silverstripe-framework/tools/CsvBulkLoader.php
Ingo Schommer f44598dc3a (merged from branches/roa. use "svn log -c <changeset> -g <module-svn-path>" for detailed commit message)
git-svn-id: svn://svn.silverstripe.com/silverstripe/open/modules/sapphire/trunk@60212 467b73ca-7a2a-4603-9d3b-597d59a354a9
2008-08-09 05:00:42 +00:00

175 lines
5.4 KiB
PHP

<?php
/**
* Uses the fgetcsv() function to process CSV input.
* The input is expected to be UTF8.
*
* @see http://rfc.net/rfc4180.html
* @package cms
* @subpackage bulkloading
* @author Ingo Schommer, Silverstripe Ltd. (<firstname>@silverstripe.com)
*/
class CsvBulkLoader extends BulkLoader {
/**
* Delimiter character (Default: comma).
*
* @var string
*/
public $delimiter = ',';
/**
* Enclosure character (Default: doublequote)
*
* @var string
*/
public $enclosure = '"';
/**
* Identifies if the loaded file has a header row.
* If a {@link self::$columnMap} is passed, we assume
* the file has no headerrow, unless explicitly noted.
*
* @var boolean
*/
public $hasHeaderRow = false;
protected function processAll($filepath, $preview = false) {
$file = fopen($filepath, 'r');
if(!$file) return false;
$return = new DataObjectSet();
// assuming that first row is column naming if no columnmap is passed
if($this->hasHeaderRow && $this->columnMap) {
$columnRow = fgetcsv($file, 0, $this->delimiter, $this->enclosure);
$columnMap = $this->columnMap;
} elseif($this->columnMap) {
$columnMap = $this->columnMap;
} else {
$columnRow = fgetcsv($file, 0, $this->delimiter, $this->enclosure);
$columnMap = array_combine($columnRow, $columnRow);
}
while (($row = fgetcsv($file, 0, $this->delimiter, $this->enclosure)) !== FALSE) {
$indexedRow = array_combine(array_values($columnMap), array_values($row));
$return->push($this->processRecord($indexedRow));
}
fclose($file);
return $return;
}
protected function processRecord($record, $preview = false) {
$class = $this->objectClass;
// find existing object, or create new one
$existingObj = $this->findExistingObject($record);
$obj = ($existingObj) ? $existingObj : new $class();
// first run: find/create any relations and store them on the object
// we can't combine runs, as other columns might rely on the relation being present
$relations = array();
foreach($record as $key => $val) {
//if($this->isNullValue($val)) continue;
// checking for existing relations
if(isset($this->relationCallbacks[$key])) {
// trigger custom search method for finding a relation based on the given value
// and write it back to the relation (or create a new object)
$relationName = $this->relationCallbacks[$key]['relationname'];
$relationObj = $obj->{$this->relationCallbacks[$key]['callback']}($val, $record);
if(!$relationObj || !$relationObj->exists()) {
$relationClass = $obj->has_one($relationName);
$relationObj = new $relationClass();
$relationObj->write();
}
$obj->setComponent($relationName, $relationObj);
$obj->{"{$relationName}ID"} = $relationObj->ID;
$obj->write();
} elseif(strpos($key, '.') !== false) {
// we have a relation column with dot notation
list($relationName,$columnName) = split('\.', $key);
$relationObj = $obj->getComponent($relationName); // always gives us an component (either empty or existing)
$obj->setComponent($relationName, $relationObj);
$relationObj->write();
$obj->{"{$relationName}ID"} = $relationObj->ID;
$obj->write();
}
$obj->flushCache(); // avoid relation caching confusion
}
$id = ($preview) ? 0 : $obj->write();
// second run: save data
foreach($record as $key => $val) {
if($obj->hasMethod("import{$key}")) {
$obj->{"import{$key}"}($val, $record);
} elseif(strpos($key, '.') !== false) {
// we have a relation column
list($relationName,$columnName) = split('\.', $key);
$relationObj = $obj->getComponent($relationName);
$relationObj->{$columnName} = $val;
$relationObj->write();
$obj->flushCache(); // avoid relation caching confusion
} elseif($obj->hasField($key) || $obj->hasMethod($key)) {
// plain old value setter
if(!$this->isNullValue($val, $key)) $obj->{$key} = $val;
}
}
$id = ($preview) ? 0 : $obj->write();
$action = 'create';
$message = '';
// memory usage
unset($existingObj);
unset($obj);
return new ArrayData(array(
'id' => $id,
'action' => $action,
'message' => $message
));
}
/**
* Find an existing objects based on one or more uniqueness
* columns specified via {@link self::$duplicateChecks}
*
* @param array $record CSV data column
* @return unknown
*/
public function findExistingObject($record) {
// checking for existing records (only if not already found)
foreach($this->duplicateChecks as $fieldName => $duplicateCheck) {
if(is_string($duplicateCheck)) {
$SQL_fieldName = Convert::raw2sql($duplicateCheck);
$SQL_fieldValue = $record[$this->columnMap[$fieldName]];
$existingRecord = DataObject::get_one($this->objectClass, "`$SQL_fieldName` = '{$SQL_fieldValue}'");
if($existingRecord) return $existingRecord;
} elseif(is_array($duplicateCheck) && isset($duplicateCheck['callback'])) {
$existingRecord = singleton($this->objectClass)->{$duplicateCheck['callback']}($val, $record);
if($existingRecord) return $existingRecord;
} else {
user_error('CsvBulkLoader:processRecord: Wrong format for $duplicateChecks', E_USER_ERROR);
}
}
return false;
}
/**
* Determine wether any loaded files should be parsed
* with a header-row (otherwise we rely on {@link self::$columnMap}.
*
* @return boolean
*/
public function hasHeaderRow() {
return ($this->hasHeaderRow || isset($this->columnMap));
}
}
?>