Code Coverage |
||||||||||
Lines |
Functions and Methods |
Classes and Traits |
||||||||
| Total | |
97.76% |
479 / 490 |
|
75.00% |
21 / 28 |
CRAP | |
0.00% |
0 / 1 |
| AbstractPart | |
97.76% |
479 / 490 |
|
75.00% |
21 / 28 |
168 | |
0.00% |
0 / 1 |
| read | n/a |
0 / 0 |
n/a |
0 / 0 |
0 | |||||
| __construct | |
100.00% |
2 / 2 |
|
100.00% |
1 / 1 |
1 | |||
| setRels | |
100.00% |
1 / 1 |
|
100.00% |
1 / 1 |
1 | |||
| setImageLoading | |
100.00% |
2 / 2 |
|
100.00% |
1 / 1 |
1 | |||
| hasImageLoading | |
100.00% |
1 / 1 |
|
100.00% |
1 / 1 |
1 | |||
| getCommentReferences | |
100.00% |
1 / 1 |
|
100.00% |
1 / 1 |
1 | |||
| setCommentReferences | |
100.00% |
2 / 2 |
|
100.00% |
1 / 1 |
1 | |||
| setCommentReference | |
88.89% |
8 / 9 |
|
0.00% |
0 / 1 |
3.01 | |||
| getCommentReference | |
66.67% |
2 / 3 |
|
0.00% |
0 / 1 |
2.15 | |||
| readParagraph | |
98.82% |
84 / 85 |
|
0.00% |
0 / 1 |
32 | |||
| readFormField | |
95.24% |
60 / 63 |
|
0.00% |
0 / 1 |
27 | |||
| getHeadingDepth | |
100.00% |
8 / 8 |
|
100.00% |
1 / 1 |
5 | |||
| readRun | |
100.00% |
16 / 16 |
|
100.00% |
1 / 1 |
7 | |||
| readRunChild | |
97.40% |
75 / 77 |
|
0.00% |
0 / 1 |
28 | |||
| readRubyProperties | |
100.00% |
12 / 12 |
|
100.00% |
1 / 1 |
1 | |||
| readTable | |
100.00% |
32 / 32 |
|
100.00% |
1 / 1 |
12 | |||
| readParagraphStyle | |
100.00% |
36 / 36 |
|
100.00% |
1 / 1 |
2 | |||
| readFontStyle | |
93.10% |
27 / 29 |
|
0.00% |
0 / 1 |
4.01 | |||
| readTableStyle | |
96.30% |
26 / 27 |
|
0.00% |
0 / 1 |
7 | |||
| readTablePosition | |
100.00% |
13 / 13 |
|
100.00% |
1 / 1 |
1 | |||
| readTableIndent | |
100.00% |
6 / 6 |
|
100.00% |
1 / 1 |
1 | |||
| readCellStyle | |
100.00% |
19 / 19 |
|
100.00% |
1 / 1 |
3 | |||
| findPossibleElement | |
100.00% |
6 / 6 |
|
100.00% |
1 / 1 |
4 | |||
| findPossibleAttribute | |
100.00% |
6 / 6 |
|
100.00% |
1 / 1 |
4 | |||
| readStyleDefs | |
100.00% |
15 / 15 |
|
100.00% |
1 / 1 |
6 | |||
| readStyleDef | |
100.00% |
10 / 10 |
|
100.00% |
1 / 1 |
5 | |||
| isOn | |
100.00% |
1 / 1 |
|
100.00% |
1 / 1 |
4 | |||
| getMediaTarget | |
100.00% |
4 / 4 |
|
100.00% |
1 / 1 |
2 | |||
| getTargetMode | |
100.00% |
4 / 4 |
|
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 | |
| 19 | namespace PhpOffice\PhpWord\Reader\Word2007; |
| 20 | |
| 21 | use DateTime; |
| 22 | use DOMElement; |
| 23 | use InvalidArgumentException; |
| 24 | use PhpOffice\Math\Reader\OfficeMathML; |
| 25 | use PhpOffice\PhpWord\ComplexType\RubyProperties; |
| 26 | use PhpOffice\PhpWord\ComplexType\TblWidth as TblWidthComplexType; |
| 27 | use PhpOffice\PhpWord\Element\AbstractContainer; |
| 28 | use PhpOffice\PhpWord\Element\AbstractElement; |
| 29 | use PhpOffice\PhpWord\Element\FormField; |
| 30 | use PhpOffice\PhpWord\Element\Ruby; |
| 31 | use PhpOffice\PhpWord\Element\Text; |
| 32 | use PhpOffice\PhpWord\Element\TextRun; |
| 33 | use PhpOffice\PhpWord\Element\TrackChange; |
| 34 | use PhpOffice\PhpWord\PhpWord; |
| 35 | use 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 | */ |
| 44 | abstract 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 | /** @phpstan-ignore offsetAccess.notFound */ |
| 373 | $formField->setText($listEntries[$formField->getValue()]); |
| 374 | } |
| 375 | |
| 376 | break; |
| 377 | case 'textInput': |
| 378 | foreach ($xmlReader->getElements('*', $node) as $ddListNode) { |
| 379 | switch ($ddListNode->localName) { |
| 380 | case 'default': |
| 381 | $formField->setDefault($xmlReader->getAttribute('w:val', $ddListNode)); |
| 382 | |
| 383 | break; |
| 384 | case 'format': |
| 385 | case 'maxLength': |
| 386 | break; |
| 387 | } |
| 388 | } |
| 389 | |
| 390 | break; |
| 391 | case 'checkBox': |
| 392 | foreach ($xmlReader->getElements('*', $node) as $ddListNode) { |
| 393 | switch ($ddListNode->localName) { |
| 394 | case 'default': |
| 395 | $formField->setDefault($xmlReader->getAttribute('w:val', $ddListNode)); |
| 396 | |
| 397 | break; |
| 398 | case 'checked': |
| 399 | $formField->setValue($xmlReader->getAttribute('w:val', $ddListNode)); |
| 400 | |
| 401 | break; |
| 402 | case 'size': |
| 403 | case 'sizeAuto': |
| 404 | break; |
| 405 | } |
| 406 | } |
| 407 | |
| 408 | break; |
| 409 | } |
| 410 | } |
| 411 | |
| 412 | if ('textinput' == $formType) { |
| 413 | $ignoreText = true; |
| 414 | $textContent = ''; |
| 415 | foreach ($domNodes as $node) { |
| 416 | if ($xmlReader->elementExists('w:fldChar', $node)) { |
| 417 | $fldCharType = $xmlReader->getAttribute('w:fldCharType', $node, 'w:fldChar'); |
| 418 | if ('separate' == $fldCharType) { |
| 419 | $ignoreText = false; |
| 420 | } elseif ('end' == $fldCharType) { |
| 421 | $ignoreText = true; |
| 422 | } |
| 423 | } |
| 424 | |
| 425 | if (false === $ignoreText) { |
| 426 | $textContent .= $xmlReader->getValue('w:t', $node); |
| 427 | } |
| 428 | } |
| 429 | $formField->setValue(htmlspecialchars($textContent, ENT_QUOTES, 'UTF-8')); |
| 430 | $formField->setText(htmlspecialchars($textContent, ENT_QUOTES, 'UTF-8')); |
| 431 | } |
| 432 | } |
| 433 | |
| 434 | /** |
| 435 | * Returns the depth of the Heading, returns 0 for a Title. |
| 436 | * |
| 437 | * @return null|number |
| 438 | */ |
| 439 | private function getHeadingDepth(?array $paragraphStyle = null) |
| 440 | { |
| 441 | if (is_array($paragraphStyle) && isset($paragraphStyle['styleName'])) { |
| 442 | if ('Title' === $paragraphStyle['styleName']) { |
| 443 | return 0; |
| 444 | } |
| 445 | |
| 446 | $headingMatches = []; |
| 447 | preg_match('/Heading(\d)/', $paragraphStyle['styleName'], $headingMatches); |
| 448 | if (!empty($headingMatches)) { |
| 449 | return $headingMatches[1]; |
| 450 | } |
| 451 | } |
| 452 | |
| 453 | return null; |
| 454 | } |
| 455 | |
| 456 | /** |
| 457 | * Read w:r. |
| 458 | * |
| 459 | * @param AbstractContainer $parent |
| 460 | * @param string $docPart |
| 461 | * @param mixed $paragraphStyle |
| 462 | * |
| 463 | * @todo Footnote paragraph style |
| 464 | */ |
| 465 | protected function readRun(XMLReader $xmlReader, DOMElement $domNode, $parent, $docPart, $paragraphStyle = null): void |
| 466 | { |
| 467 | if (in_array($domNode->nodeName, ['w:ins', 'w:del', 'w:smartTag', 'w:hyperlink', 'w:commentReference'])) { |
| 468 | $nodes = $xmlReader->getElements('*', $domNode); |
| 469 | foreach ($nodes as $node) { |
| 470 | $this->readRun($xmlReader, $node, $parent, $docPart, $paragraphStyle); |
| 471 | } |
| 472 | } elseif ($domNode->nodeName == 'w:r') { |
| 473 | $fontStyle = $this->readFontStyle($xmlReader, $domNode); |
| 474 | $nodes = $xmlReader->getElements('*', $domNode); |
| 475 | foreach ($nodes as $node) { |
| 476 | $this->readRunChild($xmlReader, $node, $parent, $docPart, $paragraphStyle, $fontStyle); |
| 477 | } |
| 478 | } |
| 479 | |
| 480 | if ($xmlReader->elementExists('.//*["commentReference"=local-name()]', $domNode)) { |
| 481 | $node = iterator_to_array($xmlReader->getElements('.//*["commentReference"=local-name()]', $domNode))[0]; |
| 482 | $attributeIdentifier = $node->attributes->getNamedItem('id'); |
| 483 | if ($attributeIdentifier) { |
| 484 | $id = $attributeIdentifier->nodeValue; |
| 485 | |
| 486 | $this->setCommentReference('start', $id, $parent->getElement($parent->countElements() - 1)); |
| 487 | $this->setCommentReference('end', $id, $parent->getElement($parent->countElements() - 1)); |
| 488 | } |
| 489 | } |
| 490 | } |
| 491 | |
| 492 | /** |
| 493 | * Parses nodes under w:r. |
| 494 | * |
| 495 | * @param string $docPart |
| 496 | * @param mixed $paragraphStyle |
| 497 | * @param mixed $fontStyle |
| 498 | */ |
| 499 | protected function readRunChild(XMLReader $xmlReader, DOMElement $node, AbstractContainer $parent, $docPart, $paragraphStyle = null, $fontStyle = null): void |
| 500 | { |
| 501 | $runParent = $node->parentNode->parentNode; |
| 502 | if ($node->nodeName == 'w:footnoteReference') { |
| 503 | // Footnote |
| 504 | $wId = $xmlReader->getAttribute('w:id', $node); |
| 505 | $footnote = $parent->addFootnote(); |
| 506 | $footnote->setRelationId($wId); |
| 507 | } elseif ($node->nodeName == 'w:endnoteReference') { |
| 508 | // Endnote |
| 509 | $wId = $xmlReader->getAttribute('w:id', $node); |
| 510 | $endnote = $parent->addEndnote(); |
| 511 | $endnote->setRelationId($wId); |
| 512 | } elseif ($node->nodeName == 'w:pict') { |
| 513 | // Image |
| 514 | $xmlReader->registerNamespace('w10', 'urn:schemas-microsoft-com:office:word'); |
| 515 | $shape = $xmlReader->getElement('v:shape', $node); |
| 516 | $rId = $xmlReader->getAttribute('r:id', $shape, 'v:imagedata'); |
| 517 | $target = $this->getMediaTarget($docPart, $rId); |
| 518 | if ($this->hasImageLoading() && null !== $shape && null !== $target) { |
| 519 | if ('External' == $this->getTargetMode($docPart, $rId)) { |
| 520 | $imageSource = $target; |
| 521 | } else { |
| 522 | $imageSource = "zip://{$this->docFile}#{$target}"; |
| 523 | } |
| 524 | $parent->addImage($imageSource, ImageStyle::readVml($xmlReader, $shape), false, null, $xmlReader->getAttribute('alt', $shape)); |
| 525 | } |
| 526 | } elseif ($node->nodeName == 'w:drawing') { |
| 527 | // Office 2011 Image |
| 528 | $xmlReader->registerNamespace('wp', 'http://schemas.openxmlformats.org/drawingml/2006/wordprocessingDrawing'); |
| 529 | $xmlReader->registerNamespace('r', 'http://schemas.openxmlformats.org/officeDocument/2006/relationships'); |
| 530 | $xmlReader->registerNamespace('pic', 'http://schemas.openxmlformats.org/drawingml/2006/picture'); |
| 531 | $xmlReader->registerNamespace('a', 'http://schemas.openxmlformats.org/drawingml/2006/main'); |
| 532 | |
| 533 | // An inline image, or a floating one |
| 534 | $frame = $xmlReader->getElement('wp:inline', $node) ?? $xmlReader->getElement('wp:anchor', $node); |
| 535 | $picture = 'a:graphic/a:graphicData/pic:pic'; |
| 536 | // The alternative text is wp:docPr/@descr; pic:cNvPr may repeat it |
| 537 | $name = $xmlReader->getAttribute('name', $frame, $picture . '/pic:nvPicPr/pic:cNvPr'); |
| 538 | $altText = $xmlReader->getAttribute('descr', $frame, 'wp:docPr') ?? $xmlReader->getAttribute('descr', $frame, $picture . '/pic:nvPicPr/pic:cNvPr'); |
| 539 | $embedId = $xmlReader->getAttribute('r:embed', $frame, $picture . '/pic:blipFill/a:blip'); |
| 540 | $target = $this->getMediaTarget($docPart, $embedId); |
| 541 | if ($this->hasImageLoading() && null !== $frame && null !== $target) { |
| 542 | $imageSource = "zip://{$this->docFile}#{$target}"; |
| 543 | $parent->addImage($imageSource, ImageStyle::readDrawing($xmlReader, $frame), false, $name, $altText); |
| 544 | } |
| 545 | } elseif ($node->nodeName == 'w:object') { |
| 546 | // Object |
| 547 | $rId = $xmlReader->getAttribute('r:id', $node, 'o:OLEObject'); |
| 548 | // $rIdIcon = $xmlReader->getAttribute('r:id', $domNode, 'w:object/v:shape/v:imagedata'); |
| 549 | $target = $this->getMediaTarget($docPart, $rId); |
| 550 | if (null !== $target) { |
| 551 | $textContent = "<Object: {$target}>"; |
| 552 | $parent->addText($textContent, $fontStyle, $paragraphStyle); |
| 553 | } |
| 554 | } elseif ($node->nodeName == 'w:br') { |
| 555 | $parent->addTextBreak(); |
| 556 | } elseif ($node->nodeName == 'w:tab') { |
| 557 | $parent->addText("\t"); |
| 558 | } elseif ($node->nodeName == 'mc:AlternateContent') { |
| 559 | if ($node->hasChildNodes()) { |
| 560 | // Get fallback instead of mc:Choice to make sure it is compatible |
| 561 | $fallbackElements = $node->getElementsByTagName('Fallback'); |
| 562 | |
| 563 | if ($fallbackElements->length) { |
| 564 | $fallback = $fallbackElements->item(0); |
| 565 | // TextRun |
| 566 | $textContent = htmlspecialchars($fallback->nodeValue ?? '', ENT_QUOTES, 'UTF-8'); |
| 567 | |
| 568 | $parent->addText($textContent, $fontStyle, $paragraphStyle); |
| 569 | } |
| 570 | } |
| 571 | } elseif ($node->nodeName == 'w:t' || $node->nodeName == 'w:delText') { |
| 572 | // TextRun |
| 573 | $textContent = htmlspecialchars($xmlReader->getValue('.', $node) ?? '', ENT_QUOTES, 'UTF-8'); |
| 574 | |
| 575 | if ($runParent->nodeName == 'w:hyperlink') { |
| 576 | $rId = $xmlReader->getAttribute('r:id', $runParent); |
| 577 | $target = $this->getMediaTarget($docPart, $rId); |
| 578 | if (null !== $target) { |
| 579 | $parent->addLink($target, $textContent, $fontStyle, $paragraphStyle); |
| 580 | } else { |
| 581 | $parent->addText($textContent, $fontStyle, $paragraphStyle); |
| 582 | } |
| 583 | } else { |
| 584 | /** @var AbstractElement $element */ |
| 585 | $element = $parent->addText($textContent, $fontStyle, $paragraphStyle); |
| 586 | if (in_array($runParent->nodeName, ['w:ins', 'w:del'])) { |
| 587 | $type = ($runParent->nodeName == 'w:del') ? TrackChange::DELETED : TrackChange::INSERTED; |
| 588 | $author = $runParent->getAttribute('w:author'); |
| 589 | $date = DateTime::createFromFormat('Y-m-d\TH:i:s\Z', $runParent->getAttribute('w:date')); |
| 590 | $date = $date instanceof DateTime ? $date : null; |
| 591 | $element->setChangeInfo($type, $author, $date); |
| 592 | } |
| 593 | } |
| 594 | } elseif ($node->nodeName == 'w:softHyphen') { |
| 595 | $element = $parent->addText("\u{200c}", $fontStyle, $paragraphStyle); |
| 596 | } elseif ($node->nodeName == 'w:ruby') { |
| 597 | $rubyPropertiesNode = $xmlReader->getElement('w:rubyPr', $node); |
| 598 | $properties = $this->readRubyProperties($xmlReader, $rubyPropertiesNode); |
| 599 | // read base text node |
| 600 | $baseText = new TextRun($paragraphStyle); |
| 601 | $baseTextNode = $xmlReader->getElement('w:rubyBase/w:r', $node); |
| 602 | $this->readRun($xmlReader, $baseTextNode, $baseText, $docPart, $paragraphStyle); |
| 603 | // read the actual ruby text (e.g. furigana in Japanese) |
| 604 | $rubyText = new TextRun($paragraphStyle); |
| 605 | $rubyTextNode = $xmlReader->getElement('w:rt/w:r', $node); |
| 606 | $this->readRun($xmlReader, $rubyTextNode, $rubyText, $docPart, $paragraphStyle); |
| 607 | // add element to parent |
| 608 | $parent->addRuby($baseText, $rubyText, $properties); |
| 609 | } |
| 610 | } |
| 611 | |
| 612 | /** |
| 613 | * Read w:rubyPr element. |
| 614 | * |
| 615 | * @param XMLReader $xmlReader reader for XML |
| 616 | * @param DOMElement $domNode w:RubyPr element |
| 617 | * |
| 618 | * @return RubyProperties ruby properties from element |
| 619 | */ |
| 620 | protected function readRubyProperties(XMLReader $xmlReader, DOMElement $domNode): RubyProperties |
| 621 | { |
| 622 | $rubyAlignment = $xmlReader->getElement('w:rubyAlign', $domNode)->getAttribute('w:val'); |
| 623 | $rubyHps = $xmlReader->getElement('w:hps', $domNode)->getAttribute('w:val'); // font face |
| 624 | $rubyHpsRaise = $xmlReader->getElement('w:hpsRaise', $domNode)->getAttribute('w:val'); // pts above base text |
| 625 | $rubyHpsBaseText = $xmlReader->getElement('w:hpsBaseText', $domNode)->getAttribute('w:val'); // base text size |
| 626 | $rubyLid = $xmlReader->getElement('w:lid', $domNode)->getAttribute('w:val'); // type of ruby |
| 627 | $properties = new RubyProperties(); |
| 628 | $properties->setAlignment($rubyAlignment); |
| 629 | $properties->setFontFaceSize((float) $rubyHps); |
| 630 | $properties->setFontPointsAboveBaseText((float) $rubyHpsRaise); |
| 631 | $properties->setFontSizeForBaseText((float) $rubyHpsBaseText); |
| 632 | $properties->setLanguageId($rubyLid); |
| 633 | |
| 634 | return $properties; |
| 635 | } |
| 636 | |
| 637 | /** |
| 638 | * Read w:tbl. |
| 639 | * |
| 640 | * @param mixed $parent |
| 641 | * @param string $docPart |
| 642 | */ |
| 643 | protected function readTable(XMLReader $xmlReader, DOMElement $domNode, $parent, $docPart = 'document'): void |
| 644 | { |
| 645 | // Table style |
| 646 | $tblStyle = null; |
| 647 | if ($xmlReader->elementExists('w:tblPr', $domNode)) { |
| 648 | $tblStyle = $this->readTableStyle($xmlReader, $domNode); |
| 649 | } |
| 650 | |
| 651 | /** @var \PhpOffice\PhpWord\Element\Table $table Type hint */ |
| 652 | $table = $parent->addTable($tblStyle); |
| 653 | $tblNodes = $xmlReader->getElements('*', $domNode); |
| 654 | foreach ($tblNodes as $tblNode) { |
| 655 | if ('w:tblGrid' == $tblNode->nodeName) { // Column |
| 656 | // @todo Do something with table columns |
| 657 | } elseif ('w:tr' == $tblNode->nodeName) { // Row |
| 658 | $rowHeight = $xmlReader->getAttribute('w:val', $tblNode, 'w:trPr/w:trHeight'); |
| 659 | $rowHRule = $xmlReader->getAttribute('w:hRule', $tblNode, 'w:trPr/w:trHeight'); |
| 660 | $rowHRule = $rowHRule == 'exact'; |
| 661 | $rowStyle = [ |
| 662 | 'tblHeader' => $xmlReader->elementExists('w:trPr/w:tblHeader', $tblNode), |
| 663 | 'cantSplit' => $xmlReader->elementExists('w:trPr/w:cantSplit', $tblNode), |
| 664 | 'exactHeight' => $rowHRule, |
| 665 | ]; |
| 666 | |
| 667 | $row = $table->addRow($rowHeight, $rowStyle); |
| 668 | $rowNodes = $xmlReader->getElements('*', $tblNode); |
| 669 | foreach ($rowNodes as $rowNode) { |
| 670 | if ('w:trPr' == $rowNode->nodeName) { // Row style |
| 671 | // @todo Do something with row style |
| 672 | } elseif ('w:tc' == $rowNode->nodeName) { // Cell |
| 673 | $cellWidth = $xmlReader->getAttribute('w:w', $rowNode, 'w:tcPr/w:tcW'); |
| 674 | $cellStyle = null; |
| 675 | if ($xmlReader->elementExists('w:tcPr', $rowNode)) { |
| 676 | $cellStyle = $this->readCellStyle($xmlReader, $rowNode); |
| 677 | } |
| 678 | |
| 679 | $cell = $row->addCell($cellWidth, $cellStyle); |
| 680 | $cellNodes = $xmlReader->getElements('*', $rowNode); |
| 681 | foreach ($cellNodes as $cellNode) { |
| 682 | if ('w:p' == $cellNode->nodeName) { // Paragraph |
| 683 | $this->readParagraph($xmlReader, $cellNode, $cell, $docPart); |
| 684 | } elseif ($cellNode->nodeName == 'w:tbl') { // Table |
| 685 | $this->readTable($xmlReader, $cellNode, $cell, $docPart); |
| 686 | } |
| 687 | } |
| 688 | } |
| 689 | } |
| 690 | } |
| 691 | } |
| 692 | } |
| 693 | |
| 694 | /** |
| 695 | * Read w:pPr. |
| 696 | * |
| 697 | * @return null|array |
| 698 | */ |
| 699 | protected function readParagraphStyle(XMLReader $xmlReader, DOMElement $domNode) |
| 700 | { |
| 701 | if (!$xmlReader->elementExists('w:pPr', $domNode)) { |
| 702 | return null; |
| 703 | } |
| 704 | |
| 705 | $styleNode = $xmlReader->getElement('w:pPr', $domNode); |
| 706 | $styleDefs = [ |
| 707 | 'styleName' => [self::READ_VALUE, ['w:pStyle', 'w:name']], |
| 708 | 'alignment' => [self::READ_VALUE, 'w:jc'], |
| 709 | 'basedOn' => [self::READ_VALUE, 'w:basedOn'], |
| 710 | 'next' => [self::READ_VALUE, 'w:next'], |
| 711 | 'indentLeft' => [self::READ_VALUE, 'w:ind', 'w:left'], |
| 712 | 'indentRight' => [self::READ_VALUE, 'w:ind', 'w:right'], |
| 713 | 'indentHanging' => [self::READ_VALUE, 'w:ind', 'w:hanging'], |
| 714 | 'indentFirstLine' => [self::READ_VALUE, 'w:ind', 'w:firstLine'], |
| 715 | 'indentFirstLineChars' => [self::READ_VALUE, 'w:ind', 'w:firstLineChars'], |
| 716 | 'spaceAfter' => [self::READ_VALUE, 'w:spacing', 'w:after'], |
| 717 | 'spaceBefore' => [self::READ_VALUE, 'w:spacing', 'w:before'], |
| 718 | 'widowControl' => [self::READ_FALSE, 'w:widowControl'], |
| 719 | 'keepNext' => [self::READ_TRUE, 'w:keepNext'], |
| 720 | 'keepLines' => [self::READ_TRUE, 'w:keepLines'], |
| 721 | 'pageBreakBefore' => [self::READ_TRUE, 'w:pageBreakBefore'], |
| 722 | 'contextualSpacing' => [self::READ_TRUE, 'w:contextualSpacing'], |
| 723 | 'bidi' => [self::READ_TRUE, 'w:bidi'], |
| 724 | 'suppressAutoHyphens' => [self::READ_TRUE, 'w:suppressAutoHyphens'], |
| 725 | 'borderTopStyle' => [self::READ_VALUE, 'w:pBdr/w:top'], |
| 726 | 'borderTopColor' => [self::READ_VALUE, 'w:pBdr/w:top', 'w:color'], |
| 727 | 'borderTopSize' => [self::READ_VALUE, 'w:pBdr/w:top', 'w:sz'], |
| 728 | 'borderRightStyle' => [self::READ_VALUE, 'w:pBdr/w:right'], |
| 729 | 'borderRightColor' => [self::READ_VALUE, 'w:pBdr/w:right', 'w:color'], |
| 730 | 'borderRightSize' => [self::READ_VALUE, 'w:pBdr/w:right', 'w:sz'], |
| 731 | 'borderBottomStyle' => [self::READ_VALUE, 'w:pBdr/w:bottom'], |
| 732 | 'borderBottomColor' => [self::READ_VALUE, 'w:pBdr/w:bottom', 'w:color'], |
| 733 | 'borderBottomSize' => [self::READ_VALUE, 'w:pBdr/w:bottom', 'w:sz'], |
| 734 | 'borderLeftStyle' => [self::READ_VALUE, 'w:pBdr/w:left'], |
| 735 | 'borderLeftColor' => [self::READ_VALUE, 'w:pBdr/w:left', 'w:color'], |
| 736 | 'borderLeftSize' => [self::READ_VALUE, 'w:pBdr/w:left', 'w:sz'], |
| 737 | ]; |
| 738 | |
| 739 | return $this->readStyleDefs($xmlReader, $styleNode, $styleDefs); |
| 740 | } |
| 741 | |
| 742 | /** |
| 743 | * Read w:rPr. |
| 744 | * |
| 745 | * @return null|array |
| 746 | */ |
| 747 | protected function readFontStyle(XMLReader $xmlReader, DOMElement $domNode) |
| 748 | { |
| 749 | if (null === $domNode) { |
| 750 | return null; |
| 751 | } |
| 752 | // Hyperlink has an extra w:r child |
| 753 | if ('w:hyperlink' == $domNode->nodeName) { |
| 754 | $domNode = $xmlReader->getElement('w:r', $domNode); |
| 755 | } |
| 756 | if (!$xmlReader->elementExists('w:rPr', $domNode)) { |
| 757 | return null; |
| 758 | } |
| 759 | |
| 760 | $styleNode = $xmlReader->getElement('w:rPr', $domNode); |
| 761 | $styleDefs = [ |
| 762 | 'styleName' => [self::READ_VALUE, 'w:rStyle'], |
| 763 | 'name' => [self::READ_VALUE, 'w:rFonts', ['w:ascii', 'w:hAnsi', 'w:eastAsia', 'w:cs']], |
| 764 | 'hint' => [self::READ_VALUE, 'w:rFonts', 'w:hint'], |
| 765 | 'size' => [self::READ_SIZE, ['w:sz', 'w:szCs']], |
| 766 | 'color' => [self::READ_VALUE, 'w:color'], |
| 767 | 'underline' => [self::READ_VALUE, 'w:u'], |
| 768 | 'bold' => [self::READ_TRUE, 'w:b'], |
| 769 | 'italic' => [self::READ_TRUE, 'w:i'], |
| 770 | 'strikethrough' => [self::READ_TRUE, 'w:strike'], |
| 771 | 'doubleStrikethrough' => [self::READ_TRUE, 'w:dstrike'], |
| 772 | 'smallCaps' => [self::READ_TRUE, 'w:smallCaps'], |
| 773 | 'allCaps' => [self::READ_TRUE, 'w:caps'], |
| 774 | 'superScript' => [self::READ_EQUAL, 'w:vertAlign', 'w:val', 'superscript'], |
| 775 | 'subScript' => [self::READ_EQUAL, 'w:vertAlign', 'w:val', 'subscript'], |
| 776 | 'fgColor' => [self::READ_VALUE, 'w:highlight'], |
| 777 | 'rtl' => [self::READ_TRUE, 'w:rtl'], |
| 778 | 'lang' => [self::READ_VALUE, 'w:lang'], |
| 779 | 'position' => [self::READ_VALUE, 'w:position'], |
| 780 | 'hidden' => [self::READ_TRUE, 'w:vanish'], |
| 781 | ]; |
| 782 | |
| 783 | return $this->readStyleDefs($xmlReader, $styleNode, $styleDefs); |
| 784 | } |
| 785 | |
| 786 | /** |
| 787 | * Read w:tblPr. |
| 788 | * |
| 789 | * @return null|array|string |
| 790 | * |
| 791 | * @todo Capture w:tblStylePr w:type="firstRow" |
| 792 | */ |
| 793 | protected function readTableStyle(XMLReader $xmlReader, DOMElement $domNode) |
| 794 | { |
| 795 | $style = null; |
| 796 | $margins = ['top', 'left', 'bottom', 'right']; |
| 797 | $borders = array_merge($margins, ['insideH', 'insideV']); |
| 798 | |
| 799 | if ($xmlReader->elementExists('w:tblPr', $domNode)) { |
| 800 | if ($xmlReader->elementExists('w:tblPr/w:tblStyle', $domNode)) { |
| 801 | $style = $xmlReader->getAttribute('w:val', $domNode, 'w:tblPr/w:tblStyle'); |
| 802 | } else { |
| 803 | $styleNode = $xmlReader->getElement('w:tblPr', $domNode); |
| 804 | $styleDefs = []; |
| 805 | foreach ($margins as $side) { |
| 806 | $ucfSide = ucfirst($side); |
| 807 | $styleDefs["cellMargin$ucfSide"] = [self::READ_VALUE, "w:tblCellMar/w:$side", 'w:w']; |
| 808 | } |
| 809 | foreach ($borders as $side) { |
| 810 | $ucfSide = ucfirst($side); |
| 811 | $styleDefs["border{$ucfSide}Size"] = [self::READ_VALUE, "w:tblBorders/w:$side", 'w:sz']; |
| 812 | $styleDefs["border{$ucfSide}Color"] = [self::READ_VALUE, "w:tblBorders/w:$side", 'w:color']; |
| 813 | $styleDefs["border{$ucfSide}Style"] = [self::READ_VALUE, "w:tblBorders/w:$side", 'w:val']; |
| 814 | } |
| 815 | $styleDefs['layout'] = [self::READ_VALUE, 'w:tblLayout', 'w:type']; |
| 816 | $styleDefs['bidiVisual'] = [self::READ_TRUE, 'w:bidiVisual']; |
| 817 | $styleDefs['cellSpacing'] = [self::READ_VALUE, 'w:tblCellSpacing', 'w:w']; |
| 818 | $style = $this->readStyleDefs($xmlReader, $styleNode, $styleDefs); |
| 819 | |
| 820 | $tablePositionNode = $xmlReader->getElement('w:tblpPr', $styleNode); |
| 821 | if ($tablePositionNode !== null) { |
| 822 | $style['position'] = $this->readTablePosition($xmlReader, $tablePositionNode); |
| 823 | } |
| 824 | |
| 825 | $indentNode = $xmlReader->getElement('w:tblInd', $styleNode); |
| 826 | if ($indentNode !== null) { |
| 827 | $style['indent'] = $this->readTableIndent($xmlReader, $indentNode); |
| 828 | } |
| 829 | } |
| 830 | } |
| 831 | |
| 832 | return $style; |
| 833 | } |
| 834 | |
| 835 | /** |
| 836 | * Read w:tblpPr. |
| 837 | * |
| 838 | * @return array |
| 839 | */ |
| 840 | private function readTablePosition(XMLReader $xmlReader, DOMElement $domNode) |
| 841 | { |
| 842 | $styleDefs = [ |
| 843 | 'leftFromText' => [self::READ_VALUE, '.', 'w:leftFromText'], |
| 844 | 'rightFromText' => [self::READ_VALUE, '.', 'w:rightFromText'], |
| 845 | 'topFromText' => [self::READ_VALUE, '.', 'w:topFromText'], |
| 846 | 'bottomFromText' => [self::READ_VALUE, '.', 'w:bottomFromText'], |
| 847 | 'vertAnchor' => [self::READ_VALUE, '.', 'w:vertAnchor'], |
| 848 | 'horzAnchor' => [self::READ_VALUE, '.', 'w:horzAnchor'], |
| 849 | 'tblpXSpec' => [self::READ_VALUE, '.', 'w:tblpXSpec'], |
| 850 | 'tblpX' => [self::READ_VALUE, '.', 'w:tblpX'], |
| 851 | 'tblpYSpec' => [self::READ_VALUE, '.', 'w:tblpYSpec'], |
| 852 | 'tblpY' => [self::READ_VALUE, '.', 'w:tblpY'], |
| 853 | ]; |
| 854 | |
| 855 | return $this->readStyleDefs($xmlReader, $domNode, $styleDefs); |
| 856 | } |
| 857 | |
| 858 | /** |
| 859 | * Read w:tblInd. |
| 860 | * |
| 861 | * @return TblWidthComplexType |
| 862 | */ |
| 863 | private function readTableIndent(XMLReader $xmlReader, DOMElement $domNode) |
| 864 | { |
| 865 | $styleDefs = [ |
| 866 | 'value' => [self::READ_VALUE, '.', 'w:w'], |
| 867 | 'type' => [self::READ_VALUE, '.', 'w:type'], |
| 868 | ]; |
| 869 | $styleDefs = $this->readStyleDefs($xmlReader, $domNode, $styleDefs); |
| 870 | |
| 871 | return new TblWidthComplexType((int) $styleDefs['value'], $styleDefs['type']); |
| 872 | } |
| 873 | |
| 874 | /** |
| 875 | * Read w:tcPr. |
| 876 | * |
| 877 | * @return null|array |
| 878 | */ |
| 879 | private function readCellStyle(XMLReader $xmlReader, DOMElement $domNode) |
| 880 | { |
| 881 | $styleDefs = [ |
| 882 | 'valign' => [self::READ_VALUE, 'w:vAlign'], |
| 883 | 'textDirection' => [self::READ_VALUE, 'w:textDirection'], |
| 884 | 'gridSpan' => [self::READ_VALUE, 'w:gridSpan'], |
| 885 | 'vMerge' => [self::READ_VALUE, 'w:vMerge', null, null, 'continue'], |
| 886 | 'bgColor' => [self::READ_VALUE, 'w:shd', 'w:fill'], |
| 887 | 'noWrap' => [self::READ_VALUE, 'w:noWrap', null, null, true], |
| 888 | ]; |
| 889 | $style = null; |
| 890 | |
| 891 | if ($xmlReader->elementExists('w:tcPr', $domNode)) { |
| 892 | $styleNode = $xmlReader->getElement('w:tcPr', $domNode); |
| 893 | |
| 894 | $borders = ['top', 'left', 'bottom', 'right']; |
| 895 | foreach ($borders as $side) { |
| 896 | $ucfSide = ucfirst($side); |
| 897 | |
| 898 | $styleDefs['border' . $ucfSide . 'Size'] = [self::READ_VALUE, 'w:tcBorders/w:' . $side, 'w:sz']; |
| 899 | $styleDefs['border' . $ucfSide . 'Color'] = [self::READ_VALUE, 'w:tcBorders/w:' . $side, 'w:color']; |
| 900 | $styleDefs['border' . $ucfSide . 'Style'] = [self::READ_VALUE, 'w:tcBorders/w:' . $side, 'w:val']; |
| 901 | } |
| 902 | |
| 903 | $style = $this->readStyleDefs($xmlReader, $styleNode, $styleDefs); |
| 904 | } |
| 905 | |
| 906 | return $style; |
| 907 | } |
| 908 | |
| 909 | /** |
| 910 | * Returns the first child element found. |
| 911 | * |
| 912 | * @param null|array|string $elements |
| 913 | * |
| 914 | * @return null|string |
| 915 | */ |
| 916 | private function findPossibleElement(XMLReader $xmlReader, ?DOMElement $parentNode = null, $elements = null) |
| 917 | { |
| 918 | if (is_array($elements)) { |
| 919 | //if element is an array, we take the first element that exists in the XML |
| 920 | foreach ($elements as $possibleElement) { |
| 921 | if ($xmlReader->elementExists($possibleElement, $parentNode)) { |
| 922 | return $possibleElement; |
| 923 | } |
| 924 | } |
| 925 | } else { |
| 926 | return $elements; |
| 927 | } |
| 928 | |
| 929 | return null; |
| 930 | } |
| 931 | |
| 932 | /** |
| 933 | * Returns the first attribute found. |
| 934 | * |
| 935 | * @param array|string $attributes |
| 936 | * |
| 937 | * @return null|string |
| 938 | */ |
| 939 | private function findPossibleAttribute(XMLReader $xmlReader, DOMElement $node, $attributes) |
| 940 | { |
| 941 | //if attribute is an array, we take the first attribute that exists in the XML |
| 942 | if (is_array($attributes)) { |
| 943 | foreach ($attributes as $possibleAttribute) { |
| 944 | if ($xmlReader->getAttribute($possibleAttribute, $node)) { |
| 945 | return $possibleAttribute; |
| 946 | } |
| 947 | } |
| 948 | |
| 949 | return null; |
| 950 | } |
| 951 | |
| 952 | return $attributes; |
| 953 | } |
| 954 | |
| 955 | /** |
| 956 | * Read style definition. |
| 957 | * |
| 958 | * @param array $styleDefs |
| 959 | * |
| 960 | * @ignoreScrutinizerPatch |
| 961 | * |
| 962 | * @return array |
| 963 | */ |
| 964 | protected function readStyleDefs(XMLReader $xmlReader, ?DOMElement $parentNode = null, $styleDefs = []) |
| 965 | { |
| 966 | $styles = []; |
| 967 | |
| 968 | foreach ($styleDefs as $styleProp => $styleVal) { |
| 969 | [$method, $element, $attribute, $expected, $default] = array_pad($styleVal, 5, null); |
| 970 | |
| 971 | $element = $this->findPossibleElement($xmlReader, $parentNode, $element); |
| 972 | if ($element === null) { |
| 973 | continue; |
| 974 | } |
| 975 | |
| 976 | if ($xmlReader->elementExists($element, $parentNode)) { |
| 977 | $node = $xmlReader->getElement($element, $parentNode); |
| 978 | |
| 979 | $attribute = $this->findPossibleAttribute($xmlReader, $node, $attribute); |
| 980 | |
| 981 | // Use w:val as default if no attribute assigned |
| 982 | $attribute = ($attribute === null) ? 'w:val' : $attribute; |
| 983 | $attributeValue = $xmlReader->getAttribute($attribute, $node) ?? $default; |
| 984 | |
| 985 | $styleValue = $this->readStyleDef($method, $attributeValue, $expected); |
| 986 | if ($styleValue !== null) { |
| 987 | $styles[$styleProp] = $styleValue; |
| 988 | } |
| 989 | } |
| 990 | } |
| 991 | |
| 992 | return $styles; |
| 993 | } |
| 994 | |
| 995 | /** |
| 996 | * Return style definition based on conversion method. |
| 997 | * |
| 998 | * @param string $method |
| 999 | * |
| 1000 | * @ignoreScrutinizerPatch |
| 1001 | * |
| 1002 | * @param null|string $attributeValue |
| 1003 | * @param mixed $expected |
| 1004 | * |
| 1005 | * @return mixed |
| 1006 | */ |
| 1007 | private function readStyleDef($method, $attributeValue, $expected) |
| 1008 | { |
| 1009 | $style = $attributeValue; |
| 1010 | |
| 1011 | if (self::READ_SIZE == $method) { |
| 1012 | $style = $attributeValue / 2; |
| 1013 | } elseif (self::READ_TRUE == $method) { |
| 1014 | $style = $this->isOn($attributeValue); |
| 1015 | } elseif (self::READ_FALSE == $method) { |
| 1016 | $style = !$this->isOn($attributeValue); |
| 1017 | } elseif (self::READ_EQUAL == $method) { |
| 1018 | $style = $attributeValue == $expected; |
| 1019 | } |
| 1020 | |
| 1021 | return $style; |
| 1022 | } |
| 1023 | |
| 1024 | /** |
| 1025 | * Parses the value of the on/off value, null is considered true as it means the w:val attribute was not present. |
| 1026 | * |
| 1027 | * @see http://www.datypic.com/sc/ooxml/t-w_ST_OnOff.html |
| 1028 | * |
| 1029 | * @param string $value |
| 1030 | * |
| 1031 | * @return bool |
| 1032 | */ |
| 1033 | private function isOn($value = null) |
| 1034 | { |
| 1035 | return $value === null || $value === '1' || $value === 'true' || $value === 'on'; |
| 1036 | } |
| 1037 | |
| 1038 | /** |
| 1039 | * Returns the target of image, object, or link as stored in ::readMainRels. |
| 1040 | * |
| 1041 | * @param string $docPart |
| 1042 | * @param string $rId |
| 1043 | * |
| 1044 | * @return null|string |
| 1045 | */ |
| 1046 | private function getMediaTarget($docPart, $rId) |
| 1047 | { |
| 1048 | $target = null; |
| 1049 | |
| 1050 | if (isset($this->rels[$docPart], $this->rels[$docPart][$rId])) { |
| 1051 | $target = $this->rels[$docPart][$rId]['target']; |
| 1052 | } |
| 1053 | |
| 1054 | return $target; |
| 1055 | } |
| 1056 | |
| 1057 | /** |
| 1058 | * Returns the target mode. |
| 1059 | * |
| 1060 | * @param string $docPart |
| 1061 | * @param string $rId |
| 1062 | * |
| 1063 | * @return null|string |
| 1064 | */ |
| 1065 | private function getTargetMode($docPart, $rId) |
| 1066 | { |
| 1067 | $mode = null; |
| 1068 | |
| 1069 | if (isset($this->rels[$docPart], $this->rels[$docPart][$rId])) { |
| 1070 | $mode = $this->rels[$docPart][$rId]['targetMode']; |
| 1071 | } |
| 1072 | |
| 1073 | return $mode; |
| 1074 | } |
| 1075 | } |