Code Coverage
 
Lines
Functions and Methods
Classes and Traits
Total
97.14% covered (success)
97.14%
476 / 490
75.00% covered (warning)
75.00%
21 / 28
CRAP
0.00% covered (danger)
0.00%
0 / 1
AbstractPart
97.14% covered (success)
97.14%
476 / 490
75.00% covered (warning)
75.00%
21 / 28
169
0.00% covered (danger)
0.00%
0 / 1
 read
n/a
0 / 0
n/a
0 / 0
0
 __construct
100.00% covered (success)
100.00%
2 / 2
100.00% covered (success)
100.00%
1 / 1
1
 setRels
100.00% covered (success)
100.00%
1 / 1
100.00% covered (success)
100.00%
1 / 1
1
 setImageLoading
100.00% covered (success)
100.00%
2 / 2
100.00% covered (success)
100.00%
1 / 1
1
 hasImageLoading
100.00% covered (success)
100.00%
1 / 1
100.00% covered (success)
100.00%
1 / 1
1
 getCommentReferences
100.00% covered (success)
100.00%
1 / 1
100.00% covered (success)
100.00%
1 / 1
1
 setCommentReferences
100.00% covered (success)
100.00%
2 / 2
100.00% covered (success)
100.00%
1 / 1
1
 setCommentReference
88.89% covered (warning)
88.89%
8 / 9
0.00% covered (danger)
0.00%
0 / 1
3.01
 getCommentReference
66.67% covered (warning)
66.67%
2 / 3
0.00% covered (danger)
0.00%
0 / 1
2.15
 readParagraph
98.82% covered (success)
98.82%
84 / 85
0.00% covered (danger)
0.00%
0 / 1
32
 readFormField
95.24% covered (success)
95.24%
60 / 63
0.00% covered (danger)
0.00%
0 / 1
27
 getHeadingDepth
100.00% covered (success)
100.00%
8 / 8
100.00% covered (success)
100.00%
1 / 1
5
 readRun
100.00% covered (success)
100.00%
16 / 16
100.00% covered (success)
100.00%
1 / 1
7
 readRunChild
93.51% covered (success)
93.51%
72 / 77
0.00% covered (danger)
0.00%
0 / 1
29.23
 readRubyProperties
100.00% covered (success)
100.00%
12 / 12
100.00% covered (success)
100.00%
1 / 1
1
 readTable
100.00% covered (success)
100.00%
32 / 32
100.00% covered (success)
100.00%
1 / 1
12
 readParagraphStyle
100.00% covered (success)
100.00%
36 / 36
100.00% covered (success)
100.00%
1 / 1
2
 readFontStyle
93.10% covered (success)
93.10%
27 / 29
0.00% covered (danger)
0.00%
0 / 1
4.01
 readTableStyle
96.30% covered (success)
96.30%
26 / 27
0.00% covered (danger)
0.00%
0 / 1
7
 readTablePosition
100.00% covered (success)
100.00%
13 / 13
100.00% covered (success)
100.00%
1 / 1
1
 readTableIndent
100.00% covered (success)
100.00%
6 / 6
100.00% covered (success)
100.00%
1 / 1
1
 readCellStyle
100.00% covered (success)
100.00%
19 / 19
100.00% covered (success)
100.00%
1 / 1
3
 findPossibleElement
100.00% covered (success)
100.00%
6 / 6
100.00% covered (success)
100.00%
1 / 1
4
 findPossibleAttribute
100.00% covered (success)
100.00%
6 / 6
100.00% covered (success)
100.00%
1 / 1
4
 readStyleDefs
100.00% covered (success)
100.00%
15 / 15
100.00% covered (success)
100.00%
1 / 1
6
 readStyleDef
100.00% covered (success)
100.00%
10 / 10
100.00% covered (success)
100.00%
1 / 1
5
 isOn
100.00% covered (success)
100.00%
1 / 1
100.00% covered (success)
100.00%
1 / 1
4
 getMediaTarget
100.00% covered (success)
100.00%
4 / 4
100.00% covered (success)
100.00%
1 / 1
2
 getTargetMode
100.00% covered (success)
100.00%
4 / 4
100.00% covered (success)
100.00%
1 / 1
2
1<?php
2
3/**
4 * This file is part of PHPWord - A pure PHP library for reading and writing
5 * word processing documents.
6 *
7 * PHPWord is free software distributed under the terms of the GNU Lesser
8 * General Public License version 3 as published by the Free Software Foundation.
9 *
10 * For the full copyright and license information, please read the LICENSE
11 * file that was distributed with this source code. For the full list of
12 * contributors, visit https://github.com/PHPOffice/PHPWord/contributors.
13 *
14 * @see         https://github.com/PHPOffice/PHPWord
15 *
16 * @license     http://www.gnu.org/licenses/lgpl.txt LGPL version 3
17 */
18
19namespace PhpOffice\PhpWord\Reader\Word2007;
20
21use DateTime;
22use DOMElement;
23use InvalidArgumentException;
24use PhpOffice\Math\Reader\OfficeMathML;
25use PhpOffice\PhpWord\ComplexType\RubyProperties;
26use PhpOffice\PhpWord\ComplexType\TblWidth as TblWidthComplexType;
27use PhpOffice\PhpWord\Element\AbstractContainer;
28use PhpOffice\PhpWord\Element\AbstractElement;
29use PhpOffice\PhpWord\Element\FormField;
30use PhpOffice\PhpWord\Element\Ruby;
31use PhpOffice\PhpWord\Element\Text;
32use PhpOffice\PhpWord\Element\TextRun;
33use PhpOffice\PhpWord\Element\TrackChange;
34use PhpOffice\PhpWord\PhpWord;
35use PhpOffice\PhpWord\Shared\XMLReader;
36
37/**
38 * Abstract part reader.
39 *
40 * This class is inherited by ODText reader
41 *
42 * @since 0.10.0
43 */
44abstract class AbstractPart
45{
46    /**
47     * Conversion method.
48     *
49     * @const int
50     */
51    const READ_VALUE = 'attributeValue';            // Read attribute value
52    const READ_EQUAL = 'attributeEquals';           // Read `true` when attribute value equals specified value
53    const READ_TRUE = 'attributeTrue';              // Read `true` when element exists
54    const READ_FALSE = 'attributeFalse';            // Read `false` when element exists
55    const READ_SIZE = 'attributeMultiplyByTwo';     // Read special attribute value for Font::$size
56
57    /**
58     * Document file.
59     *
60     * @var string
61     */
62    protected $docFile;
63
64    /**
65     * XML file.
66     *
67     * @var string
68     */
69    protected $xmlFile;
70
71    /**
72     * Part relationships.
73     *
74     * @var array
75     */
76    protected $rels = [];
77
78    /**
79     * Comment references.
80     *
81     * @var array<string, array<string, AbstractElement>>
82     */
83    protected $commentRefs = [];
84
85    /**
86     * Image Loading.
87     *
88     * @var bool
89     */
90    protected $imageLoading = true;
91
92    /**
93     * Read part.
94     */
95    abstract public function read(PhpWord $phpWord);
96
97    /**
98     * Create new instance.
99     *
100     * @param string $docFile
101     * @param string $xmlFile
102     */
103    public function __construct($docFile, $xmlFile)
104    {
105        $this->docFile = $docFile;
106        $this->xmlFile = $xmlFile;
107    }
108
109    /**
110     * Set relationships.
111     *
112     * @param array $value
113     */
114    public function setRels($value): void
115    {
116        $this->rels = $value;
117    }
118
119    public function setImageLoading(bool $value): self
120    {
121        $this->imageLoading = $value;
122
123        return $this;
124    }
125
126    public function hasImageLoading(): bool
127    {
128        return $this->imageLoading;
129    }
130
131    /**
132     * Get comment references.
133     *
134     * @return array<string, array<string, null|AbstractElement>>
135     */
136    public function getCommentReferences(): array
137    {
138        return $this->commentRefs;
139    }
140
141    /**
142     * Set comment references.
143     *
144     * @param array<string, array<string, null|AbstractElement>> $commentRefs
145     */
146    public function setCommentReferences(array $commentRefs): self
147    {
148        $this->commentRefs = $commentRefs;
149
150        return $this;
151    }
152
153    /**
154     * Set comment reference.
155     */
156    private function setCommentReference(string $type, string $id, AbstractElement $element): self
157    {
158        if (!in_array($type, ['start', 'end'])) {
159            throw new InvalidArgumentException('Type must be "start" or "end"');
160        }
161
162        if (!array_key_exists($id, $this->commentRefs)) {
163            $this->commentRefs[$id] = [
164                'start' => null,
165                'end' => null,
166            ];
167        }
168        $this->commentRefs[$id][$type] = $element;
169
170        return $this;
171    }
172
173    /**
174     * Get comment reference.
175     *
176     * @return array<string, null|AbstractElement>
177     */
178    protected function getCommentReference(string $id): array
179    {
180        if (!array_key_exists($id, $this->commentRefs)) {
181            throw new InvalidArgumentException(sprintf('Comment with id %s isn\'t referenced in document', $id));
182        }
183
184        return $this->commentRefs[$id];
185    }
186
187    /**
188     * Read w:p.
189     *
190     * @param AbstractContainer $parent
191     * @param string $docPart
192     *
193     * @todo Get font style for preserve text
194     */
195    protected function readParagraph(XMLReader $xmlReader, DOMElement $domNode, $parent, $docPart = 'document'): void
196    {
197        // Paragraph style
198        $paragraphStyle = $xmlReader->elementExists('w:pPr', $domNode) ? $this->readParagraphStyle($xmlReader, $domNode) : null;
199
200        if ($xmlReader->elementExists('w:r/w:fldChar/w:ffData', $domNode)) {
201            // FormField
202            $partOfFormField = false;
203            $formNodes = [];
204            $formType = null;
205            $textRunContainers = $xmlReader->countElements('w:r|w:ins|w:del|w:hyperlink|w:smartTag', $domNode);
206            if ($textRunContainers > 0) {
207                $nodes = $xmlReader->getElements('*', $domNode);
208                $paragraph = $parent->addTextRun($paragraphStyle);
209                foreach ($nodes as $node) {
210                    if ($xmlReader->elementExists('w:fldChar/w:ffData', $node)) {
211                        $partOfFormField = true;
212                        $formNodes[] = $node;
213                        if ($xmlReader->elementExists('w:fldChar/w:ffData/w:ddList', $node)) {
214                            $formType = 'dropdown';
215                        } elseif ($xmlReader->elementExists('w:fldChar/w:ffData/w:textInput', $node)) {
216                            $formType = 'textinput';
217                        } elseif ($xmlReader->elementExists('w:fldChar/w:ffData/w:checkBox', $node)) {
218                            $formType = 'checkbox';
219                        }
220                    } elseif ($partOfFormField &&
221                        $xmlReader->elementExists('w:fldChar', $node) &&
222                        'end' == $xmlReader->getAttribute('w:fldCharType', $node, 'w:fldChar')
223                    ) {
224                        $formNodes[] = $node;
225                        $partOfFormField = false;
226                        // Process the form fields
227                        $this->readFormField($xmlReader, $formNodes, $paragraph, $paragraphStyle, $formType);
228                    } elseif ($partOfFormField) {
229                        $formNodes[] = $node;
230                    } else {
231                        // normal runs
232                        $this->readRun($xmlReader, $node, $paragraph, $docPart, $paragraphStyle);
233                    }
234                }
235            }
236        } elseif ($xmlReader->elementExists('w:r/w:instrText', $domNode)) {
237            // PreserveText
238            $ignoreText = false;
239            $textContent = '';
240            $fontStyle = $this->readFontStyle($xmlReader, $domNode);
241            $nodes = $xmlReader->getElements('w:r', $domNode);
242            foreach ($nodes as $node) {
243                if ($xmlReader->elementExists('w:lastRenderedPageBreak', $node)) {
244                    $parent->addPageBreak();
245                }
246                $instrText = $xmlReader->getValue('w:instrText', $node);
247                if (null !== $instrText) {
248                    $textContent .= '{' . $instrText . '}';
249                } else {
250                    if ($xmlReader->elementExists('w:fldChar', $node)) {
251                        $fldCharType = $xmlReader->getAttribute('w:fldCharType', $node, 'w:fldChar');
252                        if ('begin' == $fldCharType) {
253                            $ignoreText = true;
254                        } elseif ('end' == $fldCharType) {
255                            $ignoreText = false;
256                        }
257                    }
258                    if (false === $ignoreText) {
259                        $textContent .= $xmlReader->getValue('w:t', $node);
260                    }
261                }
262            }
263            $parent->addPreserveText(htmlspecialchars($textContent, ENT_QUOTES, 'UTF-8'), $fontStyle, $paragraphStyle);
264
265            return;
266        }
267
268        // Formula
269        $xmlReader->registerNamespace('m', 'http://schemas.openxmlformats.org/officeDocument/2006/math');
270        if ($xmlReader->elementExists('m:oMath', $domNode)) {
271            $mathElement = $xmlReader->getElement('m:oMath', $domNode);
272            $mathXML = $mathElement->ownerDocument->saveXML($mathElement);
273            if (is_string($mathXML)) {
274                $reader = new OfficeMathML();
275                $math = $reader->read($mathXML);
276
277                $parent->addFormula($math);
278            }
279
280            return;
281        }
282
283        // List item
284        if ($xmlReader->elementExists('w:pPr/w:numPr', $domNode)) {
285            $numId = $xmlReader->getAttribute('w:val', $domNode, 'w:pPr/w:numPr/w:numId');
286            $levelId = $xmlReader->getAttribute('w:val', $domNode, 'w:pPr/w:numPr/w:ilvl');
287            $nodes = $xmlReader->getElements('*', $domNode);
288
289            $listItemRun = $parent->addListItemRun($levelId, "PHPWordList{$numId}", $paragraphStyle);
290
291            foreach ($nodes as $node) {
292                $this->readRun($xmlReader, $node, $listItemRun, $docPart, $paragraphStyle);
293            }
294
295            return;
296        }
297
298        // Heading or Title
299        $headingDepth = $xmlReader->elementExists('w:pPr', $domNode) ? $this->getHeadingDepth($paragraphStyle) : null;
300        if ($headingDepth !== null) {
301            $textContent = null;
302            $nodes = $xmlReader->getElements('w:r|w:hyperlink', $domNode);
303            $hasRubyElement = $xmlReader->elementExists('w:r/w:ruby', $domNode);
304            if ($nodes->length === 1 && !$hasRubyElement) {
305                $textContent = htmlspecialchars($xmlReader->getValue('w:t', $nodes->item(0)), ENT_QUOTES, 'UTF-8');
306            } else {
307                $textContent = new TextRun($paragraphStyle);
308                foreach ($nodes as $node) {
309                    $this->readRun($xmlReader, $node, $textContent, $docPart, $paragraphStyle);
310                }
311            }
312            $parent->addTitle($textContent, $headingDepth);
313
314            return;
315        }
316
317        // Text and TextRun
318        $textRunContainers = $xmlReader->countElements('w:r|w:ins|w:del|w:hyperlink|w:smartTag|w:commentReference|w:commentRangeStart|w:commentRangeEnd', $domNode);
319        if (0 === $textRunContainers) {
320            $parent->addTextBreak(1, $paragraphStyle);
321        } else {
322            $nodes = $xmlReader->getElements('*', $domNode);
323            $paragraph = $parent->addTextRun($paragraphStyle);
324            foreach ($nodes as $node) {
325                $this->readRun($xmlReader, $node, $paragraph, $docPart, $paragraphStyle);
326            }
327        }
328    }
329
330    /**
331     * @param DOMElement[] $domNodes
332     * @param AbstractContainer $parent
333     * @param mixed $paragraphStyle
334     * @param string $formType
335     */
336    private function readFormField(XMLReader $xmlReader, array $domNodes, $parent, $paragraphStyle, $formType): void
337    {
338        if (!in_array($formType, ['textinput', 'checkbox', 'dropdown'])) {
339            return;
340        }
341
342        $formField = $parent->addFormField($formType, null, $paragraphStyle);
343        $ffData = $xmlReader->getElement('w:fldChar/w:ffData', $domNodes[0]);
344
345        foreach ($xmlReader->getElements('*', $ffData) as $node) {
346            /** @var DOMElement $node */
347            switch ($node->localName) {
348                case 'name':
349                    $formField->setName($node->getAttribute('w:val'));
350
351                    break;
352                case 'ddList':
353                    $listEntries = [];
354                    foreach ($xmlReader->getElements('*', $node) as $ddListNode) {
355                        switch ($ddListNode->localName) {
356                            case 'result':
357                                $formField->setValue($xmlReader->getAttribute('w:val', $ddListNode));
358
359                                break;
360                            case 'default':
361                                $formField->setDefault($xmlReader->getAttribute('w:val', $ddListNode));
362
363                                break;
364                            case 'listEntry':
365                                $listEntries[] = $xmlReader->getAttribute('w:val', $ddListNode);
366
367                                break;
368                        }
369                    }
370                    $formField->setEntries($listEntries);
371                    if (null !== $formField->getValue()) {
372                        $formField->setText($listEntries[$formField->getValue()]);
373                    }
374
375                    break;
376                case 'textInput':
377                    foreach ($xmlReader->getElements('*', $node) as $ddListNode) {
378                        switch ($ddListNode->localName) {
379                            case 'default':
380                                $formField->setDefault($xmlReader->getAttribute('w:val', $ddListNode));
381
382                                break;
383                            case 'format':
384                            case 'maxLength':
385                                break;
386                        }
387                    }
388
389                    break;
390                case 'checkBox':
391                    foreach ($xmlReader->getElements('*', $node) as $ddListNode) {
392                        switch ($ddListNode->localName) {
393                            case 'default':
394                                $formField->setDefault($xmlReader->getAttribute('w:val', $ddListNode));
395
396                                break;
397                            case 'checked':
398                                $formField->setValue($xmlReader->getAttribute('w:val', $ddListNode));
399
400                                break;
401                            case 'size':
402                            case 'sizeAuto':
403                                break;
404                        }
405                    }
406
407                    break;
408            }
409        }
410
411        if ('textinput' == $formType) {
412            $ignoreText = true;
413            $textContent = '';
414            foreach ($domNodes as $node) {
415                if ($xmlReader->elementExists('w:fldChar', $node)) {
416                    $fldCharType = $xmlReader->getAttribute('w:fldCharType', $node, 'w:fldChar');
417                    if ('separate' == $fldCharType) {
418                        $ignoreText = false;
419                    } elseif ('end' == $fldCharType) {
420                        $ignoreText = true;
421                    }
422                }
423
424                if (false === $ignoreText) {
425                    $textContent .= $xmlReader->getValue('w:t', $node);
426                }
427            }
428            $formField->setValue(htmlspecialchars($textContent, ENT_QUOTES, 'UTF-8'));
429            $formField->setText(htmlspecialchars($textContent, ENT_QUOTES, 'UTF-8'));
430        }
431    }
432
433    /**
434     * Returns the depth of the Heading, returns 0 for a Title.
435     *
436     * @return null|number
437     */
438    private function getHeadingDepth(?array $paragraphStyle = null)
439    {
440        if (is_array($paragraphStyle) && isset($paragraphStyle['styleName'])) {
441            if ('Title' === $paragraphStyle['styleName']) {
442                return 0;
443            }
444
445            $headingMatches = [];
446            preg_match('/Heading(\d)/', $paragraphStyle['styleName'], $headingMatches);
447            if (!empty($headingMatches)) {
448                return $headingMatches[1];
449            }
450        }
451
452        return null;
453    }
454
455    /**
456     * Read w:r.
457     *
458     * @param AbstractContainer $parent
459     * @param string $docPart
460     * @param mixed $paragraphStyle
461     *
462     * @todo Footnote paragraph style
463     */
464    protected function readRun(XMLReader $xmlReader, DOMElement $domNode, $parent, $docPart, $paragraphStyle = null): void
465    {
466        if (in_array($domNode->nodeName, ['w:ins', 'w:del', 'w:smartTag', 'w:hyperlink', 'w:commentReference'])) {
467            $nodes = $xmlReader->getElements('*', $domNode);
468            foreach ($nodes as $node) {
469                $this->readRun($xmlReader, $node, $parent, $docPart, $paragraphStyle);
470            }
471        } elseif ($domNode->nodeName == 'w:r') {
472            $fontStyle = $this->readFontStyle($xmlReader, $domNode);
473            $nodes = $xmlReader->getElements('*', $domNode);
474            foreach ($nodes as $node) {
475                $this->readRunChild($xmlReader, $node, $parent, $docPart, $paragraphStyle, $fontStyle);
476            }
477        }
478
479        if ($xmlReader->elementExists('.//*["commentReference"=local-name()]', $domNode)) {
480            $node = iterator_to_array($xmlReader->getElements('.//*["commentReference"=local-name()]', $domNode))[0];
481            $attributeIdentifier = $node->attributes->getNamedItem('id');
482            if ($attributeIdentifier) {
483                $id = $attributeIdentifier->nodeValue;
484
485                $this->setCommentReference('start', $id, $parent->getElement($parent->countElements() - 1));
486                $this->setCommentReference('end', $id, $parent->getElement($parent->countElements() - 1));
487            }
488        }
489    }
490
491    /**
492     * Parses nodes under w:r.
493     *
494     * @param string $docPart
495     * @param mixed $paragraphStyle
496     * @param mixed $fontStyle
497     */
498    protected function readRunChild(XMLReader $xmlReader, DOMElement $node, AbstractContainer $parent, $docPart, $paragraphStyle = null, $fontStyle = null): void
499    {
500        $runParent = $node->parentNode->parentNode;
501        if ($node->nodeName == 'w:footnoteReference') {
502            // Footnote
503            $wId = $xmlReader->getAttribute('w:id', $node);
504            $footnote = $parent->addFootnote();
505            $footnote->setRelationId($wId);
506        } elseif ($node->nodeName == 'w:endnoteReference') {
507            // Endnote
508            $wId = $xmlReader->getAttribute('w:id', $node);
509            $endnote = $parent->addEndnote();
510            $endnote->setRelationId($wId);
511        } elseif ($node->nodeName == 'w:pict') {
512            // Image
513            $rId = $xmlReader->getAttribute('r:id', $node, 'v:shape/v:imagedata');
514            $target = $this->getMediaTarget($docPart, $rId);
515            if ($this->hasImageLoading() && null !== $target) {
516                if ('External' == $this->getTargetMode($docPart, $rId)) {
517                    $imageSource = $target;
518                } else {
519                    $imageSource = "zip://{$this->docFile}#{$target}";
520                }
521                $parent->addImage($imageSource);
522            }
523        } elseif ($node->nodeName == 'w:drawing') {
524            // Office 2011 Image
525            $xmlReader->registerNamespace('wp', 'http://schemas.openxmlformats.org/drawingml/2006/wordprocessingDrawing');
526            $xmlReader->registerNamespace('r', 'http://schemas.openxmlformats.org/officeDocument/2006/relationships');
527            $xmlReader->registerNamespace('pic', 'http://schemas.openxmlformats.org/drawingml/2006/picture');
528            $xmlReader->registerNamespace('a', 'http://schemas.openxmlformats.org/drawingml/2006/main');
529
530            $name = $xmlReader->getAttribute('name', $node, 'wp:inline/a:graphic/a:graphicData/pic:pic/pic:nvPicPr/pic:cNvPr');
531            $altText = $xmlReader->getAttribute('descr', $node, 'wp:inline/a:graphic/a:graphicData/pic:pic/pic:nvPicPr/pic:cNvPr');
532            $embedId = $xmlReader->getAttribute('r:embed', $node, 'wp:inline/a:graphic/a:graphicData/pic:pic/pic:blipFill/a:blip');
533            if ($name === null && $altText === null && $embedId === null) { // some Converters puts images on a different path
534                $name = $xmlReader->getAttribute('name', $node, 'wp:anchor/a:graphic/a:graphicData/pic:pic/pic:nvPicPr/pic:cNvPr');
535                $altText = $xmlReader->getAttribute('descr', $node, 'wp:anchor/a:graphic/a:graphicData/pic:pic/pic:nvPicPr/pic:cNvPr');
536                $embedId = $xmlReader->getAttribute('r:embed', $node, 'wp:anchor/a:graphic/a:graphicData/pic:pic/pic:blipFill/a:blip');
537            }
538            $target = $this->getMediaTarget($docPart, $embedId);
539            if ($this->hasImageLoading() && null !== $target) {
540                $imageSource = "zip://{$this->docFile}#{$target}";
541                $parent->addImage($imageSource, null, false, $name, $altText);
542            }
543        } elseif ($node->nodeName == 'w:object') {
544            // Object
545            $rId = $xmlReader->getAttribute('r:id', $node, 'o:OLEObject');
546            // $rIdIcon = $xmlReader->getAttribute('r:id', $domNode, 'w:object/v:shape/v:imagedata');
547            $target = $this->getMediaTarget($docPart, $rId);
548            if (null !== $target) {
549                $textContent = "&lt;Object: {$target}>";
550                $parent->addText($textContent, $fontStyle, $paragraphStyle);
551            }
552        } elseif ($node->nodeName == 'w:br') {
553            $parent->addTextBreak();
554        } elseif ($node->nodeName == 'w:tab') {
555            $parent->addText("\t");
556        } elseif ($node->nodeName == 'mc:AlternateContent') {
557            if ($node->hasChildNodes()) {
558                // Get fallback instead of mc:Choice to make sure it is compatible
559                $fallbackElements = $node->getElementsByTagName('Fallback');
560
561                if ($fallbackElements->length) {
562                    $fallback = $fallbackElements->item(0);
563                    // TextRun
564                    $textContent = htmlspecialchars($fallback->nodeValue, ENT_QUOTES, 'UTF-8');
565
566                    $parent->addText($textContent, $fontStyle, $paragraphStyle);
567                }
568            }
569        } elseif ($node->nodeName == 'w:t' || $node->nodeName == 'w:delText') {
570            // TextRun
571            $textContent = htmlspecialchars($xmlReader->getValue('.', $node), ENT_QUOTES, 'UTF-8');
572
573            if ($runParent->nodeName == 'w:hyperlink') {
574                $rId = $xmlReader->getAttribute('r:id', $runParent);
575                $target = $this->getMediaTarget($docPart, $rId);
576                if (null !== $target) {
577                    $parent->addLink($target, $textContent, $fontStyle, $paragraphStyle);
578                } else {
579                    $parent->addText($textContent, $fontStyle, $paragraphStyle);
580                }
581            } else {
582                /** @var AbstractElement $element */
583                $element = $parent->addText($textContent, $fontStyle, $paragraphStyle);
584                if (in_array($runParent->nodeName, ['w:ins', 'w:del'])) {
585                    $type = ($runParent->nodeName == 'w:del') ? TrackChange::DELETED : TrackChange::INSERTED;
586                    $author = $runParent->getAttribute('w:author');
587                    $date = DateTime::createFromFormat('Y-m-d\TH:i:s\Z', $runParent->getAttribute('w:date'));
588                    $date = $date instanceof DateTime ? $date : null;
589                    $element->setChangeInfo($type, $author, $date);
590                }
591            }
592        } elseif ($node->nodeName == 'w:softHyphen') {
593            $element = $parent->addText("\u{200c}", $fontStyle, $paragraphStyle);
594        } elseif ($node->nodeName == 'w:ruby') {
595            $rubyPropertiesNode = $xmlReader->getElement('w:rubyPr', $node);
596            $properties = $this->readRubyProperties($xmlReader, $rubyPropertiesNode);
597            // read base text node
598            $baseText = new TextRun($paragraphStyle);
599            $baseTextNode = $xmlReader->getElement('w:rubyBase/w:r', $node);
600            $this->readRun($xmlReader, $baseTextNode, $baseText, $docPart, $paragraphStyle);
601            // read the actual ruby text (e.g. furigana in Japanese)
602            $rubyText = new TextRun($paragraphStyle);
603            $rubyTextNode = $xmlReader->getElement('w:rt/w:r', $node);
604            $this->readRun($xmlReader, $rubyTextNode, $rubyText, $docPart, $paragraphStyle);
605            // add element to parent
606            $parent->addRuby($baseText, $rubyText, $properties);
607        }
608    }
609
610    /**
611     * Read w:rubyPr element.
612     *
613     * @param XMLReader $xmlReader reader for XML
614     * @param DOMElement $domNode w:RubyPr element
615     *
616     * @return RubyProperties ruby properties from element
617     */
618    protected function readRubyProperties(XMLReader $xmlReader, DOMElement $domNode): RubyProperties
619    {
620        $rubyAlignment = $xmlReader->getElement('w:rubyAlign', $domNode)->getAttribute('w:val');
621        $rubyHps = $xmlReader->getElement('w:hps', $domNode)->getAttribute('w:val'); // font face
622        $rubyHpsRaise = $xmlReader->getElement('w:hpsRaise', $domNode)->getAttribute('w:val'); // pts above base text
623        $rubyHpsBaseText = $xmlReader->getElement('w:hpsBaseText', $domNode)->getAttribute('w:val'); // base text size
624        $rubyLid = $xmlReader->getElement('w:lid', $domNode)->getAttribute('w:val'); // type of ruby
625        $properties = new RubyProperties();
626        $properties->setAlignment($rubyAlignment);
627        $properties->setFontFaceSize((float) $rubyHps);
628        $properties->setFontPointsAboveBaseText((float) $rubyHpsRaise);
629        $properties->setFontSizeForBaseText((float) $rubyHpsBaseText);
630        $properties->setLanguageId($rubyLid);
631
632        return $properties;
633    }
634
635    /**
636     * Read w:tbl.
637     *
638     * @param mixed $parent
639     * @param string $docPart
640     */
641    protected function readTable(XMLReader $xmlReader, DOMElement $domNode, $parent, $docPart = 'document'): void
642    {
643        // Table style
644        $tblStyle = null;
645        if ($xmlReader->elementExists('w:tblPr', $domNode)) {
646            $tblStyle = $this->readTableStyle($xmlReader, $domNode);
647        }
648
649        /** @var \PhpOffice\PhpWord\Element\Table $table Type hint */
650        $table = $parent->addTable($tblStyle);
651        $tblNodes = $xmlReader->getElements('*', $domNode);
652        foreach ($tblNodes as $tblNode) {
653            if ('w:tblGrid' == $tblNode->nodeName) { // Column
654                // @todo Do something with table columns
655            } elseif ('w:tr' == $tblNode->nodeName) { // Row
656                $rowHeight = $xmlReader->getAttribute('w:val', $tblNode, 'w:trPr/w:trHeight');
657                $rowHRule = $xmlReader->getAttribute('w:hRule', $tblNode, 'w:trPr/w:trHeight');
658                $rowHRule = $rowHRule == 'exact';
659                $rowStyle = [
660                    'tblHeader' => $xmlReader->elementExists('w:trPr/w:tblHeader', $tblNode),
661                    'cantSplit' => $xmlReader->elementExists('w:trPr/w:cantSplit', $tblNode),
662                    'exactHeight' => $rowHRule,
663                ];
664
665                $row = $table->addRow($rowHeight, $rowStyle);
666                $rowNodes = $xmlReader->getElements('*', $tblNode);
667                foreach ($rowNodes as $rowNode) {
668                    if ('w:trPr' == $rowNode->nodeName) { // Row style
669                        // @todo Do something with row style
670                    } elseif ('w:tc' == $rowNode->nodeName) { // Cell
671                        $cellWidth = $xmlReader->getAttribute('w:w', $rowNode, 'w:tcPr/w:tcW');
672                        $cellStyle = null;
673                        if ($xmlReader->elementExists('w:tcPr', $rowNode)) {
674                            $cellStyle = $this->readCellStyle($xmlReader, $rowNode);
675                        }
676
677                        $cell = $row->addCell($cellWidth, $cellStyle);
678                        $cellNodes = $xmlReader->getElements('*', $rowNode);
679                        foreach ($cellNodes as $cellNode) {
680                            if ('w:p' == $cellNode->nodeName) { // Paragraph
681                                $this->readParagraph($xmlReader, $cellNode, $cell, $docPart);
682                            } elseif ($cellNode->nodeName == 'w:tbl') { // Table
683                                $this->readTable($xmlReader, $cellNode, $cell, $docPart);
684                            }
685                        }
686                    }
687                }
688            }
689        }
690    }
691
692    /**
693     * Read w:pPr.
694     *
695     * @return null|array
696     */
697    protected function readParagraphStyle(XMLReader $xmlReader, DOMElement $domNode)
698    {
699        if (!$xmlReader->elementExists('w:pPr', $domNode)) {
700            return null;
701        }
702
703        $styleNode = $xmlReader->getElement('w:pPr', $domNode);
704        $styleDefs = [
705            'styleName' => [self::READ_VALUE, ['w:pStyle', 'w:name']],
706            'alignment' => [self::READ_VALUE, 'w:jc'],
707            'basedOn' => [self::READ_VALUE, 'w:basedOn'],
708            'next' => [self::READ_VALUE, 'w:next'],
709            'indentLeft' => [self::READ_VALUE, 'w:ind', 'w:left'],
710            'indentRight' => [self::READ_VALUE, 'w:ind', 'w:right'],
711            'indentHanging' => [self::READ_VALUE, 'w:ind', 'w:hanging'],
712            'indentFirstLine' => [self::READ_VALUE, 'w:ind', 'w:firstLine'],
713            'indentFirstLineChars' => [self::READ_VALUE, 'w:ind', 'w:firstLineChars'],
714            'spaceAfter' => [self::READ_VALUE, 'w:spacing', 'w:after'],
715            'spaceBefore' => [self::READ_VALUE, 'w:spacing', 'w:before'],
716            'widowControl' => [self::READ_FALSE, 'w:widowControl'],
717            'keepNext' => [self::READ_TRUE,  'w:keepNext'],
718            'keepLines' => [self::READ_TRUE,  'w:keepLines'],
719            'pageBreakBefore' => [self::READ_TRUE,  'w:pageBreakBefore'],
720            'contextualSpacing' => [self::READ_TRUE,  'w:contextualSpacing'],
721            'bidi' => [self::READ_TRUE,  'w:bidi'],
722            'suppressAutoHyphens' => [self::READ_TRUE,  'w:suppressAutoHyphens'],
723            'borderTopStyle' => [self::READ_VALUE, 'w:pBdr/w:top'],
724            'borderTopColor' => [self::READ_VALUE, 'w:pBdr/w:top', 'w:color'],
725            'borderTopSize' => [self::READ_VALUE, 'w:pBdr/w:top', 'w:sz'],
726            'borderRightStyle' => [self::READ_VALUE, 'w:pBdr/w:right'],
727            'borderRightColor' => [self::READ_VALUE, 'w:pBdr/w:right', 'w:color'],
728            'borderRightSize' => [self::READ_VALUE, 'w:pBdr/w:right', 'w:sz'],
729            'borderBottomStyle' => [self::READ_VALUE, 'w:pBdr/w:bottom'],
730            'borderBottomColor' => [self::READ_VALUE, 'w:pBdr/w:bottom', 'w:color'],
731            'borderBottomSize' => [self::READ_VALUE, 'w:pBdr/w:bottom', 'w:sz'],
732            'borderLeftStyle' => [self::READ_VALUE, 'w:pBdr/w:left'],
733            'borderLeftColor' => [self::READ_VALUE, 'w:pBdr/w:left', 'w:color'],
734            'borderLeftSize' => [self::READ_VALUE, 'w:pBdr/w:left', 'w:sz'],
735        ];
736
737        return $this->readStyleDefs($xmlReader, $styleNode, $styleDefs);
738    }
739
740    /**
741     * Read w:rPr.
742     *
743     * @return null|array
744     */
745    protected function readFontStyle(XMLReader $xmlReader, DOMElement $domNode)
746    {
747        if (null === $domNode) {
748            return null;
749        }
750        // Hyperlink has an extra w:r child
751        if ('w:hyperlink' == $domNode->nodeName) {
752            $domNode = $xmlReader->getElement('w:r', $domNode);
753        }
754        if (!$xmlReader->elementExists('w:rPr', $domNode)) {
755            return null;
756        }
757
758        $styleNode = $xmlReader->getElement('w:rPr', $domNode);
759        $styleDefs = [
760            'styleName' => [self::READ_VALUE, 'w:rStyle'],
761            'name' => [self::READ_VALUE, 'w:rFonts', ['w:ascii', 'w:hAnsi', 'w:eastAsia', 'w:cs']],
762            'hint' => [self::READ_VALUE, 'w:rFonts', 'w:hint'],
763            'size' => [self::READ_SIZE,  ['w:sz', 'w:szCs']],
764            'color' => [self::READ_VALUE, 'w:color'],
765            'underline' => [self::READ_VALUE, 'w:u'],
766            'bold' => [self::READ_TRUE,  'w:b'],
767            'italic' => [self::READ_TRUE,  'w:i'],
768            'strikethrough' => [self::READ_TRUE,  'w:strike'],
769            'doubleStrikethrough' => [self::READ_TRUE,  'w:dstrike'],
770            'smallCaps' => [self::READ_TRUE,  'w:smallCaps'],
771            'allCaps' => [self::READ_TRUE,  'w:caps'],
772            'superScript' => [self::READ_EQUAL, 'w:vertAlign', 'w:val', 'superscript'],
773            'subScript' => [self::READ_EQUAL, 'w:vertAlign', 'w:val', 'subscript'],
774            'fgColor' => [self::READ_VALUE, 'w:highlight'],
775            'rtl' => [self::READ_TRUE,  'w:rtl'],
776            'lang' => [self::READ_VALUE, 'w:lang'],
777            'position' => [self::READ_VALUE, 'w:position'],
778            'hidden' => [self::READ_TRUE,  'w:vanish'],
779        ];
780
781        return $this->readStyleDefs($xmlReader, $styleNode, $styleDefs);
782    }
783
784    /**
785     * Read w:tblPr.
786     *
787     * @return null|array|string
788     *
789     * @todo Capture w:tblStylePr w:type="firstRow"
790     */
791    protected function readTableStyle(XMLReader $xmlReader, DOMElement $domNode)
792    {
793        $style = null;
794        $margins = ['top', 'left', 'bottom', 'right'];
795        $borders = array_merge($margins, ['insideH', 'insideV']);
796
797        if ($xmlReader->elementExists('w:tblPr', $domNode)) {
798            if ($xmlReader->elementExists('w:tblPr/w:tblStyle', $domNode)) {
799                $style = $xmlReader->getAttribute('w:val', $domNode, 'w:tblPr/w:tblStyle');
800            } else {
801                $styleNode = $xmlReader->getElement('w:tblPr', $domNode);
802                $styleDefs = [];
803                foreach ($margins as $side) {
804                    $ucfSide = ucfirst($side);
805                    $styleDefs["cellMargin$ucfSide"] = [self::READ_VALUE, "w:tblCellMar/w:$side", 'w:w'];
806                }
807                foreach ($borders as $side) {
808                    $ucfSide = ucfirst($side);
809                    $styleDefs["border{$ucfSide}Size"] = [self::READ_VALUE, "w:tblBorders/w:$side", 'w:sz'];
810                    $styleDefs["border{$ucfSide}Color"] = [self::READ_VALUE, "w:tblBorders/w:$side", 'w:color'];
811                    $styleDefs["border{$ucfSide}Style"] = [self::READ_VALUE, "w:tblBorders/w:$side", 'w:val'];
812                }
813                $styleDefs['layout'] = [self::READ_VALUE, 'w:tblLayout', 'w:type'];
814                $styleDefs['bidiVisual'] = [self::READ_TRUE, 'w:bidiVisual'];
815                $styleDefs['cellSpacing'] = [self::READ_VALUE, 'w:tblCellSpacing', 'w:w'];
816                $style = $this->readStyleDefs($xmlReader, $styleNode, $styleDefs);
817
818                $tablePositionNode = $xmlReader->getElement('w:tblpPr', $styleNode);
819                if ($tablePositionNode !== null) {
820                    $style['position'] = $this->readTablePosition($xmlReader, $tablePositionNode);
821                }
822
823                $indentNode = $xmlReader->getElement('w:tblInd', $styleNode);
824                if ($indentNode !== null) {
825                    $style['indent'] = $this->readTableIndent($xmlReader, $indentNode);
826                }
827            }
828        }
829
830        return $style;
831    }
832
833    /**
834     * Read w:tblpPr.
835     *
836     * @return array
837     */
838    private function readTablePosition(XMLReader $xmlReader, DOMElement $domNode)
839    {
840        $styleDefs = [
841            'leftFromText' => [self::READ_VALUE, '.', 'w:leftFromText'],
842            'rightFromText' => [self::READ_VALUE, '.', 'w:rightFromText'],
843            'topFromText' => [self::READ_VALUE, '.', 'w:topFromText'],
844            'bottomFromText' => [self::READ_VALUE, '.', 'w:bottomFromText'],
845            'vertAnchor' => [self::READ_VALUE, '.', 'w:vertAnchor'],
846            'horzAnchor' => [self::READ_VALUE, '.', 'w:horzAnchor'],
847            'tblpXSpec' => [self::READ_VALUE, '.', 'w:tblpXSpec'],
848            'tblpX' => [self::READ_VALUE, '.', 'w:tblpX'],
849            'tblpYSpec' => [self::READ_VALUE, '.', 'w:tblpYSpec'],
850            'tblpY' => [self::READ_VALUE, '.', 'w:tblpY'],
851        ];
852
853        return $this->readStyleDefs($xmlReader, $domNode, $styleDefs);
854    }
855
856    /**
857     * Read w:tblInd.
858     *
859     * @return TblWidthComplexType
860     */
861    private function readTableIndent(XMLReader $xmlReader, DOMElement $domNode)
862    {
863        $styleDefs = [
864            'value' => [self::READ_VALUE, '.', 'w:w'],
865            'type' => [self::READ_VALUE, '.', 'w:type'],
866        ];
867        $styleDefs = $this->readStyleDefs($xmlReader, $domNode, $styleDefs);
868
869        return new TblWidthComplexType((int) $styleDefs['value'], $styleDefs['type']);
870    }
871
872    /**
873     * Read w:tcPr.
874     *
875     * @return null|array
876     */
877    private function readCellStyle(XMLReader $xmlReader, DOMElement $domNode)
878    {
879        $styleDefs = [
880            'valign' => [self::READ_VALUE, 'w:vAlign'],
881            'textDirection' => [self::READ_VALUE, 'w:textDirection'],
882            'gridSpan' => [self::READ_VALUE, 'w:gridSpan'],
883            'vMerge' => [self::READ_VALUE, 'w:vMerge', null, null, 'continue'],
884            'bgColor' => [self::READ_VALUE, 'w:shd', 'w:fill'],
885            'noWrap' => [self::READ_VALUE, 'w:noWrap', null, null, true],
886        ];
887        $style = null;
888
889        if ($xmlReader->elementExists('w:tcPr', $domNode)) {
890            $styleNode = $xmlReader->getElement('w:tcPr', $domNode);
891
892            $borders = ['top', 'left', 'bottom', 'right'];
893            foreach ($borders as $side) {
894                $ucfSide = ucfirst($side);
895
896                $styleDefs['border' . $ucfSide . 'Size'] = [self::READ_VALUE, 'w:tcBorders/w:' . $side, 'w:sz'];
897                $styleDefs['border' . $ucfSide . 'Color'] = [self::READ_VALUE, 'w:tcBorders/w:' . $side, 'w:color'];
898                $styleDefs['border' . $ucfSide . 'Style'] = [self::READ_VALUE, 'w:tcBorders/w:' . $side, 'w:val'];
899            }
900
901            $style = $this->readStyleDefs($xmlReader, $styleNode, $styleDefs);
902        }
903
904        return $style;
905    }
906
907    /**
908     * Returns the first child element found.
909     *
910     * @param null|array|string $elements
911     *
912     * @return null|string
913     */
914    private function findPossibleElement(XMLReader $xmlReader, ?DOMElement $parentNode = null, $elements = null)
915    {
916        if (is_array($elements)) {
917            //if element is an array, we take the first element that exists in the XML
918            foreach ($elements as $possibleElement) {
919                if ($xmlReader->elementExists($possibleElement, $parentNode)) {
920                    return $possibleElement;
921                }
922            }
923        } else {
924            return $elements;
925        }
926
927        return null;
928    }
929
930    /**
931     * Returns the first attribute found.
932     *
933     * @param array|string $attributes
934     *
935     * @return null|string
936     */
937    private function findPossibleAttribute(XMLReader $xmlReader, DOMElement $node, $attributes)
938    {
939        //if attribute is an array, we take the first attribute that exists in the XML
940        if (is_array($attributes)) {
941            foreach ($attributes as $possibleAttribute) {
942                if ($xmlReader->getAttribute($possibleAttribute, $node)) {
943                    return $possibleAttribute;
944                }
945            }
946
947            return null;
948        }
949
950        return $attributes;
951    }
952
953    /**
954     * Read style definition.
955     *
956     * @param array $styleDefs
957     *
958     * @ignoreScrutinizerPatch
959     *
960     * @return array
961     */
962    protected function readStyleDefs(XMLReader $xmlReader, ?DOMElement $parentNode = null, $styleDefs = [])
963    {
964        $styles = [];
965
966        foreach ($styleDefs as $styleProp => $styleVal) {
967            [$method, $element, $attribute, $expected, $default] = array_pad($styleVal, 5, null);
968
969            $element = $this->findPossibleElement($xmlReader, $parentNode, $element);
970            if ($element === null) {
971                continue;
972            }
973
974            if ($xmlReader->elementExists($element, $parentNode)) {
975                $node = $xmlReader->getElement($element, $parentNode);
976
977                $attribute = $this->findPossibleAttribute($xmlReader, $node, $attribute);
978
979                // Use w:val as default if no attribute assigned
980                $attribute = ($attribute === null) ? 'w:val' : $attribute;
981                $attributeValue = $xmlReader->getAttribute($attribute, $node) ?? $default;
982
983                $styleValue = $this->readStyleDef($method, $attributeValue, $expected);
984                if ($styleValue !== null) {
985                    $styles[$styleProp] = $styleValue;
986                }
987            }
988        }
989
990        return $styles;
991    }
992
993    /**
994     * Return style definition based on conversion method.
995     *
996     * @param string $method
997     *
998     * @ignoreScrutinizerPatch
999     *
1000     * @param null|string $attributeValue
1001     * @param mixed $expected
1002     *
1003     * @return mixed
1004     */
1005    private function readStyleDef($method, $attributeValue, $expected)
1006    {
1007        $style = $attributeValue;
1008
1009        if (self::READ_SIZE == $method) {
1010            $style = $attributeValue / 2;
1011        } elseif (self::READ_TRUE == $method) {
1012            $style = $this->isOn($attributeValue);
1013        } elseif (self::READ_FALSE == $method) {
1014            $style = !$this->isOn($attributeValue);
1015        } elseif (self::READ_EQUAL == $method) {
1016            $style = $attributeValue == $expected;
1017        }
1018
1019        return $style;
1020    }
1021
1022    /**
1023     * Parses the value of the on/off value, null is considered true as it means the w:val attribute was not present.
1024     *
1025     * @see http://www.datypic.com/sc/ooxml/t-w_ST_OnOff.html
1026     *
1027     * @param string $value
1028     *
1029     * @return bool
1030     */
1031    private function isOn($value = null)
1032    {
1033        return $value === null || $value === '1' || $value === 'true' || $value === 'on';
1034    }
1035
1036    /**
1037     * Returns the target of image, object, or link as stored in ::readMainRels.
1038     *
1039     * @param string $docPart
1040     * @param string $rId
1041     *
1042     * @return null|string
1043     */
1044    private function getMediaTarget($docPart, $rId)
1045    {
1046        $target = null;
1047
1048        if (isset($this->rels[$docPart], $this->rels[$docPart][$rId])) {
1049            $target = $this->rels[$docPart][$rId]['target'];
1050        }
1051
1052        return $target;
1053    }
1054
1055    /**
1056     * Returns the target mode.
1057     *
1058     * @param string $docPart
1059     * @param string $rId
1060     *
1061     * @return null|string
1062     */
1063    private function getTargetMode($docPart, $rId)
1064    {
1065        $mode = null;
1066
1067        if (isset($this->rels[$docPart], $this->rels[$docPart][$rId])) {
1068            $mode = $this->rels[$docPart][$rId]['targetMode'];
1069        }
1070
1071        return $mode;
1072    }
1073}