xref: /plugin/totext/Extractor/AbstractOoxmlExtractor.php (revision 0fbee1892eb8b7339ce48e27e48e3fc821a8695f) 
1<?php
2
3namespace dokuwiki\plugin\totext\Extractor;
4
5/**
6 * Shared metadata reader for the OOXML family (DOCX/XLSX/PPTX).
7 *
8 * All three store their package metadata in the same two parts:
9 * docProps/core.xml (Dublin Core + core properties) and docProps/app.xml
10 * (the authoring application). The element local name => canonical key maps
11 * are declared here once; subclasses only implement extractText().
12 */
13abstract class AbstractOoxmlExtractor extends AbstractZipXmlExtractor
14{
15    /**
16     * docProps/core.xml element local name => canonical metadata key.
17     *
18     * @var array<string, string>
19     */
20    protected const CORE_META_MAP = [
21        'title' => 'Title',
22        'creator' => 'Author',
23        'subject' => 'Subject',
24        'keywords' => 'Keywords',
25        'description' => 'Description',
26        'created' => 'Created',
27        'modified' => 'Modified',
28        'language' => 'Language',
29    ];
30
31    /**
32     * docProps/app.xml element local name => canonical metadata key.
33     *
34     * @var array<string, string>
35     */
36    protected const APP_META_MAP = [
37        'Application' => 'Producer',
38    ];
39
40    /** @inheritDoc */
41    protected function extractMetadata(): array
42    {
43        $meta = [];
44
45        $core = $this->readPart('docProps/core.xml');
46        if ($core !== null) {
47            $meta = $this->mapMetadataFromXml($core, self::CORE_META_MAP);
48        }
49
50        $app = $this->readPart('docProps/app.xml');
51        if ($app !== null) {
52            $meta += $this->mapMetadataFromXml($app, self::APP_META_MAP);
53        }
54
55        return $meta;
56    }
57}
58