(__CLASS__, 'cooldown');
$now = new DateTime(SS_Datetime::now()->getValue());
$now->add(new DateInterval('PT'.$cooldown.'S'));
$runat = $now->Format('Y-m-d H:i:s');
$this->addMessage("Indexing already run this request, but incomplete. Re-scheduling for {$runat}");
// Queue after the given cooldown
static::queue(false, $runat);
}
public function process()
{
// If we have already run an instance of SearchUpdateCommitJobProcessor this request, immediately
// quit this job to prevent hitting warming search limits in Solr
if (static::$has_run) {
$this->discardJob();
return true;
}
// To prevent other commit jobs from running this request
static::$has_run = true;
// Run all incompleted indexes
$indexNames = $this->getAllIndexes();
foreach ($indexNames as $name) {
$index = singleton($name);
$this->commitIndex($index);
}
$this->addMessage("All indexes committed");
return true;
}
/**
* Commits a specific index
*
* @param SolrIndex $index
* @throws Exception
*/
protected function commitIndex($index)
{
// Skip index if this is already complete
$name = get_class($index);
if (in_array($name, $this->completed)) {
$this->addMessage("Skipping already comitted index {$name}");
return;
}
// Bypass SolrIndex::commit exception handling so that queuedjobs can handle the error
$this->addMessage("Committing index {$name}");
$index->getService()->commit(false, false, false);
$this->addMessage("Committing index {$name} was successful");
// If this index is currently marked as dirty, it's now clean
if (in_array($name, static::$dirty_indexes)) {
static::$dirty_indexes = array_diff(static::$dirty_indexes, array($name));
}
// Mark complete
$this->completed[] = $name;
}
public function setup()
{
// NOOP
}
public function getJobData()
{
$data = new stdClass();
$data->totalSteps = count($this->getAllIndexes());
$data->currentStep = count($this->completed);
$data->isComplete = $this->jobFinished();
$data->messages = $this->messages;
$data->jobData = new stdClass();
$data->jobData->skipped = $this->skipped;
$data->jobData->completed = $this->completed;
$data->jobData->indexes = $this->getAllIndexes();
return $data;
}
public function setJobData($totalSteps, $currentStep, $isComplete, $jobData, $messages)
{
$this->isComplete = $isComplete;
$this->messages = $messages;
$this->skipped = $jobData->skipped;
$this->completed = $jobData->completed;
$this->indexes = $jobData->indexes;
}
public function addMessage($message, $severity='INFO')
{
$severity = strtoupper($severity);
$this->messages[] = '[' . date('Y-m-d H:i:s') . "][$severity] $message";
}
public function getMessages()
{
return $this->messages;
}
}

10
code/search/processors/SearchUpdateImmediateProcessor.php
View File

@ -1,7 +1,9 @@
<?php
class SearchUpdateImmediateProcessor extends SearchUpdateProcessor {
public function triggerProcessing() {
$this->process();
}
class SearchUpdateImmediateProcessor extends SearchUpdateProcessor
{
public function triggerProcessing()
{
$this->process();
}
}

28
code/search/processors/SearchUpdateMessageQueueProcessor.php
View File

@ -1,17 +1,19 @@
<?php
class SearchUpdateMessageQueueProcessor extends SearchUpdateProcessor {
/**
* The MessageQueue to use when processing updates
* @config
* @var string
*/
private static $reindex_queue = "search_indexing";
class SearchUpdateMessageQueueProcessor extends SearchUpdateProcessor
{
/**
* The MessageQueue to use when processing updates
* @config
* @var string
*/
private static $reindex_queue = "search_indexing";
public function triggerProcessing() {
MessageQueue::send(
Config::inst()->get('SearchMessageQueueUpdater', 'reindex_queue'),
new MethodInvocationMessage($this, "process")
);
}
public function triggerProcessing()
{
MessageQueue::send(
Config::inst()->get('SearchMessageQueueUpdater', 'reindex_queue'),
new MethodInvocationMessage($this, "process")
);
}
}

257
code/search/processors/SearchUpdateProcessor.php
View File

@ -1,137 +1,146 @@
<?php
abstract class SearchUpdateProcessor {
/**
* List of dirty records to process in format
*
* array(
* '$BaseClass' => array(
* '$State Key' => array(
* 'state' => array(
* 'key1' => 'value',
* 'key2' => 'value'
* ),
* 'ids' => array(
* '*id*' => array(
* '*Index Name 1*',
* '*Index Name 2*'
* )
* )
* )
* )
* )
*
* @var array
*/
protected $dirty;
public function __construct() {
$this->dirty = array();
}
abstract class SearchUpdateProcessor
{
/**
* List of dirty records to process in format
*
* array(
* '$BaseClass' => array(
* '$State Key' => array(
* 'state' => array(
* 'key1' => 'value',
* 'key2' => 'value'
* ),
* 'ids' => array(
* '*id*' => array(
* '*Index Name 1*',
* '*Index Name 2*'
* )
* )
* )
* )
* )
*
* @var array
*/
protected $dirty;
public function __construct()
{
$this->dirty = array();
}
public function addDirtyIDs($class, $statefulids, $index) {
$base = ClassInfo::baseDataClass($class);
$forclass = isset($this->dirty[$base]) ? $this->dirty[$base] : array();
public function addDirtyIDs($class, $statefulids, $index)
{
$base = ClassInfo::baseDataClass($class);
$forclass = isset($this->dirty[$base]) ? $this->dirty[$base] : array();
foreach ($statefulids as $statefulid) {
$id = $statefulid['id'];
$state = $statefulid['state']; $statekey = serialize($state);
foreach ($statefulids as $statefulid) {
$id = $statefulid['id'];
$state = $statefulid['state'];
$statekey = serialize($state);
if (!isset($forclass[$statekey])) {
$forclass[$statekey] = array('state' => $state, 'ids' => array($id => array($index)));
}
else if (!isset($forclass[$statekey]['ids'][$id])) {
$forclass[$statekey]['ids'][$id] = array($index);
}
else if (array_search($index, $forclass[$statekey]['ids'][$id]) === false) {
$forclass[$statekey]['ids'][$id][] = $index;
// dirty count stays the same
}
}
if (!isset($forclass[$statekey])) {
$forclass[$statekey] = array('state' => $state, 'ids' => array($id => array($index)));
} elseif (!isset($forclass[$statekey]['ids'][$id])) {
$forclass[$statekey]['ids'][$id] = array($index);
} elseif (array_search($index, $forclass[$statekey]['ids'][$id]) === false) {
$forclass[$statekey]['ids'][$id][] = $index;
// dirty count stays the same
}
}
$this->dirty[$base] = $forclass;
}
/**
* Generates the list of indexes to process for the dirty items
*
* @return array
*/
protected function prepareIndexes() {
$originalState = SearchVariant::current_state();
$dirtyIndexes = array();
$dirty = $this->getSource();
$indexes = FullTextSearch::get_indexes();
foreach ($dirty as $base => $statefulids) {
if (!$statefulids) continue;
$this->dirty[$base] = $forclass;
}
/**
* Generates the list of indexes to process for the dirty items
*
* @return array
*/
protected function prepareIndexes()
{
$originalState = SearchVariant::current_state();
$dirtyIndexes = array();
$dirty = $this->getSource();
$indexes = FullTextSearch::get_indexes();
foreach ($dirty as $base => $statefulids) {
if (!$statefulids) {
continue;
}
foreach ($statefulids as $statefulid) {
$state = $statefulid['state'];
$ids = $statefulid['ids'];
foreach ($statefulids as $statefulid) {
$state = $statefulid['state'];
$ids = $statefulid['ids'];
SearchVariant::activate_state($state);
SearchVariant::activate_state($state);
// Ensure that indexes for all new / updated objects are included
$objs = DataObject::get($base)->byIDs(array_keys($ids));
foreach ($objs as $obj) {
foreach ($ids[$obj->ID] as $index) {
if (!$indexes[$index]->variantStateExcluded($state)) {
$indexes[$index]->add($obj);
$dirtyIndexes[$index] = $indexes[$index];
}
}
unset($ids[$obj->ID]);
}
// Ensure that indexes for all new / updated objects are included
$objs = DataObject::get($base)->byIDs(array_keys($ids));
foreach ($objs as $obj) {
foreach ($ids[$obj->ID] as $index) {
if (!$indexes[$index]->variantStateExcluded($state)) {
$indexes[$index]->add($obj);
$dirtyIndexes[$index] = $indexes[$index];
}
}
unset($ids[$obj->ID]);
}
// Generate list of records that do not exist and should be removed
foreach ($ids as $id => $fromindexes) {
foreach ($fromindexes as $index) {
if (!$indexes[$index]->variantStateExcluded($state)) {
$indexes[$index]->delete($base, $id, $state);
$dirtyIndexes[$index] = $indexes[$index];
}
}
}
}
}
SearchVariant::activate_state($originalState);
return $dirtyIndexes;
}
/**
* Commits the specified index to the Solr service
*
* @param SolrIndex $index Index object
* @return bool Flag indicating success
*/
protected function commitIndex($index) {
return $index->commit() !== false;
}
/**
* Gets the record data source to process
*
* @return array
*/
protected function getSource() {
return $this->dirty;
}
// Generate list of records that do not exist and should be removed
foreach ($ids as $id => $fromindexes) {
foreach ($fromindexes as $index) {
if (!$indexes[$index]->variantStateExcluded($state)) {
$indexes[$index]->delete($base, $id, $state);
$dirtyIndexes[$index] = $indexes[$index];
}
}
}
}
}
SearchVariant::activate_state($originalState);
return $dirtyIndexes;
}
/**
* Commits the specified index to the Solr service
*
* @param SolrIndex $index Index object
* @return bool Flag indicating success
*/
protected function commitIndex($index)
{
return $index->commit() !== false;
}
/**
* Gets the record data source to process
*
* @return array
*/
protected function getSource()
{
return $this->dirty;
}
/**
* Process all indexes, returning true if successful
*
* @return bool Flag indicating success
*/
public function process() {
// Generate and commit all instances
$indexes = $this->prepareIndexes();
foreach ($indexes as $index) {
if(!$this->commitIndex($index)) return false;
}
return true;
}
/**
* Process all indexes, returning true if successful
*
* @return bool Flag indicating success
*/
public function process()
{
// Generate and commit all instances
$indexes = $this->prepareIndexes();
foreach ($indexes as $index) {
if (!$this->commitIndex($index)) {
return false;
}
}
return true;
}
abstract public function triggerProcessing();
abstract public function triggerProcessing();
}

182
code/search/processors/SearchUpdateQueuedJobProcessor.php
View File

@ -1,87 +1,101 @@
<?php
if(!interface_exists('QueuedJob')) return;
class SearchUpdateQueuedJobProcessor extends SearchUpdateBatchedProcessor implements QueuedJob {
/**
* The QueuedJob queue to use when processing updates
* @config
* @var int
*/
private static $reindex_queue = 2; // QueuedJob::QUEUED;
protected $messages = array();
public function triggerProcessing() {
parent::triggerProcessing();
singleton('QueuedJobService')->queueJob($this);
}
public function getTitle() {
return "FullTextSearch Update Job";
}
public function getSignature() {
return md5(get_class($this) . time() . mt_rand(0, 100000));
}
public function getJobType() {
return Config::inst()->get('SearchUpdateQueuedJobProcessor', 'reindex_queue');
}
public function jobFinished() {
return $this->currentBatch >= count($this->batches);
}
public function setup() {
// NOP
}
public function prepareForRestart() {
// NOP
}
public function afterComplete() {
// Once indexing is complete, commit later in order to avoid solr limits
// see http://stackoverflow.com/questions/7512945/how-to-fix-exceeded-limit-of-maxwarmingsearchers
SearchUpdateCommitJobProcessor::queue();
}
public function getJobData() {
$data = new stdClass();
$data->totalSteps = count($this->batches);
$data->currentStep = $this->currentBatch;
$data->isComplete = $this->jobFinished();
$data->messages = $this->messages;
$data->jobData = new stdClass();
$data->jobData->batches = $this->batches;
$data->jobData->currentBatch = $this->currentBatch;
return $data;
}
public function setJobData($totalSteps, $currentStep, $isComplete, $jobData, $messages) {
$this->isComplete = $isComplete;
$this->messages = $messages;
$this->batches = $jobData->batches;
$this->currentBatch = $jobData->currentBatch;
}
public function addMessage($message, $severity='INFO') {
$severity = strtoupper($severity);
$this->messages[] = '[' . date('Y-m-d H:i:s') . "][$severity] $message";
}
public function process() {
$result = parent::process();
if($this->jobFinished()) {
$this->addMessage("All batched updates complete. Queuing commit job");
}
return $result;
}
if (!interface_exists('QueuedJob')) {
return;
}
class SearchUpdateQueuedJobProcessor extends SearchUpdateBatchedProcessor implements QueuedJob
{
/**
* The QueuedJob queue to use when processing updates
* @config
* @var int
*/
private static $reindex_queue = 2; // QueuedJob::QUEUED;
protected $messages = array();
public function triggerProcessing()
{
parent::triggerProcessing();
singleton('QueuedJobService')->queueJob($this);
}
public function getTitle()
{
return "FullTextSearch Update Job";
}
public function getSignature()
{
return md5(get_class($this) . time() . mt_rand(0, 100000));
}
public function getJobType()
{
return Config::inst()->get('SearchUpdateQueuedJobProcessor', 'reindex_queue');
}
public function jobFinished()
{
return $this->currentBatch >= count($this->batches);
}
public function setup()
{
// NOP
}
public function prepareForRestart()
{
// NOP
}
public function afterComplete()
{
// Once indexing is complete, commit later in order to avoid solr limits
// see http://stackoverflow.com/questions/7512945/how-to-fix-exceeded-limit-of-maxwarmingsearchers
SearchUpdateCommitJobProcessor::queue();
}
public function getJobData()
{
$data = new stdClass();
$data->totalSteps = count($this->batches);
$data->currentStep = $this->currentBatch;
$data->isComplete = $this->jobFinished();
$data->messages = $this->messages;
$data->jobData = new stdClass();
$data->jobData->batches = $this->batches;
$data->jobData->currentBatch = $this->currentBatch;
return $data;
}
public function setJobData($totalSteps, $currentStep, $isComplete, $jobData, $messages)
{
$this->isComplete = $isComplete;
$this->messages = $messages;
$this->batches = $jobData->batches;
$this->currentBatch = $jobData->currentBatch;
}
public function addMessage($message, $severity='INFO')
{
$severity = strtoupper($severity);
$this->messages[] = '[' . date('Y-m-d H:i:s') . "][$severity] $message";
}
public function process()
{
$result = parent::process();
if ($this->jobFinished()) {
$this->addMessage("All batched updates complete. Queuing commit job");
}
return $result;
}
}

676
code/solr/Solr.php
View File

@ -5,284 +5,296 @@ use Monolog\Handler\StreamHandler;
use Monolog\Logger;
use Psr\Log\LoggerInterface;
class Solr {
class Solr
{
/**
* Configuration on where to find the solr server and how to get new index configurations into it.
*
* Required fields:
* host (default: localhost) - The host or IP Solr is listening on
* port (default: 8983) - The port Solr is listening on
* path (default: /solr) - The suburl the solr service is available on
*
* Optional fields:
* version (default: 4) - The Solr server version. Currently supports 3 and 4 (you can add a sub-version like 4.5 if
* you like, but currently it has no effect)
* service (default: depends on version, Solr3Service for 3, Solr4Service for 4)
* the class that provides actual communcation to the Solr server
* extraspath (default: <basefolder>/fulltextsearch/conf/solr/{version}/extras/) - Absolute path to
* the folder containing templates which are used for generating the schema and field definitions.
* templates (default: <basefolder>/fulltextsearch/conf/solr/{version}/templates/) - Absolute path to
* the configuration default files, e.g. solrconfig.xml.
*
* indexstore => an array with
*
* mode - a classname which implements SolrConfigStore, or 'file' or 'webdav'
*
* When mode == SolrConfigStore_File or file (indexes should be written on a local filesystem)
* path - The (locally accessible) path to write the index configurations to.
* remotepath (default: the same as indexpath) - The path that the Solr server will read the index configurations from
*
* When mode == SolrConfigStore_WebDAV or webdav (indexes should stored on a remote Solr server via webdav)
* auth (default: none) - A username:password pair string to use to auth against the webdav server
* path (default: /solrindex) - The suburl on the solr host that is set up to accept index configurations via webdav
* remotepath - The path that the Solr server will read the index configurations from
*/
protected static $solr_options = array();
/**
* Configuration on where to find the solr server and how to get new index configurations into it.
*
* Required fields:
* host (default: localhost) - The host or IP Solr is listening on
* port (default: 8983) - The port Solr is listening on
* path (default: /solr) - The suburl the solr service is available on
*
* Optional fields:
* version (default: 4) - The Solr server version. Currently supports 3 and 4 (you can add a sub-version like 4.5 if
* you like, but currently it has no effect)
* service (default: depends on version, Solr3Service for 3, Solr4Service for 4)
* the class that provides actual communcation to the Solr server
* extraspath (default: <basefolder>/fulltextsearch/conf/solr/{version}/extras/) - Absolute path to
* the folder containing templates which are used for generating the schema and field definitions.
* templates (default: <basefolder>/fulltextsearch/conf/solr/{version}/templates/) - Absolute path to
* the configuration default files, e.g. solrconfig.xml.
*
* indexstore => an array with
*
* mode - a classname which implements SolrConfigStore, or 'file' or 'webdav'
*
* When mode == SolrConfigStore_File or file (indexes should be written on a local filesystem)
* path - The (locally accessible) path to write the index configurations to.
* remotepath (default: the same as indexpath) - The path that the Solr server will read the index configurations from
*
* When mode == SolrConfigStore_WebDAV or webdav (indexes should stored on a remote Solr server via webdav)
* auth (default: none) - A username:password pair string to use to auth against the webdav server
* path (default: /solrindex) - The suburl on the solr host that is set up to accept index configurations via webdav
* remotepath - The path that the Solr server will read the index configurations from
*/
protected static $solr_options = array();
/** A cache of solr_options with the defaults all merged in */
protected static $merged_solr_options = null;
/** A cache of solr_options with the defaults all merged in */
protected static $merged_solr_options = null;
/**
* Update the configuration for Solr. See $solr_options for a discussion of the accepted array keys
* @param array $options - The options to update
*/
public static function configure_server($options = array())
{
self::$solr_options = array_merge(self::$solr_options, $options);
self::$merged_solr_options = null;
/**
* Update the configuration for Solr. See $solr_options for a discussion of the accepted array keys
* @param array $options - The options to update
*/
static function configure_server($options = array()) {
self::$solr_options = array_merge(self::$solr_options, $options);
self::$merged_solr_options = null;
self::$service_singleton = null;
self::$service_core_singletons = array();
}
self::$service_singleton = null;
self::$service_core_singletons = array();
}
/**
* Get the configured Solr options with the defaults all merged in
* @return array - The merged options
*/
public static function solr_options()
{
if (self::$merged_solr_options) {
return self::$merged_solr_options;
}
/**
* Get the configured Solr options with the defaults all merged in
* @return array - The merged options
*/
static function solr_options() {
if (self::$merged_solr_options) return self::$merged_solr_options;
$defaults = array(
'host' => 'localhost',
'port' => 8983,
'path' => '/solr',
'version' => '4'
);
$defaults = array(
'host' => 'localhost',
'port' => 8983,
'path' => '/solr',
'version' => '4'
);
// Build some by-version defaults
$version = isset(self::$solr_options['version']) ? self::$solr_options['version'] : $defaults['version'];
// Build some by-version defaults
$version = isset(self::$solr_options['version']) ? self::$solr_options['version'] : $defaults['version'];
if (version_compare($version, '4', '>=')) {
$versionDefaults = array(
'service' => 'Solr4Service',
'extraspath' => Director::baseFolder().'/fulltextsearch/conf/solr/4/extras/',
'templatespath' => Director::baseFolder().'/fulltextsearch/conf/solr/4/templates/',
);
} else {
$versionDefaults = array(
'service' => 'Solr3Service',
'extraspath' => Director::baseFolder().'/fulltextsearch/conf/solr/3/extras/',
'templatespath' => Director::baseFolder().'/fulltextsearch/conf/solr/3/templates/',
);
}
if (version_compare($version, '4', '>=')){
$versionDefaults = array(
'service' => 'Solr4Service',
'extraspath' => Director::baseFolder().'/fulltextsearch/conf/solr/4/extras/',
'templatespath' => Director::baseFolder().'/fulltextsearch/conf/solr/4/templates/',
);
}
else {
$versionDefaults = array(
'service' => 'Solr3Service',
'extraspath' => Director::baseFolder().'/fulltextsearch/conf/solr/3/extras/',
'templatespath' => Director::baseFolder().'/fulltextsearch/conf/solr/3/templates/',
);
}
return (self::$merged_solr_options = array_merge($defaults, $versionDefaults, self::$solr_options));
}
return (self::$merged_solr_options = array_merge($defaults, $versionDefaults, self::$solr_options));
}
static function set_service_class($class) {
user_error('set_service_class is deprecated - pass as part of $options to configure_server', E_USER_WARNING);
self::configure_server(array('service' => $class));
}
public static function set_service_class($class)
{
user_error('set_service_class is deprecated - pass as part of $options to configure_server', E_USER_WARNING);
self::configure_server(array('service' => $class));
}
/** @var SolrService | null - The instance of SolrService for core management */
static protected $service_singleton = null;
/** @var [SolrService_Core] - The instances of SolrService_Core for each core */
static protected $service_core_singletons = array();
/** @var SolrService | null - The instance of SolrService for core management */
protected static $service_singleton = null;
/** @var [SolrService_Core] - The instances of SolrService_Core for each core */
protected static $service_core_singletons = array();
/**
* Get a SolrService
*
* @param string $core Optional core name
* @return SolrService_Core
*/
static function service($core = null) {
$options = self::solr_options();
/**
* Get a SolrService
*
* @param string $core Optional core name
* @return SolrService_Core
*/
public static function service($core = null)
{
$options = self::solr_options();
if (!self::$service_singleton) {
self::$service_singleton = Object::create(
$options['service'], $options['host'], $options['port'], $options['path']
);
}
if (!self::$service_singleton) {
self::$service_singleton = Object::create(
$options['service'], $options['host'], $options['port'], $options['path']
);
}
if ($core) {
if (!isset(self::$service_core_singletons[$core])) {
self::$service_core_singletons[$core] = self::$service_singleton->serviceForCore(
singleton($core)->getIndexName()
);
}
if ($core) {
if (!isset(self::$service_core_singletons[$core])) {
self::$service_core_singletons[$core] = self::$service_singleton->serviceForCore(
singleton($core)->getIndexName()
);
}
return self::$service_core_singletons[$core];
} else {
return self::$service_singleton;
}
}
return self::$service_core_singletons[$core];
} else {
return self::$service_singleton;
}
}
static function get_indexes() {
return FullTextSearch::get_indexes('SolrIndex');
}
public static function get_indexes()
{
return FullTextSearch::get_indexes('SolrIndex');
}
/**
* Include the thirdparty Solr client api library. Done this way to avoid issues where code is called in
* mysite/_config before fulltextsearch/_config has a change to update the include path.
*/
static function include_client_api() {
static $included = false;
/**
* Include the thirdparty Solr client api library. Done this way to avoid issues where code is called in
* mysite/_config before fulltextsearch/_config has a change to update the include path.
*/
public static function include_client_api()
{
static $included = false;
if (!$included) {
set_include_path(get_include_path() . PATH_SEPARATOR . Director::baseFolder() . '/fulltextsearch/thirdparty/solr-php-client');
require_once('Apache/Solr/Service.php');
require_once('Apache/Solr/Document.php');
if (!$included) {
set_include_path(get_include_path() . PATH_SEPARATOR . Director::baseFolder() . '/fulltextsearch/thirdparty/solr-php-client');
require_once('Apache/Solr/Service.php');
require_once('Apache/Solr/Document.php');
$included = true;
}
}
$included = true;
}
}
}
/**
* Abstract class for build tasks
*/
class Solr_BuildTask extends BuildTask {
class Solr_BuildTask extends BuildTask
{
protected $enabled = false;
protected $enabled = false;
/**
* Logger
*
* @var LoggerInterface
*/
protected $logger = null;
/**
* Logger
*
* @var LoggerInterface
*/
protected $logger = null;
/**
* Get the current logger
*
* @return LoggerInterface
*/
public function getLogger()
{
return $this->logger;
}
/**
* Get the current logger
*
* @return LoggerInterface
*/
public function getLogger() {
return $this->logger;
}
/**
* Assign a new logger
*
* @param LoggerInterface $logger
*/
public function setLogger(LoggerInterface $logger)
{
$this->logger = $logger;
}
/**
* Assign a new logger
*
* @param LoggerInterface $logger
*/
public function setLogger(LoggerInterface $logger) {
$this->logger = $logger;
}
/**
* @return SearchLogFactory
*/
protected function getLoggerFactory()
{
return Injector::inst()->get('SearchLogFactory');
}
/**
* @return SearchLogFactory
*/
protected function getLoggerFactory() {
return Injector::inst()->get('SearchLogFactory');
}
/**
* Setup task
*
* @param SS_HTTPReqest $request
*/
public function run($request)
{
$name = get_class($this);
$verbose = $request->getVar('verbose');
/**
* Setup task
*
* @param SS_HTTPReqest $request
*/
public function run($request) {
$name = get_class($this);
$verbose = $request->getVar('verbose');
// Set new logger
$logger = $this
->getLoggerFactory()
->getOutputLogger($name, $verbose);
$this->setLogger($logger);
}
// Set new logger
$logger = $this
->getLoggerFactory()
->getOutputLogger($name, $verbose);
$this->setLogger($logger);
}
}
class Solr_Configure extends Solr_BuildTask {
class Solr_Configure extends Solr_BuildTask
{
protected $enabled = true;
protected $enabled = true;
public function run($request)
{
parent::run($request);
// Find the IndexStore handler, which will handle uploading config files to Solr
$store = $this->getSolrConfigStore();
$indexes = Solr::get_indexes();
foreach ($indexes as $instance) {
try {
$this->updateIndex($instance, $store);
} catch (Exception $e) {
// We got an exception. Warn, but continue to next index.
$this
->getLogger()
->error("Failure: " . $e->getMessage());
}
}
}
/**
* Update the index on the given store
*
* @param SolrIndex $instance Instance
* @param SolrConfigStore $store
*/
protected function updateIndex($instance, $store)
{
$index = $instance->getIndexName();
$this->getLogger()->info("Configuring $index.");
// Upload the config files for this index
$this->getLogger()->info("Uploading configuration ...");
$instance->uploadConfig($store);
// Then tell Solr to use those config files
$service = Solr::service();
if ($service->coreIsActive($index)) {
$this->getLogger()->info("Reloading core ...");
$service->coreReload($index);
} else {
$this->getLogger()->info("Creating core ...");
$service->coreCreate($index, $store->instanceDir($index));
}
$this->getLogger()->info("Done");
}
public function run($request) {
parent::run($request);
// Find the IndexStore handler, which will handle uploading config files to Solr
$store = $this->getSolrConfigStore();
$indexes = Solr::get_indexes();
foreach ($indexes as $instance) {
/**
* Get config store
*
* @return SolrConfigStore
*/
protected function getSolrConfigStore()
{
$options = Solr::solr_options();
if (!isset($options['indexstore']) || !($indexstore = $options['indexstore'])) {
user_error('No index configuration for Solr provided', E_USER_ERROR);
}
// Find the IndexStore handler, which will handle uploading config files to Solr
$mode = $indexstore['mode'];
try {
$this->updateIndex($instance, $store);
} catch(Exception $e) {
// We got an exception. Warn, but continue to next index.
$this
->getLogger()
->error("Failure: " . $e->getMessage());
}
}
}
/**
* Update the index on the given store
*
* @param SolrIndex $instance Instance
* @param SolrConfigStore $store
*/
protected function updateIndex($instance, $store) {
$index = $instance->getIndexName();
$this->getLogger()->info("Configuring $index.");
// Upload the config files for this index
$this->getLogger()->info("Uploading configuration ...");
$instance->uploadConfig($store);
// Then tell Solr to use those config files
$service = Solr::service();
if ($service->coreIsActive($index)) {
$this->getLogger()->info("Reloading core ...");
$service->coreReload($index);
} else {
$this->getLogger()->info("Creating core ...");
$service->coreCreate($index, $store->instanceDir($index));
}
$this->getLogger()->info("Done");
}
/**
* Get config store
*
* @return SolrConfigStore
*/
protected function getSolrConfigStore() {
$options = Solr::solr_options();
if (!isset($options['indexstore']) || !($indexstore = $options['indexstore'])) {
user_error('No index configuration for Solr provided', E_USER_ERROR);
}
// Find the IndexStore handler, which will handle uploading config files to Solr
$mode = $indexstore['mode'];
if ($mode == 'file') {
return new SolrConfigStore_File($indexstore);
} elseif ($mode == 'webdav') {
return new SolrConfigStore_WebDAV($indexstore);
} elseif (ClassInfo::exists($mode) && ClassInfo::classImplements($mode, 'SolrConfigStore')) {
return new $mode($indexstore);
} else {
user_error('Unknown Solr index mode '.$indexstore['mode'], E_USER_ERROR);
}
}
if ($mode == 'file') {
return new SolrConfigStore_File($indexstore);
} elseif ($mode == 'webdav') {
return new SolrConfigStore_WebDAV($indexstore);
} elseif (ClassInfo::exists($mode) && ClassInfo::classImplements($mode, 'SolrConfigStore')) {
return new $mode($indexstore);
} else {
user_error('Unknown Solr index mode '.$indexstore['mode'], E_USER_ERROR);
}
}
}
/**
@ -300,106 +312,110 @@ class Solr_Configure extends Solr_BuildTask {
* - variantstate
* - verbose (optional)
*/
class Solr_Reindex extends Solr_BuildTask {
class Solr_Reindex extends Solr_BuildTask
{
protected $enabled = true;
protected $enabled = true;
/**
* Number of records to load and index per request
*
* @var int
* @config
*/
private static $recordsPerRequest = 200;
/**
* Number of records to load and index per request
*
* @var int
* @config
*/
private static $recordsPerRequest = 200;
/**
* Get the reindex handler
*
* @return SolrReindexHandler
*/
protected function getHandler()
{
return Injector::inst()->get('SolrReindexHandler');
}
/**
* Get the reindex handler
*
* @return SolrReindexHandler
*/
protected function getHandler() {
return Injector::inst()->get('SolrReindexHandler');
}
/**
* @param SS_HTTPRequest $request
*/
public function run($request)
{
parent::run($request);
// Reset state
$originalState = SearchVariant::current_state();
$this->doReindex($request);
SearchVariant::activate_state($originalState);
}
/**
* @param SS_HTTPRequest $request
*/
public function run($request) {
parent::run($request);
// Reset state
$originalState = SearchVariant::current_state();
$this->doReindex($request);
SearchVariant::activate_state($originalState);
}
/**
* @param SS_HTTPRequest $request
*/
protected function doReindex($request)
{
$class = $request->getVar('class');
/**
* @param SS_HTTPRequest $request
*/
protected function doReindex($request) {
$class = $request->getVar('class');
// Deprecated reindex mechanism
$start = $request->getVar('start');
if ($start !== null) {
// Run single batch directly
$indexInstance = singleton($request->getVar('index'));
$state = json_decode($request->getVar('variantstate'), true);
$this->runFrom($indexInstance, $class, $start, $state);
return;
}
// Deprecated reindex mechanism
$start = $request->getVar('start');
if ($start !== null) {
// Run single batch directly
$indexInstance = singleton($request->getVar('index'));
$state = json_decode($request->getVar('variantstate'), true);
$this->runFrom($indexInstance, $class, $start, $state);
return;
}
// Check if we are re-indexing a single group
// If not using queuedjobs, we need to invoke Solr_Reindex as a separate process
// Otherwise each group is processed via a SolrReindexGroupJob
$groups = $request->getVar('groups');
$handler = $this->getHandler();
if ($groups) {
// Run grouped batches (id % groups = group)
$group = $request->getVar('group');
$indexInstance = singleton($request->getVar('index'));
$state = json_decode($request->getVar('variantstate'), true);
// Check if we are re-indexing a single group
// If not using queuedjobs, we need to invoke Solr_Reindex as a separate process
// Otherwise each group is processed via a SolrReindexGroupJob
$groups = $request->getVar('groups');
$handler = $this->getHandler();
if($groups) {
// Run grouped batches (id % groups = group)
$group = $request->getVar('group');
$indexInstance = singleton($request->getVar('index'));
$state = json_decode($request->getVar('variantstate'), true);
$handler->runGroup($this->getLogger(), $indexInstance, $state, $class, $groups, $group);
return;
}
$handler->runGroup($this->getLogger(), $indexInstance, $state, $class, $groups, $group);
return;
}
// If run at the top level, delegate to appropriate handler
$self = get_class($this);
$handler->triggerReindex($this->getLogger(), $this->config()->recordsPerRequest, $self, $class);
}
// If run at the top level, delegate to appropriate handler
$self = get_class($this);
$handler->triggerReindex($this->getLogger(), $this->config()->recordsPerRequest, $self, $class);
}
/**
* @deprecated since version 2.0.0
*/
protected function runFrom($index, $class, $start, $variantstate)
{
DeprecationTest_Deprecation::notice('2.0.0', 'Solr_Reindex now uses a new grouping mechanism');
// Set time limit and state
increase_time_limit_to();
SearchVariant::activate_state($variantstate);
/**
* @deprecated since version 2.0.0
*/
protected function runFrom($index, $class, $start, $variantstate) {
DeprecationTest_Deprecation::notice('2.0.0', 'Solr_Reindex now uses a new grouping mechanism');
// Set time limit and state
increase_time_limit_to();
SearchVariant::activate_state($variantstate);
// Generate filtered list
$items = DataList::create($class)
->limit($this->config()->recordsPerRequest, $start);
// Generate filtered list
$items = DataList::create($class)
->limit($this->config()->recordsPerRequest, $start);
// Add child filter
$classes = $index->getClasses();
$options = $classes[$class];
if (!$options['include_children']) {
$items = $items->filter('ClassName', $class);
}
// Add child filter
$classes = $index->getClasses();
$options = $classes[$class];
if(!$options['include_children']) {
$items = $items->filter('ClassName', $class);
}
// Process selected records in this class
$this->getLogger()->info("Adding $class");
foreach ($items->sort("ID") as $item) {
$this->getLogger()->debug($item->ID);
// Process selected records in this class
$this->getLogger()->info("Adding $class");
foreach ($items->sort("ID") as $item) {
$this->getLogger()->debug($item->ID);
// See SearchUpdater_ObjectHandler::triggerReindex
$item->triggerReindex();
$item->destroy();
}
// See SearchUpdater_ObjectHandler::triggerReindex
$item->triggerReindex();
$item->destroy();
}
$this->getLogger()->info("Done");
}
$this->getLogger()->info("Done");
}
}

8
code/solr/Solr3Service.php
View File

@ -1,8 +1,10 @@
<?php
class Solr3Service_Core extends SolrService_Core {
class Solr3Service_Core extends SolrService_Core
{
}
class Solr3Service extends SolrService {
private static $core_class = 'Solr3Service_Core';
class Solr3Service extends SolrService
{
private static $core_class = 'Solr3Service_Core';
}

95
code/solr/Solr4Service.php
View File

@ -1,57 +1,58 @@
<?php
class Solr4Service_Core extends SolrService_Core {
class Solr4Service_Core extends SolrService_Core
{
/**
* Replace underlying commit function to remove waitFlush in 4.0+, since it's been deprecated and 4.4 throws errors
* if you pass it
*/
public function commit($expungeDeletes = false, $waitFlush = null, $waitSearcher = true, $timeout = 3600)
{
if ($waitFlush) {
user_error('waitFlush must be false when using Solr 4.0+' . E_USER_ERROR);
}
/**
* Replace underlying commit function to remove waitFlush in 4.0+, since it's been deprecated and 4.4 throws errors
* if you pass it
*/
public function commit($expungeDeletes = false, $waitFlush = null, $waitSearcher = true, $timeout = 3600) {
if ($waitFlush) {
user_error('waitFlush must be false when using Solr 4.0+' . E_USER_ERROR);
}
$expungeValue = $expungeDeletes ? 'true' : 'false';
$searcherValue = $waitSearcher ? 'true' : 'false';
$expungeValue = $expungeDeletes ? 'true' : 'false';
$searcherValue = $waitSearcher ? 'true' : 'false';
$rawPost = '<commit expungeDeletes="' . $expungeValue . '" waitSearcher="' . $searcherValue . '" />';
return $this->_sendRawPost($this->_updateUrl, $rawPost, $timeout);
}
/**
* @inheritdoc
* @see Solr4Service_Core::addDocuments
*/
public function addDocument(Apache_Solr_Document $document, $allowDups = false,
$overwritePending = true, $overwriteCommitted = true, $commitWithin = 0
) {
return $this->addDocuments(array($document), $allowDups, $overwritePending, $overwriteCommitted, $commitWithin);
}
$rawPost = '<commit expungeDeletes="' . $expungeValue . '" waitSearcher="' . $searcherValue . '" />';
return $this->_sendRawPost($this->_updateUrl, $rawPost, $timeout);
}
/**
* @inheritdoc
* @see Solr4Service_Core::addDocuments
*/
public function addDocument(Apache_Solr_Document $document, $allowDups = false,
$overwritePending = true, $overwriteCommitted = true, $commitWithin = 0
) {
return $this->addDocuments(array($document), $allowDups, $overwritePending, $overwriteCommitted, $commitWithin);
}
/**
* Solr 4.0 compat http://wiki.apache.org/solr/UpdateXmlMessages#Optional_attributes_for_.22add.22
* Remove allowDups, overwritePending and overwriteComitted
*/
public function addDocuments($documents, $allowDups = false, $overwritePending = true,
$overwriteCommitted = true, $commitWithin = 0
) {
$overwriteVal = $allowDups ? 'false' : 'true';
$commitWithin = (int) $commitWithin;
$commitWithinString = $commitWithin > 0 ? " commitWithin=\"{$commitWithin}\"" : '';
/**
* Solr 4.0 compat http://wiki.apache.org/solr/UpdateXmlMessages#Optional_attributes_for_.22add.22
* Remove allowDups, overwritePending and overwriteComitted
*/
public function addDocuments($documents, $allowDups = false, $overwritePending = true,
$overwriteCommitted = true, $commitWithin = 0
) {
$overwriteVal = $allowDups ? 'false' : 'true';
$commitWithin = (int) $commitWithin;
$commitWithinString = $commitWithin > 0 ? " commitWithin=\"{$commitWithin}\"" : '';
$rawPost = "<add overwrite=\"{$overwriteVal}\"{$commitWithinString}>";
foreach ($documents as $document) {
if ($document instanceof Apache_Solr_Document) {
$rawPost .= $this->_documentToXmlFragment($document);
}
}
$rawPost .= '</add>';
$rawPost = "<add overwrite=\"{$overwriteVal}\"{$commitWithinString}>";
foreach ($documents as $document) {
if ($document instanceof Apache_Solr_Document) {
$rawPost .= $this->_documentToXmlFragment($document);
}
}
$rawPost .= '</add>';
return $this->add($rawPost);
}
return $this->add($rawPost);
}
}
class Solr4Service extends SolrService {
private static $core_class = 'Solr4Service_Core';
class Solr4Service extends SolrService
{
private static $core_class = 'Solr4Service_Core';
}

173
code/solr/SolrConfigStore.php
View File

@ -5,29 +5,30 @@
*
* The interface Solr_Configure uses to upload configuration files to Solr
*/
interface SolrConfigStore {
/**
* Upload a file to Solr for index $index
* @param $index string - The name of an index (which is also used as the name of the Solr core for the index)
* @param $file string - A path to a file to upload. The base name of the file will be used on the remote side
* @return null
*/
function uploadFile($index, $file);
interface SolrConfigStore
{
/**
* Upload a file to Solr for index $index
* @param $index string - The name of an index (which is also used as the name of the Solr core for the index)
* @param $file string - A path to a file to upload. The base name of the file will be used on the remote side
* @return null
*/
public function uploadFile($index, $file);
/**
* Upload a file to Solr from a string for index $index
* @param $index string - The name of an index (which is also used as the name of the Solr core for the index)
* @param $filename string - The base name of the file to use on the remote side
* @param $strong string - The contents of the file
* @return null
*/
function uploadString($index, $filename, $string);
/**
* Upload a file to Solr from a string for index $index
* @param $index string - The name of an index (which is also used as the name of the Solr core for the index)
* @param $filename string - The base name of the file to use on the remote side
* @param $strong string - The contents of the file
* @return null
*/
public function uploadString($index, $filename, $string);
/**
* Get the instanceDir to tell Solr to use for index $index
* @param $index string - The name of an index (which is also used as the name of the Solr core for the index)
*/
function instanceDir($index);
/**
* Get the instanceDir to tell Solr to use for index $index
* @param $index string - The name of an index (which is also used as the name of the Solr core for the index)
*/
public function instanceDir($index);
}
/**
@ -36,41 +37,47 @@ interface SolrConfigStore {
* A ConfigStore that uploads files to a Solr instance on a locally accessible filesystem
* by just using file copies
*/
class SolrConfigStore_File implements SolrConfigStore {
function __construct($config) {
$this->local = $config['path'];
$this->remote = isset($config['remotepath']) ? $config['remotepath'] : $config['path'];
}
class SolrConfigStore_File implements SolrConfigStore
{
public function __construct($config)
{
$this->local = $config['path'];
$this->remote = isset($config['remotepath']) ? $config['remotepath'] : $config['path'];
}
function getTargetDir($index) {
$targetDir = "{$this->local}/{$index}/conf";
public function getTargetDir($index)
{
$targetDir = "{$this->local}/{$index}/conf";
if (!is_dir($targetDir)) {
$worked = @mkdir($targetDir, 0770, true);
if (!is_dir($targetDir)) {
$worked = @mkdir($targetDir, 0770, true);
if(!$worked) {
throw new RuntimeException(
sprintf('Failed creating target directory %s, please check permissions', $targetDir)
);
}
}
if (!$worked) {
throw new RuntimeException(
sprintf('Failed creating target directory %s, please check permissions', $targetDir)
);
}
}
return $targetDir;
}
return $targetDir;
}
function uploadFile($index, $file) {
$targetDir = $this->getTargetDir($index);
copy($file, $targetDir.'/'.basename($file));
}
public function uploadFile($index, $file)
{
$targetDir = $this->getTargetDir($index);
copy($file, $targetDir.'/'.basename($file));
}
function uploadString($index, $filename, $string) {
$targetDir = $this->getTargetDir($index);
file_put_contents("$targetDir/$filename", $string);
}
public function uploadString($index, $filename, $string)
{
$targetDir = $this->getTargetDir($index);
file_put_contents("$targetDir/$filename", $string);
}
function instanceDir($index) {
return $this->remote.'/'.$index;
}
public function instanceDir($index)
{
return $this->remote.'/'.$index;
}
}
/**
@ -78,40 +85,50 @@ class SolrConfigStore_File implements SolrConfigStore {
*
* A ConfigStore that uploads files to a Solr instance via a WebDAV server
*/
class SolrConfigStore_WebDAV implements SolrConfigStore {
function __construct($config) {
$options = Solr::solr_options();
class SolrConfigStore_WebDAV implements SolrConfigStore
{
public function __construct($config)
{
$options = Solr::solr_options();
$this->url = implode('', array(
'http://',
isset($config['auth']) ? $config['auth'].'@' : '',
$options['host'].':'.$options['port'],
$config['path']
));
$this->remote = $config['remotepath'];
}
$this->url = implode('', array(
'http://',
isset($config['auth']) ? $config['auth'].'@' : '',
$options['host'].':'.$options['port'],
$config['path']
));
$this->remote = $config['remotepath'];
}
function getTargetDir($index) {
$indexdir = "{$this->url}/$index";
if (!WebDAV::exists($indexdir)) WebDAV::mkdir($indexdir);
public function getTargetDir($index)
{
$indexdir = "{$this->url}/$index";
if (!WebDAV::exists($indexdir)) {
WebDAV::mkdir($indexdir);
}
$targetDir = "{$this->url}/$index/conf";
if (!WebDAV::exists($targetDir)) WebDAV::mkdir($targetDir);
$targetDir = "{$this->url}/$index/conf";
if (!WebDAV::exists($targetDir)) {
WebDAV::mkdir($targetDir);
}
return $targetDir;
}
return $targetDir;
}
function uploadFile($index, $file) {
$targetDir = $this->getTargetDir($index);
WebDAV::upload_from_file($file, $targetDir.'/'.basename($file));
}
public function uploadFile($index, $file)
{
$targetDir = $this->getTargetDir($index);
WebDAV::upload_from_file($file, $targetDir.'/'.basename($file));
}
function uploadString($index, $filename, $string) {
$targetDir = $this->getTargetDir($index);
WebDAV::upload_from_string($string, "$targetDir/$filename");
}
public function uploadString($index, $filename, $string)
{
$targetDir = $this->getTargetDir($index);
WebDAV::upload_from_string($string, "$targetDir/$filename");
}
function instanceDir($index) {
return $this->remote ? "{$this->remote}/$index" : $index;
}
public function instanceDir($index)
{
return $this->remote ? "{$this->remote}/$index" : $index;
}
}

1806
code/solr/SolrIndex.php
View File

@ -2,868 +2,946 @@
Solr::include_client_api();
abstract class SolrIndex extends SearchIndex {
static $fulltextTypeMap = array(
'*' => 'text',
'HTMLVarchar' => 'htmltext',
'HTMLText' => 'htmltext'
);
static $filterTypeMap = array(
'*' => 'string',
'Boolean' => 'boolean',
'Date' => 'tdate',
'SSDatetime' => 'tdate',
'SS_Datetime' => 'tdate',
'ForeignKey' => 'tint',
'Int' => 'tint',
'Float' => 'tfloat',
'Double' => 'tdouble'
);
static $sortTypeMap = array();
protected $analyzerFields = array();
protected $copyFields = array();
protected $extrasPath = null;
protected $templatesPath = null;
/**
* List of boosted fields
*
* @var array
*/
protected $boostedFields = array();
/**
* Name of default field
*
* @var string
* @config
*/
private static $default_field = '_text';
/**
* List of copy fields all fulltext fields should be copied into.
* This will fallback to default_field if not specified
*
* @var array
*/
private static $copy_fields = array();
/**
* @return String Absolute path to the folder containing
* templates which are used for generating the schema and field definitions.
*/
function getTemplatesPath() {
$globalOptions = Solr::solr_options();
return $this->templatesPath ? $this->templatesPath : $globalOptions['templatespath'];
}
/**
* @return String Absolute path to the configuration default files,
* e.g. solrconfig.xml.
*/
function getExtrasPath() {
$globalOptions = Solr::solr_options();
return $this->extrasPath ? $this->extrasPath : $globalOptions['extraspath'];
}
function generateSchema() {
return $this->renderWith($this->getTemplatesPath() . '/schema.ss');
}
function getIndexName() {
return get_class($this);
}
function getTypes() {
return $this->renderWith($this->getTemplatesPath() . '/types.ss');
}
/**
* Index-time analyzer which is applied to a specific field.
* Can be used to remove HTML tags, apply stemming, etc.
*
* @see http://wiki.apache.org/solr/AnalyzersTokenizersTokenFilters#solr.WhitespaceTokenizerFactory
*
* @param String $field
* @param String $type
* @param Array $params Parameters for the analyzer, usually at least a "class"
*/
function addAnalyzer($field, $type, $params) {
$fullFields = $this->fieldData($field);
if($fullFields) foreach($fullFields as $fullField => $spec) {
if(!isset($this->analyzerFields[$fullField])) $this->analyzerFields[$fullField] = array();
$this->analyzerFields[$fullField][$type] = $params;
}
}
/**
* Get the default text field, normally '_text'
*
* @return string
*/
public function getDefaultField() {
return $this->config()->default_field;
}
/**
* Get list of fields each text field should be copied into.
* This will fallback to the default field if omitted.
*
* @return array
*/
protected function getCopyDestinations() {
$copyFields = $this->config()->copy_fields;
if($copyFields) {
return $copyFields;
}
// Fallback to default field
$df = $this->getDefaultField();
return array($df);
}
public function getFieldDefinitions() {
$xml = array();
$stored = $this->getStoredDefault();
$xml[] = "";
// Add the hardcoded field definitions
$xml[] = "<field name='_documentid' type='string' indexed='true' stored='true' required='true' />";
$xml[] = "<field name='ID' type='tint' indexed='true' stored='true' required='true' />";
$xml[] = "<field name='ClassName' type='string' indexed='true' stored='true' required='true' />";
$xml[] = "<field name='ClassHierarchy' type='string' indexed='true' stored='true' required='true' multiValued='true' />";
// Add the fulltext collation field
$df = $this->getDefaultField();
$xml[] = "<field name='{$df}' type='htmltext' indexed='true' stored='{$stored}' multiValued='true' />" ;
// Add the user-specified fields
foreach ($this->fulltextFields as $name => $field) {
$xml[] = $this->getFieldDefinition($name, $field, self::$fulltextTypeMap);
}
foreach ($this->filterFields as $name => $field) {
if ($field['fullfield'] == 'ID' || $field['fullfield'] == 'ClassName') continue;
$xml[] = $this->getFieldDefinition($name, $field);
}
foreach ($this->sortFields as $name => $field) {
if ($field['fullfield'] == 'ID' || $field['fullfield'] == 'ClassName') continue;
$xml[] = $this->getFieldDefinition($name, $field);
}
return implode("\n\t\t", $xml);
}
/**
* Extract first suggestion text from collated values
*
* @param mixed $collation
* @return string
*/
protected function getCollatedSuggestion($collation = '') {
if(is_string($collation)) {
return $collation;
}
if(is_object($collation)) {
if(isset($collation->misspellingsAndCorrections)) {
foreach($collation->misspellingsAndCorrections as $key => $value) {
return $value;
}
}
}
return '';
}
/**
* Extract a human friendly spelling suggestion from a Solr spellcheck collation string.
* @param String $collation
* @return String
*/
protected function getNiceSuggestion($collation = '') {
$collationParts = explode(' ', $collation);
// Remove advanced query params from the beginning of each collation part.
foreach ($collationParts as $key => &$part) {
$part = ltrim($part, '+');
}
return implode(' ', $collationParts);
}
/**
* Extract a query string from a Solr spellcheck collation string.
* Useful for constructing 'Did you mean?' links, for example:
* <a href="http://example.com/search?q=$SuggestionQueryString">$SuggestionNice</a>
* @param String $collation
* @return String
*/
protected function getSuggestionQueryString($collation = '') {
return str_replace(' ', '+', $this->getNiceSuggestion($collation));
}
/**
* Add a field that should be stored
*
* @param string $field The field to add
* @param string $forceType The type to force this field as (required in some cases, when not
* detectable from metadata)
* @param array $extraOptions Dependent on search implementation
*/
public function addStoredField($field, $forceType = null, $extraOptions = array()) {
$options = array_merge($extraOptions, array('stored' => 'true'));
$this->addFulltextField($field, $forceType, $options);
}
/**
* Add a fulltext field with a boosted value
*
* @param string $field The field to add
* @param string $forceType The type to force this field as (required in some cases, when not
* detectable from metadata)
* @param array $extraOptions Dependent on search implementation
* @param float $boost Numeric boosting value (defaults to 2)
*/
public function addBoostedField($field, $forceType = null, $extraOptions = array(), $boost = 2) {
$options = array_merge($extraOptions, array('boost' => $boost));
$this->addFulltextField($field, $forceType, $options);
}
public function fieldData($field, $forceType = null, $extraOptions = array()) {
// Ensure that 'boost' is recorded here without being captured by solr
$boost = null;
if(array_key_exists('boost', $extraOptions)) {
$boost = $extraOptions['boost'];
unset($extraOptions['boost']);
}
$data = parent::fieldData($field, $forceType, $extraOptions);
// Boost all fields with this name
if(isset($boost)) {
foreach($data as $fieldName => $fieldInfo) {
$this->boostedFields[$fieldName] = $boost;
}
}
return $data;
}
/**
* Set the default boosting level for a specific field.
* Will control the default value for qf param (Query Fields), but will not
* override a query-specific value.
*
* Fields must be added before having a field boosting specified
*
* @param string $field Full field key (Model_Field)
* @param float|null $level Numeric boosting value. Set to null to clear boost
*/
public function setFieldBoosting($field, $level) {
if(!isset($this->fulltextFields[$field])) {
throw new InvalidArgumentException("No fulltext field $field exists on ".$this->getIndexName());
}
if($level === null) {
unset($this->boostedFields[$field]);
} else {
$this->boostedFields[$field] = $level;
}
}
/**
* Get all boosted fields
*
* @return array
*/
public function getBoostedFields() {
return $this->boostedFields;
}
/**
* Determine the best default value for the 'qf' parameter
*
* @return array|null List of query fields, or null if not specified
*/
public function getQueryFields() {
// Not necessary to specify this unless boosting
if(empty($this->boostedFields)) {
return null;
}
$queryFields = array();
foreach ($this->boostedFields as $fieldName => $boost) {
$queryFields[] = $fieldName . '^' . $boost;
}
// If any fields are queried, we must always include the default field, otherwise it will be excluded
$df = $this->getDefaultField();
if($queryFields && !isset($this->boostedFields[$df])) {
$queryFields[] = $df;
}
return $queryFields;
}
/**
* Gets the default 'stored' value for fields in this index
*
* @return string A default value for the 'stored' field option, either 'true' or 'false'
*/
protected function getStoredDefault() {
return Director::isDev() ? 'true' : 'false';
}
/**
* @param String $name
* @param Array $spec
* @param Array $typeMap
* @return String XML
*/
protected function getFieldDefinition($name, $spec, $typeMap = null) {
if(!$typeMap) $typeMap = self::$filterTypeMap;
$multiValued = (isset($spec['multi_valued']) && $spec['multi_valued']) ? "true" : '';
$type = isset($typeMap[$spec['type']]) ? $typeMap[$spec['type']] : $typeMap['*'];
$analyzerXml = '';
if(isset($this->analyzerFields[$name])) {
foreach($this->analyzerFields[$name] as $analyzerType => $analyzerParams) {
$analyzerXml .= $this->toXmlTag($analyzerType, $analyzerParams);
}
}
$fieldParams = array_merge(
array(
'name' => $name,
'type' => $type,
'indexed' => 'true',
'stored' => $this->getStoredDefault(),
'multiValued' => $multiValued
),
isset($spec['extra_options']) ? $spec['extra_options'] : array()
);
return $this->toXmlTag(
"field",
$fieldParams,
$analyzerXml ? "<analyzer>$analyzerXml</analyzer>" : null
);
}
/**
* Convert definition to XML tag
*
* @param String $tag
* @param String $attrs Map of attributes
* @param String $content Inner content
* @return String XML tag
*/
protected function toXmlTag($tag, $attrs, $content = null) {
$xml = "<$tag ";
if($attrs) {
$attrStrs = array();
foreach($attrs as $attrName => $attrVal) $attrStrs[] = "$attrName='$attrVal'";
$xml .= $attrStrs ? implode(' ', $attrStrs) : '';
}
$xml .= $content ? ">$content</$tag>" : '/>';
return $xml;
}
/**
* @param String $source Composite field name (<class>_<fieldname>)
* @param String $dest
*/
function addCopyField($source, $dest, $extraOptions = array()) {
if(!isset($this->copyFields[$source])) $this->copyFields[$source] = array();
$this->copyFields[$source][] = array_merge(
array('source' => $source, 'dest' => $dest),
$extraOptions
);
}
/**
* Generate XML for copy field definitions
*
* @return string
*/
public function getCopyFieldDefinitions() {
$xml = array();
// Default copy fields
foreach($this->getCopyDestinations() as $copyTo) {
foreach ($this->fulltextFields as $name => $field) {
$xml[] = "<copyField source='{$name}' dest='{$copyTo}' />";
}
}
// Explicit copy fields
foreach ($this->copyFields as $source => $fields) {
foreach($fields as $fieldAttrs) {
$xml[] = $this->toXmlTag('copyField', $fieldAttrs);
}
}
return implode("\n\t", $xml);
}
protected function _addField($doc, $object, $field) {
$class = get_class($object);
if ($class != $field['origin'] && !is_subclass_of($class, $field['origin'])) return;
$value = $this->_getFieldValue($object, $field);
$type = isset(self::$filterTypeMap[$field['type']]) ? self::$filterTypeMap[$field['type']] : self::$filterTypeMap['*'];
if (is_array($value)) foreach($value as $sub) {
/* Solr requires dates in the form 1995-12-31T23:59:59Z */
if ($type == 'tdate') {
if(!$sub) continue;
$sub = gmdate('Y-m-d\TH:i:s\Z', strtotime($sub));
}
/* Solr requires numbers to be valid if presented, not just empty */
if (($type == 'tint' || $type == 'tfloat' || $type == 'tdouble') && !is_numeric($sub)) continue;
$doc->addField($field['name'], $sub);
}
else {
/* Solr requires dates in the form 1995-12-31T23:59:59Z */
if ($type == 'tdate') {
if(!$value) return;
$value = gmdate('Y-m-d\TH:i:s\Z', strtotime($value));
}
/* Solr requires numbers to be valid if presented, not just empty */
if (($type == 'tint' || $type == 'tfloat' || $type == 'tdouble') && !is_numeric($value)) return;
$doc->setField($field['name'], $value);
}
}
protected function _addAs($object, $base, $options) {
$includeSubs = $options['include_children'];
$doc = new Apache_Solr_Document();
// Always present fields
$doc->setField('_documentid', $this->getDocumentID($object, $base, $includeSubs));
$doc->setField('ID', $object->ID);
$doc->setField('ClassName', $object->ClassName);
foreach (SearchIntrospection::hierarchy(get_class($object), false) as $class) $doc->addField('ClassHierarchy', $class);
// Add the user-specified fields
foreach ($this->getFieldsIterator() as $name => $field) {
if ($field['base'] == $base) $this->_addField($doc, $object, $field);
}
try {
$this->getService()->addDocument($doc);
} catch (Exception $e) {
SS_Log::log($e, SS_Log::WARN);
return false;
}
return $doc;
}
function add($object) {
$class = get_class($object);
$docs = array();
foreach ($this->getClasses() as $searchclass => $options) {
if ($searchclass == $class || ($options['include_children'] && is_subclass_of($class, $searchclass))) {
$base = ClassInfo::baseDataClass($searchclass);
$docs[] = $this->_addAs($object, $base, $options);
}
}
return $docs;
}
function canAdd($class) {
foreach ($this->classes as $searchclass => $options) {
if ($searchclass == $class || ($options['include_children'] && is_subclass_of($class, $searchclass))) return true;
}
return false;
}
function delete($base, $id, $state) {
$documentID = $this->getDocumentIDForState($base, $id, $state);
try {
$this->getService()->deleteById($documentID);
} catch (Exception $e) {
SS_Log::log($e, SS_Log::WARN);
return false;
}
}
/**
* Clear all records which do not match the given classname whitelist.
*
* Can also be used to trim an index when reducing to a narrower set of classes.
*
* Ignores current state / variant.
*
* @param array $classes List of non-obsolete classes in the same format as SolrIndex::getClasses()
* @return bool Flag if successful
*/
public function clearObsoleteClasses($classes) {
if(empty($classes)) {
return false;
}
// Delete all records which do not match the necessary classname rules
$conditions = array();
foreach ($classes as $class => $options) {
if ($options['include_children']) {
$conditions[] = "ClassHierarchy:{$class}";
} else {
$conditions[] = "ClassName:{$class}";
}
}
// Delete records which don't match any of these conditions in this index
$deleteQuery = "-(" . implode(' ', $conditions) . ")";
$this
->getService()
->deleteByQuery($deleteQuery);
return true;
}
function commit() {
try {
$this->getService()->commit(false, false, false);
} catch (Exception $e) {
SS_Log::log($e, SS_Log::WARN);
return false;
}
}
/**
* @param SearchQuery $query
* @param integer $offset
* @param integer $limit
* @param array $params Extra request parameters passed through to Solr
* @return ArrayData Map with the following keys:
* - 'Matches': ArrayList of the matched object instances
*/
public function search(SearchQuery $query, $offset = -1, $limit = -1, $params = array()) {
$service = $this->getService();
$searchClass = count($query->classes) == 1
? $query->classes[0]['class']
: null;
SearchVariant::with($searchClass)
->call('alterQuery', $query, $this);
$q = array(); // Query
$fq = array(); // Filter query
$qf = array(); // Query fields
$hlq = array(); // Highlight query
// Build the search itself
$q = $this->getQueryComponent($query, $hlq);
// If using boosting, set the clean term separately for highlighting.
// See https://issues.apache.org/jira/browse/SOLR-2632
if(array_key_exists('hl', $params) && !array_key_exists('hl.q', $params)) {
$params['hl.q'] = implode(' ', $hlq);
}
// Filter by class if requested
$classq = array();
foreach ($query->classes as $class) {
if (!empty($class['includeSubclasses'])) {
$classq[] = 'ClassHierarchy:'.$class['class'];
}
else $classq[] = 'ClassName:'.$class['class'];
}
if ($classq) $fq[] = '+('.implode(' ', $classq).')';
// Filter by filters
$fq = array_merge($fq, $this->getFiltersComponent($query));
// Prepare query fields unless specified explicitly
if(isset($params['qf'])) {
$qf = $params['qf'];
} else {
$qf = $this->getQueryFields();
}
if(is_array($qf)) {
$qf = implode(' ', $qf);
}
if($qf) {
$params['qf'] = $qf;
}
if(!headers_sent() && !Director::isLive()) {
if ($q) header('X-Query: '.implode(' ', $q));
if ($fq) header('X-Filters: "'.implode('", "', $fq).'"');
if ($qf) header('X-QueryFields: '.$qf);
}
if ($offset == -1) $offset = $query->start;
if ($limit == -1) $limit = $query->limit;
if ($limit == -1) $limit = SearchQuery::$default_page_size;
$params = array_merge($params, array('fq' => implode(' ', $fq)));
$res = $service->search(
$q ? implode(' ', $q) : '*:*',
$offset,
$limit,
$params,
Apache_Solr_Service::METHOD_POST
);
$results = new ArrayList();
if($res->getHttpStatus() >= 200 && $res->getHttpStatus() < 300) {
foreach ($res->response->docs as $doc) {
$result = DataObject::get_by_id($doc->ClassName, $doc->ID);
if($result) {
$results->push($result);
// Add highlighting (optional)
$docId = $doc->_documentid;
if($res->highlighting && $res->highlighting->$docId) {
// TODO Create decorator class for search results rather than adding arbitrary object properties
// TODO Allow specifying highlighted field, and lazy loading
// in case the search API needs another query (similar to SphinxSearchable->buildExcerpt()).
$combinedHighlights = array();
foreach($res->highlighting->$docId as $field => $highlights) {
$combinedHighlights = array_merge($combinedHighlights, $highlights);
}
// Remove entity-encoded U+FFFD replacement character. It signifies non-displayable characters,
// and shows up as an encoding error in browsers.
$result->Excerpt = DBField::create_field(
'HTMLText',
str_replace(
'&#65533;',
'',
implode(' ... ', $combinedHighlights)
)
);
}
}
}
$numFound = $res->response->numFound;
} else {
$numFound = 0;
}
$ret = array();
$ret['Matches'] = new PaginatedList($results);
$ret['Matches']->setLimitItems(false);
// Tell PaginatedList how many results there are
$ret['Matches']->setTotalItems($numFound);
// Results for current page start at $offset
$ret['Matches']->setPageStart($offset);
// Results per page
$ret['Matches']->setPageLength($limit);
// Include spellcheck and suggestion data. Requires spellcheck=true in $params
if(isset($res->spellcheck)) {
// Expose all spellcheck data, for custom handling.
$ret['Spellcheck'] = $res->spellcheck;
// Suggestions. Requires spellcheck.collate=true in $params
if(isset($res->spellcheck->suggestions->collation)) {
// Extract string suggestion
$suggestion = $this->getCollatedSuggestion($res->spellcheck->suggestions->collation);
// The collation, including advanced query params (e.g. +), suitable for making another query programmatically.
$ret['Suggestion'] = $suggestion;
// A human friendly version of the suggestion, suitable for 'Did you mean $SuggestionNice?' display.
$ret['SuggestionNice'] = $this->getNiceSuggestion($suggestion);
// A string suitable for appending to an href as a query string.
// For example <a href="http://example.com/search?q=$SuggestionQueryString">$SuggestionNice</a>
$ret['SuggestionQueryString'] = $this->getSuggestionQueryString($suggestion);
}
}
return new ArrayData($ret);
}
/**
* Get the query (q) component for this search
*
* @param SearchQuery $searchQuery
* @param array &$hlq Highlight query returned by reference
* @return array
*/
protected function getQueryComponent(SearchQuery $searchQuery, &$hlq = array()) {
$q = array();
foreach ($searchQuery->search as $search) {
$text = $search['text'];
preg_match_all('/"[^"]*"|\S+/', $text, $parts);
$fuzzy = $search['fuzzy'] ? '~' : '';
foreach ($parts[0] as $part) {
$fields = (isset($search['fields'])) ? $search['fields'] : array();
if(isset($search['boost'])) {
$fields = array_merge($fields, array_keys($search['boost']));
}
if ($fields) {
$searchq = array();
foreach ($fields as $field) {
$boost = (isset($search['boost'][$field])) ? '^' . $search['boost'][$field] : '';
$searchq[] = "{$field}:".$part.$fuzzy.$boost;
}
$q[] = '+('.implode(' OR ', $searchq).')';
}
else {
$q[] = '+'.$part.$fuzzy;
}
$hlq[] = $part;
}
}
return $q;
}
/**
* Parse all require constraints for inclusion in a filter query
*
* @param SearchQuery $searchQuery
* @return array List of parsed string values for each require
*/
protected function getRequireFiltersComponent(SearchQuery $searchQuery) {
$fq = array();
foreach ($searchQuery->require as $field => $values) {
$requireq = array();
foreach ($values as $value) {
if ($value === SearchQuery::$missing) {
$requireq[] = "(*:* -{$field}:[* TO *])";
}
else if ($value === SearchQuery::$present) {
$requireq[] = "{$field}:[* TO *]";
}
else if ($value instanceof SearchQuery_Range) {
$start = $value->start;
if ($start === null) {
$start = '*';
}
$end = $value->end;
if ($end === null) {
$end = '*';
}
$requireq[] = "$field:[$start TO $end]";
}
else {
$requireq[] = $field.':"'.$value.'"';
}
}
$fq[] = '+('.implode(' ', $requireq).')';
}
return $fq;
}
/**
* Parse all exclude constraints for inclusion in a filter query
*
* @param SearchQuery $searchQuery
* @return array List of parsed string values for each exclusion
*/
protected function getExcludeFiltersComponent(SearchQuery $searchQuery) {
$fq = array();
foreach ($searchQuery->exclude as $field => $values) {
$excludeq = array();
$missing = false;
foreach ($values as $value) {
if ($value === SearchQuery::$missing) {
$missing = true;
}
else if ($value === SearchQuery::$present) {
$excludeq[] = "{$field}:[* TO *]";
}
else if ($value instanceof SearchQuery_Range) {
$start = $value->start;
if ($start === null) {
$start = '*';
}
$end = $value->end;
if ($end === null) {
$end = '*';
}
$excludeq[] = "$field:[$start TO $end]";
}
else {
$excludeq[] = $field.':"'.$value.'"';
}
}
$fq[] = ($missing ? "+{$field}:[* TO *] " : '') . '-('.implode(' ', $excludeq).')';
}
return $fq;
}
/**
* Get all filter conditions for this search
*
* @param SearchQuery $searchQuery
* @return array
*/
public function getFiltersComponent(SearchQuery $searchQuery) {
return array_merge(
$this->getRequireFiltersComponent($searchQuery),
$this->getExcludeFiltersComponent($searchQuery)
);
}
protected $service;
/**
* @return SolrService
*/
public function getService() {
if(!$this->service) $this->service = Solr::service(get_class($this));
return $this->service;
}
public function setService(SolrService $service) {
$this->service = $service;
return $this;
}
/**
* Upload config for this index to the given store
*
* @param SolrConfigStore $store
*/
public function uploadConfig($store) {
// Upload the config files for this index
$store->uploadString(
$this->getIndexName(),
'schema.xml',
(string)$this->generateSchema()
);
// Upload additional files
foreach (glob($this->getExtrasPath().'/*') as $file) {
if (is_file($file)) {
$store->uploadFile($this->getIndexName(), $file);
}
}
}
abstract class SolrIndex extends SearchIndex
{
public static $fulltextTypeMap = array(
'*' => 'text',
'HTMLVarchar' => 'htmltext',
'HTMLText' => 'htmltext'
);
public static $filterTypeMap = array(
'*' => 'string',
'Boolean' => 'boolean',
'Date' => 'tdate',
'SSDatetime' => 'tdate',
'SS_Datetime' => 'tdate',
'ForeignKey' => 'tint',
'Int' => 'tint',
'Float' => 'tfloat',
'Double' => 'tdouble'
);
public static $sortTypeMap = array();
protected $analyzerFields = array();
protected $copyFields = array();
protected $extrasPath = null;
protected $templatesPath = null;
/**
* List of boosted fields
*
* @var array
*/
protected $boostedFields = array();
/**
* Name of default field
*
* @var string
* @config
*/
private static $default_field = '_text';
/**
* List of copy fields all fulltext fields should be copied into.
* This will fallback to default_field if not specified
*
* @var array
*/
private static $copy_fields = array();
/**
* @return String Absolute path to the folder containing
* templates which are used for generating the schema and field definitions.
*/
public function getTemplatesPath()
{
$globalOptions = Solr::solr_options();
return $this->templatesPath ? $this->templatesPath : $globalOptions['templatespath'];
}
/**
* @return String Absolute path to the configuration default files,
* e.g. solrconfig.xml.
*/
public function getExtrasPath()
{
$globalOptions = Solr::solr_options();
return $this->extrasPath ? $this->extrasPath : $globalOptions['extraspath'];
}
public function generateSchema()
{
return $this->renderWith($this->getTemplatesPath() . '/schema.ss');
}
public function getIndexName()
{
return get_class($this);
}
public function getTypes()
{
return $this->renderWith($this->getTemplatesPath() . '/types.ss');
}
/**
* Index-time analyzer which is applied to a specific field.
* Can be used to remove HTML tags, apply stemming, etc.
*
* @see http://wiki.apache.org/solr/AnalyzersTokenizersTokenFilters#solr.WhitespaceTokenizerFactory
*
* @param String $field
* @param String $type
* @param Array $params Parameters for the analyzer, usually at least a "class"
*/
public function addAnalyzer($field, $type, $params)
{
$fullFields = $this->fieldData($field);
if ($fullFields) {
foreach ($fullFields as $fullField => $spec) {
if (!isset($this->analyzerFields[$fullField])) {
$this->analyzerFields[$fullField] = array();
}
$this->analyzerFields[$fullField][$type] = $params;
}
}
}
/**
* Get the default text field, normally '_text'
*
* @return string
*/
public function getDefaultField()
{
return $this->config()->default_field;
}
/**
* Get list of fields each text field should be copied into.
* This will fallback to the default field if omitted.
*
* @return array
*/
protected function getCopyDestinations()
{
$copyFields = $this->config()->copy_fields;
if ($copyFields) {
return $copyFields;
}
// Fallback to default field
$df = $this->getDefaultField();
return array($df);
}
public function getFieldDefinitions()
{
$xml = array();
$stored = $this->getStoredDefault();
$xml[] = "";
// Add the hardcoded field definitions
$xml[] = "<field name='_documentid' type='string' indexed='true' stored='true' required='true' />";
$xml[] = "<field name='ID' type='tint' indexed='true' stored='true' required='true' />";
$xml[] = "<field name='ClassName' type='string' indexed='true' stored='true' required='true' />";
$xml[] = "<field name='ClassHierarchy' type='string' indexed='true' stored='true' required='true' multiValued='true' />";
// Add the fulltext collation field
$df = $this->getDefaultField();
$xml[] = "<field name='{$df}' type='htmltext' indexed='true' stored='{$stored}' multiValued='true' />" ;
// Add the user-specified fields
foreach ($this->fulltextFields as $name => $field) {
$xml[] = $this->getFieldDefinition($name, $field, self::$fulltextTypeMap);
}
foreach ($this->filterFields as $name => $field) {
if ($field['fullfield'] == 'ID' || $field['fullfield'] == 'ClassName') {
continue;
}
$xml[] = $this->getFieldDefinition($name, $field);
}
foreach ($this->sortFields as $name => $field) {
if ($field['fullfield'] == 'ID' || $field['fullfield'] == 'ClassName') {
continue;
}
$xml[] = $this->getFieldDefinition($name, $field);
}
return implode("\n\t\t", $xml);
}
/**
* Extract first suggestion text from collated values
*
* @param mixed $collation
* @return string
*/
protected function getCollatedSuggestion($collation = '')
{
if (is_string($collation)) {
return $collation;
}
if (is_object($collation)) {
if (isset($collation->misspellingsAndCorrections)) {
foreach ($collation->misspellingsAndCorrections as $key => $value) {
return $value;
}
}
}
return '';
}
/**
* Extract a human friendly spelling suggestion from a Solr spellcheck collation string.
* @param String $collation
* @return String
*/
protected function getNiceSuggestion($collation = '')
{
$collationParts = explode(' ', $collation);
// Remove advanced query params from the beginning of each collation part.
foreach ($collationParts as $key => &$part) {
$part = ltrim($part, '+');
}
return implode(' ', $collationParts);
}
/**
* Extract a query string from a Solr spellcheck collation string.
* Useful for constructing 'Did you mean?' links, for example:
* <a href="http://example.com/search?q=$SuggestionQueryString">$SuggestionNice</a>
* @param String $collation
* @return String
*/
protected function getSuggestionQueryString($collation = '')
{
return str_replace(' ', '+', $this->getNiceSuggestion($collation));
}
/**
* Add a field that should be stored
*
* @param string $field The field to add
* @param string $forceType The type to force this field as (required in some cases, when not
* detectable from metadata)
* @param array $extraOptions Dependent on search implementation
*/
public function addStoredField($field, $forceType = null, $extraOptions = array())
{
$options = array_merge($extraOptions, array('stored' => 'true'));
$this->addFulltextField($field, $forceType, $options);
}
/**
* Add a fulltext field with a boosted value
*
* @param string $field The field to add
* @param string $forceType The type to force this field as (required in some cases, when not
* detectable from metadata)
* @param array $extraOptions Dependent on search implementation
* @param float $boost Numeric boosting value (defaults to 2)
*/
public function addBoostedField($field, $forceType = null, $extraOptions = array(), $boost = 2)
{
$options = array_merge($extraOptions, array('boost' => $boost));
$this->addFulltextField($field, $forceType, $options);
}
public function fieldData($field, $forceType = null, $extraOptions = array())
{
// Ensure that 'boost' is recorded here without being captured by solr
$boost = null;
if (array_key_exists('boost', $extraOptions)) {
$boost = $extraOptions['boost'];
unset($extraOptions['boost']);
}
$data = parent::fieldData($field, $forceType, $extraOptions);
// Boost all fields with this name
if (isset($boost)) {
foreach ($data as $fieldName => $fieldInfo) {
$this->boostedFields[$fieldName] = $boost;
}
}
return $data;
}
/**
* Set the default boosting level for a specific field.
* Will control the default value for qf param (Query Fields), but will not
* override a query-specific value.
*
* Fields must be added before having a field boosting specified
*
* @param string $field Full field key (Model_Field)
* @param float|null $level Numeric boosting value. Set to null to clear boost
*/
public function setFieldBoosting($field, $level)
{
if (!isset($this->fulltextFields[$field])) {
throw new InvalidArgumentException("No fulltext field $field exists on ".$this->getIndexName());
}
if ($level === null) {
unset($this->boostedFields[$field]);
} else {
$this->boostedFields[$field] = $level;
}
}
/**
* Get all boosted fields
*
* @return array
*/
public function getBoostedFields()
{
return $this->boostedFields;
}
/**
* Determine the best default value for the 'qf' parameter
*
* @return array|null List of query fields, or null if not specified
*/
public function getQueryFields()
{
// Not necessary to specify this unless boosting
if (empty($this->boostedFields)) {
return null;
}
$queryFields = array();
foreach ($this->boostedFields as $fieldName => $boost) {
$queryFields[] = $fieldName . '^' . $boost;
}
// If any fields are queried, we must always include the default field, otherwise it will be excluded
$df = $this->getDefaultField();
if ($queryFields && !isset($this->boostedFields[$df])) {
$queryFields[] = $df;
}
return $queryFields;
}
/**
* Gets the default 'stored' value for fields in this index
*
* @return string A default value for the 'stored' field option, either 'true' or 'false'
*/
protected function getStoredDefault()
{
return Director::isDev() ? 'true' : 'false';
}
/**
* @param String $name
* @param Array $spec
* @param Array $typeMap
* @return String XML
*/
protected function getFieldDefinition($name, $spec, $typeMap = null)
{
if (!$typeMap) {
$typeMap = self::$filterTypeMap;
}
$multiValued = (isset($spec['multi_valued']) && $spec['multi_valued']) ? "true" : '';
$type = isset($typeMap[$spec['type']]) ? $typeMap[$spec['type']] : $typeMap['*'];
$analyzerXml = '';
if (isset($this->analyzerFields[$name])) {
foreach ($this->analyzerFields[$name] as $analyzerType => $analyzerParams) {
$analyzerXml .= $this->toXmlTag($analyzerType, $analyzerParams);
}
}
$fieldParams = array_merge(
array(
'name' => $name,
'type' => $type,
'indexed' => 'true',
'stored' => $this->getStoredDefault(),
'multiValued' => $multiValued
),
isset($spec['extra_options']) ? $spec['extra_options'] : array()
);
return $this->toXmlTag(
"field",
$fieldParams,
$analyzerXml ? "<analyzer>$analyzerXml</analyzer>" : null
);
}
/**
* Convert definition to XML tag
*
* @param String $tag
* @param String $attrs Map of attributes
* @param String $content Inner content
* @return String XML tag
*/
protected function toXmlTag($tag, $attrs, $content = null)
{
$xml = "<$tag ";
if ($attrs) {
$attrStrs = array();
foreach ($attrs as $attrName => $attrVal) {
$attrStrs[] = "$attrName='$attrVal'";
}
$xml .= $attrStrs ? implode(' ', $attrStrs) : '';
}
$xml .= $content ? ">$content</$tag>" : '/>';
return $xml;
}
/**
* @param String $source Composite field name (<class>_<fieldname>)
* @param String $dest
*/
public function addCopyField($source, $dest, $extraOptions = array())
{
if (!isset($this->copyFields[$source])) {
$this->copyFields[$source] = array();
}
$this->copyFields[$source][] = array_merge(
array('source' => $source, 'dest' => $dest),
$extraOptions
);
}
/**
* Generate XML for copy field definitions
*
* @return string
*/
public function getCopyFieldDefinitions()
{
$xml = array();
// Default copy fields
foreach ($this->getCopyDestinations() as $copyTo) {
foreach ($this->fulltextFields as $name => $field) {
$xml[] = "<copyField source='{$name}' dest='{$copyTo}' />";
}
}
// Explicit copy fields
foreach ($this->copyFields as $source => $fields) {
foreach ($fields as $fieldAttrs) {
$xml[] = $this->toXmlTag('copyField', $fieldAttrs);
}
}
return implode("\n\t", $xml);
}
protected function _addField($doc, $object, $field)
{
$class = get_class($object);
if ($class != $field['origin'] && !is_subclass_of($class, $field['origin'])) {
return;
}
$value = $this->_getFieldValue($object, $field);
$type = isset(self::$filterTypeMap[$field['type']]) ? self::$filterTypeMap[$field['type']] : self::$filterTypeMap['*'];
if (is_array($value)) {
foreach ($value as $sub) {
/* Solr requires dates in the form 1995-12-31T23:59:59Z */
if ($type == 'tdate') {
if (!$sub) {
continue;
}
$sub = gmdate('Y-m-d\TH:i:s\Z', strtotime($sub));
}
/* Solr requires numbers to be valid if presented, not just empty */
if (($type == 'tint' || $type == 'tfloat' || $type == 'tdouble') && !is_numeric($sub)) {
continue;
}
$doc->addField($field['name'], $sub);
}
} else {
/* Solr requires dates in the form 1995-12-31T23:59:59Z */
if ($type == 'tdate') {
if (!$value) {
return;
}
$value = gmdate('Y-m-d\TH:i:s\Z', strtotime($value));
}
/* Solr requires numbers to be valid if presented, not just empty */
if (($type == 'tint' || $type == 'tfloat' || $type == 'tdouble') && !is_numeric($value)) {
return;
}
$doc->setField($field['name'], $value);
}
}
protected function _addAs($object, $base, $options)
{
$includeSubs = $options['include_children'];
$doc = new Apache_Solr_Document();
// Always present fields
$doc->setField('_documentid', $this->getDocumentID($object, $base, $includeSubs));
$doc->setField('ID', $object->ID);
$doc->setField('ClassName', $object->ClassName);
foreach (SearchIntrospection::hierarchy(get_class($object), false) as $class) {
$doc->addField('ClassHierarchy', $class);
}
// Add the user-specified fields
foreach ($this->getFieldsIterator() as $name => $field) {
if ($field['base'] == $base) {
$this->_addField($doc, $object, $field);
}
}
try {
$this->getService()->addDocument($doc);
} catch (Exception $e) {
SS_Log::log($e, SS_Log::WARN);
return false;
}
return $doc;
}
public function add($object)
{
$class = get_class($object);
$docs = array();
foreach ($this->getClasses() as $searchclass => $options) {
if ($searchclass == $class || ($options['include_children'] && is_subclass_of($class, $searchclass))) {
$base = ClassInfo::baseDataClass($searchclass);
$docs[] = $this->_addAs($object, $base, $options);
}
}
return $docs;
}
public function canAdd($class)
{
foreach ($this->classes as $searchclass => $options) {
if ($searchclass == $class || ($options['include_children'] && is_subclass_of($class, $searchclass))) {
return true;
}
}
return false;
}
public function delete($base, $id, $state)
{
$documentID = $this->getDocumentIDForState($base, $id, $state);
try {
$this->getService()->deleteById($documentID);
} catch (Exception $e) {
SS_Log::log($e, SS_Log::WARN);
return false;
}
}
/**
* Clear all records which do not match the given classname whitelist.
*
* Can also be used to trim an index when reducing to a narrower set of classes.
*
* Ignores current state / variant.
*
* @param array $classes List of non-obsolete classes in the same format as SolrIndex::getClasses()
* @return bool Flag if successful
*/
public function clearObsoleteClasses($classes)
{
if (empty($classes)) {
return false;
}
// Delete all records which do not match the necessary classname rules
$conditions = array();
foreach ($classes as $class => $options) {
if ($options['include_children']) {
$conditions[] = "ClassHierarchy:{$class}";
} else {
$conditions[] = "ClassName:{$class}";
}
}
// Delete records which don't match any of these conditions in this index
$deleteQuery = "-(" . implode(' ', $conditions) . ")";
$this
->getService()
->deleteByQuery($deleteQuery);
return true;
}
public function commit()
{
try {
$this->getService()->commit(false, false, false);
} catch (Exception $e) {
SS_Log::log($e, SS_Log::WARN);
return false;
}
}
/**
* @param SearchQuery $query
* @param integer $offset
* @param integer $limit
* @param array $params Extra request parameters passed through to Solr
* @return ArrayData Map with the following keys:
* - 'Matches': ArrayList of the matched object instances
*/
public function search(SearchQuery $query, $offset = -1, $limit = -1, $params = array())
{
$service = $this->getService();
$searchClass = count($query->classes) == 1
? $query->classes[0]['class']
: null;
SearchVariant::with($searchClass)
->call('alterQuery', $query, $this);
$q = array(); // Query
$fq = array(); // Filter query
$qf = array(); // Query fields
$hlq = array(); // Highlight query
// Build the search itself
$q = $this->getQueryComponent($query, $hlq);
// If using boosting, set the clean term separately for highlighting.
// See https://issues.apache.org/jira/browse/SOLR-2632
if (array_key_exists('hl', $params) && !array_key_exists('hl.q', $params)) {
$params['hl.q'] = implode(' ', $hlq);
}
// Filter by class if requested
$classq = array();
foreach ($query->classes as $class) {
if (!empty($class['includeSubclasses'])) {
$classq[] = 'ClassHierarchy:'.$class['class'];
} else {
$classq[] = 'ClassName:'.$class['class'];
}
}
if ($classq) {
$fq[] = '+('.implode(' ', $classq).')';
}
// Filter by filters
$fq = array_merge($fq, $this->getFiltersComponent($query));
// Prepare query fields unless specified explicitly
if (isset($params['qf'])) {
$qf = $params['qf'];
} else {
$qf = $this->getQueryFields();
}
if (is_array($qf)) {
$qf = implode(' ', $qf);
}
if ($qf) {
$params['qf'] = $qf;
}
if (!headers_sent() && !Director::isLive()) {
if ($q) {
header('X-Query: '.implode(' ', $q));
}
if ($fq) {
header('X-Filters: "'.implode('", "', $fq).'"');
}
if ($qf) {
header('X-QueryFields: '.$qf);
}
}
if ($offset == -1) {
$offset = $query->start;
}
if ($limit == -1) {
$limit = $query->limit;
}
if ($limit == -1) {
$limit = SearchQuery::$default_page_size;
}
$params = array_merge($params, array('fq' => implode(' ', $fq)));
$res = $service->search(
$q ? implode(' ', $q) : '*:*',
$offset,
$limit,
$params,
Apache_Solr_Service::METHOD_POST
);
$results = new ArrayList();
if ($res->getHttpStatus() >= 200 && $res->getHttpStatus() < 300) {
foreach ($res->response->docs as $doc) {
$result = DataObject::get_by_id($doc->ClassName, $doc->ID);
if ($result) {
$results->push($result);
// Add highlighting (optional)
$docId = $doc->_documentid;
if ($res->highlighting && $res->highlighting->$docId) {
// TODO Create decorator class for search results rather than adding arbitrary object properties
// TODO Allow specifying highlighted field, and lazy loading
// in case the search API needs another query (similar to SphinxSearchable->buildExcerpt()).
$combinedHighlights = array();
foreach ($res->highlighting->$docId as $field => $highlights) {
$combinedHighlights = array_merge($combinedHighlights, $highlights);
}
// Remove entity-encoded U+FFFD replacement character. It signifies non-displayable characters,
// and shows up as an encoding error in browsers.
$result->Excerpt = DBField::create_field(
'HTMLText',
str_replace(
'&#65533;',
'',
implode(' ... ', $combinedHighlights)
)
);
}
}
}
$numFound = $res->response->numFound;
} else {
$numFound = 0;
}
$ret = array();
$ret['Matches'] = new PaginatedList($results);
$ret['Matches']->setLimitItems(false);
// Tell PaginatedList how many results there are
$ret['Matches']->setTotalItems($numFound);
// Results for current page start at $offset
$ret['Matches']->setPageStart($offset);
// Results per page
$ret['Matches']->setPageLength($limit);
// Include spellcheck and suggestion data. Requires spellcheck=true in $params
if (isset($res->spellcheck)) {
// Expose all spellcheck data, for custom handling.
$ret['Spellcheck'] = $res->spellcheck;
// Suggestions. Requires spellcheck.collate=true in $params
if (isset($res->spellcheck->suggestions->collation)) {
// Extract string suggestion
$suggestion = $this->getCollatedSuggestion($res->spellcheck->suggestions->collation);
// The collation, including advanced query params (e.g. +), suitable for making another query programmatically.
$ret['Suggestion'] = $suggestion;
// A human friendly version of the suggestion, suitable for 'Did you mean $SuggestionNice?' display.
$ret['SuggestionNice'] = $this->getNiceSuggestion($suggestion);
// A string suitable for appending to an href as a query string.
// For example <a href="http://example.com/search?q=$SuggestionQueryString">$SuggestionNice</a>
$ret['SuggestionQueryString'] = $this->getSuggestionQueryString($suggestion);
}
}
return new ArrayData($ret);
}
/**
* Get the query (q) component for this search
*
* @param SearchQuery $searchQuery
* @param array &$hlq Highlight query returned by reference
* @return array
*/
protected function getQueryComponent(SearchQuery $searchQuery, &$hlq = array())
{
$q = array();
foreach ($searchQuery->search as $search) {
$text = $search['text'];
preg_match_all('/"[^"]*"|\S+/', $text, $parts);
$fuzzy = $search['fuzzy'] ? '~' : '';
foreach ($parts[0] as $part) {
$fields = (isset($search['fields'])) ? $search['fields'] : array();
if (isset($search['boost'])) {
$fields = array_merge($fields, array_keys($search['boost']));
}
if ($fields) {
$searchq = array();
foreach ($fields as $field) {
$boost = (isset($search['boost'][$field])) ? '^' . $search['boost'][$field] : '';
$searchq[] = "{$field}:".$part.$fuzzy.$boost;
}
$q[] = '+('.implode(' OR ', $searchq).')';
} else {
$q[] = '+'.$part.$fuzzy;
}
$hlq[] = $part;
}
}
return $q;
}
/**
* Parse all require constraints for inclusion in a filter query
*
* @param SearchQuery $searchQuery
* @return array List of parsed string values for each require
*/
protected function getRequireFiltersComponent(SearchQuery $searchQuery)
{
$fq = array();
foreach ($searchQuery->require as $field => $values) {
$requireq = array();
foreach ($values as $value) {
if ($value === SearchQuery::$missing) {
$requireq[] = "(*:* -{$field}:[* TO *])";
} elseif ($value === SearchQuery::$present) {
$requireq[] = "{$field}:[* TO *]";
} elseif ($value instanceof SearchQuery_Range) {
$start = $value->start;
if ($start === null) {
$start = '*';
}
$end = $value->end;
if ($end === null) {
$end = '*';
}
$requireq[] = "$field:[$start TO $end]";
} else {
$requireq[] = $field.':"'.$value.'"';
}
}
$fq[] = '+('.implode(' ', $requireq).')';
}
return $fq;
}
/**
* Parse all exclude constraints for inclusion in a filter query
*
* @param SearchQuery $searchQuery
* @return array List of parsed string values for each exclusion
*/
protected function getExcludeFiltersComponent(SearchQuery $searchQuery)
{
$fq = array();
foreach ($searchQuery->exclude as $field => $values) {
$excludeq = array();
$missing = false;
foreach ($values as $value) {
if ($value === SearchQuery::$missing) {
$missing = true;
} elseif ($value === SearchQuery::$present) {
$excludeq[] = "{$field}:[* TO *]";
} elseif ($value instanceof SearchQuery_Range) {
$start = $value->start;
if ($start === null) {
$start = '*';
}
$end = $value->end;
if ($end === null) {
$end = '*';
}
$excludeq[] = "$field:[$start TO $end]";
} else {
$excludeq[] = $field.':"'.$value.'"';
}
}
$fq[] = ($missing ? "+{$field}:[* TO *] " : '') . '-('.implode(' ', $excludeq).')';
}
return $fq;
}
/**
* Get all filter conditions for this search
*
* @param SearchQuery $searchQuery
* @return array
*/
public function getFiltersComponent(SearchQuery $searchQuery)
{
return array_merge(
$this->getRequireFiltersComponent($searchQuery),
$this->getExcludeFiltersComponent($searchQuery)
);
}
protected $service;
/**
* @return SolrService
*/
public function getService()
{
if (!$this->service) {
$this->service = Solr::service(get_class($this));
}
return $this->service;
}
public function setService(SolrService $service)
{
$this->service = $service;
return $this;
}
/**
* Upload config for this index to the given store
*
* @param SolrConfigStore $store
*/
public function uploadConfig($store)
{
// Upload the config files for this index
$store->uploadString(
$this->getIndexName(),
'schema.xml',
(string)$this->generateSchema()
);
// Upload additional files
foreach (glob($this->getExtrasPath().'/*') as $file) {
if (is_file($file)) {
$store->uploadFile($this->getIndexName(), $file);
}
}
}
}

121
code/solr/SolrService.php
View File

@ -5,7 +5,8 @@ Solr::include_client_api();
/**
* The API for accessing a specific core of a Solr server. Exactly the same as Apache_Solr_Service for now.
*/
class SolrService_Core extends Apache_Solr_Service {
class SolrService_Core extends Apache_Solr_Service
{
}
/**
@ -13,65 +14,77 @@ class SolrService_Core extends Apache_Solr_Service {
* plus extra methods for interrogating, creating, reloading and getting SolrService_Core instances
* for Solr cores.
*/
class SolrService extends SolrService_Core {
private static $core_class = 'SolrService_Core';
class SolrService extends SolrService_Core
{
private static $core_class = 'SolrService_Core';
/**
* Handle encoding the GET parameters and making the HTTP call to execute a core command
*/
protected function coreCommand($command, $core, $params=array()) {
$command = strtoupper($command);
/**
* Handle encoding the GET parameters and making the HTTP call to execute a core command
*/
protected function coreCommand($command, $core, $params=array())
{
$command = strtoupper($command);
$params = array_merge($params, array('action' => $command, 'wt' => 'json'));
$params[$command == 'CREATE' ? 'name' : 'core'] = $core;
$params = array_merge($params, array('action' => $command, 'wt' => 'json'));
$params[$command == 'CREATE' ? 'name' : 'core'] = $core;
return $this->_sendRawGet($this->_constructUrl('admin/cores', $params));
}
return $this->_sendRawGet($this->_constructUrl('admin/cores', $params));
}
/**
* Is the passed core active?
* @param $core string - The name of the core
* @return boolean - True if that core exists & is active
*/
public function coreIsActive($core) {
$result = $this->coreCommand('STATUS', $core);
return isset($result->status->$core->uptime);
}
/**
* Is the passed core active?
* @param $core string - The name of the core
* @return boolean - True if that core exists & is active
*/
public function coreIsActive($core)
{
$result = $this->coreCommand('STATUS', $core);
return isset($result->status->$core->uptime);
}
/**
* Create a new core
* @param $core string - The name of the core
* @param $instancedir string - The base path of the core on the server
* @param $config string - The filename of solrconfig.xml on the server. Default is $instancedir/solrconfig.xml
* @param $schema string - The filename of schema.xml on the server. Default is $instancedir/schema.xml
* @param $datadir string - The path to store data for this core on the server. Default depends on solrconfig.xml
* @return Apache_Solr_Response
*/
public function coreCreate($core, $instancedir, $config=null, $schema=null, $datadir=null) {
$args = array('instanceDir' => $instancedir);
if ($config) $args['config'] = $config;
if ($schema) $args['schema'] = $schema;
if ($datadir) $args['dataDir'] = $datadir;
/**
* Create a new core
* @param $core string - The name of the core
* @param $instancedir string - The base path of the core on the server
* @param $config string - The filename of solrconfig.xml on the server. Default is $instancedir/solrconfig.xml
* @param $schema string - The filename of schema.xml on the server. Default is $instancedir/schema.xml
* @param $datadir string - The path to store data for this core on the server. Default depends on solrconfig.xml
* @return Apache_Solr_Response
*/
public function coreCreate($core, $instancedir, $config=null, $schema=null, $datadir=null)
{
$args = array('instanceDir' => $instancedir);
if ($config) {
$args['config'] = $config;
}
if ($schema) {
$args['schema'] = $schema;
}
if ($datadir) {
$args['dataDir'] = $datadir;
}
return $this->coreCommand('CREATE', $core, $args);
}
return $this->coreCommand('CREATE', $core, $args);
}
/**
* Reload a core
* @param $core string - The name of the core
* @return Apache_Solr_Response
*/
public function coreReload($core) {
return $this->coreCommand('RELOAD', $core);
}
/**
* Reload a core
* @param $core string - The name of the core
* @return Apache_Solr_Response
*/
public function coreReload($core)
{
return $this->coreCommand('RELOAD', $core);
}
/**
* Create a new Solr3Service_Core instance for the passed core
* @param $core string - The name of the core
* @return Solr3Service_Core
*/
public function serviceForCore($core) {
$klass = Config::inst()->get(get_called_class(), 'core_class');
return new $klass($this->_host, $this->_port, $this->_path.$core, $this->_httpTransport);
}
/**
* Create a new Solr3Service_Core instance for the passed core
* @param $core string - The name of the core
* @return Solr3Service_Core
*/
public function serviceForCore($core)
{
$klass = Config::inst()->get(get_called_class(), 'core_class');
return new $klass($this->_host, $this->_port, $this->_path.$core, $this->_httpTransport);
}
}

404
code/solr/reindex/handlers/SolrReindexBase.php
View File

@ -5,226 +5,230 @@ use Psr\Log\LoggerInterface;
/**
* Base class for re-indexing of solr content
*/
abstract class SolrReindexBase implements SolrReindexHandler {
abstract class SolrReindexBase implements SolrReindexHandler
{
public function runReindex(LoggerInterface $logger, $batchSize, $taskName, $classes = null)
{
foreach (Solr::get_indexes() as $indexInstance) {
$this->processIndex($logger, $indexInstance, $batchSize, $taskName, $classes);
}
}
/**
* Process index for a single SolrIndex instance
*
* @param LoggerInterface $logger
* @param SolrIndex $indexInstance
* @param int $batchSize
* @param string $taskName
* @param string $classes
*/
protected function processIndex(
LoggerInterface $logger, SolrIndex $indexInstance, $batchSize, $taskName, $classes = null
) {
// Filter classes for this index
$indexClasses = $this->getClassesForIndex($indexInstance, $classes);
public function runReindex(LoggerInterface $logger, $batchSize, $taskName, $classes = null) {
foreach (Solr::get_indexes() as $indexInstance) {
$this->processIndex($logger, $indexInstance, $batchSize, $taskName, $classes);
}
}
/**
* Process index for a single SolrIndex instance
*
* @param LoggerInterface $logger
* @param SolrIndex $indexInstance
* @param int $batchSize
* @param string $taskName
* @param string $classes
*/
protected function processIndex(
LoggerInterface $logger, SolrIndex $indexInstance, $batchSize, $taskName, $classes = null
) {
// Filter classes for this index
$indexClasses = $this->getClassesForIndex($indexInstance, $classes);
// Clear all records in this index which do not contain the given classes
$logger->info("Clearing obsolete classes from ".$indexInstance->getIndexName());
$indexInstance->clearObsoleteClasses($indexClasses);
// Clear all records in this index which do not contain the given classes
$logger->info("Clearing obsolete classes from ".$indexInstance->getIndexName());
$indexInstance->clearObsoleteClasses($indexClasses);
// Build queue for each class
foreach ($indexClasses as $class => $options) {
$includeSubclasses = $options['include_children'];
// Build queue for each class
foreach ($indexClasses as $class => $options) {
$includeSubclasses = $options['include_children'];
foreach (SearchVariant::reindex_states($class, $includeSubclasses) as $state) {
$this->processVariant($logger, $indexInstance, $state, $class, $includeSubclasses, $batchSize, $taskName);
}
}
}
foreach (SearchVariant::reindex_states($class, $includeSubclasses) as $state) {
$this->processVariant($logger, $indexInstance, $state, $class, $includeSubclasses, $batchSize, $taskName);
}
}
}
/**
* Get valid classes and options for an index with an optional filter
*
* @param SolrIndex $index
* @param string|array $filterClasses Optional class or classes to limit to
* @return array List of classes, where the key is the classname and value is list of options
*/
protected function getClassesForIndex(SolrIndex $index, $filterClasses = null)
{
// Get base classes
$classes = $index->getClasses();
if (!$filterClasses) {
return $classes;
}
/**
* Get valid classes and options for an index with an optional filter
*
* @param SolrIndex $index
* @param string|array $filterClasses Optional class or classes to limit to
* @return array List of classes, where the key is the classname and value is list of options
*/
protected function getClassesForIndex(SolrIndex $index, $filterClasses = null) {
// Get base classes
$classes = $index->getClasses();
if(!$filterClasses) {
return $classes;
}
// Apply filter
if (!is_array($filterClasses)) {
$filterClasses = explode(',', $filterClasses);
}
return array_intersect_key($classes, array_combine($filterClasses, $filterClasses));
}
// Apply filter
if(!is_array($filterClasses)) {
$filterClasses = explode(',', $filterClasses);
}
return array_intersect_key($classes, array_combine($filterClasses, $filterClasses));
}
/**
* Process re-index for a given variant state and class
*
* @param LoggerInterface $logger
* @param SolrIndex $indexInstance
* @param array $state Variant state
* @param string $class
* @param bool $includeSubclasses
* @param int $batchSize
* @param string $taskName
*/
protected function processVariant(
LoggerInterface $logger, SolrIndex $indexInstance, $state,
$class, $includeSubclasses, $batchSize, $taskName
) {
// Set state
SearchVariant::activate_state($state);
/**
* Process re-index for a given variant state and class
*
* @param LoggerInterface $logger
* @param SolrIndex $indexInstance
* @param array $state Variant state
* @param string $class
* @param bool $includeSubclasses
* @param int $batchSize
* @param string $taskName
*/
protected function processVariant(
LoggerInterface $logger, SolrIndex $indexInstance, $state,
$class, $includeSubclasses, $batchSize, $taskName
) {
// Set state
SearchVariant::activate_state($state);
// Count records
$query = $class::get();
if (!$includeSubclasses) {
$query = $query->filter('ClassName', $class);
}
$total = $query->count();
// Count records
$query = $class::get();
if(!$includeSubclasses) {
$query = $query->filter('ClassName', $class);
}
$total = $query->count();
// Skip this variant if nothing to process, or if there are no records
if ($total == 0 || $indexInstance->variantStateExcluded($state)) {
// Remove all records in the current state, since there are no groups to process
$logger->info("Clearing all records of type {$class} in the current state: " . json_encode($state));
$this->clearRecords($indexInstance, $class);
return;
}
// Skip this variant if nothing to process, or if there are no records
if ($total == 0 || $indexInstance->variantStateExcluded($state)) {
// Remove all records in the current state, since there are no groups to process
$logger->info("Clearing all records of type {$class} in the current state: " . json_encode($state));
$this->clearRecords($indexInstance, $class);
return;
}
// For each group, run processing
$groups = (int)(($total + $batchSize - 1) / $batchSize);
for ($group = 0; $group < $groups; $group++) {
$this->processGroup($logger, $indexInstance, $state, $class, $groups, $group, $taskName);
}
}
// For each group, run processing
$groups = (int)(($total + $batchSize - 1) / $batchSize);
for ($group = 0; $group < $groups; $group++) {
$this->processGroup($logger, $indexInstance, $state, $class, $groups, $group, $taskName);
}
}
/**
* Initiate the processing of a single group
*
* @param LoggerInterface $logger
* @param SolrIndex $indexInstance Index instance
* @param array $state Variant state
* @param string $class Class to index
* @param int $groups Total groups
* @param int $group Index of group to process
* @param string $taskName Name of task script to run
*/
abstract protected function processGroup(
LoggerInterface $logger, SolrIndex $indexInstance, $state, $class, $groups, $group, $taskName
);
/**
* Initiate the processing of a single group
*
* @param LoggerInterface $logger
* @param SolrIndex $indexInstance Index instance
* @param array $state Variant state
* @param string $class Class to index
* @param int $groups Total groups
* @param int $group Index of group to process
* @param string $taskName Name of task script to run
*/
abstract protected function processGroup(
LoggerInterface $logger, SolrIndex $indexInstance, $state, $class, $groups, $group, $taskName
);
/**
* Explicitly invoke the process that performs the group
* processing. Can be run either by a background task or a queuedjob.
*
* Does not commit changes to the index, so this must be controlled externally.
*
* @param LoggerInterface $logger
* @param SolrIndex $indexInstance
* @param array $state
* @param string $class
* @param int $groups
* @param int $group
*/
public function runGroup(
LoggerInterface $logger, SolrIndex $indexInstance, $state, $class, $groups, $group
) {
// Set time limit and state
increase_time_limit_to();
SearchVariant::activate_state($state);
$logger->info("Adding $class");
/**
* Explicitly invoke the process that performs the group
* processing. Can be run either by a background task or a queuedjob.
*
* Does not commit changes to the index, so this must be controlled externally.
*
* @param LoggerInterface $logger
* @param SolrIndex $indexInstance
* @param array $state
* @param string $class
* @param int $groups
* @param int $group
*/
public function runGroup(
LoggerInterface $logger, SolrIndex $indexInstance, $state, $class, $groups, $group
) {
// Set time limit and state
increase_time_limit_to();
SearchVariant::activate_state($state);
$logger->info("Adding $class");
// Prior to adding these records to solr, delete existing solr records
$this->clearRecords($indexInstance, $class, $groups, $group);
// Prior to adding these records to solr, delete existing solr records
$this->clearRecords($indexInstance, $class, $groups, $group);
// Process selected records in this class
$items = $this->getRecordsInGroup($indexInstance, $class, $groups, $group);
$processed = array();
foreach ($items as $item) {
$processed[] = $item->ID;
// Process selected records in this class
$items = $this->getRecordsInGroup($indexInstance, $class, $groups, $group);
$processed = array();
foreach ($items as $item) {
$processed[] = $item->ID;
// By this point, obsolete classes/states have been removed in processVariant
// and obsolete records have been removed in clearRecords
$indexInstance->add($item);
$item->destroy();
}
$logger->info("Updated ".implode(',', $processed));
// By this point, obsolete classes/states have been removed in processVariant
// and obsolete records have been removed in clearRecords
$indexInstance->add($item);
$item->destroy();
}
$logger->info("Updated ".implode(',', $processed));
// This will slow down things a tiny bit, but it is done so that we don't timeout to the database during a reindex
DB::query('SELECT 1');
$logger->info("Done");
}
// This will slow down things a tiny bit, but it is done so that we don't timeout to the database during a reindex
DB::query('SELECT 1');
$logger->info("Done");
}
/**
* Gets the datalist of records in the given group in the current state
*
* Assumes that the desired variant state is in effect.
*
* @param SolrIndex $indexInstance
* @param string $class
* @param int $groups
* @param int $group
* @return DataList
*/
protected function getRecordsInGroup(SolrIndex $indexInstance, $class, $groups, $group)
{
// Generate filtered list of local records
$baseClass = ClassInfo::baseDataClass($class);
$items = DataList::create($class)
->where(sprintf(
'"%s"."ID" %% \'%d\' = \'%d\'',
$baseClass,
intval($groups),
intval($group)
))
->sort("ID");
/**
* Gets the datalist of records in the given group in the current state
*
* Assumes that the desired variant state is in effect.
*
* @param SolrIndex $indexInstance
* @param string $class
* @param int $groups
* @param int $group
* @return DataList
*/
protected function getRecordsInGroup(SolrIndex $indexInstance, $class, $groups, $group) {
// Generate filtered list of local records
$baseClass = ClassInfo::baseDataClass($class);
$items = DataList::create($class)
->where(sprintf(
'"%s"."ID" %% \'%d\' = \'%d\'',
$baseClass,
intval($groups),
intval($group)
))
->sort("ID");
// Add child filter
$classes = $indexInstance->getClasses();
$options = $classes[$class];
if (!$options['include_children']) {
$items = $items->filter('ClassName', $class);
}
// Add child filter
$classes = $indexInstance->getClasses();
$options = $classes[$class];
if(!$options['include_children']) {
$items = $items->filter('ClassName', $class);
}
return $items;
}
return $items;
}
/**
* Clear all records of the given class in the current state ONLY.
*
* Optionally delete from a given group (where the group is defined as the ID % total groups)
*
* @param SolrIndex $indexInstance Index instance
* @param string $class Class name
* @param int $groups Number of groups, if clearing from a striped group
* @param int $group Group number, if clearing from a striped group
*/
protected function clearRecords(SolrIndex $indexInstance, $class, $groups = null, $group = null)
{
// Clear by classname
$conditions = array("+(ClassHierarchy:{$class})");
/**
* Clear all records of the given class in the current state ONLY.
*
* Optionally delete from a given group (where the group is defined as the ID % total groups)
*
* @param SolrIndex $indexInstance Index instance
* @param string $class Class name
* @param int $groups Number of groups, if clearing from a striped group
* @param int $group Group number, if clearing from a striped group
*/
protected function clearRecords(SolrIndex $indexInstance, $class, $groups = null, $group = null) {
// Clear by classname
$conditions = array("+(ClassHierarchy:{$class})");
// If grouping, delete from this group only
if ($groups) {
$conditions[] = "+_query_:\"{!frange l={$group} u={$group}}mod(ID, {$groups})\"";
}
// If grouping, delete from this group only
if($groups) {
$conditions[] = "+_query_:\"{!frange l={$group} u={$group}}mod(ID, {$groups})\"";
}
// Also filter by state (suffix on document ID)
$query = new SearchQuery();
SearchVariant::with($class)
->call('alterQuery', $query, $indexInstance);
if ($query->isfiltered()) {
$conditions = array_merge($conditions, $indexInstance->getFiltersComponent($query));
}
// Also filter by state (suffix on document ID)
$query = new SearchQuery();
SearchVariant::with($class)
->call('alterQuery', $query, $indexInstance);
if($query->isfiltered()) {
$conditions = array_merge($conditions, $indexInstance->getFiltersComponent($query));
}
// Invoke delete on index
$deleteQuery = implode(' ', $conditions);
$indexInstance
->getService()
->deleteByQuery($deleteQuery);
}
// Invoke delete on index
$deleteQuery = implode(' ', $conditions);
$indexInstance
->getService()
->deleteByQuery($deleteQuery);
}
}

64
code/solr/reindex/handlers/SolrReindexHandler.php
View File

@ -5,38 +5,38 @@ use Psr\Log\LoggerInterface;
/**
* Provides interface for queueing a solr reindex
*/
interface SolrReindexHandler {
interface SolrReindexHandler
{
/**
* Trigger a solr-reindex
*
* @param LoggerInterface $logger
* @param int $batchSize Records to run each process
* @param string $taskName Name of devtask to run
* @param string|array|null $classes Optional class or classes to limit index to
*/
public function triggerReindex(LoggerInterface $logger, $batchSize, $taskName, $classes = null);
/**
* Trigger a solr-reindex
*
* @param LoggerInterface $logger
* @param int $batchSize Records to run each process
* @param string $taskName Name of devtask to run
* @param string|array|null $classes Optional class or classes to limit index to
*/
public function triggerReindex(LoggerInterface $logger, $batchSize, $taskName, $classes = null);
/**
* Begin an immediate re-index
*
* @param LoggerInterface $logger
* @param int $batchSize Records to run each process
* @param string $taskName Name of devtask to run
* @param string|array|null $classes Optional class or classes to limit index to
*/
public function runReindex(LoggerInterface $logger, $batchSize, $taskName, $classes = null);
/**
* Begin an immediate re-index
*
* @param LoggerInterface $logger
* @param int $batchSize Records to run each process
* @param string $taskName Name of devtask to run
* @param string|array|null $classes Optional class or classes to limit index to
*/
public function runReindex(LoggerInterface $logger, $batchSize, $taskName, $classes = null);
/**
* Do an immediate re-index on the given group, where the group is defined as the list of items
* where ID mod $groups = $group, in the given $state and optional $class filter.
*
* @param LoggerInterface $logger
* @param SolrIndex $indexInstance
* @param array $state
* @param string $class
* @param int $groups
* @param int $group
*/
public function runGroup(LoggerInterface $logger, SolrIndex $indexInstance, $state, $class, $groups, $group);
/**
* Do an immediate re-index on the given group, where the group is defined as the list of items
* where ID mod $groups = $group, in the given $state and optional $class filter.
*
* @param LoggerInterface $logger
* @param SolrIndex $indexInstance
* @param array $state
* @param string $class
* @param int $groups
* @param int $group
*/
public function runGroup(LoggerInterface $logger, SolrIndex $indexInstance, $state, $class, $groups, $group);
}

115
code/solr/reindex/handlers/SolrReindexImmediateHandler.php
View File

@ -7,68 +7,69 @@ use Psr\Log\LoggerInterface;
*
* Internally batches of records will be invoked via shell tasks in the background
*/
class SolrReindexImmediateHandler extends SolrReindexBase {
class SolrReindexImmediateHandler extends SolrReindexBase
{
public function triggerReindex(LoggerInterface $logger, $batchSize, $taskName, $classes = null)
{
$this->runReindex($logger, $batchSize, $taskName, $classes);
}
public function triggerReindex(LoggerInterface $logger, $batchSize, $taskName, $classes = null) {
$this->runReindex($logger, $batchSize, $taskName, $classes);
}
protected function processIndex(
LoggerInterface $logger, SolrIndex $indexInstance, $batchSize, $taskName, $classes = null
) {
parent::processIndex($logger, $indexInstance, $batchSize, $taskName, $classes);
protected function processIndex(
LoggerInterface $logger, SolrIndex $indexInstance, $batchSize, $taskName, $classes = null
) {
parent::processIndex($logger, $indexInstance, $batchSize, $taskName, $classes);
// Immediate processor needs to immediately commit after each index
$indexInstance->getService()->commit();
}
// Immediate processor needs to immediately commit after each index
$indexInstance->getService()->commit();
}
/**
* Process a single group.
*
* Without queuedjobs, it's necessary to shell this out to a background task as this is
* very memory intensive.
*
* The sub-process will then invoke $processor->runGroup() in {@see Solr_Reindex::doReindex}
*
* @param LoggerInterface $logger
* @param SolrIndex $indexInstance Index instance
* @param array $state Variant state
* @param string $class Class to index
* @param int $groups Total groups
* @param int $group Index of group to process
* @param string $taskName Name of task script to run
*/
protected function processGroup(
LoggerInterface $logger, SolrIndex $indexInstance, $state, $class, $groups, $group, $taskName
) {
// Build state
$statevar = json_encode($state);
if (strpos(PHP_OS, "WIN") !== false) {
$statevar = '"'.str_replace('"', '\\"', $statevar).'"';
} else {
$statevar = "'".$statevar."'";
}
/**
* Process a single group.
*
* Without queuedjobs, it's necessary to shell this out to a background task as this is
* very memory intensive.
*
* The sub-process will then invoke $processor->runGroup() in {@see Solr_Reindex::doReindex}
*
* @param LoggerInterface $logger
* @param SolrIndex $indexInstance Index instance
* @param array $state Variant state
* @param string $class Class to index
* @param int $groups Total groups
* @param int $group Index of group to process
* @param string $taskName Name of task script to run
*/
protected function processGroup(
LoggerInterface $logger, SolrIndex $indexInstance, $state, $class, $groups, $group, $taskName
) {
// Build state
$statevar = json_encode($state);
if (strpos(PHP_OS, "WIN") !== false) {
$statevar = '"'.str_replace('"', '\\"', $statevar).'"';
} else {
$statevar = "'".$statevar."'";
}
// Build script
$indexName = $indexInstance->getIndexName();
$scriptPath = sprintf("%s%sframework%scli-script.php", BASE_PATH, DIRECTORY_SEPARATOR, DIRECTORY_SEPARATOR);
$scriptTask = "php {$scriptPath} dev/tasks/{$taskName}";
$cmd = "{$scriptTask} index={$indexName} class={$class} group={$group} groups={$groups} variantstate={$statevar}";
$cmd .= " verbose=1 2>&1";
$logger->info("Running '$cmd'");
// Build script
$indexName = $indexInstance->getIndexName();
$scriptPath = sprintf("%s%sframework%scli-script.php", BASE_PATH, DIRECTORY_SEPARATOR, DIRECTORY_SEPARATOR);
$scriptTask = "php {$scriptPath} dev/tasks/{$taskName}";
$cmd = "{$scriptTask} index={$indexName} class={$class} group={$group} groups={$groups} variantstate={$statevar}";
$cmd .= " verbose=1 2>&1";
$logger->info("Running '$cmd'");
// Execute script via shell
$res = $logger ? passthru($cmd) : `$cmd`;
if ($logger) {
$logger->info(preg_replace('/\r\n|\n/', '$0 ', $res));
}
// Execute script via shell
$res = $logger ? passthru($cmd) : `$cmd`;
if($logger) {
$logger->info(preg_replace('/\r\n|\n/', '$0 ', $res));
}
// If we're in dev mode, commit more often for fun and profit
if (Director::isDev()) {
Solr::service($indexName)->commit();
}
// If we're in dev mode, commit more often for fun and profit
if (Director::isDev()) {
Solr::service($indexName)->commit();
}
// This will slow down things a tiny bit, but it is done so that we don't timeout to the database during a reindex
DB::query('SELECT 1');
}
// This will slow down things a tiny bit, but it is done so that we don't timeout to the database during a reindex
DB::query('SELECT 1');
}
}

74
code/solr/reindex/handlers/SolrReindexMessageHandler.php
View File

@ -2,39 +2,43 @@
use Psr\Log\LoggerInterface;
if(!class_exists('MessageQueue')) return;
class SolrReindexMessageHandler extends SolrReindexImmediateHandler {
/**
* The MessageQueue to use when processing updates
* @config
* @var string
*/
private static $reindex_queue = "search_indexing";
public function triggerReindex(LoggerInterface $logger, $batchSize, $taskName, $classes = null) {
$queue = Config::inst()->get(__CLASS__, 'reindex_queue');
$logger->info('Queuing message');
MessageQueue::send(
$queue,
new MethodInvocationMessage('SolrReindexMessageHandler', 'run_reindex', $batchSize, $taskName, $classes)
);
}
/**
* Entry point for message queue
*
* @param int $batchSize
* @param string $taskName
* @param array|string|null $classes
*/
public static function run_reindex($batchSize, $taskName, $classes = null) {
// @todo Logger for message queue?
$logger = Injector::inst()->createWithArgs('Monolog\Logger', array(strtolower(get_class())));
$inst = Injector::inst()->get(get_class());
$inst->runReindex($logger, $batchSize, $taskName, $classes);
}
if (!class_exists('MessageQueue')) {
return;
}
class SolrReindexMessageHandler extends SolrReindexImmediateHandler
{
/**
* The MessageQueue to use when processing updates
* @config
* @var string
*/
private static $reindex_queue = "search_indexing";
public function triggerReindex(LoggerInterface $logger, $batchSize, $taskName, $classes = null)
{
$queue = Config::inst()->get(__CLASS__, 'reindex_queue');
$logger->info('Queuing message');
MessageQueue::send(
$queue,
new MethodInvocationMessage('SolrReindexMessageHandler', 'run_reindex', $batchSize, $taskName, $classes)
);
}
/**
* Entry point for message queue
*
* @param int $batchSize
* @param string $taskName
* @param array|string|null $classes
*/
public static function run_reindex($batchSize, $taskName, $classes = null)
{
// @todo Logger for message queue?
$logger = Injector::inst()->createWithArgs('Monolog\Logger', array(strtolower(get_class())));
$inst = Injector::inst()->get(get_class());
$inst->runReindex($logger, $batchSize, $taskName, $classes);
}
}

158
code/solr/reindex/handlers/SolrReindexQueuedHandler.php
View File

@ -2,93 +2,97 @@
use Psr\Log\LoggerInterface;
if(!interface_exists('QueuedJob')) return;
if (!interface_exists('QueuedJob')) {
return;
}
/**
* Represents a queued task to start the reindex job
*/
class SolrReindexQueuedHandler extends SolrReindexBase {
class SolrReindexQueuedHandler extends SolrReindexBase
{
/**
* @return QueuedJobService
*/
protected function getQueuedJobService()
{
return singleton('QueuedJobService');
}
/**
* @return QueuedJobService
*/
protected function getQueuedJobService() {
return singleton('QueuedJobService');
}
/**
* Cancel any cancellable jobs
*
* @param string $type Type of job to cancel
* @return int Number of jobs cleared
*/
protected function cancelExistingJobs($type)
{
$clearable = array(
// Paused jobs need to be discarded
QueuedJob::STATUS_PAUSED,
// These types would be automatically started
QueuedJob::STATUS_NEW,
QueuedJob::STATUS_WAIT,
/**
* Cancel any cancellable jobs
*
* @param string $type Type of job to cancel
* @return int Number of jobs cleared
*/
protected function cancelExistingJobs($type) {
$clearable = array(
// Paused jobs need to be discarded
QueuedJob::STATUS_PAUSED,
// These types would be automatically started
QueuedJob::STATUS_NEW,
QueuedJob::STATUS_WAIT,
// Cancel any in-progress job
QueuedJob::STATUS_INIT,
QueuedJob::STATUS_RUN
);
DB::query(sprintf(
'UPDATE "QueuedJobDescriptor" '
. ' SET "JobStatus" = \'%s\''
. ' WHERE "JobStatus" IN (\'%s\')'
. ' AND "Implementation" = \'%s\'',
Convert::raw2sql(QueuedJob::STATUS_CANCELLED),
implode("','", Convert::raw2sql($clearable)),
Convert::raw2sql($type)
));
return DB::affectedRows();
}
// Cancel any in-progress job
QueuedJob::STATUS_INIT,
QueuedJob::STATUS_RUN
);
DB::query(sprintf(
'UPDATE "QueuedJobDescriptor" '
. ' SET "JobStatus" = \'%s\''
. ' WHERE "JobStatus" IN (\'%s\')'
. ' AND "Implementation" = \'%s\'',
Convert::raw2sql(QueuedJob::STATUS_CANCELLED),
implode("','", Convert::raw2sql($clearable)),
Convert::raw2sql($type)
));
return DB::affectedRows();
}
public function triggerReindex(LoggerInterface $logger, $batchSize, $taskName, $classes = null)
{
// Cancel existing jobs
$queues = $this->cancelExistingJobs('SolrReindexQueuedJob');
$groups = $this->cancelExistingJobs('SolrReindexGroupQueuedJob');
$logger->info("Cancelled {$queues} re-index tasks and {$groups} re-index groups");
public function triggerReindex(LoggerInterface $logger, $batchSize, $taskName, $classes = null) {
// Cancel existing jobs
$queues = $this->cancelExistingJobs('SolrReindexQueuedJob');
$groups = $this->cancelExistingJobs('SolrReindexGroupQueuedJob');
$logger->info("Cancelled {$queues} re-index tasks and {$groups} re-index groups");
// Although this class is used as a service (singleton) it may also be instantiated
// as a queuedjob
$job = Injector::inst()->create('SolrReindexQueuedJob', $batchSize, $taskName, $classes);
$this
->getQueuedJobService()
->queueJob($job);
// Although this class is used as a service (singleton) it may also be instantiated
// as a queuedjob
$job = Injector::inst()->create('SolrReindexQueuedJob', $batchSize, $taskName, $classes);
$this
->getQueuedJobService()
->queueJob($job);
$title = $job->getTitle();
$logger->info("Queued {$title}");
}
$title = $job->getTitle();
$logger->info("Queued {$title}");
}
protected function processGroup(
LoggerInterface $logger, SolrIndex $indexInstance, $state, $class, $groups, $group, $taskName
) {
// Trigger another job for this group
$job = Injector::inst()->create(
'SolrReindexGroupQueuedJob',
$indexInstance->getIndexName(), $state, $class, $groups, $group
);
$this
->getQueuedJobService()
->queueJob($job);
$title = $job->getTitle();
$logger->info("Queued {$title}");
}
protected function processGroup(
LoggerInterface $logger, SolrIndex $indexInstance, $state, $class, $groups, $group, $taskName
) {
// Trigger another job for this group
$job = Injector::inst()->create(
'SolrReindexGroupQueuedJob',
$indexInstance->getIndexName(), $state, $class, $groups, $group
);
$this
->getQueuedJobService()
->queueJob($job);
$title = $job->getTitle();
$logger->info("Queued {$title}");
}
public function runGroup(
LoggerInterface $logger, SolrIndex $indexInstance, $state, $class, $groups, $group
) {
parent::runGroup($logger, $indexInstance, $state, $class, $groups, $group);
// After any changes have been made, mark all indexes as dirty for commit
// see http://stackoverflow.com/questions/7512945/how-to-fix-exceeded-limit-of-maxwarmingsearchers
$logger->info("Queuing commit on all changes");
SearchUpdateCommitJobProcessor::queue();
}
public function runGroup(
LoggerInterface $logger, SolrIndex $indexInstance, $state, $class, $groups, $group
) {
parent::runGroup($logger, $indexInstance, $state, $class, $groups, $group);
// After any changes have been made, mark all indexes as dirty for commit
// see http://stackoverflow.com/questions/7512945/how-to-fix-exceeded-limit-of-maxwarmingsearchers
$logger->info("Queuing commit on all changes");
SearchUpdateCommitJobProcessor::queue();
}
}

187
code/solr/reindex/jobs/SolrReindexGroupQueuedJob.php
View File

@ -1,6 +1,8 @@
<?php
if(!interface_exists('QueuedJob')) return;
if (!interface_exists('QueuedJob')) {
return;
}
/**
* Queuedjob to re-index a small group within an index.
@ -11,107 +13,112 @@ if(!interface_exists('QueuedJob')) return;
* list of IDs. Instead groups are segmented by ID. Additionally, this task does incremental
* deletions of records.
*/
class SolrReindexGroupQueuedJob extends SolrReindexQueuedJobBase {
class SolrReindexGroupQueuedJob extends SolrReindexQueuedJobBase
{
/**
* Name of index to reindex
*
* @var string
*/
protected $indexName;
/**
* Name of index to reindex
*
* @var string
*/
protected $indexName;
/**
* Variant state that this group belongs to
*
* @var type
*/
protected $state;
/**
* Variant state that this group belongs to
*
* @var type
*/
protected $state;
/**
* Single class name to index
*
* @var string
*/
protected $class;
/**
* Single class name to index
*
* @var string
*/
protected $class;
/**
* Total number of groups
*
* @var int
*/
protected $groups;
/**
* Total number of groups
*
* @var int
*/
protected $groups;
/**
* Group index
*
* @var int
*/
protected $group;
/**
* Group index
*
* @var int
*/
protected $group;
public function __construct($indexName = null, $state = null, $class = null, $groups = null, $group = null)
{
parent::__construct();
$this->indexName = $indexName;
$this->state = $state;
$this->class = $class;
$this->groups = $groups;
$this->group = $group;
}
public function __construct($indexName = null, $state = null, $class = null, $groups = null, $group = null) {
parent::__construct();
$this->indexName = $indexName;
$this->state = $state;
$this->class = $class;
$this->groups = $groups;
$this->group = $group;
}
public function getJobData()
{
$data = parent::getJobData();
public function getJobData() {
$data = parent::getJobData();
// Custom data
$data->jobData->indexName = $this->indexName;
$data->jobData->state = $this->state;
$data->jobData->class = $this->class;
$data->jobData->groups = $this->groups;
$data->jobData->group = $this->group;
return $data;
}
// Custom data
$data->jobData->indexName = $this->indexName;
$data->jobData->state = $this->state;
$data->jobData->class = $this->class;
$data->jobData->groups = $this->groups;
$data->jobData->group = $this->group;
return $data;
}
public function setJobData($totalSteps, $currentStep, $isComplete, $jobData, $messages)
{
parent::setJobData($totalSteps, $currentStep, $isComplete, $jobData, $messages);
public function setJobData($totalSteps, $currentStep, $isComplete, $jobData, $messages) {
parent::setJobData($totalSteps, $currentStep, $isComplete, $jobData, $messages);
// Custom data
$this->indexName = $jobData->indexName;
$this->state = $jobData->state;
$this->class = $jobData->class;
$this->groups = $jobData->groups;
$this->group = $jobData->group;
}
// Custom data
$this->indexName = $jobData->indexName;
$this->state = $jobData->state;
$this->class = $jobData->class;
$this->groups = $jobData->groups;
$this->group = $jobData->group;
}
public function getSignature()
{
return md5(get_class($this) . time() . mt_rand(0, 100000));
}
public function getSignature() {
return md5(get_class($this) . time() . mt_rand(0, 100000));
}
public function getTitle()
{
return sprintf(
'Solr Reindex Group (%d/%d) of %s in %s',
($this->group+1),
$this->groups,
$this->class,
json_encode($this->state)
);
}
public function getTitle() {
return sprintf(
'Solr Reindex Group (%d/%d) of %s in %s',
($this->group+1),
$this->groups,
$this->class,
json_encode($this->state)
);
}
public function process()
{
$logger = $this->getLogger();
if ($this->jobFinished()) {
$logger->notice("reindex group already complete");
return;
}
public function process() {
$logger = $this->getLogger();
if($this->jobFinished()) {
$logger->notice("reindex group already complete");
return;
}
// Get instance of index
$indexInstance = singleton($this->indexName);
// Send back to processor
$logger->info("Beginning reindex group");
$this
->getHandler()
->runGroup($logger, $indexInstance, $this->state, $this->class, $this->groups, $this->group);
$logger->info("Completed reindex group");
$this->isComplete = true;
}
// Get instance of index
$indexInstance = singleton($this->indexName);
// Send back to processor
$logger->info("Beginning reindex group");
$this
->getHandler()
->runGroup($logger, $indexInstance, $this->state, $this->class, $this->groups, $this->group);
$logger->info("Completed reindex group");
$this->isComplete = true;
}
}

153
code/solr/reindex/jobs/SolrReindexQueuedJob.php
View File

@ -1,91 +1,100 @@
<?php
if(!interface_exists('QueuedJob')) return;
if (!interface_exists('QueuedJob')) {
return;
}
/**
* Represents a queuedjob which invokes a reindex
*/
class SolrReindexQueuedJob extends SolrReindexQueuedJobBase {
class SolrReindexQueuedJob extends SolrReindexQueuedJobBase
{
/**
* Size of each batch to run
*
* @var int
*/
protected $batchSize;
/**
* Size of each batch to run
*
* @var int
*/
protected $batchSize;
/**
* Name of devtask Which invoked this
* Not necessary for re-index processing performed entirely by queuedjobs
*
* @var string
*/
protected $taskName;
/**
* Name of devtask Which invoked this
* Not necessary for re-index processing performed entirely by queuedjobs
*
* @var string
*/
protected $taskName;
/**
* List of classes to filter
*
* @var array|string
*/
protected $classes;
/**
* List of classes to filter
*
* @var array|string
*/
protected $classes;
public function __construct($batchSize = null, $taskName = null, $classes = null)
{
$this->batchSize = $batchSize;
$this->taskName = $taskName;
$this->classes = $classes;
parent::__construct();
}
public function __construct($batchSize = null, $taskName = null, $classes = null) {
$this->batchSize = $batchSize;
$this->taskName = $taskName;
$this->classes = $classes;
parent::__construct();
}
public function getJobData()
{
$data = parent::getJobData();
public function getJobData() {
$data = parent::getJobData();
// Custom data
$data->jobData->batchSize = $this->batchSize;
$data->jobData->taskName = $this->taskName;
$data->jobData->classes = $this->classes;
return $data;
}
// Custom data
$data->jobData->batchSize = $this->batchSize;
$data->jobData->taskName = $this->taskName;
$data->jobData->classes = $this->classes;
return $data;
}
public function setJobData($totalSteps, $currentStep, $isComplete, $jobData, $messages)
{
parent::setJobData($totalSteps, $currentStep, $isComplete, $jobData, $messages);
public function setJobData($totalSteps, $currentStep, $isComplete, $jobData, $messages) {
parent::setJobData($totalSteps, $currentStep, $isComplete, $jobData, $messages);
// Custom data
$this->batchSize = $jobData->batchSize;
$this->taskName = $jobData->taskName;
$this->classes = $jobData->classes;
}
// Custom data
$this->batchSize = $jobData->batchSize;
$this->taskName = $jobData->taskName;
$this->classes = $jobData->classes;
}
public function getSignature()
{
return __CLASS__;
}
public function getSignature() {
return __CLASS__;
}
public function getTitle()
{
return 'Solr Reindex Job';
}
public function getTitle() {
return 'Solr Reindex Job';
}
public function process()
{
$logger = $this->getLogger();
if ($this->jobFinished()) {
$logger->notice("reindex already complete");
return;
}
public function process() {
$logger = $this->getLogger();
if($this->jobFinished()) {
$logger->notice("reindex already complete");
return;
}
// Send back to processor
$logger->info("Beginning init of reindex");
$this
->getHandler()
->runReindex($logger, $this->batchSize, $this->taskName, $this->classes);
$logger->info("Completed init of reindex");
$this->isComplete = true;
}
// Send back to processor
$logger->info("Beginning init of reindex");
$this
->getHandler()
->runReindex($logger, $this->batchSize, $this->taskName, $this->classes);
$logger->info("Completed init of reindex");
$this->isComplete = true;
}
/**
* Get size of batch
*
* @return int
*/
public function getBatchSize() {
return $this->batchSize;
}
/**
* Get size of batch
*
* @return int
*/
public function getBatchSize()
{
return $this->batchSize;
}
}

205
code/solr/reindex/jobs/SolrReindexQueuedJobBase.php
View File

@ -3,121 +3,136 @@
use Monolog\Logger;
use Psr\Log\LoggerInterface;
if(!interface_exists('QueuedJob')) return;
if (!interface_exists('QueuedJob')) {
return;
}
/**
* Base class for jobs which perform re-index
*/
abstract class SolrReindexQueuedJobBase implements QueuedJob {
abstract class SolrReindexQueuedJobBase implements QueuedJob
{
/**
* Flag whether this job is done
*
* @var bool
*/
protected $isComplete;
/**
* Flag whether this job is done
*
* @var bool
*/
protected $isComplete;
/**
* List of messages
*
* @var array
*/
protected $messages;
/**
* List of messages
*
* @var array
*/
protected $messages;
/**
* Logger to use for this job
*
* @var LoggerInterface
*/
protected $logger;
/**
* Logger to use for this job
*
* @var LoggerInterface
*/
protected $logger;
public function __construct()
{
$this->isComplete = false;
$this->messages = array();
}
public function __construct() {
$this->isComplete = false;
$this->messages = array();
}
/**
* @return SearchLogFactory
*/
protected function getLoggerFactory()
{
return Injector::inst()->get('SearchLogFactory');
}
/**
* @return SearchLogFactory
*/
protected function getLoggerFactory() {
return Injector::inst()->get('SearchLogFactory');
}
/**
* Gets a logger for this job
*
* @return LoggerInterface
*/
protected function getLogger()
{
if ($this->logger) {
return $this->logger;
}
/**
* Gets a logger for this job
*
* @return LoggerInterface
*/
protected function getLogger() {
if($this->logger) {
return $this->logger;
}
// Set logger for this job
$this->logger = $this
->getLoggerFactory()
->getQueuedJobLogger($this);
return $this->logger;
}
// Set logger for this job
$this->logger = $this
->getLoggerFactory()
->getQueuedJobLogger($this);
return $this->logger;
}
/**
* Assign custom logger for this job
*
* @param LoggerInterface $logger
*/
public function setLogger($logger)
{
$this->logger = $logger;
}
/**
* Assign custom logger for this job
*
* @param LoggerInterface $logger
*/
public function setLogger($logger) {
$this->logger = $logger;
}
public function getJobData()
{
$data = new stdClass();
public function getJobData() {
$data = new stdClass();
// Standard fields
$data->totalSteps = 1;
$data->currentStep = $this->isComplete ? 0 : 1;
$data->isComplete = $this->isComplete;
$data->messages = $this->messages;
// Standard fields
$data->totalSteps = 1;
$data->currentStep = $this->isComplete ? 0 : 1;
$data->isComplete = $this->isComplete;
$data->messages = $this->messages;
// Custom data
$data->jobData = new stdClass();
return $data;
}
// Custom data
$data->jobData = new stdClass();
return $data;
}
public function setJobData($totalSteps, $currentStep, $isComplete, $jobData, $messages)
{
$this->isComplete = $isComplete;
$this->messages = $messages;
}
public function setJobData($totalSteps, $currentStep, $isComplete, $jobData, $messages) {
$this->isComplete = $isComplete;
$this->messages = $messages;
}
/**
* Get the reindex handler
*
* @return SolrReindexHandler
*/
protected function getHandler()
{
return Injector::inst()->get('SolrReindexHandler');
}
/**
* Get the reindex handler
*
* @return SolrReindexHandler
*/
protected function getHandler() {
return Injector::inst()->get('SolrReindexHandler');
}
public function jobFinished()
{
return $this->isComplete;
}
public function jobFinished() {
return $this->isComplete;
}
public function prepareForRestart()
{
// NOOP
}
public function prepareForRestart() {
// NOOP
}
public function setup()
{
// NOOP
}
public function setup() {
// NOOP
}
public function afterComplete()
{
// NOOP
}
public function afterComplete() {
// NOOP
}
public function getJobType()
{
return QueuedJob::QUEUED;
}
public function getJobType() {
return QueuedJob::QUEUED;
}
public function addMessage($message) {
$this->messages[] = $message;
}
public function addMessage($message)
{
$this->messages[] = $message;
}
}

120
code/utils/CombinationsArrayIterator.php
View File

@ -1,66 +1,80 @@
<?php
class CombinationsArrayIterator implements Iterator {
protected $arrays;
protected $keys;
protected $numArrays;
class CombinationsArrayIterator implements Iterator
{
protected $arrays;
protected $keys;
protected $numArrays;
protected $isValid = false;
protected $k = 0;
protected $isValid = false;
protected $k = 0;
function __construct($args) {
$this->arrays = array();
$this->keys = array();
public function __construct($args)
{
$this->arrays = array();
$this->keys = array();
$keys = array_keys($args);
$values = array_values($args);
$keys = array_keys($args);
$values = array_values($args);
foreach ($values as $i => $arg) {
if (is_array($arg) && count($arg)) {
$this->arrays[] = $arg;
$this->keys[] = $keys[$i];
}
}
foreach ($values as $i => $arg) {
if (is_array($arg) && count($arg)) {
$this->arrays[] = $arg;
$this->keys[] = $keys[$i];
}
}
$this->numArrays = count($this->arrays);
$this->rewind();
}
$this->numArrays = count($this->arrays);
$this->rewind();
}
function rewind() {
if (!$this->numArrays) {
$this->isValid = false;
}
else {
$this->isValid = true;
$this->k = 0;
for ($i = 0; $i < $this->numArrays; $i++) reset($this->arrays[$i]);
}
}
public function rewind()
{
if (!$this->numArrays) {
$this->isValid = false;
} else {
$this->isValid = true;
$this->k = 0;
for ($i = 0; $i < $this->numArrays; $i++) {
reset($this->arrays[$i]);
}
}
}
function valid() {
return $this->isValid;
}
public function valid()
{
return $this->isValid;
}
function next() {
$this->k++;
public function next()
{
$this->k++;
for ($i = 0; $i < $this->numArrays; $i++) {
if (next($this->arrays[$i]) === false) {
if ($i == $this->numArrays-1) $this->isValid = false;
else reset($this->arrays[$i]);
}
else break;
}
}
for ($i = 0; $i < $this->numArrays; $i++) {
if (next($this->arrays[$i]) === false) {
if ($i == $this->numArrays-1) {
$this->isValid = false;
} else {
reset($this->arrays[$i]);
}
} else {
break;
}
}
}
function current() {
$res = array();
for ($i = 0; $i < $this->numArrays; $i++) $res[$this->keys[$i]] = current($this->arrays[$i]);
return $res;
}
public function current()
{
$res = array();
for ($i = 0; $i < $this->numArrays; $i++) {
$res[$this->keys[$i]] = current($this->arrays[$i]);
}
return $res;
}
function key() {
return $this->k;
}
}
public function key()
{
return $this->k;
}
}

78
code/utils/MultipleArrayIterator.php
View File

@ -1,44 +1,58 @@
<?php
class MultipleArrayIterator implements Iterator {
class MultipleArrayIterator implements Iterator
{
protected $arrays;
protected $active;
protected $arrays;
protected $active;
public function __construct()
{
$args = func_get_args();
function __construct() {
$args = func_get_args();
$this->arrays = array();
foreach ($args as $arg) {
if (is_array($arg) && count($arg)) {
$this->arrays[] = $arg;
}
}
$this->arrays = array();
foreach ($args as $arg) {
if (is_array($arg) && count($arg)) $this->arrays[] = $arg;
}
$this->rewind();
}
$this->rewind();
}
public function rewind()
{
$this->active = $this->arrays;
if ($this->active) {
reset($this->active[0]);
}
}
function rewind() {
$this->active = $this->arrays;
if ($this->active) reset($this->active[0]);
}
public function current()
{
return $this->active ? current($this->active[0]) : false;
}
function current() {
return $this->active ? current($this->active[0]) : false;
}
public function key()
{
return $this->active ? key($this->active[0]) : false;
}
function key() {
return $this->active ? key($this->active[0]) : false;
}
public function next()
{
if (!$this->active) {
return;
}
function next() {
if (!$this->active) return;
if (next($this->active[0]) === false) {
array_shift($this->active);
if ($this->active) {
reset($this->active[0]);
}
}
}
if (next($this->active[0]) === false) {
array_shift($this->active);
if ($this->active) reset($this->active[0]);
}
}
function valid() {
return $this->active && (current($this->active[0]) !== false);
}
public function valid()
{
return $this->active && (current($this->active[0]) !== false);
}
}

106
code/utils/WebDAV.php
View File

@ -1,68 +1,76 @@
<?php
class WebDAV {
class WebDAV
{
public static function curl_init($url, $method)
{
$ch = curl_init($url);
static function curl_init($url, $method) {
$ch = curl_init($url);
curl_setopt($ch, CURLOPT_RETURNTRANSFER, true);
curl_setopt($ch, CURLOPT_HTTPAUTH, CURLAUTH_ANY);
curl_setopt($ch, CURLOPT_CUSTOMREQUEST, $method);
curl_setopt($ch, CURLOPT_RETURNTRANSFER, true);
curl_setopt($ch, CURLOPT_HTTPAUTH, CURLAUTH_ANY);
curl_setopt($ch, CURLOPT_CUSTOMREQUEST, $method);
return $ch;
}
return $ch;
}
public static function exists($url)
{
// WebDAV expects that checking a directory exists has a trailing slash
if (substr($url, -1) != '/') {
$url .= '/';
}
static function exists($url) {
// WebDAV expects that checking a directory exists has a trailing slash
if (substr($url, -1) != '/') {
$url .= '/';
}
$ch = self::curl_init($url, 'PROPFIND');
$res = curl_exec($ch);
$code = curl_getinfo($ch, CURLINFO_HTTP_CODE);
$ch = self::curl_init($url, 'PROPFIND');
$res = curl_exec($ch);
$code = curl_getinfo($ch, CURLINFO_HTTP_CODE);
if ($code == 404) {
return false;
}
if ($code == 200 || $code == 207) {
return true;
}
if ($code == 404) return false;
if ($code == 200 || $code == 207) return true;
user_error("Got error from webdav server - ".$code, E_USER_ERROR);
}
user_error("Got error from webdav server - ".$code, E_USER_ERROR);
}
public static function mkdir($url)
{
$ch = self::curl_init(rtrim($url, '/').'/', 'MKCOL');
static function mkdir($url) {
$ch = self::curl_init(rtrim($url, '/').'/', 'MKCOL');
$res = curl_exec($ch);
$code = curl_getinfo($ch, CURLINFO_HTTP_CODE);
$res = curl_exec($ch);
$code = curl_getinfo($ch, CURLINFO_HTTP_CODE);
return $code == 201;
}
return $code == 201;
}
public static function put($handle, $url)
{
$ch = curl_init($url);
curl_setopt($ch, CURLOPT_HTTPAUTH, CURLAUTH_ANY);
static function put($handle, $url) {
$ch = curl_init($url);
curl_setopt($ch, CURLOPT_HTTPAUTH, CURLAUTH_ANY);
curl_setopt($ch, CURLOPT_PUT, true);
curl_setopt($ch, CURLOPT_BINARYTRANSFER, true);
curl_setopt($ch, CURLOPT_PUT, true);
curl_setopt($ch, CURLOPT_BINARYTRANSFER, true);
curl_setopt($ch, CURLOPT_INFILE, $handle);
curl_setopt($ch, CURLOPT_INFILE, $handle);
$res = curl_exec($ch);
fclose($handle);
$res = curl_exec($ch);
fclose($handle);
return curl_getinfo($ch, CURLINFO_HTTP_CODE);
}
static function upload_from_string($string, $url) {
$fh = tmpfile();
fwrite($fh, $string);
fseek($fh, 0);
return self::put($fh, $url);
}
static function upload_from_file($string, $url) {
return self::put(fopen($string, 'rb'), $url);
}
return curl_getinfo($ch, CURLINFO_HTTP_CODE);
}
public static function upload_from_string($string, $url)
{
$fh = tmpfile();
fwrite($fh, $string);
fseek($fh, 0);
return self::put($fh, $url);
}
public static function upload_from_file($string, $url)
{
return self::put(fopen($string, 'rb'), $url);
}
}

166
code/utils/logging/MonologFactory.php
View File

@ -8,91 +8,97 @@ use Monolog\Logger;
/**
* Provides logging based on monolog
*/
class MonologFactory implements SearchLogFactory {
class MonologFactory implements SearchLogFactory
{
public function getOutputLogger($name, $verbose)
{
$logger = $this->getLoggerFor($name);
$formatter = $this->getFormatter();
public function getOutputLogger($name, $verbose) {
$logger = $this->getLoggerFor($name);
$formatter = $this->getFormatter();
// Notice handling
if ($verbose) {
$messageHandler = $this->getStreamHandler($formatter, 'php://stdout', Logger::INFO);
$logger->pushHandler($messageHandler);
}
// Notice handling
if($verbose) {
$messageHandler = $this->getStreamHandler($formatter, 'php://stdout', Logger::INFO);
$logger->pushHandler($messageHandler);
}
// Error handling. buble is false so that errors aren't logged twice
$errorHandler = $this->getStreamHandler($formatter, 'php://stderr', Logger::ERROR, false);
$logger->pushHandler($errorHandler);
return $logger;
}
// Error handling. buble is false so that errors aren't logged twice
$errorHandler = $this->getStreamHandler($formatter, 'php://stderr', Logger::ERROR, false);
$logger->pushHandler($errorHandler);
return $logger;
}
public function getQueuedJobLogger($job)
{
$logger = $this->getLoggerFor(get_class($job));
$handler = $this->getJobHandler($job);
$logger->pushHandler($handler);
return $logger;
}
public function getQueuedJobLogger($job) {
$logger = $this->getLoggerFor(get_class($job));
$handler = $this->getJobHandler($job);
$logger->pushHandler($handler);
return $logger;
}
/**
* Generate a handler for the given stream
*
* @param FormatterInterface $formatter
* @param string $stream Name of preferred stream
* @param int $level
* @param bool $bubble
* @return HandlerInterface
*/
protected function getStreamHandler(FormatterInterface $formatter, $stream, $level = Logger::DEBUG, $bubble = true)
{
// Unless cli, force output to php://output
$stream = Director::is_cli() ? $stream : 'php://output';
$handler = Injector::inst()->createWithArgs(
'Monolog\Handler\StreamHandler',
array($stream, $level, $bubble)
);
$handler->setFormatter($formatter);
return $handler;
}
/**
* Generate a handler for the given stream
*
* @param FormatterInterface $formatter
* @param string $stream Name of preferred stream
* @param int $level
* @param bool $bubble
* @return HandlerInterface
*/
protected function getStreamHandler(FormatterInterface $formatter, $stream, $level = Logger::DEBUG, $bubble = true) {
// Unless cli, force output to php://output
$stream = Director::is_cli() ? $stream : 'php://output';
$handler = Injector::inst()->createWithArgs(
'Monolog\Handler\StreamHandler',
array($stream, $level, $bubble)
);
$handler->setFormatter($formatter);
return $handler;
}
/**
* Gets a formatter for standard output
*
* @return FormatterInterface
*/
protected function getFormatter()
{
// Get formatter
$format = LineFormatter::SIMPLE_FORMAT;
if (!Director::is_cli()) {
$format = "<p>$format</p>";
}
return Injector::inst()->createWithArgs(
'Monolog\Formatter\LineFormatter',
array($format)
);
}
/**
* Gets a formatter for standard output
*
* @return FormatterInterface
*/
protected function getFormatter() {
// Get formatter
$format = LineFormatter::SIMPLE_FORMAT;
if(!Director::is_cli()) {
$format = "<p>$format</p>";
}
return Injector::inst()->createWithArgs(
'Monolog\Formatter\LineFormatter',
array($format)
);
}
/**
* Get a logger for a named class
*
* @param string $name
* @return Logger
*/
protected function getLoggerFor($name)
{
return Injector::inst()->createWithArgs(
'Monolog\Logger',
array(strtolower($name))
);
}
/**
* Get a logger for a named class
*
* @param string $name
* @return Logger
*/
protected function getLoggerFor($name) {
return Injector::inst()->createWithArgs(
'Monolog\Logger',
array(strtolower($name))
);
}
/**
* Generate handler for a job object
*
* @param QueuedJob $job
* @return HandlerInterface
*/
protected function getJobHandler($job) {
return Injector::inst()->createWithArgs(
'QueuedJobLogHandler',
array($job, Logger::INFO)
);
}
/**
* Generate handler for a job object
*
* @param QueuedJob $job
* @return HandlerInterface
*/
protected function getJobHandler($job)
{
return Injector::inst()->createWithArgs(
'QueuedJobLogHandler',
array($job, Logger::INFO)
);
}
}

83
code/utils/logging/QueuedJobLogHandler.php
View File

@ -3,51 +3,56 @@
use Monolog\Handler\AbstractProcessingHandler;
use Monolog\Logger;
if(!interface_exists('QueuedJob')) return;
if (!interface_exists('QueuedJob')) {
return;
}
/**
* Handler for logging events into QueuedJob message data
*/
class QueuedJobLogHandler extends AbstractProcessingHandler {
class QueuedJobLogHandler extends AbstractProcessingHandler
{
/**
* Job to log to
*
* @var QueuedJob
*/
protected $queuedJob;
/**
* Job to log to
*
* @var QueuedJob
*/
protected $queuedJob;
/**
* @param QueuedJob $queuedJob Job to log to
* @param integer $level The minimum logging level at which this handler will be triggered
* @param Boolean $bubble Whether the messages that are handled can bubble up the stack or not
*/
public function __construct(QueuedJob $queuedJob, $level = Logger::DEBUG, $bubble = true)
{
parent::__construct($level, $bubble);
$this->setQueuedJob($queuedJob);
}
/**
* @param QueuedJob $queuedJob Job to log to
* @param integer $level The minimum logging level at which this handler will be triggered
* @param Boolean $bubble Whether the messages that are handled can bubble up the stack or not
*/
public function __construct(QueuedJob $queuedJob, $level = Logger::DEBUG, $bubble = true) {
parent::__construct($level, $bubble);
$this->setQueuedJob($queuedJob);
}
/**
* Set a new queuedjob
*
* @param QueuedJob $queuedJob
*/
public function setQueuedJob(QueuedJob $queuedJob)
{
$this->queuedJob = $queuedJob;
}
/**
* Set a new queuedjob
*
* @param QueuedJob $queuedJob
*/
public function setQueuedJob(QueuedJob $queuedJob) {
$this->queuedJob = $queuedJob;
}
/**
* Get queuedjob
*
* @return QueuedJob
*/
public function getQueuedJob() {
return $this->queuedJob;
}
protected function write(array $record) {
// Write formatted message
$this->getQueuedJob()->addMessage($record['formatted']);
}
/**
* Get queuedjob
*
* @return QueuedJob
*/
public function getQueuedJob()
{
return $this->queuedJob;
}
protected function write(array $record)
{
// Write formatted message
$this->getQueuedJob()->addMessage($record['formatted']);
}
}

34
code/utils/logging/SearchLogFactory.php
View File

@ -2,22 +2,22 @@
use Psr\Log;
interface SearchLogFactory {
interface SearchLogFactory
{
/**
* Make a logger for a queuedjob
*
* @param QueuedJob $job
* @return Log
*/
public function getQueuedJobLogger($job);
/**
* Make a logger for a queuedjob
*
* @param QueuedJob $job
* @return Log
*/
public function getQueuedJobLogger($job);
/**
* Get an output logger with the given verbosity
*
* @param string $name
* @param bool $verbose
* @return Log
*/
public function getOutputLogger($name, $verbose);
/**
* Get an output logger with the given verbosity
*
* @param string $name
* @param bool $verbose
* @return Log
*/
public function getOutputLogger($name, $verbose);
}

439
tests/BatchedProcessorTest.php
View File

@ -1,246 +1,259 @@
<?php
class BatchedProcessorTest_Object extends SiteTree implements TestOnly {
private static $db = array(
'TestText' => 'Varchar'
);
class BatchedProcessorTest_Object extends SiteTree implements TestOnly
{
private static $db = array(
'TestText' => 'Varchar'
);
}
class BatchedProcessorTest_Index extends SearchIndex_Recording implements TestOnly {
function init() {
$this->addClass('BatchedProcessorTest_Object');
$this->addFilterField('TestText');
}
class BatchedProcessorTest_Index extends SearchIndex_Recording implements TestOnly
{
public function init()
{
$this->addClass('BatchedProcessorTest_Object');
$this->addFilterField('TestText');
}
}
class BatchedProcessor_QueuedJobService {
protected $jobs = array();
class BatchedProcessor_QueuedJobService
{
protected $jobs = array();
public function queueJob(QueuedJob $job, $startAfter = null, $userId = null, $queueName = null) {
$this->jobs[] = array(
'job' => $job,
'startAfter' => $startAfter
);
return $job;
}
public function queueJob(QueuedJob $job, $startAfter = null, $userId = null, $queueName = null)
{
$this->jobs[] = array(
'job' => $job,
'startAfter' => $startAfter
);
return $job;
}
public function getJobs() {
return $this->jobs;
}
public function getJobs()
{
return $this->jobs;
}
}
/**
* Tests {@see SearchUpdateQueuedJobProcessor}
*/
class BatchedProcessorTest extends SapphireTest {
protected $oldProcessor;
protected $extraDataObjects = array(
'BatchedProcessorTest_Object'
);
class BatchedProcessorTest extends SapphireTest
{
protected $oldProcessor;
protected $extraDataObjects = array(
'BatchedProcessorTest_Object'
);
protected $illegalExtensions = array(
'SiteTree' => array(
'SiteTreeSubsites',
'Translatable'
)
);
protected $illegalExtensions = array(
'SiteTree' => array(
'SiteTreeSubsites',
'Translatable'
)
);
public function setUpOnce() {
// Disable illegal extensions if skipping this test
if(class_exists('Subsite') || !interface_exists('QueuedJob')) {
$this->illegalExtensions = array();
}
parent::setUpOnce();
}
public function setUpOnce()
{
// Disable illegal extensions if skipping this test
if (class_exists('Subsite') || !interface_exists('QueuedJob')) {
$this->illegalExtensions = array();
}
parent::setUpOnce();
}
public function setUp() {
parent::setUp();
Config::nest();
if (!interface_exists('QueuedJob')) {
$this->skipTest = true;
$this->markTestSkipped("These tests need the QueuedJobs module installed to run");
}
public function setUp()
{
parent::setUp();
Config::nest();
if (!interface_exists('QueuedJob')) {
$this->skipTest = true;
$this->markTestSkipped("These tests need the QueuedJobs module installed to run");
}
if(class_exists('Subsite')) {
$this->skipTest = true;
$this->markTestSkipped(get_class() . ' skipped when running with subsites');
}
if (class_exists('Subsite')) {
$this->skipTest = true;
$this->markTestSkipped(get_class() . ' skipped when running with subsites');
}
SS_Datetime::set_mock_now('2015-05-07 06:00:00');
Config::inst()->update('SearchUpdateBatchedProcessor', 'batch_size', 5);
Config::inst()->update('SearchUpdateBatchedProcessor', 'batch_soft_cap', 0);
Config::inst()->update('SearchUpdateCommitJobProcessor', 'cooldown', 600);
Versioned::reading_stage("Stage");
SS_Datetime::set_mock_now('2015-05-07 06:00:00');
Config::inst()->update('SearchUpdateBatchedProcessor', 'batch_size', 5);
Config::inst()->update('SearchUpdateBatchedProcessor', 'batch_soft_cap', 0);
Config::inst()->update('SearchUpdateCommitJobProcessor', 'cooldown', 600);
Versioned::reading_stage("Stage");
Injector::inst()->registerService(new BatchedProcessor_QueuedJobService(), 'QueuedJobService');
Injector::inst()->registerService(new BatchedProcessor_QueuedJobService(), 'QueuedJobService');
FullTextSearch::force_index_list('BatchedProcessorTest_Index');
FullTextSearch::force_index_list('BatchedProcessorTest_Index');
SearchUpdateCommitJobProcessor::$dirty_indexes = array();
SearchUpdateCommitJobProcessor::$has_run = false;
$this->oldProcessor = SearchUpdater::$processor;
SearchUpdater::$processor = new SearchUpdateQueuedJobProcessor();
}
SearchUpdateCommitJobProcessor::$dirty_indexes = array();
SearchUpdateCommitJobProcessor::$has_run = false;
$this->oldProcessor = SearchUpdater::$processor;
SearchUpdater::$processor = new SearchUpdateQueuedJobProcessor();
}
public function tearDown() {
if($this->oldProcessor) {
SearchUpdater::$processor = $this->oldProcessor;
}
Config::unnest();
Injector::inst()->unregisterNamedObject('QueuedJobService');
FullTextSearch::force_index_list();
parent::tearDown();
}
public function tearDown()
{
if ($this->oldProcessor) {
SearchUpdater::$processor = $this->oldProcessor;
}
Config::unnest();
Injector::inst()->unregisterNamedObject('QueuedJobService');
FullTextSearch::force_index_list();
parent::tearDown();
}
/**
* @return SearchUpdateQueuedJobProcessor
*/
protected function generateDirtyIds() {
$processor = SearchUpdater::$processor;
for($id = 1; $id <= 42; $id++) {
// Save to db
$object = new BatchedProcessorTest_Object();
$object->TestText = 'Object ' . $id;
$object->write();
// Add to index manually
$processor->addDirtyIDs(
'BatchedProcessorTest_Object',
array(array(
'id' => $id,
'state' => array('SearchVariantVersioned' => 'Stage')
)),
'BatchedProcessorTest_Index'
);
}
$processor->batchData();
return $processor;
}
/**
* Tests that large jobs are broken up into a suitable number of batches
*/
public function testBatching() {
$index = singleton('BatchedProcessorTest_Index');
$index->reset();
$processor = $this->generateDirtyIds();
// Check initial state
$data = $processor->getJobData();
$this->assertEquals(9, $data->totalSteps);
$this->assertEquals(0, $data->currentStep);
$this->assertEmpty($data->isComplete);
$this->assertEquals(0, count($index->getAdded()));
// Advance state
for($pass = 1; $pass <= 8; $pass++) {
$processor->process();
$data = $processor->getJobData();
$this->assertEquals($pass, $data->currentStep);
$this->assertEquals($pass * 5, count($index->getAdded()));
}
// Last run should have two hanging items
$processor->process();
$data = $processor->getJobData();
$this->assertEquals(9, $data->currentStep);
$this->assertEquals(42, count($index->getAdded()));
$this->assertTrue($data->isComplete);
/**
* @return SearchUpdateQueuedJobProcessor
*/
protected function generateDirtyIds()
{
$processor = SearchUpdater::$processor;
for ($id = 1; $id <= 42; $id++) {
// Save to db
$object = new BatchedProcessorTest_Object();
$object->TestText = 'Object ' . $id;
$object->write();
// Add to index manually
$processor->addDirtyIDs(
'BatchedProcessorTest_Object',
array(array(
'id' => $id,
'state' => array('SearchVariantVersioned' => 'Stage')
)),
'BatchedProcessorTest_Index'
);
}
$processor->batchData();
return $processor;
}
/**
* Tests that large jobs are broken up into a suitable number of batches
*/
public function testBatching()
{
$index = singleton('BatchedProcessorTest_Index');
$index->reset();
$processor = $this->generateDirtyIds();
// Check initial state
$data = $processor->getJobData();
$this->assertEquals(9, $data->totalSteps);
$this->assertEquals(0, $data->currentStep);
$this->assertEmpty($data->isComplete);
$this->assertEquals(0, count($index->getAdded()));
// Advance state
for ($pass = 1; $pass <= 8; $pass++) {
$processor->process();
$data = $processor->getJobData();
$this->assertEquals($pass, $data->currentStep);
$this->assertEquals($pass * 5, count($index->getAdded()));
}
// Last run should have two hanging items
$processor->process();
$data = $processor->getJobData();
$this->assertEquals(9, $data->currentStep);
$this->assertEquals(42, count($index->getAdded()));
$this->assertTrue($data->isComplete);
// Check any additional queued jobs
$processor->afterComplete();
$service = singleton('QueuedJobService');
$jobs = $service->getJobs();
$this->assertEquals(1, count($jobs));
$this->assertInstanceOf('SearchUpdateCommitJobProcessor', $jobs[0]['job']);
}
// Check any additional queued jobs
$processor->afterComplete();
$service = singleton('QueuedJobService');
$jobs = $service->getJobs();
$this->assertEquals(1, count($jobs));
$this->assertInstanceOf('SearchUpdateCommitJobProcessor', $jobs[0]['job']);
}
/**
* Test creation of multiple commit jobs
*/
public function testMultipleCommits() {
$index = singleton('BatchedProcessorTest_Index');
$index->reset();
/**
* Test creation of multiple commit jobs
*/
public function testMultipleCommits()
{
$index = singleton('BatchedProcessorTest_Index');
$index->reset();
// Test that running a commit immediately after submitting to the indexes
// correctly commits
$first = SearchUpdateCommitJobProcessor::queue();
$second = SearchUpdateCommitJobProcessor::queue();
// Test that running a commit immediately after submitting to the indexes
// correctly commits
$first = SearchUpdateCommitJobProcessor::queue();
$second = SearchUpdateCommitJobProcessor::queue();
$this->assertFalse($index->getIsCommitted());
$this->assertFalse($index->getIsCommitted());
// First process will cause the commit
$this->assertFalse($first->jobFinished());
$first->process();
$allMessages = $first->getMessages();
$this->assertTrue($index->getIsCommitted());
$this->assertTrue($first->jobFinished());
$this->assertStringEndsWith('All indexes committed', $allMessages[2]);
// First process will cause the commit
$this->assertFalse($first->jobFinished());
$first->process();
$allMessages = $first->getMessages();
$this->assertTrue($index->getIsCommitted());
$this->assertTrue($first->jobFinished());
$this->assertStringEndsWith('All indexes committed', $allMessages[2]);
// Executing the subsequent processor should not re-trigger a commit
$index->reset();
$this->assertFalse($second->jobFinished());
$second->process();
$allMessages = $second->getMessages();
$this->assertFalse($index->getIsCommitted());
$this->assertTrue($second->jobFinished());
$this->assertStringEndsWith('Indexing already completed this request: Discarding this job', $allMessages[0]);
// Executing the subsequent processor should not re-trigger a commit
$index->reset();
$this->assertFalse($second->jobFinished());
$second->process();
$allMessages = $second->getMessages();
$this->assertFalse($index->getIsCommitted());
$this->assertTrue($second->jobFinished());
$this->assertStringEndsWith('Indexing already completed this request: Discarding this job', $allMessages[0]);
// Given that a third job is created, and the indexes are dirtied, attempting to run this job
// should result in a delay
$index->reset();
$third = SearchUpdateCommitJobProcessor::queue();
$this->assertFalse($third->jobFinished());
$third->process();
$this->assertTrue($third->jobFinished());
$allMessages = $third->getMessages();
$this->assertStringEndsWith(
'Indexing already run this request, but incomplete. Re-scheduling for 2015-05-07 06:10:00',
$allMessages[0]
);
}
// Given that a third job is created, and the indexes are dirtied, attempting to run this job
// should result in a delay
$index->reset();
$third = SearchUpdateCommitJobProcessor::queue();
$this->assertFalse($third->jobFinished());
$third->process();
$this->assertTrue($third->jobFinished());
$allMessages = $third->getMessages();
$this->assertStringEndsWith(
'Indexing already run this request, but incomplete. Re-scheduling for 2015-05-07 06:10:00',
$allMessages[0]
);
}
/**
* Tests that the batch_soft_cap setting is properly respected
*/
public function testSoftCap() {
$index = singleton('BatchedProcessorTest_Index');
$index->reset();
$processor = $this->generateDirtyIds();
// Test that increasing the soft cap to 2 will reduce the number of batches
Config::inst()->update('SearchUpdateBatchedProcessor', 'batch_soft_cap', 2);
$processor->batchData();
$data = $processor->getJobData();
//Debug::dump($data);die;
$this->assertEquals(8, $data->totalSteps);
// A soft cap of 1 should not fit in the hanging two items
Config::inst()->update('SearchUpdateBatchedProcessor', 'batch_soft_cap', 1);
$processor->batchData();
$data = $processor->getJobData();
$this->assertEquals(9, $data->totalSteps);
// Extra large soft cap should fit both items
Config::inst()->update('SearchUpdateBatchedProcessor', 'batch_soft_cap', 4);
$processor->batchData();
$data = $processor->getJobData();
$this->assertEquals(8, $data->totalSteps);
// Process all data and ensure that all are processed adequately
for($pass = 1; $pass <= 8; $pass++) {
$processor->process();
}
$data = $processor->getJobData();
$this->assertEquals(8, $data->currentStep);
$this->assertEquals(42, count($index->getAdded()));
$this->assertTrue($data->isComplete);
}
/**
* Tests that the batch_soft_cap setting is properly respected
*/
public function testSoftCap()
{
$index = singleton('BatchedProcessorTest_Index');
$index->reset();
$processor = $this->generateDirtyIds();
// Test that increasing the soft cap to 2 will reduce the number of batches
Config::inst()->update('SearchUpdateBatchedProcessor', 'batch_soft_cap', 2);
$processor->batchData();
$data = $processor->getJobData();
//Debug::dump($data);die;
$this->assertEquals(8, $data->totalSteps);
// A soft cap of 1 should not fit in the hanging two items
Config::inst()->update('SearchUpdateBatchedProcessor', 'batch_soft_cap', 1);
$processor->batchData();
$data = $processor->getJobData();
$this->assertEquals(9, $data->totalSteps);
// Extra large soft cap should fit both items
Config::inst()->update('SearchUpdateBatchedProcessor', 'batch_soft_cap', 4);
$processor->batchData();
$data = $processor->getJobData();
$this->assertEquals(8, $data->totalSteps);
// Process all data and ensure that all are processed adequately
for ($pass = 1; $pass <= 8; $pass++) {
$processor->process();
}
$data = $processor->getJobData();
$this->assertEquals(8, $data->currentStep);
$this->assertEquals(42, count($index->getAdded()));
$this->assertTrue($data->isComplete);
}
}

318
tests/SearchUpdaterTest.php
View File

@ -1,215 +1,229 @@
<?php
class SearchUpdaterTest_Container extends DataObject {
private static $db = array(
'Field1' => 'Varchar',
'Field2' => 'Varchar',
'MyDate' => 'Date',
);
class SearchUpdaterTest_Container extends DataObject
{
private static $db = array(
'Field1' => 'Varchar',
'Field2' => 'Varchar',
'MyDate' => 'Date',
);
private static $has_one = array(
'HasOneObject' => 'SearchUpdaterTest_HasOne'
);
private static $has_one = array(
'HasOneObject' => 'SearchUpdaterTest_HasOne'
);
private static $has_many = array(
'HasManyObjects' => 'SearchUpdaterTest_HasMany'
);
private static $has_many = array(
'HasManyObjects' => 'SearchUpdaterTest_HasMany'
);
private static $many_many = array(
'ManyManyObjects' => 'SearchUpdaterTest_ManyMany'
);
private static $many_many = array(
'ManyManyObjects' => 'SearchUpdaterTest_ManyMany'
);
}
class SearchUpdaterTest_HasOne extends DataObject {
private static $db = array(
'Field1' => 'Varchar',
'Field2' => 'Varchar'
);
class SearchUpdaterTest_HasOne extends DataObject
{
private static $db = array(
'Field1' => 'Varchar',
'Field2' => 'Varchar'
);
private static $has_many = array(
'HasManyContainers' => 'SearchUpdaterTest_Container'
);
private static $has_many = array(
'HasManyContainers' => 'SearchUpdaterTest_Container'
);
}
class SearchUpdaterTest_HasMany extends DataObject {
private static $db = array(
'Field1' => 'Varchar',
'Field2' => 'Varchar'
);
class SearchUpdaterTest_HasMany extends DataObject
{
private static $db = array(
'Field1' => 'Varchar',
'Field2' => 'Varchar'
);
private static $has_one = array(
'HasManyContainer' => 'SearchUpdaterTest_Container'
);
private static $has_one = array(
'HasManyContainer' => 'SearchUpdaterTest_Container'
);
}
class SearchUpdaterTest_ManyMany extends DataObject {
private static $db = array(
'Field1' => 'Varchar',
'Field2' => 'Varchar'
);
class SearchUpdaterTest_ManyMany extends DataObject
{
private static $db = array(
'Field1' => 'Varchar',
'Field2' => 'Varchar'
);
private static $belongs_many_many = array(
'ManyManyContainer' => 'SearchUpdaterTest_Container'
);
private static $belongs_many_many = array(
'ManyManyContainer' => 'SearchUpdaterTest_Container'
);
}
class SearchUpdaterTest_Index extends SearchIndex_Recording {
function init() {
$this->addClass('SearchUpdaterTest_Container');
class SearchUpdaterTest_Index extends SearchIndex_Recording
{
public function init()
{
$this->addClass('SearchUpdaterTest_Container');
$this->addFilterField('Field1');
$this->addFilterField('HasOneObject.Field1');
$this->addFilterField('HasManyObjects.Field1');
}
$this->addFilterField('Field1');
$this->addFilterField('HasOneObject.Field1');
$this->addFilterField('HasManyObjects.Field1');
}
}
class SearchUpdaterTest extends SapphireTest {
class SearchUpdaterTest extends SapphireTest
{
protected $usesDatabase = true;
protected $usesDatabase = true;
private static $index = null;
public function setUp()
{
parent::setUp();
private static $index = null;
function setUp() {
parent::setUp();
if (self::$index === null) {
self::$index = singleton(get_class($this).'_Index');
} else {
self::$index->reset();
}
if (self::$index === null) self::$index = singleton(get_class($this).'_Index');
else self::$index->reset();
SearchUpdater::bind_manipulation_capture();
SearchUpdater::bind_manipulation_capture();
Config::nest();
Config::nest();
Config::inst()->update('Injector', 'SearchUpdateProcessor', array(
'class' => 'SearchUpdateImmediateProcessor'
));
Config::inst()->update('Injector', 'SearchUpdateProcessor', array(
'class' => 'SearchUpdateImmediateProcessor'
));
FullTextSearch::force_index_list(self::$index);
SearchUpdater::clear_dirty_indexes();
}
FullTextSearch::force_index_list(self::$index);
SearchUpdater::clear_dirty_indexes();
}
public function tearDown()
{
Config::unnest();
function tearDown() {
Config::unnest();
parent::tearDown();
}
parent::tearDown();
}
public function testBasic()
{
$item = new SearchUpdaterTest_Container();
$item->write();
function testBasic() {
$item = new SearchUpdaterTest_Container();
$item->write();
// TODO: Make sure changing field1 updates item.
// TODO: Get updating just field2 to not update item (maybe not possible - variants complicate)
}
// TODO: Make sure changing field1 updates item.
// TODO: Get updating just field2 to not update item (maybe not possible - variants complicate)
}
public function testHasOneHook()
{
$hasOne = new SearchUpdaterTest_HasOne();
$hasOne->write();
function testHasOneHook() {
$hasOne = new SearchUpdaterTest_HasOne();
$hasOne->write();
$alternateHasOne = new SearchUpdaterTest_HasOne();
$alternateHasOne->write();
$alternateHasOne = new SearchUpdaterTest_HasOne();
$alternateHasOne->write();
$container1 = new SearchUpdaterTest_Container();
$container1->HasOneObjectID = $hasOne->ID;
$container1->write();
$container1 = new SearchUpdaterTest_Container();
$container1->HasOneObjectID = $hasOne->ID;
$container1->write();
$container2 = new SearchUpdaterTest_Container();
$container2->HasOneObjectID = $hasOne->ID;
$container2->write();
$container2 = new SearchUpdaterTest_Container();
$container2->HasOneObjectID = $hasOne->ID;
$container2->write();
$container3 = new SearchUpdaterTest_Container();
$container3->HasOneObjectID = $alternateHasOne->ID;
$container3->write();
$container3 = new SearchUpdaterTest_Container();
$container3->HasOneObjectID = $alternateHasOne->ID;
$container3->write();
// Check the default "writing a document updates the document"
SearchUpdater::flush_dirty_indexes();
// Check the default "writing a document updates the document"
SearchUpdater::flush_dirty_indexes();
$added = self::$index->getAdded(array('ID'));
// Some databases don't output $added in a consistent order; that's okay
usort($added, function($a,$b) {return $a['ID']-$b['ID']; });
$added = self::$index->getAdded(array('ID'));
// Some databases don't output $added in a consistent order; that's okay
usort($added, function ($a, $b) {return $a['ID']-$b['ID']; });
$this->assertEquals($added, array(
array('ID' => $container1->ID),
array('ID' => $container2->ID),
array('ID' => $container3->ID)
));
$this->assertEquals($added, array(
array('ID' => $container1->ID),
array('ID' => $container2->ID),
array('ID' => $container3->ID)
));
// Check writing a has_one tracks back to the origin documents
// Check writing a has_one tracks back to the origin documents
self::$index->reset();
self::$index->reset();
$hasOne->Field1 = "Updated";
$hasOne->write();
$hasOne->Field1 = "Updated";
$hasOne->write();
SearchUpdater::flush_dirty_indexes();
$added = self::$index->getAdded(array('ID'));
// Some databases don't output $added in a consistent order; that's okay
usort($added, function($a,$b) {return $a['ID']-$b['ID']; });
SearchUpdater::flush_dirty_indexes();
$added = self::$index->getAdded(array('ID'));
// Some databases don't output $added in a consistent order; that's okay
usort($added, function ($a, $b) {return $a['ID']-$b['ID']; });
$this->assertEquals($added, array(
array('ID' => $container1->ID),
array('ID' => $container2->ID)
));
$this->assertEquals($added, array(
array('ID' => $container1->ID),
array('ID' => $container2->ID)
));
// Check updating an unrelated field doesn't track back
// Check updating an unrelated field doesn't track back
self::$index->reset();
self::$index->reset();
$hasOne->Field2 = "Updated";
$hasOne->write();
$hasOne->Field2 = "Updated";
$hasOne->write();
SearchUpdater::flush_dirty_indexes();
$this->assertEquals(self::$index->getAdded(array('ID')), array());
SearchUpdater::flush_dirty_indexes();
$this->assertEquals(self::$index->getAdded(array('ID')), array());
// Check writing a has_one tracks back to the origin documents
// Check writing a has_one tracks back to the origin documents
self::$index->reset();
self::$index->reset();
$alternateHasOne->Field1= "Updated";
$alternateHasOne->write();
$alternateHasOne->Field1= "Updated";
$alternateHasOne->write();
SearchUpdater::flush_dirty_indexes();
$this->assertEquals(self::$index->getAdded(array('ID')), array(
array('ID' => $container3->ID)
));
}
SearchUpdater::flush_dirty_indexes();
$this->assertEquals(self::$index->getAdded(array('ID')), array(
array('ID' => $container3->ID)
));
}
function testHasManyHook() {
$container1 = new SearchUpdaterTest_Container();
$container1->write();
public function testHasManyHook()
{
$container1 = new SearchUpdaterTest_Container();
$container1->write();
$container2 = new SearchUpdaterTest_Container();
$container2->write();
$container2 = new SearchUpdaterTest_Container();
$container2->write();
//self::$index->reset();
//SearchUpdater::clear_dirty_indexes();
//self::$index->reset();
//SearchUpdater::clear_dirty_indexes();
$hasMany1 = new SearchUpdaterTest_HasMany();
$hasMany1->HasManyContainerID = $container1->ID;
$hasMany1->write();
$hasMany1 = new SearchUpdaterTest_HasMany();
$hasMany1->HasManyContainerID = $container1->ID;
$hasMany1->write();
$hasMany2 = new SearchUpdaterTest_HasMany();
$hasMany2->HasManyContainerID = $container1->ID;
$hasMany2->write();
$hasMany2 = new SearchUpdaterTest_HasMany();
$hasMany2->HasManyContainerID = $container1->ID;
$hasMany2->write();
SearchUpdater::flush_dirty_indexes();
SearchUpdater::flush_dirty_indexes();
$this->assertEquals(self::$index->getAdded(array('ID')), array(
array('ID' => $container1->ID),
array('ID' => $container2->ID)
));
$this->assertEquals(self::$index->getAdded(array('ID')), array(
array('ID' => $container1->ID),
array('ID' => $container2->ID)
));
self::$index->reset();
self::$index->reset();
$hasMany1->Field1 = 'Updated';
$hasMany1->write();
$hasMany1->Field1 = 'Updated';
$hasMany1->write();
$hasMany2->Field1 = 'Updated';
$hasMany2->write();
$hasMany2->Field1 = 'Updated';
$hasMany2->write();
SearchUpdater::flush_dirty_indexes();
$this->assertEquals(self::$index->getAdded(array('ID')), array(
array('ID' => $container1->ID)
));
}
SearchUpdater::flush_dirty_indexes();
$this->assertEquals(self::$index->getAdded(array('ID')), array(
array('ID' => $container1->ID)
));
}
}

119
tests/SearchVariantSiteTreeSubsitesPolyhomeTest.php
View File

@ -1,76 +1,83 @@
<?php
class SearchVariantSiteTreeSubsitesPolyhomeTest_Item extends SiteTree {
// TODO: Currently theres a failure if you addClass a non-table class
private static $db = array(
'TestText' => 'Varchar'
);
class SearchVariantSiteTreeSubsitesPolyhomeTest_Item extends SiteTree
{
// TODO: Currently theres a failure if you addClass a non-table class
private static $db = array(
'TestText' => 'Varchar'
);
}
class SearchVariantSiteTreeSubsitesPolyhomeTest_Index extends SearchIndex_Recording {
function init() {
$this->addClass('SearchVariantSiteTreeSubsitesPolyhomeTest_Item');
$this->addFilterField('TestText');
}
class SearchVariantSiteTreeSubsitesPolyhomeTest_Index extends SearchIndex_Recording
{
public function init()
{
$this->addClass('SearchVariantSiteTreeSubsitesPolyhomeTest_Item');
$this->addFilterField('TestText');
}
}
class SearchVariantSiteTreeSubsitesPolyhomeTest extends SapphireTest {
class SearchVariantSiteTreeSubsitesPolyhomeTest extends SapphireTest
{
private static $index = null;
private static $index = null;
private static $subsite_a = null;
private static $subsite_b = null;
private static $subsite_a = null;
private static $subsite_b = null;
public function setUp()
{
parent::setUp();
function setUp() {
parent::setUp();
// Check subsites installed
if (!class_exists('Subsite') || !class_exists('SubsitePolyhome')) {
return $this->markTestSkipped('The subsites polyhome module is not installed');
}
// Check subsites installed
if(!class_exists('Subsite') || !class_exists('SubsitePolyhome')) {
return $this->markTestSkipped('The subsites polyhome module is not installed');
}
if (self::$index === null) {
self::$index = singleton('SearchVariantSiteTreeSubsitesPolyhomeTest_Index');
}
if (self::$index === null) self::$index = singleton('SearchVariantSiteTreeSubsitesPolyhomeTest_Index');
if (self::$subsite_a === null) {
self::$subsite_a = new Subsite();
self::$subsite_a->write();
self::$subsite_b = new Subsite();
self::$subsite_b->write();
}
if (self::$subsite_a === null) {
self::$subsite_a = new Subsite(); self::$subsite_a->write();
self::$subsite_b = new Subsite(); self::$subsite_b->write();
}
FullTextSearch::force_index_list(self::$index);
SearchUpdater::clear_dirty_indexes();
}
FullTextSearch::force_index_list(self::$index);
SearchUpdater::clear_dirty_indexes();
}
public function testSavingDirect()
{
// Initial add
function testSavingDirect() {
// Initial add
$item = new SearchVariantSiteTreeSubsitesPolyhomeTest_Item();
$item->write();
$item = new SearchVariantSiteTreeSubsitesPolyhomeTest_Item();
$item->write();
SearchUpdater::flush_dirty_indexes();
$this->assertEquals(self::$index->getAdded(array('ID', '_subsite')), array(
array('ID' => $item->ID, '_subsite' => 0)
));
SearchUpdater::flush_dirty_indexes();
$this->assertEquals(self::$index->getAdded(array('ID', '_subsite')), array(
array('ID' => $item->ID, '_subsite' => 0)
));
// Check that adding to subsites works
// Check that adding to subsites works
self::$index->reset();
self::$index->reset();
$item->setField('AddToSubsite[0]', 1);
$item->setField('AddToSubsite['.(self::$subsite_a->ID).']', 1);
$item->setField('AddToSubsite[0]', 1);
$item->setField('AddToSubsite['.(self::$subsite_a->ID).']', 1);
$item->write();
$item->write();
SearchUpdater::flush_dirty_indexes();
$this->assertEquals(self::$index->getAdded(array('ID', '_subsite')), array(
array('ID' => $item->ID, '_subsite' => 0),
array('ID' => $item->ID, '_subsite' => self::$subsite_a->ID)
));
$this->assertEquals(self::$index->deleted, array(
array('base' => 'SiteTree', 'id' => $item->ID, 'state' => array(
'SearchVariantVersioned' => 'Stage', 'SearchVariantSiteTreeSubsitesPolyhome' => self::$subsite_b->ID
))
));
}
}
SearchUpdater::flush_dirty_indexes();
$this->assertEquals(self::$index->getAdded(array('ID', '_subsite')), array(
array('ID' => $item->ID, '_subsite' => 0),
array('ID' => $item->ID, '_subsite' => self::$subsite_a->ID)
));
$this->assertEquals(self::$index->deleted, array(
array('base' => 'SiteTree', 'id' => $item->ID, 'state' => array(
'SearchVariantVersioned' => 'Stage', 'SearchVariantSiteTreeSubsitesPolyhome' => self::$subsite_b->ID
))
));
}
}

175
tests/SearchVariantVersionedTest.php
View File

@ -1,114 +1,125 @@
<?php
class SearchVariantVersionedTest extends SapphireTest {
class SearchVariantVersionedTest extends SapphireTest
{
private static $index = null;
private static $index = null;
protected $extraDataObjects = array(
'SearchVariantVersionedTest_Item'
);
protected $extraDataObjects = array(
'SearchVariantVersionedTest_Item'
);
public function setUp()
{
parent::setUp();
function setUp() {
parent::setUp();
// Check versioned available
if (!class_exists('Versioned')) {
return $this->markTestSkipped('The versioned decorator is not installed');
}
// Check versioned available
if(!class_exists('Versioned')) {
return $this->markTestSkipped('The versioned decorator is not installed');
}
if (self::$index === null) {
self::$index = singleton('SearchVariantVersionedTest_Index');
}
if (self::$index === null) self::$index = singleton('SearchVariantVersionedTest_Index');
SearchUpdater::bind_manipulation_capture();
SearchUpdater::bind_manipulation_capture();
Config::nest();
Config::nest();
Config::inst()->update('Injector', 'SearchUpdateProcessor', array(
'class' => 'SearchUpdateImmediateProcessor'
));
Config::inst()->update('Injector', 'SearchUpdateProcessor', array(
'class' => 'SearchUpdateImmediateProcessor'
));
FullTextSearch::force_index_list(self::$index);
SearchUpdater::clear_dirty_indexes();
}
FullTextSearch::force_index_list(self::$index);
SearchUpdater::clear_dirty_indexes();
}
public function tearDown()
{
Config::unnest();
function tearDown() {
Config::unnest();
parent::tearDown();
}
parent::tearDown();
}
public function testPublishing()
{
// Check that write updates Stage
function testPublishing() {
// Check that write updates Stage
$item = new SearchVariantVersionedTest_Item(array('TestText' => 'Foo'));
$item->write();
$item = new SearchVariantVersionedTest_Item(array('TestText' => 'Foo'));
$item->write();
SearchUpdater::flush_dirty_indexes();
$this->assertEquals(self::$index->getAdded(array('ID', '_versionedstage')), array(
array('ID' => $item->ID, '_versionedstage' => 'Stage')
));
SearchUpdater::flush_dirty_indexes();
$this->assertEquals(self::$index->getAdded(array('ID', '_versionedstage')), array(
array('ID' => $item->ID, '_versionedstage' => 'Stage')
));
// Check that publish updates Live
// Check that publish updates Live
self::$index->reset();
self::$index->reset();
$item->publish("Stage", "Live");
$item->publish("Stage", "Live");
SearchUpdater::flush_dirty_indexes();
$this->assertEquals(self::$index->getAdded(array('ID', '_versionedstage')), array(
array('ID' => $item->ID, '_versionedstage' => 'Live')
));
SearchUpdater::flush_dirty_indexes();
$this->assertEquals(self::$index->getAdded(array('ID', '_versionedstage')), array(
array('ID' => $item->ID, '_versionedstage' => 'Live')
));
// Just update a SiteTree field, and check it updates Stage
// Just update a SiteTree field, and check it updates Stage
self::$index->reset();
self::$index->reset();
$item->Title = "Pow!";
$item->write();
$item->Title = "Pow!";
$item->write();
SearchUpdater::flush_dirty_indexes();
$this->assertEquals(self::$index->getAdded(array('ID', '_versionedstage')), array(
array('ID' => $item->ID, '_versionedstage' => 'Stage')
));
}
SearchUpdater::flush_dirty_indexes();
$this->assertEquals(self::$index->getAdded(array('ID', '_versionedstage')), array(
array('ID' => $item->ID, '_versionedstage' => 'Stage')
));
}
function testExcludeVariantState() {
$index = singleton('SearchVariantVersionedTest_IndexNoStage');
FullTextSearch::force_index_list($index);
public function testExcludeVariantState()
{
$index = singleton('SearchVariantVersionedTest_IndexNoStage');
FullTextSearch::force_index_list($index);
// Check that write doesn't update stage
$item = new SearchVariantVersionedTest_Item(array('TestText' => 'Foo'));
$item->write();
SearchUpdater::flush_dirty_indexes();
$this->assertEquals($index->getAdded(array('ID', '_versionedstage')), array());
// Check that write doesn't update stage
$item = new SearchVariantVersionedTest_Item(array('TestText' => 'Foo'));
$item->write();
SearchUpdater::flush_dirty_indexes();
$this->assertEquals($index->getAdded(array('ID', '_versionedstage')), array());
// Check that publish updates Live
$index->reset();
$item->publish("Stage", "Live");
SearchUpdater::flush_dirty_indexes();
$this->assertEquals($index->getAdded(array('ID', '_versionedstage')), array(
array('ID' => $item->ID, '_versionedstage' => 'Live')
));
}
// Check that publish updates Live
$index->reset();
$item->publish("Stage", "Live");
SearchUpdater::flush_dirty_indexes();
$this->assertEquals($index->getAdded(array('ID', '_versionedstage')), array(
array('ID' => $item->ID, '_versionedstage' => 'Live')
));
}
}
class SearchVariantVersionedTest_Item extends SiteTree implements TestOnly {
// TODO: Currently theres a failure if you addClass a non-table class
private static $db = array(
'TestText' => 'Varchar'
);
class SearchVariantVersionedTest_Item extends SiteTree implements TestOnly
{
// TODO: Currently theres a failure if you addClass a non-table class
private static $db = array(
'TestText' => 'Varchar'
);
}
class SearchVariantVersionedTest_Index extends SearchIndex_Recording {
function init() {
$this->addClass('SearchVariantVersionedTest_Item');
$this->addFilterField('TestText');
}
class SearchVariantVersionedTest_Index extends SearchIndex_Recording
{
public function init()
{
$this->addClass('SearchVariantVersionedTest_Item');
$this->addFilterField('TestText');
}
}
class SearchVariantVersionedTest_IndexNoStage extends SearchIndex_Recording {
function init() {
$this->addClass('SearchVariantVersionedTest_Item');
$this->addFilterField('TestText');
$this->excludeVariantState(array('SearchVariantVersioned' => 'Stage'));
}
}
class SearchVariantVersionedTest_IndexNoStage extends SearchIndex_Recording
{
public function init()
{
$this->addClass('SearchVariantVersionedTest_Item');
$this->addFilterField('TestText');
$this->excludeVariantState(array('SearchVariantVersioned' => 'Stage'));
}
}

123
tests/Solr4ServiceTest.php
View File

@ -3,64 +3,71 @@
/**
* Test solr 4.0 compatibility
*/
class Solr4ServiceTest extends SapphireTest {
/**
*
* @return Solr4ServiceTest_RecordingService
*/
protected function getMockService() {
return new Solr4ServiceTest_RecordingService();
}
protected function getMockDocument($id) {
$document = new Apache_Solr_Document();
$document->setField('id', $id);
$document->setField('title', "Item $id");
return $document;
}
public function testAddDocument() {
$service = $this->getMockService();
$sent = $service->addDocument($this->getMockDocument('A'), false);
$this->assertEquals(
'<add overwrite="true"><doc><field name="id">A</field><field name="title">Item A</field></doc></add>',
$sent
);
$sent = $service->addDocument($this->getMockDocument('B'), true);
$this->assertEquals(
'<add overwrite="false"><doc><field name="id">B</field><field name="title">Item B</field></doc></add>',
$sent
);
}
public function testAddDocuments() {
$service = $this->getMockService();
$sent = $service->addDocuments(array(
$this->getMockDocument('C'),
$this->getMockDocument('D')
), false);
$this->assertEquals(
'<add overwrite="true"><doc><field name="id">C</field><field name="title">Item C</field></doc><doc><field name="id">D</field><field name="title">Item D</field></doc></add>',
$sent
);
$sent = $service->addDocuments(array(
$this->getMockDocument('E'),
$this->getMockDocument('F')
), true);
$this->assertEquals(
'<add overwrite="false"><doc><field name="id">E</field><field name="title">Item E</field></doc><doc><field name="id">F</field><field name="title">Item F</field></doc></add>',
$sent
);
}
class Solr4ServiceTest extends SapphireTest
{
/**
*
* @return Solr4ServiceTest_RecordingService
*/
protected function getMockService()
{
return new Solr4ServiceTest_RecordingService();
}
protected function getMockDocument($id)
{
$document = new Apache_Solr_Document();
$document->setField('id', $id);
$document->setField('title', "Item $id");
return $document;
}
public function testAddDocument()
{
$service = $this->getMockService();
$sent = $service->addDocument($this->getMockDocument('A'), false);
$this->assertEquals(
'<add overwrite="true"><doc><field name="id">A</field><field name="title">Item A</field></doc></add>',
$sent
);
$sent = $service->addDocument($this->getMockDocument('B'), true);
$this->assertEquals(
'<add overwrite="false"><doc><field name="id">B</field><field name="title">Item B</field></doc></add>',
$sent
);
}
public function testAddDocuments()
{
$service = $this->getMockService();
$sent = $service->addDocuments(array(
$this->getMockDocument('C'),
$this->getMockDocument('D')
), false);
$this->assertEquals(
'<add overwrite="true"><doc><field name="id">C</field><field name="title">Item C</field></doc><doc><field name="id">D</field><field name="title">Item D</field></doc></add>',
$sent
);
$sent = $service->addDocuments(array(
$this->getMockDocument('E'),
$this->getMockDocument('F')
), true);
$this->assertEquals(
'<add overwrite="false"><doc><field name="id">E</field><field name="title">Item E</field></doc><doc><field name="id">F</field><field name="title">Item F</field></doc></add>',
$sent
);
}
}
class Solr4ServiceTest_RecordingService extends Solr4Service_Core {
protected function _sendRawPost($url, $rawPost, $timeout = FALSE, $contentType = 'text/xml; charset=UTF-8') {
return $rawPost;
}
protected function _sendRawGet($url, $timeout = FALSE) {
return $url;
}
class Solr4ServiceTest_RecordingService extends Solr4Service_Core
{
protected function _sendRawPost($url, $rawPost, $timeout = false, $contentType = 'text/xml; charset=UTF-8')
{
return $rawPost;
}
protected function _sendRawGet($url, $timeout = false)
{
return $url;
}
}

531
tests/SolrIndexTest.php
View File

@ -1,307 +1,330 @@
<?php
class SolrIndexTest extends SapphireTest {
class SolrIndexTest extends SapphireTest
{
public function setUpOnce()
{
parent::setUpOnce();
function setUpOnce() {
parent::setUpOnce();
if (class_exists('Phockito')) {
Phockito::include_hamcrest();
}
}
if (class_exists('Phockito')) Phockito::include_hamcrest();
}
public function setUp()
{
if (!class_exists('Phockito')) {
$this->markTestSkipped("These tests need the Phockito module installed to run");
$this->skipTest = true;
}
function setUp() {
if (!class_exists('Phockito')) {
$this->markTestSkipped("These tests need the Phockito module installed to run");
$this->skipTest = true;
}
parent::setUp();
}
parent::setUp();
}
public function testFieldDataHasOne()
{
$index = new SolrIndexTest_FakeIndex();
$data = $index->fieldData('HasOneObject.Field1');
$data = $data['SearchUpdaterTest_Container_HasOneObject_Field1'];
function testFieldDataHasOne() {
$index = new SolrIndexTest_FakeIndex();
$data = $index->fieldData('HasOneObject.Field1');
$data = $data['SearchUpdaterTest_Container_HasOneObject_Field1'];
$this->assertEquals('SearchUpdaterTest_Container', $data['origin']);
$this->assertEquals('SearchUpdaterTest_Container', $data['base']);
$this->assertEquals('SearchUpdaterTest_HasOne', $data['class']);
}
$this->assertEquals('SearchUpdaterTest_Container', $data['origin']);
$this->assertEquals('SearchUpdaterTest_Container', $data['base']);
$this->assertEquals('SearchUpdaterTest_HasOne', $data['class']);
}
public function testFieldDataHasMany()
{
$index = new SolrIndexTest_FakeIndex();
$data = $index->fieldData('HasManyObjects.Field1');
$data = $data['SearchUpdaterTest_Container_HasManyObjects_Field1'];
function testFieldDataHasMany() {
$index = new SolrIndexTest_FakeIndex();
$data = $index->fieldData('HasManyObjects.Field1');
$data = $data['SearchUpdaterTest_Container_HasManyObjects_Field1'];
$this->assertEquals('SearchUpdaterTest_Container', $data['origin']);
$this->assertEquals('SearchUpdaterTest_Container', $data['base']);
$this->assertEquals('SearchUpdaterTest_HasMany', $data['class']);
}
$this->assertEquals('SearchUpdaterTest_Container', $data['origin']);
$this->assertEquals('SearchUpdaterTest_Container', $data['base']);
$this->assertEquals('SearchUpdaterTest_HasMany', $data['class']);
}
public function testFieldDataManyMany()
{
$index = new SolrIndexTest_FakeIndex();
$data = $index->fieldData('ManyManyObjects.Field1');
$data = $data['SearchUpdaterTest_Container_ManyManyObjects_Field1'];
function testFieldDataManyMany() {
$index = new SolrIndexTest_FakeIndex();
$data = $index->fieldData('ManyManyObjects.Field1');
$data = $data['SearchUpdaterTest_Container_ManyManyObjects_Field1'];
$this->assertEquals('SearchUpdaterTest_Container', $data['origin']);
$this->assertEquals('SearchUpdaterTest_Container', $data['base']);
$this->assertEquals('SearchUpdaterTest_ManyMany', $data['class']);
}
$this->assertEquals('SearchUpdaterTest_Container', $data['origin']);
$this->assertEquals('SearchUpdaterTest_Container', $data['base']);
$this->assertEquals('SearchUpdaterTest_ManyMany', $data['class']);
}
/**
* Test boosting on SearchQuery
*/
public function testBoostedQuery()
{
$serviceMock = $this->getServiceMock();
Phockito::when($serviceMock)->search(anything(), anything(), anything(), anything(), anything())->return($this->getFakeRawSolrResponse());
/**
* Test boosting on SearchQuery
*/
function testBoostedQuery() {
$serviceMock = $this->getServiceMock();
Phockito::when($serviceMock)->search(anything(), anything(), anything(), anything(), anything())->return($this->getFakeRawSolrResponse());
$index = new SolrIndexTest_FakeIndex();
$index->setService($serviceMock);
$index = new SolrIndexTest_FakeIndex();
$index->setService($serviceMock);
$query = new SearchQuery();
$query->search(
'term',
null,
array('Field1' => 1.5, 'HasOneObject_Field1' => 3)
);
$index->search($query);
$query = new SearchQuery();
$query->search(
'term',
null,
array('Field1' => 1.5, 'HasOneObject_Field1' => 3)
);
$index->search($query);
Phockito::verify($serviceMock)->search('+(Field1:term^1.5 OR HasOneObject_Field1:term^3)', anything(), anything(), anything(), anything());
}
/**
* Test boosting on field schema (via queried fields parameter)
*/
public function testBoostedField()
{
$serviceMock = $this->getServiceMock();
Phockito::when($serviceMock)
->search(anything(), anything(), anything(), anything(), anything())
->return($this->getFakeRawSolrResponse());
Phockito::verify($serviceMock)->search('+(Field1:term^1.5 OR HasOneObject_Field1:term^3)', anything(), anything(), anything(), anything());
}
/**
* Test boosting on field schema (via queried fields parameter)
*/
public function testBoostedField() {
$serviceMock = $this->getServiceMock();
Phockito::when($serviceMock)
->search(anything(), anything(), anything(), anything(), anything())
->return($this->getFakeRawSolrResponse());
$index = new SolrIndexTest_BoostedIndex();
$index->setService($serviceMock);
$index = new SolrIndexTest_BoostedIndex();
$index->setService($serviceMock);
$query = new SearchQuery();
$query->search('term');
$index->search($query);
$query = new SearchQuery();
$query->search('term');
$index->search($query);
// Ensure matcher contains correct boost in 'qf' parameter
$matcher = new Hamcrest_Array_IsArrayContainingKeyValuePair(
new Hamcrest_Core_IsEqual('qf'),
new Hamcrest_Core_IsEqual('SearchUpdaterTest_Container_Field1^1.5 SearchUpdaterTest_Container_Field2^2.1 _text')
);
Phockito::verify($serviceMock)
->search('+term', anything(), anything(), $matcher, anything());
}
// Ensure matcher contains correct boost in 'qf' parameter
$matcher = new Hamcrest_Array_IsArrayContainingKeyValuePair(
new Hamcrest_Core_IsEqual('qf'),
new Hamcrest_Core_IsEqual('SearchUpdaterTest_Container_Field1^1.5 SearchUpdaterTest_Container_Field2^2.1 _text')
);
Phockito::verify($serviceMock)
->search('+term', anything(), anything(), $matcher, anything());
}
public function testHighlightQueryOnBoost()
{
$serviceMock = $this->getServiceMock();
Phockito::when($serviceMock)->search(anything(), anything(), anything(), anything(), anything())->return($this->getFakeRawSolrResponse());
function testHighlightQueryOnBoost() {
$serviceMock = $this->getServiceMock();
Phockito::when($serviceMock)->search(anything(), anything(), anything(), anything(), anything())->return($this->getFakeRawSolrResponse());
$index = new SolrIndexTest_FakeIndex();
$index->setService($serviceMock);
$index = new SolrIndexTest_FakeIndex();
$index->setService($serviceMock);
// Search without highlighting
$query = new SearchQuery();
$query->search(
'term',
null,
array('Field1' => 1.5, 'HasOneObject_Field1' => 3)
);
$index->search($query);
Phockito::verify(
$serviceMock)->search(
'+(Field1:term^1.5 OR HasOneObject_Field1:term^3)',
anything(),
anything(),
not(hasKeyInArray('hl.q')),
anything()
);
// Search without highlighting
$query = new SearchQuery();
$query->search(
'term',
null,
array('Field1' => 1.5, 'HasOneObject_Field1' => 3)
);
$index->search($query);
Phockito::verify(
$serviceMock)->search(
'+(Field1:term^1.5 OR HasOneObject_Field1:term^3)',
anything(),
anything(),
not(hasKeyInArray('hl.q')),
anything()
);
// Search with highlighting
$query = new SearchQuery();
$query->search(
'term',
null,
array('Field1' => 1.5, 'HasOneObject_Field1' => 3)
);
$index->search($query, -1, -1, array('hl' => true));
Phockito::verify(
$serviceMock)->search(
'+(Field1:term^1.5 OR HasOneObject_Field1:term^3)',
anything(),
anything(),
hasKeyInArray('hl.q'),
anything()
);
}
// Search with highlighting
$query = new SearchQuery();
$query->search(
'term',
null,
array('Field1' => 1.5, 'HasOneObject_Field1' => 3)
);
$index->search($query, -1, -1, array('hl' => true));
Phockito::verify(
$serviceMock)->search(
'+(Field1:term^1.5 OR HasOneObject_Field1:term^3)',
anything(),
anything(),
hasKeyInArray('hl.q'),
anything()
);
}
public function testIndexExcludesNullValues()
{
$serviceMock = $this->getServiceMock();
$index = new SolrIndexTest_FakeIndex();
$index->setService($serviceMock);
$obj = new SearchUpdaterTest_Container();
function testIndexExcludesNullValues() {
$serviceMock = $this->getServiceMock();
$index = new SolrIndexTest_FakeIndex();
$index->setService($serviceMock);
$obj = new SearchUpdaterTest_Container();
$obj->Field1 = 'Field1 val';
$obj->Field2 = null;
$obj->MyDate = null;
$docs = $index->add($obj);
$value = $docs[0]->getField('SearchUpdaterTest_Container_Field1');
$this->assertEquals('Field1 val', $value['value'], 'Writes non-NULL string fields');
$value = $docs[0]->getField('SearchUpdaterTest_Container_Field2');
$this->assertFalse($value, 'Ignores string fields if they are NULL');
$value = $docs[0]->getField('SearchUpdaterTest_Container_MyDate');
$this->assertFalse($value, 'Ignores date fields if they are NULL');
$obj->Field1 = 'Field1 val';
$obj->Field2 = null;
$obj->MyDate = null;
$docs = $index->add($obj);
$value = $docs[0]->getField('SearchUpdaterTest_Container_Field1');
$this->assertEquals('Field1 val', $value['value'], 'Writes non-NULL string fields');
$value = $docs[0]->getField('SearchUpdaterTest_Container_Field2');
$this->assertFalse($value, 'Ignores string fields if they are NULL');
$value = $docs[0]->getField('SearchUpdaterTest_Container_MyDate');
$this->assertFalse($value, 'Ignores date fields if they are NULL');
$obj->MyDate = '2010-12-30';
$docs = $index->add($obj);
$value = $docs[0]->getField('SearchUpdaterTest_Container_MyDate');
$this->assertEquals('2010-12-30T00:00:00Z', $value['value'], 'Writes non-NULL dates');
}
$obj->MyDate = '2010-12-30';
$docs = $index->add($obj);
$value = $docs[0]->getField('SearchUpdaterTest_Container_MyDate');
$this->assertEquals('2010-12-30T00:00:00Z', $value['value'], 'Writes non-NULL dates');
}
public function testAddFieldExtraOptions()
{
Config::inst()->nest();
Config::inst()->update('Director', 'environment_type', 'live'); // dev mode sets stored=true for everything
function testAddFieldExtraOptions() {
Config::inst()->nest();
Config::inst()->update('Director', 'environment_type', 'live'); // dev mode sets stored=true for everything
$index = new SolrIndexTest_FakeIndex();
$index = new SolrIndexTest_FakeIndex();
$defs = simplexml_load_string('<fields>' . $index->getFieldDefinitions() . '</fields>');
$defField1 = $defs->xpath('field[@name="SearchUpdaterTest_Container_Field1"]');
$this->assertEquals((string)$defField1[0]['stored'], 'false');
$defs = simplexml_load_string('<fields>' . $index->getFieldDefinitions() . '</fields>');
$defField1 = $defs->xpath('field[@name="SearchUpdaterTest_Container_Field1"]');
$this->assertEquals((string)$defField1[0]['stored'], 'false');
$index->addFilterField('Field1', null, array('stored' => 'true'));
$defs = simplexml_load_string('<fields>' . $index->getFieldDefinitions() . '</fields>');
$defField1 = $defs->xpath('field[@name="SearchUpdaterTest_Container_Field1"]');
$this->assertEquals((string)$defField1[0]['stored'], 'true');
$index->addFilterField('Field1', null, array('stored' => 'true'));
$defs = simplexml_load_string('<fields>' . $index->getFieldDefinitions() . '</fields>');
$defField1 = $defs->xpath('field[@name="SearchUpdaterTest_Container_Field1"]');
$this->assertEquals((string)$defField1[0]['stored'], 'true');
Config::inst()->unnest();
}
Config::inst()->unnest();
}
public function testAddAnalyzer()
{
$index = new SolrIndexTest_FakeIndex();
function testAddAnalyzer() {
$index = new SolrIndexTest_FakeIndex();
$defs = simplexml_load_string('<fields>' . $index->getFieldDefinitions() . '</fields>');
$defField1 = $defs->xpath('field[@name="SearchUpdaterTest_Container_Field1"]');
$analyzers = $defField1[0]->analyzer;
$this->assertFalse((bool)$analyzers);
$defs = simplexml_load_string('<fields>' . $index->getFieldDefinitions() . '</fields>');
$defField1 = $defs->xpath('field[@name="SearchUpdaterTest_Container_Field1"]');
$analyzers = $defField1[0]->analyzer;
$this->assertFalse((bool)$analyzers);
$index->addAnalyzer('Field1', 'charFilter', array('class' => 'solr.HTMLStripCharFilterFactory'));
$defs = simplexml_load_string('<fields>' . $index->getFieldDefinitions() . '</fields>');
$defField1 = $defs->xpath('field[@name="SearchUpdaterTest_Container_Field1"]');
$analyzers = $defField1[0]->analyzer;
$this->assertTrue((bool)$analyzers);
$this->assertEquals('solr.HTMLStripCharFilterFactory', $analyzers[0]->charFilter[0]['class']);
}
$index->addAnalyzer('Field1', 'charFilter', array('class' => 'solr.HTMLStripCharFilterFactory'));
$defs = simplexml_load_string('<fields>' . $index->getFieldDefinitions() . '</fields>');
$defField1 = $defs->xpath('field[@name="SearchUpdaterTest_Container_Field1"]');
$analyzers = $defField1[0]->analyzer;
$this->assertTrue((bool)$analyzers);
$this->assertEquals('solr.HTMLStripCharFilterFactory', $analyzers[0]->charFilter[0]['class']);
}
public function testAddCopyField()
{
$index = new SolrIndexTest_FakeIndex();
$index->addCopyField('sourceField', 'destField');
function testAddCopyField() {
$index = new SolrIndexTest_FakeIndex();
$index->addCopyField('sourceField', 'destField');
$defs = simplexml_load_string('<fields>' . $index->getCopyFieldDefinitions() . '</fields>');
$copyField = $defs->xpath('copyField');
$defs = simplexml_load_string('<fields>' . $index->getCopyFieldDefinitions() . '</fields>');
$copyField = $defs->xpath('copyField');
$this->assertEquals('sourceField', $copyField[0]['source']);
$this->assertEquals('destField', $copyField[0]['dest']);
}
$this->assertEquals('sourceField', $copyField[0]['source']);
$this->assertEquals('destField', $copyField[0]['dest']);
}
/**
* Tests the setting of the 'stored' flag
*/
public function testStoredFields()
{
// Test two fields
$index = new SolrIndexTest_FakeIndex2();
$index->addStoredField('Field1');
$index->addFulltextField('Field2');
$schema = $index->getFieldDefinitions();
$this->assertContains(
"<field name='SearchUpdaterTest_Container_Field1' type='text' indexed='true' stored='true'",
$schema
);
$this->assertContains(
"<field name='SearchUpdaterTest_Container_Field2' type='text' indexed='true' stored='false'",
$schema
);
/**
* Tests the setting of the 'stored' flag
*/
public function testStoredFields() {
// Test two fields
$index = new SolrIndexTest_FakeIndex2();
$index->addStoredField('Field1');
$index->addFulltextField('Field2');
$schema = $index->getFieldDefinitions();
$this->assertContains(
"<field name='SearchUpdaterTest_Container_Field1' type='text' indexed='true' stored='true'",
$schema
);
$this->assertContains(
"<field name='SearchUpdaterTest_Container_Field2' type='text' indexed='true' stored='false'",
$schema
);
// Test with addAllFulltextFields
$index2 = new SolrIndexTest_FakeIndex2();
$index2->addAllFulltextFields();
$index2->addStoredField('Field2');
$schema2 = $index2->getFieldDefinitions();
$this->assertContains(
"<field name='SearchUpdaterTest_Container_Field1' type='text' indexed='true' stored='false'",
$schema2
);
$this->assertContains(
"<field name='SearchUpdaterTest_Container_Field2' type='text' indexed='true' stored='true'",
$schema2
);
}
// Test with addAllFulltextFields
$index2 = new SolrIndexTest_FakeIndex2();
$index2->addAllFulltextFields();
$index2->addStoredField('Field2');
$schema2 = $index2->getFieldDefinitions();
$this->assertContains(
"<field name='SearchUpdaterTest_Container_Field1' type='text' indexed='true' stored='false'",
$schema2
);
$this->assertContains(
"<field name='SearchUpdaterTest_Container_Field2' type='text' indexed='true' stored='true'",
$schema2
);
}
/**
* @return Solr3Service
*/
protected function getServiceMock()
{
return Phockito::mock('Solr3Service');
}
/**
* @return Solr3Service
*/
protected function getServiceMock() {
return Phockito::mock('Solr3Service');
}
protected function getServiceSpy()
{
$serviceSpy = Phockito::spy('Solr3Service');
Phockito::when($serviceSpy)->_sendRawPost()->return($this->getFakeRawSolrResponse());
protected function getServiceSpy() {
$serviceSpy = Phockito::spy('Solr3Service');
Phockito::when($serviceSpy)->_sendRawPost()->return($this->getFakeRawSolrResponse());
return $serviceSpy;
}
return $serviceSpy;
}
protected function getFakeRawSolrResponse() {
return new Apache_Solr_Response(
new Apache_Solr_HttpTransport_Response(
null,
null,
'{}'
)
);
}
protected function getFakeRawSolrResponse()
{
return new Apache_Solr_Response(
new Apache_Solr_HttpTransport_Response(
null,
null,
'{}'
)
);
}
}
class SolrIndexTest_FakeIndex extends SolrIndex {
function init() {
$this->addClass('SearchUpdaterTest_Container');
class SolrIndexTest_FakeIndex extends SolrIndex
{
public function init()
{
$this->addClass('SearchUpdaterTest_Container');
$this->addFilterField('Field1');
$this->addFilterField('MyDate', 'Date');
$this->addFilterField('HasOneObject.Field1');
$this->addFilterField('HasManyObjects.Field1');
$this->addFilterField('ManyManyObjects.Field1');
}
$this->addFilterField('Field1');
$this->addFilterField('MyDate', 'Date');
$this->addFilterField('HasOneObject.Field1');
$this->addFilterField('HasManyObjects.Field1');
$this->addFilterField('ManyManyObjects.Field1');
}
}
class SolrIndexTest_FakeIndex2 extends SolrIndex {
protected function getStoredDefault() {
// Override isDev defaulting to stored
return 'false';
}
class SolrIndexTest_FakeIndex2 extends SolrIndex
{
protected function getStoredDefault()
{
// Override isDev defaulting to stored
return 'false';
}
function init() {
$this->addClass('SearchUpdaterTest_Container');
$this->addFilterField('MyDate', 'Date');
$this->addFilterField('HasOneObject.Field1');
$this->addFilterField('HasManyObjects.Field1');
$this->addFilterField('ManyManyObjects.Field1');
}
public function init()
{
$this->addClass('SearchUpdaterTest_Container');
$this->addFilterField('MyDate', 'Date');
$this->addFilterField('HasOneObject.Field1');
$this->addFilterField('HasManyObjects.Field1');
$this->addFilterField('ManyManyObjects.Field1');
}
}
class SolrIndexTest_BoostedIndex extends SolrIndex {
protected function getStoredDefault() {
// Override isDev defaulting to stored
return 'false';
}
class SolrIndexTest_BoostedIndex extends SolrIndex
{
protected function getStoredDefault()
{
// Override isDev defaulting to stored
return 'false';
}
function init() {
$this->addClass('SearchUpdaterTest_Container');
$this->addAllFulltextFields();
$this->setFieldBoosting('SearchUpdaterTest_Container_Field1', 1.5);
$this->addBoostedField('Field2', null, array(), 2.1);
}
public function init()
{
$this->addClass('SearchUpdaterTest_Container');
$this->addAllFulltextFields();
$this->setFieldBoosting('SearchUpdaterTest_Container_Field1', 1.5);
$this->addBoostedField('Field2', null, array(), 2.1);
}
}

314
tests/SolrIndexVersionedTest.php
View File

@ -1,165 +1,183 @@
<?php
if (class_exists('Phockito')) Phockito::include_hamcrest();
if (class_exists('Phockito')) {
Phockito::include_hamcrest();
}
class SolrIndexVersionedTest extends SapphireTest {
protected $oldMode = null;
protected static $index = null;
protected $extraDataObjects = array(
'SearchVariantVersionedTest_Item'
);
class SolrIndexVersionedTest extends SapphireTest
{
protected $oldMode = null;
protected static $index = null;
protected $extraDataObjects = array(
'SearchVariantVersionedTest_Item'
);
public function setUp() {
public function setUp()
{
parent::setUp();
if (!class_exists('Phockito')) {
$this->skipTest = true;
return $this->markTestSkipped("These tests need the Phockito module installed to run");
}
parent::setUp();
if (!class_exists('Phockito')) {
$this->skipTest = true;
return $this->markTestSkipped("These tests need the Phockito module installed to run");
}
// Check versioned available
if (!class_exists('Versioned')) {
$this->skipTest = true;
return $this->markTestSkipped('The versioned decorator is not installed');
}
// Check versioned available
if(!class_exists('Versioned')) {
$this->skipTest = true;
return $this->markTestSkipped('The versioned decorator is not installed');
}
if (self::$index === null) {
self::$index = singleton('SolrVersionedTest_Index');
}
if (self::$index === null) self::$index = singleton('SolrVersionedTest_Index');
SearchUpdater::bind_manipulation_capture();
SearchUpdater::bind_manipulation_capture();
Config::nest();
Config::nest();
Config::inst()->update('Injector', 'SearchUpdateProcessor', array(
'class' => 'SearchUpdateImmediateProcessor'
));
Config::inst()->update('Injector', 'SearchUpdateProcessor', array(
'class' => 'SearchUpdateImmediateProcessor'
));
FullTextSearch::force_index_list(self::$index);
SearchUpdater::clear_dirty_indexes();
$this->oldMode = Versioned::get_reading_mode();
Versioned::reading_stage('Stage');
}
public function tearDown()
{
Versioned::set_reading_mode($this->oldMode);
Config::unnest();
parent::tearDown();
}
FullTextSearch::force_index_list(self::$index);
SearchUpdater::clear_dirty_indexes();
$this->oldMode = Versioned::get_reading_mode();
Versioned::reading_stage('Stage');
}
public function tearDown() {
Versioned::set_reading_mode($this->oldMode);
Config::unnest();
parent::tearDown();
}
protected function getServiceMock() {
return Phockito::mock('Solr3Service');
}
protected function getExpectedDocumentId($id, $stage) {
// Prevent subsites from breaking tests
$subsites = class_exists('Subsite') ? '"SearchVariantSubsites":"0",' : '';
return $id.'-SiteTree-{'.$subsites.'"SearchVariantVersioned":"'.$stage.'"}';
}
public function testPublishing() {
// Setup mocks
$serviceMock = $this->getServiceMock();
self::$index->setService($serviceMock);
// Check that write updates Stage
Versioned::reading_stage('Stage');
Phockito::reset($serviceMock);
$item = new SearchVariantVersionedTest_Item(array('Title' => 'Foo'));
$item->write();
SearchUpdater::flush_dirty_indexes();
$doc = new SolrDocumentMatcher(array(
'_documentid' => $this->getExpectedDocumentId($item->ID, 'Stage'),
'ClassName' => 'SearchVariantVersionedTest_Item'
));
Phockito::verify($serviceMock)->addDocument($doc);
// Check that write updates Live
Versioned::reading_stage('Stage');
Phockito::reset($serviceMock);
$item = new SearchVariantVersionedTest_Item(array('Title' => 'Bar'));
$item->write();
$item->publish('Stage', 'Live');
SearchUpdater::flush_dirty_indexes();
$doc = new SolrDocumentMatcher(array(
'_documentid' => $this->getExpectedDocumentId($item->ID, 'Live'),
'ClassName' => 'SearchVariantVersionedTest_Item'
));
Phockito::verify($serviceMock)->addDocument($doc);
}
public function testDelete() {
// Setup mocks
$serviceMock = $this->getServiceMock();
self::$index->setService($serviceMock);
// Delete the live record (not the stage)
Versioned::reading_stage('Stage');
Phockito::reset($serviceMock);
$item = new SearchVariantVersionedTest_Item(array('Title' => 'Too'));
$item->write();
$item->publish('Stage', 'Live');
Versioned::reading_stage('Live');
$id = $item->ID;
$item->delete();
SearchUpdater::flush_dirty_indexes();
Phockito::verify($serviceMock, 1)
->deleteById($this->getExpectedDocumentId($id, 'Live'));
Phockito::verify($serviceMock, 0)
->deleteById($this->getExpectedDocumentId($id, 'Stage'));
// Delete the stage record
Versioned::reading_stage('Stage');
Phockito::reset($serviceMock);
$item = new SearchVariantVersionedTest_Item(array('Title' => 'Too'));
$item->write();
$item->publish('Stage', 'Live');
$id = $item->ID;
$item->delete();
SearchUpdater::flush_dirty_indexes();
Phockito::verify($serviceMock, 1)
->deleteById($this->getExpectedDocumentId($id, 'Stage'));
Phockito::verify($serviceMock, 0)
->deleteById($this->getExpectedDocumentId($id, 'Live'));
}
protected function getServiceMock()
{
return Phockito::mock('Solr3Service');
}
protected function getExpectedDocumentId($id, $stage)
{
// Prevent subsites from breaking tests
$subsites = class_exists('Subsite') ? '"SearchVariantSubsites":"0",' : '';
return $id.'-SiteTree-{'.$subsites.'"SearchVariantVersioned":"'.$stage.'"}';
}
public function testPublishing()
{
// Setup mocks
$serviceMock = $this->getServiceMock();
self::$index->setService($serviceMock);
// Check that write updates Stage
Versioned::reading_stage('Stage');
Phockito::reset($serviceMock);
$item = new SearchVariantVersionedTest_Item(array('Title' => 'Foo'));
$item->write();
SearchUpdater::flush_dirty_indexes();
$doc = new SolrDocumentMatcher(array(
'_documentid' => $this->getExpectedDocumentId($item->ID, 'Stage'),
'ClassName' => 'SearchVariantVersionedTest_Item'
));
Phockito::verify($serviceMock)->addDocument($doc);
// Check that write updates Live
Versioned::reading_stage('Stage');
Phockito::reset($serviceMock);
$item = new SearchVariantVersionedTest_Item(array('Title' => 'Bar'));
$item->write();
$item->publish('Stage', 'Live');
SearchUpdater::flush_dirty_indexes();
$doc = new SolrDocumentMatcher(array(
'_documentid' => $this->getExpectedDocumentId($item->ID, 'Live'),
'ClassName' => 'SearchVariantVersionedTest_Item'
));
Phockito::verify($serviceMock)->addDocument($doc);
}
public function testDelete()
{
// Setup mocks
$serviceMock = $this->getServiceMock();
self::$index->setService($serviceMock);
// Delete the live record (not the stage)
Versioned::reading_stage('Stage');
Phockito::reset($serviceMock);
$item = new SearchVariantVersionedTest_Item(array('Title' => 'Too'));
$item->write();
$item->publish('Stage', 'Live');
Versioned::reading_stage('Live');
$id = $item->ID;
$item->delete();
SearchUpdater::flush_dirty_indexes();
Phockito::verify($serviceMock, 1)
->deleteById($this->getExpectedDocumentId($id, 'Live'));
Phockito::verify($serviceMock, 0)
->deleteById($this->getExpectedDocumentId($id, 'Stage'));
// Delete the stage record
Versioned::reading_stage('Stage');
Phockito::reset($serviceMock);
$item = new SearchVariantVersionedTest_Item(array('Title' => 'Too'));
$item->write();
$item->publish('Stage', 'Live');
$id = $item->ID;
$item->delete();
SearchUpdater::flush_dirty_indexes();
Phockito::verify($serviceMock, 1)
->deleteById($this->getExpectedDocumentId($id, 'Stage'));
Phockito::verify($serviceMock, 0)
->deleteById($this->getExpectedDocumentId($id, 'Live'));
}
}
class SolrVersionedTest_Index extends SolrIndex {
function init() {
$this->addClass('SearchVariantVersionedTest_Item');
$this->addFilterField('TestText');
}
class SolrVersionedTest_Index extends SolrIndex
{
public function init()
{
$this->addClass('SearchVariantVersionedTest_Item');
$this->addFilterField('TestText');
}
}
if (!class_exists('Phockito')) return;
class SolrDocumentMatcher extends Hamcrest_BaseMatcher {
protected $properties;
public function __construct($properties) {
$this->properties = $properties;
}
public function describeTo(\Hamcrest_Description $description) {
$description->appendText('Apache_Solr_Document with properties '.var_export($this->properties, true));
}
public function matches($item) {
if(! ($item instanceof Apache_Solr_Document)) return false;
foreach($this->properties as $key => $value) {
if($item->{$key} != $value) return false;
}
return true;
}
if (!class_exists('Phockito')) {
return;
}
class SolrDocumentMatcher extends Hamcrest_BaseMatcher
{
protected $properties;
public function __construct($properties)
{
$this->properties = $properties;
}
public function describeTo(\Hamcrest_Description $description)
{
$description->appendText('Apache_Solr_Document with properties '.var_export($this->properties, true));
}
public function matches($item)
{
if (! ($item instanceof Apache_Solr_Document)) {
return false;
}
foreach ($this->properties as $key => $value) {
if ($item->{$key} != $value) {
return false;
}
}
return true;
}
}

370
tests/SolrReindexQueuedTest.php
View File

@ -3,217 +3,227 @@
/**
* Additional tests of solr reindexing processes when run with queuedjobs
*/
class SolrReindexQueuedTest extends SapphireTest {
class SolrReindexQueuedTest extends SapphireTest
{
protected $usesDatabase = true;
protected $usesDatabase = true;
protected $extraDataObjects = array(
'SolrReindexTest_Item'
);
protected $extraDataObjects = array(
'SolrReindexTest_Item'
);
/**
* Forced index for testing
*
* @var SolrReindexTest_Index
*/
protected $index = null;
/**
* Forced index for testing
*
* @var SolrReindexTest_Index
*/
protected $index = null;
/**
* Mock service
*
* @var SolrService
*/
protected $service = null;
/**
* Mock service
*
* @var SolrService
*/
protected $service = null;
public function setUp()
{
parent::setUp();
public function setUp() {
parent::setUp();
if (!class_exists('Phockito')) {
$this->skipTest = true;
return $this->markTestSkipped("These tests need the Phockito module installed to run");
}
if (!class_exists('Phockito')) {
$this->skipTest = true;
return $this->markTestSkipped("These tests need the Phockito module installed to run");
}
if (!interface_exists('QueuedJob')) {
$this->skipTest = true;
return $this->markTestSkipped("These tests need the QueuedJobs module installed to run");
}
if(!interface_exists('QueuedJob')) {
$this->skipTest = true;
return $this->markTestSkipped("These tests need the QueuedJobs module installed to run");
}
// Set queued handler for reindex
Config::inst()->update('Injector', 'SolrReindexHandler', array(
'class' => 'SolrReindexQueuedHandler'
));
Injector::inst()->registerService(new SolrReindexQueuedHandler(), 'SolrReindexHandler');
// Set queued handler for reindex
Config::inst()->update('Injector', 'SolrReindexHandler', array(
'class' => 'SolrReindexQueuedHandler'
));
Injector::inst()->registerService(new SolrReindexQueuedHandler(), 'SolrReindexHandler');
// Set test variant
SolrReindexTest_Variant::enable();
// Set test variant
SolrReindexTest_Variant::enable();
// Set index list
$this->service = $this->getServiceMock();
$this->index = singleton('SolrReindexTest_Index');
$this->index->setService($this->service);
FullTextSearch::force_index_list($this->index);
}
// Set index list
$this->service = $this->getServiceMock();
$this->index = singleton('SolrReindexTest_Index');
$this->index->setService($this->service);
FullTextSearch::force_index_list($this->index);
}
/**
* Populate database with dummy dataset
*
* @param int $number Number of records to create in each variant
*/
protected function createDummyData($number)
{
// Populate dataobjects. Use truncate to generate predictable IDs
DB::query('TRUNCATE "SolrReindexTest_Item"');
/**
* Populate database with dummy dataset
*
* @param int $number Number of records to create in each variant
*/
protected function createDummyData($number) {
// Populate dataobjects. Use truncate to generate predictable IDs
DB::query('TRUNCATE "SolrReindexTest_Item"');
// Note that we don't create any records in variant = 2, to represent a variant
// that should be cleared without any re-indexes performed
foreach (array(0, 1) as $variant) {
for ($i = 1; $i <= $number; $i++) {
$item = new SolrReindexTest_Item();
$item->Variant = $variant;
$item->Title = "Item $variant / $i";
$item->write();
}
}
}
// Note that we don't create any records in variant = 2, to represent a variant
// that should be cleared without any re-indexes performed
foreach(array(0, 1) as $variant) {
for($i = 1; $i <= $number; $i++) {
$item = new SolrReindexTest_Item();
$item->Variant = $variant;
$item->Title = "Item $variant / $i";
$item->write();
}
}
}
/**
* Mock service
*
* @return SolrService
*/
protected function getServiceMock()
{
return Phockito::mock('Solr4Service');
}
/**
* Mock service
*
* @return SolrService
*/
protected function getServiceMock() {
return Phockito::mock('Solr4Service');
}
public function tearDown()
{
FullTextSearch::force_index_list();
SolrReindexTest_Variant::disable();
parent::tearDown();
}
public function tearDown() {
FullTextSearch::force_index_list();
SolrReindexTest_Variant::disable();
parent::tearDown();
}
/**
* Get the reindex handler
*
* @return SolrReindexHandler
*/
protected function getHandler()
{
return Injector::inst()->get('SolrReindexHandler');
}
/**
* Get the reindex handler
*
* @return SolrReindexHandler
*/
protected function getHandler() {
return Injector::inst()->get('SolrReindexHandler');
}
/**
* @return SolrReindexQueuedTest_Service
*/
protected function getQueuedJobService()
{
return singleton('SolrReindexQueuedTest_Service');
}
/**
* @return SolrReindexQueuedTest_Service
*/
protected function getQueuedJobService() {
return singleton('SolrReindexQueuedTest_Service');
}
/**
* Test that reindex will generate a top top level queued job, and executing this will perform
* the necessary initialisation of the grouped queued jobs
*/
public function testReindexSegmentsGroups()
{
$this->createDummyData(18);
/**
* Test that reindex will generate a top top level queued job, and executing this will perform
* the necessary initialisation of the grouped queued jobs
*/
public function testReindexSegmentsGroups() {
$this->createDummyData(18);
// Create pre-existing jobs
$this->getQueuedJobService()->queueJob(new SolrReindexQueuedJob());
$this->getQueuedJobService()->queueJob(new SolrReindexGroupQueuedJob());
$this->getQueuedJobService()->queueJob(new SolrReindexGroupQueuedJob());
// Create pre-existing jobs
$this->getQueuedJobService()->queueJob(new SolrReindexQueuedJob());
$this->getQueuedJobService()->queueJob(new SolrReindexGroupQueuedJob());
$this->getQueuedJobService()->queueJob(new SolrReindexGroupQueuedJob());
// Initiate re-index
$logger = new SolrReindexTest_RecordingLogger();
$this->getHandler()->triggerReindex($logger, 6, 'Solr_Reindex');
// Initiate re-index
$logger = new SolrReindexTest_RecordingLogger();
$this->getHandler()->triggerReindex($logger, 6, 'Solr_Reindex');
// Old jobs should be cancelled
$this->assertEquals(1, $logger->countMessages('Cancelled 1 re-index tasks and 2 re-index groups'));
$this->assertEquals(1, $logger->countMessages('Queued Solr Reindex Job'));
// Old jobs should be cancelled
$this->assertEquals(1, $logger->countMessages('Cancelled 1 re-index tasks and 2 re-index groups'));
$this->assertEquals(1, $logger->countMessages('Queued Solr Reindex Job'));
// Next job should be queue job
$job = $this->getQueuedJobService()->getNextJob();
$this->assertInstanceOf('SolrReindexQueuedJob', $job);
$this->assertEquals(6, $job->getBatchSize());
// Next job should be queue job
$job = $this->getQueuedJobService()->getNextJob();
$this->assertInstanceOf('SolrReindexQueuedJob', $job);
$this->assertEquals(6, $job->getBatchSize());
// Test that necessary items are created
$logger->clear();
$job->setLogger($logger);
$job->process();
// Test that necessary items are created
$logger->clear();
$job->setLogger($logger);
$job->process();
// Deletes are performed in the main task prior to individual groups being processed
// 18 records means 3 groups of 6 in each variant (6 total)
Phockito::verify($this->service, 2)
->deleteByQuery(anything());
$this->assertEquals(1, $logger->countMessages('Beginning init of reindex'));
$this->assertEquals(6, $logger->countMessages('Queued Solr Reindex Group '));
$this->assertEquals(3, $logger->countMessages(' of SolrReindexTest_Item in {"SolrReindexTest_Variant":"0"}'));
$this->assertEquals(3, $logger->countMessages(' of SolrReindexTest_Item in {"SolrReindexTest_Variant":"1"}'));
$this->assertEquals(1, $logger->countMessages('Completed init of reindex'));
// Deletes are performed in the main task prior to individual groups being processed
// 18 records means 3 groups of 6 in each variant (6 total)
Phockito::verify($this->service, 2)
->deleteByQuery(anything());
$this->assertEquals(1, $logger->countMessages('Beginning init of reindex'));
$this->assertEquals(6, $logger->countMessages('Queued Solr Reindex Group '));
$this->assertEquals(3, $logger->countMessages(' of SolrReindexTest_Item in {"SolrReindexTest_Variant":"0"}'));
$this->assertEquals(3, $logger->countMessages(' of SolrReindexTest_Item in {"SolrReindexTest_Variant":"1"}'));
$this->assertEquals(1, $logger->countMessages('Completed init of reindex'));
// Test that invalid classes are removed
$this->assertNotEmpty($logger->getMessages('Clearing obsolete classes from SolrReindexTest_Index'));
Phockito::verify($this->service, 1)
->deleteByQuery('-(ClassHierarchy:SolrReindexTest_Item)');
// Test that invalid classes are removed
$this->assertNotEmpty($logger->getMessages('Clearing obsolete classes from SolrReindexTest_Index'));
Phockito::verify($this->service, 1)
->deleteByQuery('-(ClassHierarchy:SolrReindexTest_Item)');
// Test that valid classes in invalid variants are removed
$this->assertNotEmpty($logger->getMessages(
'Clearing all records of type SolrReindexTest_Item in the current state: {"SolrReindexTest_Variant":"2"}'
));
Phockito::verify($this->service, 1)
->deleteByQuery('+(ClassHierarchy:SolrReindexTest_Item) +(_testvariant:"2")');
}
// Test that valid classes in invalid variants are removed
$this->assertNotEmpty($logger->getMessages(
'Clearing all records of type SolrReindexTest_Item in the current state: {"SolrReindexTest_Variant":"2"}'
));
Phockito::verify($this->service, 1)
->deleteByQuery('+(ClassHierarchy:SolrReindexTest_Item) +(_testvariant:"2")');
}
/**
* Test index processing on individual groups
*/
public function testRunGroup()
{
$this->createDummyData(18);
/**
* Test index processing on individual groups
*/
public function testRunGroup() {
$this->createDummyData(18);
// Just do what the SolrReindexQueuedJob would do to create each sub
$logger = new SolrReindexTest_RecordingLogger();
$this->getHandler()->runReindex($logger, 6, 'Solr_Reindex');
// Just do what the SolrReindexQueuedJob would do to create each sub
$logger = new SolrReindexTest_RecordingLogger();
$this->getHandler()->runReindex($logger, 6, 'Solr_Reindex');
// Assert jobs are created
$this->assertEquals(6, $logger->countMessages('Queued Solr Reindex Group'));
// Assert jobs are created
$this->assertEquals(6, $logger->countMessages('Queued Solr Reindex Group'));
// Check next job is a group queued job
$job = $this->getQueuedJobService()->getNextJob();
$this->assertInstanceOf('SolrReindexGroupQueuedJob', $job);
$this->assertEquals(
'Solr Reindex Group (1/3) of SolrReindexTest_Item in {"SolrReindexTest_Variant":"0"}',
$job->getTitle()
);
// Check next job is a group queued job
$job = $this->getQueuedJobService()->getNextJob();
$this->assertInstanceOf('SolrReindexGroupQueuedJob', $job);
$this->assertEquals(
'Solr Reindex Group (1/3) of SolrReindexTest_Item in {"SolrReindexTest_Variant":"0"}',
$job->getTitle()
);
// Running this job performs the necessary reindex
$logger->clear();
$job->setLogger($logger);
$job->process();
// Running this job performs the necessary reindex
$logger->clear();
$job->setLogger($logger);
$job->process();
// Check tasks completed (as per non-queuedjob version)
$this->assertEquals(1, $logger->countMessages('Beginning reindex group'));
$this->assertEquals(1, $logger->countMessages('Adding SolrReindexTest_Item'));
$this->assertEquals(1, $logger->countMessages('Queuing commit on all changes'));
$this->assertEquals(1, $logger->countMessages('Completed reindex group'));
// Check tasks completed (as per non-queuedjob version)
$this->assertEquals(1, $logger->countMessages('Beginning reindex group'));
$this->assertEquals(1, $logger->countMessages('Adding SolrReindexTest_Item'));
$this->assertEquals(1, $logger->countMessages('Queuing commit on all changes'));
$this->assertEquals(1, $logger->countMessages('Completed reindex group'));
// Check IDs
$idMessage = $logger->filterMessages('Updated ');
$this->assertNotEmpty(preg_match('/^Updated (?<ids>[,\d]+)/i', $idMessage[0], $matches));
$ids = array_unique(explode(',', $matches['ids']));
$this->assertEquals(6, count($ids));
foreach($ids as $id) {
// Each id should be % 3 == 0
$this->assertEquals(0, $id % 3, "ID $id Should match pattern ID % 3 = 0");
}
}
// Check IDs
$idMessage = $logger->filterMessages('Updated ');
$this->assertNotEmpty(preg_match('/^Updated (?<ids>[,\d]+)/i', $idMessage[0], $matches));
$ids = array_unique(explode(',', $matches['ids']));
$this->assertEquals(6, count($ids));
foreach ($ids as $id) {
// Each id should be % 3 == 0
$this->assertEquals(0, $id % 3, "ID $id Should match pattern ID % 3 = 0");
}
}
}
if(!class_exists('QueuedJobService')) return;
if (!class_exists('QueuedJobService')) {
return;
}
class SolrReindexQueuedTest_Service extends QueuedJobService implements TestOnly {
/**
* @return QueuedJob
*/
public function getNextJob() {
$job = $this->getNextPendingJob();
return $this->initialiseJob($job);
}
}
class SolrReindexQueuedTest_Service extends QueuedJobService implements TestOnly
{
/**
* @return QueuedJob
*/
public function getNextJob()
{
$job = $this->getNextPendingJob();
return $this->initialiseJob($job);
}
}

893
tests/SolrReindexTest.php
View File

@ -5,335 +5,348 @@ use Monolog\Handler\HandlerInterface;
use Monolog\Logger;
use Psr\Log\LoggerInterface;
if (class_exists('Phockito')) Phockito::include_hamcrest();
if (class_exists('Phockito')) {
Phockito::include_hamcrest();
}
class SolrReindexTest extends SapphireTest {
class SolrReindexTest extends SapphireTest
{
protected $usesDatabase = true;
protected $usesDatabase = true;
protected $extraDataObjects = array(
'SolrReindexTest_Item'
);
protected $extraDataObjects = array(
'SolrReindexTest_Item'
);
/**
* Forced index for testing
*
* @var SolrReindexTest_Index
*/
protected $index = null;
/**
* Forced index for testing
*
* @var SolrReindexTest_Index
*/
protected $index = null;
/**
* Mock service
*
* @var SolrService
*/
protected $service = null;
/**
* Mock service
*
* @var SolrService
*/
protected $service = null;
public function setUp()
{
parent::setUp();
public function setUp() {
parent::setUp();
if (!class_exists('Phockito')) {
$this->skipTest = true;
return $this->markTestSkipped("These tests need the Phockito module installed to run");
}
// Set test handler for reindex
Config::inst()->update('Injector', 'SolrReindexHandler', array(
'class' => 'SolrReindexTest_TestHandler'
));
Injector::inst()->registerService(new SolrReindexTest_TestHandler(), 'SolrReindexHandler');
if (!class_exists('Phockito')) {
$this->skipTest = true;
return $this->markTestSkipped("These tests need the Phockito module installed to run");
}
// Set test handler for reindex
Config::inst()->update('Injector', 'SolrReindexHandler', array(
'class' => 'SolrReindexTest_TestHandler'
));
Injector::inst()->registerService(new SolrReindexTest_TestHandler(), 'SolrReindexHandler');
// Set test variant
SolrReindexTest_Variant::enable();
// Set test variant
SolrReindexTest_Variant::enable();
// Set index list
$this->service = $this->getServiceMock();
$this->index = singleton('SolrReindexTest_Index');
$this->index->setService($this->service);
FullTextSearch::force_index_list($this->index);
}
// Set index list
$this->service = $this->getServiceMock();
$this->index = singleton('SolrReindexTest_Index');
$this->index->setService($this->service);
FullTextSearch::force_index_list($this->index);
}
/**
* Populate database with dummy dataset
*
* @param int $number Number of records to create in each variant
*/
protected function createDummyData($number)
{
// Populate dataobjects. Use truncate to generate predictable IDs
DB::query('TRUNCATE "SolrReindexTest_Item"');
/**
* Populate database with dummy dataset
*
* @param int $number Number of records to create in each variant
*/
protected function createDummyData($number) {
// Populate dataobjects. Use truncate to generate predictable IDs
DB::query('TRUNCATE "SolrReindexTest_Item"');
// Note that we don't create any records in variant = 2, to represent a variant
// that should be cleared without any re-indexes performed
foreach (array(0, 1) as $variant) {
for ($i = 1; $i <= $number; $i++) {
$item = new SolrReindexTest_Item();
$item->Variant = $variant;
$item->Title = "Item $variant / $i";
$item->write();
}
}
}
// Note that we don't create any records in variant = 2, to represent a variant
// that should be cleared without any re-indexes performed
foreach(array(0, 1) as $variant) {
for($i = 1; $i <= $number; $i++) {
$item = new SolrReindexTest_Item();
$item->Variant = $variant;
$item->Title = "Item $variant / $i";
$item->write();
}
}
}
/**
* Mock service
*
* @return SolrService
*/
protected function getServiceMock()
{
return Phockito::mock('Solr4Service');
}
/**
* Mock service
*
* @return SolrService
*/
protected function getServiceMock() {
return Phockito::mock('Solr4Service');
}
public function tearDown()
{
FullTextSearch::force_index_list();
SolrReindexTest_Variant::disable();
parent::tearDown();
}
public function tearDown() {
FullTextSearch::force_index_list();
SolrReindexTest_Variant::disable();
parent::tearDown();
}
/**
* Get the reindex handler
*
* @return SolrReindexHandler
*/
protected function getHandler()
{
return Injector::inst()->get('SolrReindexHandler');
}
/**
* Get the reindex handler
*
* @return SolrReindexHandler
*/
protected function getHandler() {
return Injector::inst()->get('SolrReindexHandler');
}
/**
* Ensure the test variant is up and running properly
*/
public function testVariant()
{
// State defaults to 0
$variant = SearchVariant::current_state();
$this->assertEquals(
array(
"SolrReindexTest_Variant" => "0"
),
$variant
);
/**
* Ensure the test variant is up and running properly
*/
public function testVariant() {
// State defaults to 0
$variant = SearchVariant::current_state();
$this->assertEquals(
array(
"SolrReindexTest_Variant" => "0"
),
$variant
);
// All states enumerated
$allStates = iterator_to_array(SearchVariant::reindex_states());
$this->assertEquals(
array(
array(
"SolrReindexTest_Variant" => "0"
),
array(
"SolrReindexTest_Variant" => "1"
),
array(
"SolrReindexTest_Variant" => "2"
)
),
$allStates
);
// All states enumerated
$allStates = iterator_to_array(SearchVariant::reindex_states());
$this->assertEquals(
array(
array(
"SolrReindexTest_Variant" => "0"
),
array(
"SolrReindexTest_Variant" => "1"
),
array(
"SolrReindexTest_Variant" => "2"
)
),
$allStates
);
// Check correct items created and that filtering on variant works
$this->createDummyData(120);
SolrReindexTest_Variant::set_current(2);
$this->assertEquals(0, SolrReindexTest_Item::get()->count());
SolrReindexTest_Variant::set_current(1);
$this->assertEquals(120, SolrReindexTest_Item::get()->count());
SolrReindexTest_Variant::set_current(0);
$this->assertEquals(120, SolrReindexTest_Item::get()->count());
SolrReindexTest_Variant::disable();
$this->assertEquals(240, SolrReindexTest_Item::get()->count());
}
// Check correct items created and that filtering on variant works
$this->createDummyData(120);
SolrReindexTest_Variant::set_current(2);
$this->assertEquals(0, SolrReindexTest_Item::get()->count());
SolrReindexTest_Variant::set_current(1);
$this->assertEquals(120, SolrReindexTest_Item::get()->count());
SolrReindexTest_Variant::set_current(0);
$this->assertEquals(120, SolrReindexTest_Item::get()->count());
SolrReindexTest_Variant::disable();
$this->assertEquals(240, SolrReindexTest_Item::get()->count());
}
/**
* Given the invocation of a new re-index with a given set of data, ensure that the necessary
* list of groups are created and segmented for each state
*
* Test should work fine with any variants (versioned, subsites, etc) specified
*/
public function testReindexSegmentsGroups() {
$this->createDummyData(120);
/**
* Given the invocation of a new re-index with a given set of data, ensure that the necessary
* list of groups are created and segmented for each state
*
* Test should work fine with any variants (versioned, subsites, etc) specified
*/
public function testReindexSegmentsGroups()
{
$this->createDummyData(120);
// Initiate re-index
$logger = new SolrReindexTest_RecordingLogger();
$this->getHandler()->runReindex($logger, 21, 'Solr_Reindex');
// Initiate re-index
$logger = new SolrReindexTest_RecordingLogger();
$this->getHandler()->runReindex($logger, 21, 'Solr_Reindex');
// Test that invalid classes are removed
$this->assertNotEmpty($logger->getMessages('Clearing obsolete classes from SolrReindexTest_Index'));
Phockito::verify($this->service, 1)
->deleteByQuery('-(ClassHierarchy:SolrReindexTest_Item)');
// Test that invalid classes are removed
$this->assertNotEmpty($logger->getMessages('Clearing obsolete classes from SolrReindexTest_Index'));
Phockito::verify($this->service, 1)
->deleteByQuery('-(ClassHierarchy:SolrReindexTest_Item)');
// Test that valid classes in invalid variants are removed
$this->assertNotEmpty($logger->getMessages(
'Clearing all records of type SolrReindexTest_Item in the current state: {"SolrReindexTest_Variant":"2"}'
));
Phockito::verify($this->service, 1)
->deleteByQuery('+(ClassHierarchy:SolrReindexTest_Item) +(_testvariant:"2")');
// Test that valid classes in invalid variants are removed
$this->assertNotEmpty($logger->getMessages(
'Clearing all records of type SolrReindexTest_Item in the current state: {"SolrReindexTest_Variant":"2"}'
));
Phockito::verify($this->service, 1)
->deleteByQuery('+(ClassHierarchy:SolrReindexTest_Item) +(_testvariant:"2")');
// 120x2 grouped into groups of 21 results in 12 groups
$this->assertEquals(12, $logger->countMessages('Called processGroup with '));
$this->assertEquals(6, $logger->countMessages('{"SolrReindexTest_Variant":"0"}'));
$this->assertEquals(6, $logger->countMessages('{"SolrReindexTest_Variant":"1"}'));
// 120x2 grouped into groups of 21 results in 12 groups
$this->assertEquals(12, $logger->countMessages('Called processGroup with '));
$this->assertEquals(6, $logger->countMessages('{"SolrReindexTest_Variant":"0"}'));
$this->assertEquals(6, $logger->countMessages('{"SolrReindexTest_Variant":"1"}'));
// Given that there are two variants, there should be two group ids of each number
$this->assertEquals(2, $logger->countMessages(' SolrReindexTest_Item, group 0 of 6'));
$this->assertEquals(2, $logger->countMessages(' SolrReindexTest_Item, group 1 of 6'));
$this->assertEquals(2, $logger->countMessages(' SolrReindexTest_Item, group 2 of 6'));
$this->assertEquals(2, $logger->countMessages(' SolrReindexTest_Item, group 3 of 6'));
$this->assertEquals(2, $logger->countMessages(' SolrReindexTest_Item, group 4 of 6'));
$this->assertEquals(2, $logger->countMessages(' SolrReindexTest_Item, group 5 of 6'));
// Given that there are two variants, there should be two group ids of each number
$this->assertEquals(2, $logger->countMessages(' SolrReindexTest_Item, group 0 of 6'));
$this->assertEquals(2, $logger->countMessages(' SolrReindexTest_Item, group 1 of 6'));
$this->assertEquals(2, $logger->countMessages(' SolrReindexTest_Item, group 2 of 6'));
$this->assertEquals(2, $logger->countMessages(' SolrReindexTest_Item, group 3 of 6'));
$this->assertEquals(2, $logger->countMessages(' SolrReindexTest_Item, group 4 of 6'));
$this->assertEquals(2, $logger->countMessages(' SolrReindexTest_Item, group 5 of 6'));
// Check various group sizes
$logger->clear();
$this->getHandler()->runReindex($logger, 120, 'Solr_Reindex');
$this->assertEquals(2, $logger->countMessages('Called processGroup with '));
$logger->clear();
$this->getHandler()->runReindex($logger, 119, 'Solr_Reindex');
$this->assertEquals(4, $logger->countMessages('Called processGroup with '));
$logger->clear();
$this->getHandler()->runReindex($logger, 121, 'Solr_Reindex');
$this->assertEquals(2, $logger->countMessages('Called processGroup with '));
$logger->clear();
$this->getHandler()->runReindex($logger, 2, 'Solr_Reindex');
$this->assertEquals(120, $logger->countMessages('Called processGroup with '));
}
// Check various group sizes
$logger->clear();
$this->getHandler()->runReindex($logger, 120, 'Solr_Reindex');
$this->assertEquals(2, $logger->countMessages('Called processGroup with '));
$logger->clear();
$this->getHandler()->runReindex($logger, 119, 'Solr_Reindex');
$this->assertEquals(4, $logger->countMessages('Called processGroup with '));
$logger->clear();
$this->getHandler()->runReindex($logger, 121, 'Solr_Reindex');
$this->assertEquals(2, $logger->countMessages('Called processGroup with '));
$logger->clear();
$this->getHandler()->runReindex($logger, 2, 'Solr_Reindex');
$this->assertEquals(120, $logger->countMessages('Called processGroup with '));
}
/**
* Test index processing on individual groups
*/
public function testRunGroup() {
$this->createDummyData(120);
$logger = new SolrReindexTest_RecordingLogger();
/**
* Test index processing on individual groups
*/
public function testRunGroup()
{
$this->createDummyData(120);
$logger = new SolrReindexTest_RecordingLogger();
// Initiate re-index of third group (index 2 of 6)
$state = array('SolrReindexTest_Variant' => '1');
$this->getHandler()->runGroup($logger, $this->index, $state, 'SolrReindexTest_Item', 6, 2);
$idMessage = $logger->filterMessages('Updated ');
$this->assertNotEmpty(preg_match('/^Updated (?<ids>[,\d]+)/i', $idMessage[0], $matches));
$ids = array_unique(explode(',', $matches['ids']));
// Initiate re-index of third group (index 2 of 6)
$state = array('SolrReindexTest_Variant' => '1');
$this->getHandler()->runGroup($logger, $this->index, $state, 'SolrReindexTest_Item', 6, 2);
$idMessage = $logger->filterMessages('Updated ');
$this->assertNotEmpty(preg_match('/^Updated (?<ids>[,\d]+)/i', $idMessage[0], $matches));
$ids = array_unique(explode(',', $matches['ids']));
// Test successful
$this->assertNotEmpty($logger->getMessages('Adding SolrReindexTest_Item'));
$this->assertNotEmpty($logger->getMessages('Done'));
// Test successful
$this->assertNotEmpty($logger->getMessages('Adding SolrReindexTest_Item'));
$this->assertNotEmpty($logger->getMessages('Done'));
// Test that items in this variant / group are cleared from solr
Phockito::verify($this->service, 1)->deleteByQuery(
'+(ClassHierarchy:SolrReindexTest_Item) +_query_:"{!frange l=2 u=2}mod(ID, 6)" +(_testvariant:"1")'
);
// Test that items in this variant / group are cleared from solr
Phockito::verify($this->service, 1)->deleteByQuery(
'+(ClassHierarchy:SolrReindexTest_Item) +_query_:"{!frange l=2 u=2}mod(ID, 6)" +(_testvariant:"1")'
);
// Test that items in this variant / group are re-indexed
// 120 divided into 6 groups should be 20 at least (max 21)
$this->assertEquals(21, count($ids), 'Group size is about 20', 1);
foreach($ids as $id) {
// Each id should be % 6 == 2
$this->assertEquals(2, $id % 6, "ID $id Should match pattern ID % 6 = 2");
}
}
// Test that items in this variant / group are re-indexed
// 120 divided into 6 groups should be 20 at least (max 21)
$this->assertEquals(21, count($ids), 'Group size is about 20', 1);
foreach ($ids as $id) {
// Each id should be % 6 == 2
$this->assertEquals(2, $id % 6, "ID $id Should match pattern ID % 6 = 2");
}
}
/**
* Test that running all groups covers the entire range of dataobject IDs
*/
public function testRunAllGroups() {
$this->createDummyData(120);
$logger = new SolrReindexTest_RecordingLogger();
/**
* Test that running all groups covers the entire range of dataobject IDs
*/
public function testRunAllGroups()
{
$this->createDummyData(120);
$logger = new SolrReindexTest_RecordingLogger();
// Test that running all groups covers the complete set of ids
$state = array('SolrReindexTest_Variant' => '1');
for($i = 0; $i < 6; $i++) {
// See testReindexSegmentsGroups for test that each of these states is invoked during a full reindex
$this
->getHandler()
->runGroup($logger, $this->index, $state, 'SolrReindexTest_Item', 6, $i);
}
// Test that running all groups covers the complete set of ids
$state = array('SolrReindexTest_Variant' => '1');
for ($i = 0; $i < 6; $i++) {
// See testReindexSegmentsGroups for test that each of these states is invoked during a full reindex
$this
->getHandler()
->runGroup($logger, $this->index, $state, 'SolrReindexTest_Item', 6, $i);
}
// Count all ids updated
$ids = array();
foreach($logger->filterMessages('Updated ') as $message) {
$this->assertNotEmpty(preg_match('/^Updated (?<ids>[,\d]+)/', $message, $matches));
$ids = array_unique(array_merge($ids, explode(',', $matches['ids'])));
}
// Count all ids updated
$ids = array();
foreach ($logger->filterMessages('Updated ') as $message) {
$this->assertNotEmpty(preg_match('/^Updated (?<ids>[,\d]+)/', $message, $matches));
$ids = array_unique(array_merge($ids, explode(',', $matches['ids'])));
}
// Check ids
$this->assertEquals(120, count($ids));
Phockito::verify($this->service, 6)->deleteByQuery(anything());
Phockito::verify($this->service, 1)->deleteByQuery(
'+(ClassHierarchy:SolrReindexTest_Item) +_query_:"{!frange l=0 u=0}mod(ID, 6)" +(_testvariant:"1")'
);
Phockito::verify($this->service, 1)->deleteByQuery(
'+(ClassHierarchy:SolrReindexTest_Item) +_query_:"{!frange l=1 u=1}mod(ID, 6)" +(_testvariant:"1")'
);
Phockito::verify($this->service, 1)->deleteByQuery(
'+(ClassHierarchy:SolrReindexTest_Item) +_query_:"{!frange l=2 u=2}mod(ID, 6)" +(_testvariant:"1")'
);
Phockito::verify($this->service, 1)->deleteByQuery(
'+(ClassHierarchy:SolrReindexTest_Item) +_query_:"{!frange l=3 u=3}mod(ID, 6)" +(_testvariant:"1")'
);
Phockito::verify($this->service, 1)->deleteByQuery(
'+(ClassHierarchy:SolrReindexTest_Item) +_query_:"{!frange l=4 u=4}mod(ID, 6)" +(_testvariant:"1")'
);
Phockito::verify($this->service, 1)->deleteByQuery(
'+(ClassHierarchy:SolrReindexTest_Item) +_query_:"{!frange l=5 u=5}mod(ID, 6)" +(_testvariant:"1")'
);
}
// Check ids
$this->assertEquals(120, count($ids));
Phockito::verify($this->service, 6)->deleteByQuery(anything());
Phockito::verify($this->service, 1)->deleteByQuery(
'+(ClassHierarchy:SolrReindexTest_Item) +_query_:"{!frange l=0 u=0}mod(ID, 6)" +(_testvariant:"1")'
);
Phockito::verify($this->service, 1)->deleteByQuery(
'+(ClassHierarchy:SolrReindexTest_Item) +_query_:"{!frange l=1 u=1}mod(ID, 6)" +(_testvariant:"1")'
);
Phockito::verify($this->service, 1)->deleteByQuery(
'+(ClassHierarchy:SolrReindexTest_Item) +_query_:"{!frange l=2 u=2}mod(ID, 6)" +(_testvariant:"1")'
);
Phockito::verify($this->service, 1)->deleteByQuery(
'+(ClassHierarchy:SolrReindexTest_Item) +_query_:"{!frange l=3 u=3}mod(ID, 6)" +(_testvariant:"1")'
);
Phockito::verify($this->service, 1)->deleteByQuery(
'+(ClassHierarchy:SolrReindexTest_Item) +_query_:"{!frange l=4 u=4}mod(ID, 6)" +(_testvariant:"1")'
);
Phockito::verify($this->service, 1)->deleteByQuery(
'+(ClassHierarchy:SolrReindexTest_Item) +_query_:"{!frange l=5 u=5}mod(ID, 6)" +(_testvariant:"1")'
);
}
}
/**
* Provides a wrapper for testing SolrReindexBase
*/
class SolrReindexTest_TestHandler extends SolrReindexBase {
public function processGroup(
LoggerInterface $logger, SolrIndex $indexInstance, $state, $class, $groups, $group, $taskName
) {
$indexName = $indexInstance->getIndexName();
$stateName = json_encode($state);
$logger->info("Called processGroup with {$indexName}, {$stateName}, {$class}, group {$group} of {$groups}");
}
public function triggerReindex(LoggerInterface $logger, $batchSize, $taskName, $classes = null) {
$logger->info("Called triggerReindex");
}
class SolrReindexTest_TestHandler extends SolrReindexBase
{
public function processGroup(
LoggerInterface $logger, SolrIndex $indexInstance, $state, $class, $groups, $group, $taskName
) {
$indexName = $indexInstance->getIndexName();
$stateName = json_encode($state);
$logger->info("Called processGroup with {$indexName}, {$stateName}, {$class}, group {$group} of {$groups}");
}
public function triggerReindex(LoggerInterface $logger, $batchSize, $taskName, $classes = null)
{
$logger->info("Called triggerReindex");
}
}
class SolrReindexTest_Index extends SolrIndex implements TestOnly {
public function init() {
$this->addClass('SolrReindexTest_Item');
$this->addAllFulltextFields();
}
class SolrReindexTest_Index extends SolrIndex implements TestOnly
{
public function init()
{
$this->addClass('SolrReindexTest_Item');
$this->addAllFulltextFields();
}
}
/**
* Does not have any variant extensions
*/
class SolrReindexTest_Item extends DataObject implements TestOnly {
private static $extensions = array(
'SolrReindexTest_ItemExtension'
);
private static $db = array(
'Title' => 'Varchar(255)',
'Variant' => 'Int(0)'
);
class SolrReindexTest_Item extends DataObject implements TestOnly
{
private static $extensions = array(
'SolrReindexTest_ItemExtension'
);
private static $db = array(
'Title' => 'Varchar(255)',
'Variant' => 'Int(0)'
);
}
/**
* Select only records in the current variant
*/
class SolrReindexTest_ItemExtension extends DataExtension implements TestOnly {
/**
* Filter records on the current variant
*
* @param SQLQuery $query
* @param DataQuery $dataQuery
*/
public function augmentSQL(SQLQuery &$query, DataQuery &$dataQuery = null) {
$variant = SolrReindexTest_Variant::get_current();
if($variant !== null && !$query->filtersOnID()) {
$sqlVariant = Convert::raw2sql($variant);
$query->addWhere("\"Variant\" = '{$sqlVariant}'");
}
}
class SolrReindexTest_ItemExtension extends DataExtension implements TestOnly
{
/**
* Filter records on the current variant
*
* @param SQLQuery $query
* @param DataQuery $dataQuery
*/
public function augmentSQL(SQLQuery &$query, DataQuery &$dataQuery = null)
{
$variant = SolrReindexTest_Variant::get_current();
if ($variant !== null && !$query->filtersOnID()) {
$sqlVariant = Convert::raw2sql($variant);
$query->addWhere("\"Variant\" = '{$sqlVariant}'");
}
}
}
@ -342,190 +355,208 @@ class SolrReindexTest_ItemExtension extends DataExtension implements TestOnly {
*
* Variant states are 0 and 1, or null if disabled
*/
class SolrReindexTest_Variant extends SearchVariant implements TestOnly {
class SolrReindexTest_Variant extends SearchVariant implements TestOnly
{
/**
* Value of this variant (either null, 0, or 1)
*
* @var int|null
*/
protected static $current = null;
/**
* Value of this variant (either null, 0, or 1)
*
* @var int|null
*/
protected static $current = null;
/**
* Activate this variant
*/
public static function enable()
{
self::disable();
/**
* Activate this variant
*/
public static function enable() {
self::disable();
self::$current = 0;
self::$variants = array(
'SolrReindexTest_Variant' => singleton('SolrReindexTest_Variant')
);
}
self::$current = 0;
self::$variants = array(
'SolrReindexTest_Variant' => singleton('SolrReindexTest_Variant')
);
}
/**
* Disable this variant and reset
*/
public static function disable()
{
self::$current = null;
self::$variants = null;
self::$class_variants = array();
self::$call_instances = array();
}
/**
* Disable this variant and reset
*/
public static function disable() {
self::$current = null;
self::$variants = null;
self::$class_variants = array();
self::$call_instances = array();
}
public function activateState($state)
{
self::set_current($state);
}
public function activateState($state) {
self::set_current($state);
}
/**
* Set the current variant to the given state
*
* @param int $current 0, 1, 2, or null (disabled)
*/
public static function set_current($current)
{
self::$current = $current;
}
/**
* Set the current variant to the given state
*
* @param int $current 0, 1, 2, or null (disabled)
*/
public static function set_current($current) {
self::$current = $current;
}
/**
* Get the current state
*
* @return string|null
*/
public static function get_current()
{
// Always use string values for states for consistent json_encode value
if (isset(self::$current)) {
return (string)self::$current;
}
}
/**
* Get the current state
*
* @return string|null
*/
public static function get_current() {
// Always use string values for states for consistent json_encode value
if(isset(self::$current)) {
return (string)self::$current;
}
}
public function alterDefinition($base, $index)
{
$self = get_class($this);
function alterDefinition($base, $index) {
$self = get_class($this);
$index->filterFields['_testvariant'] = array(
'name' => '_testvariant',
'field' => '_testvariant',
'fullfield' => '_testvariant',
'base' => $base,
'origin' => $base,
'type' => 'Int',
'lookup_chain' => array(array('call' => 'variant', 'variant' => $self, 'method' => 'currentState'))
);
}
$index->filterFields['_testvariant'] = array(
'name' => '_testvariant',
'field' => '_testvariant',
'fullfield' => '_testvariant',
'base' => $base,
'origin' => $base,
'type' => 'Int',
'lookup_chain' => array(array('call' => 'variant', 'variant' => $self, 'method' => 'currentState'))
);
}
public function alterQuery($query, $index)
{
// I guess just calling it _testvariant is ok?
$query->filter('_testvariant', $this->currentState());
}
public function alterQuery($query, $index) {
// I guess just calling it _testvariant is ok?
$query->filter('_testvariant', $this->currentState());
}
public function appliesTo($class, $includeSubclasses)
{
return $class === 'SolrReindexTest_Item' ||
($includeSubclasses && is_subclass_of($class, 'SolrReindexTest_Item', true));
}
public function appliesTo($class, $includeSubclasses) {
return $class === 'SolrReindexTest_Item' ||
($includeSubclasses && is_subclass_of($class, 'SolrReindexTest_Item', true));
}
public function appliesToEnvironment()
{
// Set to null to disable
return self::$current !== null;
}
public function appliesToEnvironment() {
// Set to null to disable
return self::$current !== null;
}
public function currentState() {
return self::get_current();
}
public function reindexStates() {
// Always use string values for states for consistent json_encode value
return array('0', '1', '2');
}
public function currentState()
{
return self::get_current();
}
public function reindexStates()
{
// Always use string values for states for consistent json_encode value
return array('0', '1', '2');
}
}
/**
* Test logger for recording messages
*/
class SolrReindexTest_RecordingLogger extends Logger implements TestOnly {
class SolrReindexTest_RecordingLogger extends Logger implements TestOnly
{
/**
* @var SolrReindexTest_Handler
*/
protected $testHandler = null;
/**
* @var SolrReindexTest_Handler
*/
protected $testHandler = null;
public function __construct($name = 'testlogger', array $handlers = array(), array $processors = array())
{
parent::__construct($name, $handlers, $processors);
public function __construct($name = 'testlogger', array $handlers = array(), array $processors = array()) {
parent::__construct($name, $handlers, $processors);
$this->testHandler = new SolrReindexTest_Handler();
$this->pushHandler($this->testHandler);
}
$this->testHandler = new SolrReindexTest_Handler();
$this->pushHandler($this->testHandler);
}
/**
* @return array
*/
public function getMessages()
{
return $this->testHandler->getMessages();
}
/**
* @return array
*/
public function getMessages() {
return $this->testHandler->getMessages();
}
/**
* Clear all messages
*/
public function clear()
{
$this->testHandler->clear();
}
/**
* Clear all messages
*/
public function clear() {
$this->testHandler->clear();
}
/**
* Get messages with the given filter
*
* @param string $containing
* @return array Filtered array
*/
public function filterMessages($containing)
{
return array_values(array_filter(
$this->getMessages(),
function ($content) use ($containing) {
return stripos($content, $containing) !== false;
}
));
}
/**
* Get messages with the given filter
*
* @param string $containing
* @return array Filtered array
*/
public function filterMessages($containing) {
return array_values(array_filter(
$this->getMessages(),
function($content) use ($containing) {
return stripos($content, $containing) !== false;
}
));
}
/**
* Count all messages containing the given substring
*
* @param string $containing Message to filter by
* @return int
*/
public function countMessages($containing = null) {
if($containing) {
$messages = $this->filterMessages($containing);
} else {
$messages = $this->getMessages();
}
return count($messages);
}
/**
* Count all messages containing the given substring
*
* @param string $containing Message to filter by
* @return int
*/
public function countMessages($containing = null)
{
if ($containing) {
$messages = $this->filterMessages($containing);
} else {
$messages = $this->getMessages();
}
return count($messages);
}
}
/**
* Logger for recording messages for later retrieval
*/
class SolrReindexTest_Handler extends AbstractProcessingHandler implements TestOnly {
class SolrReindexTest_Handler extends AbstractProcessingHandler implements TestOnly
{
/**
* Messages
*
* @var array
*/
protected $messages = array();
/**
* Messages
*
* @var array
*/
protected $messages = array();
/**
* Get all messages
*
* @return array
*/
public function getMessages()
{
return $this->messages;
}
/**
* Get all messages
*
* @return array
*/
public function getMessages() {
return $this->messages;
}
public function clear()
{
$this->messages = array();
}
public function clear() {
$this->messages = array();
}
protected function write(array $record) {
$this->messages[] = $record['message'];
}
}
protected function write(array $record)
{
$this->messages[] = $record['message'];
}
}