common/helpers/XliffHelper.php - dev - v.bolshakov/AIEcosystem-Testing

common/helpers/XliffHelper.php
154 строки5 KB
Developer
Initial commit
03 авг 2026, 17:43
03 авг 2026, 17:437433917
<?php

namespace common\helpers;

use common\models\memoq\Document;
use common\models\memoq\TerSync;
use common\models\memoq\XliffSegment;
use Yii;
use yii\helpers\VarDumper;

/**
 * Вспомогательный класс для работы MemoQ
 *
 * @author Dmitry E. Semenov <sde.tomsk@gmail.com>
 * @copyright Self (c) 2025
 */
class XliffHelper
{
    public static function parseXliff($document_id, $xliff_content)
    {
        $content = str_replace([
            'mq:insertedmatch',
            'mq:minorversions',
            'mq:historical-unit',
            'mq:translatorcommitusername',
            'mq:reviewer2commitusername',
            'mq:reviewer1commitusername'
        ], [
            'mq_insertedmatch',
            'mq_minorversions',
            'mq_historical-unit',
            'mq_translatorcommitusername',
            'mq_reviewer2commitusername',
            'mq_reviewer1commitusername'
        ], $xliff_content);

        $dom = new \DOMDocument();
        libxml_use_internal_errors(true);
        $dom->loadXML($content);
        libxml_clear_errors();

        // Convert to SimpleXML if needed
        $document = simplexml_import_dom($dom);

        /** @var \SimpleXMLElement $body */
        $body = $document->file->body;

        $documents = [];

        /** @var SimpleXMLElement $row */
        foreach ($body->{"trans-unit"} as $row) {

            $id = (string)$row->attributes()->id;
            // Оригинал
            $source = (string)$row->source;
            // Итог
            $human_final = (string)$row->target;

            // Провайдер
            $a = $row->{'mq_insertedmatch'}->attributes();
            $mt_provider = $a ? (string)$a->source : null;
            $pre_translation = (string)$row->{'mq_insertedmatch'}->target;

            $hist_versions = [];
            foreach ($row->mq_minorversions->{'mq_historical-unit'} as $version) {
                $a = $version->attributes();
                // Переводчик
                $editor_1 = $a->mq_translatorcommitusername;
                // Редактор
                $editor_2 = $a->mq_reviewer2commitusername ?: $a->mq_reviewer1commitusername;

                $item = [
                    'editor1' => (string)$editor_1,
                    'editor2' => (string)$editor_2,
                    'target'  => (string)$version->target
                ];

                $hist_versions[] = $item;
//                array_unshift($hist_versions, $item);
            }

            $cnt = count($hist_versions);

            $editor_1_user = null;
            $editor_1_text = null;
            $editor_2_user = null;
            $editor_2_text = null;

            if ($cnt > 0) {
                $the = $hist_versions[0];
                $editor_1_user = ArrayHelper::getValue($the, 'editor1');
                if ($editor_1_user) {
                    $editor_1_text = ArrayHelper::getValue($the, 'target', $mt_provider);
                }
            }
            if ($cnt > 1) {
                $the = $hist_versions[1];
                $editor_2_user = ArrayHelper::getValue($the, 'editor2');
                if ($editor_2_user) {
                    $editor_2_text = ArrayHelper::getValue($the, 'target', $human_final);
                }
            }

            $wordCount = TextHelper::countWordsWithFilter($human_final);

            $documents[] = [
                'document_id'     => $document_id,
                'segment'         => $id,
                'source'          => $source,
                'pre_translation' => $pre_translation,
                'ai_provider'     => $mt_provider,
                'editor_1_text'   => $editor_1_text,
                'editor_1_user'   => $editor_1_user,
                'editor_2_text'   => $editor_2_text,
                'editor_2_user'   => $editor_2_user,
                'human_final'     => $human_final,
                'word_count'      => $wordCount
            ];
        }

        XliffSegment::deleteAll(['document_id' => $document_id]);

        /** @var \yii\mongodb\Connection $mongodb */
        $mongodb = Yii::$app->mongodb;
        $collectionName = XliffSegment::collectionName();
        $command = $mongodb->createCommand();

        Yii::debug(VarDumper::dumpAsString([
            'documents' => $documents
        ]));

        $segment_count = count($documents);

        Document::updateAll([
            'segment_count' => $segment_count
        ], [
            'id' => $document_id
        ]);

        $result = $command->batchInsert($collectionName, $documents);

        if (YII_DEBUG) {
            Yii::debug(VarDumper::dumpAsString([
                'result' => $result
            ]));
        }

        if ($segment_count > 0) {
            TerHelper::updateTer($document_id);
        }

        return true;
    }
}