Все переписано на фасады

2024-09-03 15:41:45 +03:00
parent 74ba48620c
commit 48b4d6cccf
34 changed files with 1036 additions and 807 deletions
--- a/ContingentParser/Parser/ContingentFacade.php
+++ b/ContingentParser/Parser/ContingentFacade.php
@ -0,0 +1,71 @@
+<?php
+namespace ContingentParser\Parser;
+
+class ContingentFacade
+{
+    /**
+     * Получить данные о численности
+     * @param string $html
+     * Разметка сайта вуза
+     * @param mixed $specialties
+     * Массив специальностей
+     * @param int $orgId
+     * Идентификатор организации
+     * @return array
+     */
+    public function getContingent(
+        string $html,
+        array $specialties,
+        int $orgId
+    ) : array {
+        $parser = new ContingentParser($html);
+        $contingent = $parser->getDataTable();
+        $this->addSpecId($contingent, $specialties);
+        $this->addOrgId($contingent, $orgId);
+
+        return $contingent;
+    }
+    /**
+     * Проверка на валидность записи численнести
+     * @param array $contingent
+     * Массив численности по специальностям
+     * @return bool
+     */
+    public function isValidContingent(array $contingent) : bool
+    {
+        $count = 0;
+        foreach ($contingent as $value) {
+            $count += $value['contingent'];
+        }
+        return $count ? true : false;
+    }
+    /**
+     * Добавить идентификатор специальности в запись численности
+     * @param array $contingent
+     * Массив численности по специальностям
+     * @param array $specialties
+     * Массив специальностей
+     * @return void
+     */
+    private function addSpecId(array &$contingent, array $specialties) : void
+    {
+        $specIdMap = array_column($specialties, 'spec_id', 'spec_code');
+        foreach ($contingent as $key => $con) {
+            $contingent[$key]['spec_id'] = $specIdMap[$con['spec_code']] ?? null;
+        }
+    }
+    /**
+     * Добавить идентификатор организации в запись численности
+     * @param array $contingent
+     * Массив численности по специальностям
+     * @param int $orgId
+     * Идентифиактор организации
+     * @return void
+     */
+    private function addOrgId(array &$contingent, int $orgId): void
+    {
+        foreach ($contingent as &$con) {
+            $con['org_id'] = $orgId;
+        }
+    }
+}
--- a/ContingentParser/Parser/ContingentParser.php
+++ b/ContingentParser/Parser/ContingentParser.php
@ -0,0 +1,99 @@
+<?php
+namespace ContingentParser\Parser;
+
+use DOMDocument;
+use DOMXPath;
+
+class ContingentParser
+{
+    private ?DOMXPath $xpath;
+    private const TEMPLATE = '//tr[@itemprop="eduChislen"]//';
+    private const ENCODING = "UTF-8"; 
+    private const FIELDS = [
+        "eduCode" => "td",
+        "eduName" => "td",
+        "eduLevel" => "td",
+        "eduForm" => "td",
+        "numberAll" => ["th", "td"]
+    ];
+
+    public function __construct(string $html)
+    {
+        libxml_use_internal_errors(true);
+        $dom = new DOMDocument(
+            encoding: self::ENCODING
+        );
+        if (empty($html)) {
+            $this->xpath = null;
+        } else {
+            $this->setEncoding($html);
+            $dom->loadHTML($html);
+            $this->xpath = new DOMXPath($dom);
+        }
+    }
+    
+    private function setEncoding(string &$html) : void
+    {
+        $encoding = mb_detect_encoding($html, 'UTF-8, windows-1251');
+        if ($encoding != self::ENCODING) {
+            $html = mb_convert_encoding(
+                $html,
+                self::ENCODING,
+                $encoding
+            );
+            $html = str_replace('windows-1251',self::ENCODING, $html); 
+        }
+        $html = mb_convert_encoding($html,'HTML-ENTITIES','UTF-8');
+    }
+    private function parse() : array
+    {
+        $data = [];
+        foreach (self::FIELDS as $field => $tag) {
+            if (!is_array($tag)) {
+                $data[$field] = $this->xpath->query(
+                    self::TEMPLATE . $tag . "[@itemprop=\"$field\"]"
+                );
+            } else {
+                $th = $this->xpath->query(
+                    self::TEMPLATE . $tag[0] . "[@itemprop=\"$field\"]"
+                );
+                $td = $this->xpath->query(
+                    self::TEMPLATE . $tag[1] . "[@itemprop=\"$field\"]"
+                );
+                $data[$field] = $th->length > $td->length ? $th : $td;
+            }
+        }
+        return $data;
+    }
+
+    public function getDataTable() : array
+    {
+        if (empty($this->xpath)) return [];
+        
+        $data = $this->parse();
+        $records = [];
+        if ($data == null) return [];
+
+        $equal = $data['eduName']->length;
+        foreach ($data as $field) {
+            if ($field->length == 0) {
+                return [];
+            }
+            if ($field->length != $equal) {
+                return [];
+            }
+        }
+
+        for ($i = 0; $i < $data['eduCode']->length; $i++) { 
+            $contingentRow = new ContingentRow(
+                $data['eduCode']->item($i)->textContent,
+                $data['eduName']->item($i)->textContent,
+                $data['eduLevel']->item($i)->textContent,
+                $data['eduForm']->item($i)->textContent,
+                (int)$data['numberAll']->item($i)->textContent
+            );
+            $records[] = $contingentRow->getData();
+        }
+        return $records;
+    }
+}
--- a/ContingentParser/Parser/ContingentRow.php
+++ b/ContingentParser/Parser/ContingentRow.php
@ -0,0 +1,33 @@
+<?php
+namespace ContingentParser\Parser;
+
+class ContingentRow
+{
+    public function __construct(
+        private string $eduCode,
+        private string $eduName,
+        private string $eduLevel,
+        private string $eduForm,
+        private int $contingent
+    ) {
+        if ($contingent < 0) {
+            throw new \Exception("Недействительная численность обучающихся!");
+        }
+        $this->eduCode = trim($eduCode);
+        $this->eduName = trim($eduName);
+        $this->eduLevel = trim($eduLevel);
+        $this->eduForm = trim($eduForm);
+        $this->contingent = $contingent;
+    }
+
+    public function getData() : array
+    {
+        return [
+            "spec_code" => $this->eduCode,
+            "spec_name" => $this->eduName,
+            "edu_level" => $this->eduLevel,
+            "edu_forms"=> $this->eduForm,
+            "contingent" => $this->contingent
+        ];
+    }
+}