CharacterReference.php
1 year ago
DOMTreeBuilder.php
1 year ago
EventHandler.php
1 year ago
FileInputStream.php
1 year ago
InputStream.php
1 year ago
ParseError.php
1 year ago
README.md
1 year ago
Scanner.php
1 year ago
StringInputStream.php
1 year ago
Tokenizer.php
1 year ago
TreeBuildingRules.php
1 year ago
UTF8Utils.php
1 year ago
DOMTreeBuilder.php
716 lines
| 1 | <?php |
| 2 | |
| 3 | namespace Masterminds\HTML5\Parser; |
| 4 | |
| 5 | use Masterminds\HTML5\Elements; |
| 6 | use Masterminds\HTML5\InstructionProcessor; |
| 7 | |
| 8 | /** |
| 9 | * Create an HTML5 DOM tree from events. |
| 10 | * |
| 11 | * This attempts to create a DOM from events emitted by a parser. This |
| 12 | * attempts (but does not guarantee) to up-convert older HTML documents |
| 13 | * to HTML5. It does this by applying HTML5's rules, but it will not |
| 14 | * change the architecture of the document itself. |
| 15 | * |
| 16 | * Many of the error correction and quirks features suggested in the specification |
| 17 | * are implemented herein; however, not all of them are. Since we do not |
| 18 | * assume a graphical user agent, no presentation-specific logic is conducted |
| 19 | * during tree building. |
| 20 | * |
| 21 | * FIXME: The present tree builder does not exactly follow the state machine rules |
| 22 | * for insert modes as outlined in the HTML5 spec. The processor needs to be |
| 23 | * re-written to accomodate this. See, for example, the Go language HTML5 |
| 24 | * parser. |
| 25 | */ |
| 26 | class DOMTreeBuilder implements EventHandler |
| 27 | { |
| 28 | /** |
| 29 | * Defined in http://www.w3.org/TR/html51/infrastructure.html#html-namespace-0. |
| 30 | */ |
| 31 | const NAMESPACE_HTML = 'http://www.w3.org/1999/xhtml'; |
| 32 | |
| 33 | const NAMESPACE_MATHML = 'http://www.w3.org/1998/Math/MathML'; |
| 34 | |
| 35 | const NAMESPACE_SVG = 'http://www.w3.org/2000/svg'; |
| 36 | |
| 37 | const NAMESPACE_XLINK = 'http://www.w3.org/1999/xlink'; |
| 38 | |
| 39 | const NAMESPACE_XML = 'http://www.w3.org/XML/1998/namespace'; |
| 40 | |
| 41 | const NAMESPACE_XMLNS = 'http://www.w3.org/2000/xmlns/'; |
| 42 | |
| 43 | const OPT_DISABLE_HTML_NS = 'disable_html_ns'; |
| 44 | |
| 45 | const OPT_TARGET_DOC = 'target_document'; |
| 46 | |
| 47 | const OPT_IMPLICIT_NS = 'implicit_namespaces'; |
| 48 | |
| 49 | /** |
| 50 | * Holds the HTML5 element names that causes a namespace switch. |
| 51 | * |
| 52 | * @var array |
| 53 | */ |
| 54 | protected $nsRoots = array( |
| 55 | 'html' => self::NAMESPACE_HTML, |
| 56 | 'svg' => self::NAMESPACE_SVG, |
| 57 | 'math' => self::NAMESPACE_MATHML, |
| 58 | ); |
| 59 | |
| 60 | /** |
| 61 | * Holds the always available namespaces (which does not require the XMLNS declaration). |
| 62 | * |
| 63 | * @var array |
| 64 | */ |
| 65 | protected $implicitNamespaces = array( |
| 66 | 'xml' => self::NAMESPACE_XML, |
| 67 | 'xmlns' => self::NAMESPACE_XMLNS, |
| 68 | 'xlink' => self::NAMESPACE_XLINK, |
| 69 | ); |
| 70 | |
| 71 | /** |
| 72 | * Holds a stack of currently active namespaces. |
| 73 | * |
| 74 | * @var array |
| 75 | */ |
| 76 | protected $nsStack = array(); |
| 77 | |
| 78 | /** |
| 79 | * Holds the number of namespaces declared by a node. |
| 80 | * |
| 81 | * @var array |
| 82 | */ |
| 83 | protected $pushes = array(); |
| 84 | |
| 85 | /** |
| 86 | * Defined in 8.2.5. |
| 87 | */ |
| 88 | const IM_INITIAL = 0; |
| 89 | |
| 90 | const IM_BEFORE_HTML = 1; |
| 91 | |
| 92 | const IM_BEFORE_HEAD = 2; |
| 93 | |
| 94 | const IM_IN_HEAD = 3; |
| 95 | |
| 96 | const IM_IN_HEAD_NOSCRIPT = 4; |
| 97 | |
| 98 | const IM_AFTER_HEAD = 5; |
| 99 | |
| 100 | const IM_IN_BODY = 6; |
| 101 | |
| 102 | const IM_TEXT = 7; |
| 103 | |
| 104 | const IM_IN_TABLE = 8; |
| 105 | |
| 106 | const IM_IN_TABLE_TEXT = 9; |
| 107 | |
| 108 | const IM_IN_CAPTION = 10; |
| 109 | |
| 110 | const IM_IN_COLUMN_GROUP = 11; |
| 111 | |
| 112 | const IM_IN_TABLE_BODY = 12; |
| 113 | |
| 114 | const IM_IN_ROW = 13; |
| 115 | |
| 116 | const IM_IN_CELL = 14; |
| 117 | |
| 118 | const IM_IN_SELECT = 15; |
| 119 | |
| 120 | const IM_IN_SELECT_IN_TABLE = 16; |
| 121 | |
| 122 | const IM_AFTER_BODY = 17; |
| 123 | |
| 124 | const IM_IN_FRAMESET = 18; |
| 125 | |
| 126 | const IM_AFTER_FRAMESET = 19; |
| 127 | |
| 128 | const IM_AFTER_AFTER_BODY = 20; |
| 129 | |
| 130 | const IM_AFTER_AFTER_FRAMESET = 21; |
| 131 | |
| 132 | const IM_IN_SVG = 22; |
| 133 | |
| 134 | const IM_IN_MATHML = 23; |
| 135 | |
| 136 | protected $options = array(); |
| 137 | |
| 138 | protected $stack = array(); |
| 139 | |
| 140 | protected $current; // Pointer in the tag hierarchy. |
| 141 | protected $rules; |
| 142 | protected $doc; |
| 143 | |
| 144 | protected $frag; |
| 145 | |
| 146 | protected $processor; |
| 147 | |
| 148 | protected $insertMode = 0; |
| 149 | |
| 150 | /** |
| 151 | * Track if we are in an element that allows only inline child nodes. |
| 152 | * |
| 153 | * @var string|null |
| 154 | */ |
| 155 | protected $onlyInline; |
| 156 | |
| 157 | /** |
| 158 | * Quirks mode is enabled by default. |
| 159 | * Any document that is missing the DT will be considered to be in quirks mode. |
| 160 | */ |
| 161 | protected $quirks = true; |
| 162 | |
| 163 | protected $errors = array(); |
| 164 | |
| 165 | public function __construct($isFragment = false, array $options = array()) |
| 166 | { |
| 167 | $this->options = $options; |
| 168 | |
| 169 | if (isset($options[self::OPT_TARGET_DOC])) { |
| 170 | $this->doc = $options[self::OPT_TARGET_DOC]; |
| 171 | } else { |
| 172 | $impl = new \DOMImplementation(); |
| 173 | // XXX: |
| 174 | // Create the doctype. For now, we are always creating HTML5 |
| 175 | // documents, and attempting to up-convert any older DTDs to HTML5. |
| 176 | $dt = $impl->createDocumentType('html'); |
| 177 | // $this->doc = \DOMImplementation::createDocument(NULL, 'html', $dt); |
| 178 | $this->doc = $impl->createDocument(null, '', $dt); |
| 179 | $this->doc->encoding = !empty($options['encoding']) ? $options['encoding'] : 'UTF-8'; |
| 180 | } |
| 181 | |
| 182 | $this->errors = array(); |
| 183 | |
| 184 | $this->current = $this->doc; // ->documentElement; |
| 185 | |
| 186 | // Create a rules engine for tags. |
| 187 | $this->rules = new TreeBuildingRules(); |
| 188 | |
| 189 | $implicitNS = array(); |
| 190 | if (isset($this->options[self::OPT_IMPLICIT_NS])) { |
| 191 | $implicitNS = $this->options[self::OPT_IMPLICIT_NS]; |
| 192 | } elseif (isset($this->options['implicitNamespaces'])) { |
| 193 | $implicitNS = $this->options['implicitNamespaces']; |
| 194 | } |
| 195 | |
| 196 | // Fill $nsStack with the defalut HTML5 namespaces, plus the "implicitNamespaces" array taken form $options |
| 197 | array_unshift($this->nsStack, $implicitNS + array('' => self::NAMESPACE_HTML) + $this->implicitNamespaces); |
| 198 | |
| 199 | if ($isFragment) { |
| 200 | $this->insertMode = static::IM_IN_BODY; |
| 201 | $this->frag = $this->doc->createDocumentFragment(); |
| 202 | $this->current = $this->frag; |
| 203 | } |
| 204 | } |
| 205 | |
| 206 | /** |
| 207 | * Get the document. |
| 208 | */ |
| 209 | public function document() |
| 210 | { |
| 211 | return $this->doc; |
| 212 | } |
| 213 | |
| 214 | /** |
| 215 | * Get the DOM fragment for the body. |
| 216 | * |
| 217 | * This returns a DOMNodeList because a fragment may have zero or more |
| 218 | * DOMNodes at its root. |
| 219 | * |
| 220 | * @see http://www.w3.org/TR/2012/CR-html5-20121217/syntax.html#concept-frag-parse-context |
| 221 | * |
| 222 | * @return \DOMDocumentFragment |
| 223 | */ |
| 224 | public function fragment() |
| 225 | { |
| 226 | return $this->frag; |
| 227 | } |
| 228 | |
| 229 | /** |
| 230 | * Provide an instruction processor. |
| 231 | * |
| 232 | * This is used for handling Processor Instructions as they are |
| 233 | * inserted. If omitted, PI's are inserted directly into the DOM tree. |
| 234 | * |
| 235 | * @param InstructionProcessor $proc |
| 236 | */ |
| 237 | public function setInstructionProcessor(InstructionProcessor $proc) |
| 238 | { |
| 239 | $this->processor = $proc; |
| 240 | } |
| 241 | |
| 242 | public function doctype($name, $idType = 0, $id = null, $quirks = false) |
| 243 | { |
| 244 | // This is used solely for setting quirks mode. Currently we don't |
| 245 | // try to preserve the inbound DT. We convert it to HTML5. |
| 246 | $this->quirks = $quirks; |
| 247 | |
| 248 | if ($this->insertMode > static::IM_INITIAL) { |
| 249 | $this->parseError('Illegal placement of DOCTYPE tag. Ignoring: ' . $name); |
| 250 | |
| 251 | return; |
| 252 | } |
| 253 | |
| 254 | $this->insertMode = static::IM_BEFORE_HTML; |
| 255 | } |
| 256 | |
| 257 | /** |
| 258 | * Process the start tag. |
| 259 | * |
| 260 | * @todo - XMLNS namespace handling (we need to parse, even if it's not valid) |
| 261 | * - XLink, MathML and SVG namespace handling |
| 262 | * - Omission rules: 8.1.2.4 Optional tags |
| 263 | * |
| 264 | * @param string $name |
| 265 | * @param array $attributes |
| 266 | * @param bool $selfClosing |
| 267 | * |
| 268 | * @return int |
| 269 | */ |
| 270 | public function startTag($name, $attributes = array(), $selfClosing = false) |
| 271 | { |
| 272 | $lname = $this->normalizeTagName($name); |
| 273 | |
| 274 | // Make sure we have an html element. |
| 275 | if (!$this->doc->documentElement && 'html' !== $name && !$this->frag) { |
| 276 | $this->startTag('html'); |
| 277 | } |
| 278 | |
| 279 | // Set quirks mode if we're at IM_INITIAL with no doctype. |
| 280 | if ($this->insertMode === static::IM_INITIAL) { |
| 281 | $this->quirks = true; |
| 282 | $this->parseError('No DOCTYPE specified.'); |
| 283 | } |
| 284 | |
| 285 | // SPECIAL TAG HANDLING: |
| 286 | // Spec says do this, and "don't ask." |
| 287 | // find the spec where this is defined... looks problematic |
| 288 | if ('image' === $name && !($this->insertMode === static::IM_IN_SVG || $this->insertMode === static::IM_IN_MATHML)) { |
| 289 | $name = 'img'; |
| 290 | } |
| 291 | |
| 292 | // Autoclose p tags where appropriate. |
| 293 | if ($this->insertMode >= static::IM_IN_BODY && Elements::isA($name, Elements::AUTOCLOSE_P)) { |
| 294 | $this->autoclose('p'); |
| 295 | } |
| 296 | |
| 297 | // Set insert mode: |
| 298 | switch ($name) { |
| 299 | case 'html': |
| 300 | $this->insertMode = static::IM_BEFORE_HEAD; |
| 301 | break; |
| 302 | case 'head': |
| 303 | if ($this->insertMode > static::IM_BEFORE_HEAD) { |
| 304 | $this->parseError('Unexpected head tag outside of head context.'); |
| 305 | } else { |
| 306 | $this->insertMode = static::IM_IN_HEAD; |
| 307 | } |
| 308 | break; |
| 309 | case 'body': |
| 310 | $this->insertMode = static::IM_IN_BODY; |
| 311 | break; |
| 312 | case 'svg': |
| 313 | $this->insertMode = static::IM_IN_SVG; |
| 314 | break; |
| 315 | case 'math': |
| 316 | $this->insertMode = static::IM_IN_MATHML; |
| 317 | break; |
| 318 | case 'noscript': |
| 319 | if ($this->insertMode === static::IM_IN_HEAD) { |
| 320 | $this->insertMode = static::IM_IN_HEAD_NOSCRIPT; |
| 321 | } |
| 322 | break; |
| 323 | } |
| 324 | |
| 325 | // Special case handling for SVG. |
| 326 | if ($this->insertMode === static::IM_IN_SVG) { |
| 327 | $lname = Elements::normalizeSvgElement($lname); |
| 328 | } |
| 329 | |
| 330 | $pushes = 0; |
| 331 | // when we found a tag thats appears inside $nsRoots, we have to switch the defalut namespace |
| 332 | if (isset($this->nsRoots[$lname]) && $this->nsStack[0][''] !== $this->nsRoots[$lname]) { |
| 333 | array_unshift($this->nsStack, array( |
| 334 | '' => $this->nsRoots[$lname], |
| 335 | ) + $this->nsStack[0]); |
| 336 | ++$pushes; |
| 337 | } |
| 338 | $needsWorkaround = false; |
| 339 | if (isset($this->options['xmlNamespaces']) && $this->options['xmlNamespaces']) { |
| 340 | // when xmlNamespaces is true a and we found a 'xmlns' or 'xmlns:*' attribute, we should add a new item to the $nsStack |
| 341 | foreach ($attributes as $aName => $aVal) { |
| 342 | if ('xmlns' === $aName) { |
| 343 | $needsWorkaround = $aVal; |
| 344 | array_unshift($this->nsStack, array( |
| 345 | '' => $aVal, |
| 346 | ) + $this->nsStack[0]); |
| 347 | ++$pushes; |
| 348 | } elseif ('xmlns' === (($pos = strpos($aName, ':')) ? substr($aName, 0, $pos) : '')) { |
| 349 | array_unshift($this->nsStack, array( |
| 350 | substr($aName, $pos + 1) => $aVal, |
| 351 | ) + $this->nsStack[0]); |
| 352 | ++$pushes; |
| 353 | } |
| 354 | } |
| 355 | } |
| 356 | |
| 357 | if ($this->onlyInline && Elements::isA($lname, Elements::BLOCK_TAG)) { |
| 358 | $this->autoclose($this->onlyInline); |
| 359 | $this->onlyInline = null; |
| 360 | } |
| 361 | |
| 362 | // some elements as table related tags might have optional end tags that force us to auto close multiple tags |
| 363 | // https://www.w3.org/TR/html401/struct/tables.html |
| 364 | if ($this->current instanceof \DOMElement && isset(Elements::$optionalEndElementsParentsToClose[$lname])) { |
| 365 | foreach (Elements::$optionalEndElementsParentsToClose[$lname] as $parentElName) { |
| 366 | if ($this->current instanceof \DOMElement && $this->current->tagName === $parentElName) { |
| 367 | $this->autoclose($parentElName); |
| 368 | } |
| 369 | } |
| 370 | } |
| 371 | |
| 372 | try { |
| 373 | $prefix = ($pos = strpos($lname, ':')) ? substr($lname, 0, $pos) : ''; |
| 374 | |
| 375 | if (false !== $needsWorkaround) { |
| 376 | $xml = "<$lname xmlns=\"$needsWorkaround\" " . (strlen($prefix) && isset($this->nsStack[0][$prefix]) ? ("xmlns:$prefix=\"" . $this->nsStack[0][$prefix] . '"') : '') . '/>'; |
| 377 | |
| 378 | $frag = new \DOMDocument('1.0', 'UTF-8'); |
| 379 | $frag->loadXML($xml); |
| 380 | |
| 381 | $ele = $this->doc->importNode($frag->documentElement, true); |
| 382 | } else { |
| 383 | if (!isset($this->nsStack[0][$prefix]) || ('' === $prefix && isset($this->options[self::OPT_DISABLE_HTML_NS]) && $this->options[self::OPT_DISABLE_HTML_NS])) { |
| 384 | $ele = $this->doc->createElement($lname); |
| 385 | } else { |
| 386 | $ele = $this->doc->createElementNS($this->nsStack[0][$prefix], $lname); |
| 387 | } |
| 388 | } |
| 389 | } catch (\DOMException $e) { |
| 390 | $this->parseError("Illegal tag name: <$lname>. Replaced with <invalid>."); |
| 391 | $ele = $this->doc->createElement('invalid'); |
| 392 | } |
| 393 | |
| 394 | if (Elements::isA($lname, Elements::BLOCK_ONLY_INLINE)) { |
| 395 | $this->onlyInline = $lname; |
| 396 | } |
| 397 | |
| 398 | // When we add some namespacess, we have to track them. Later, when "endElement" is invoked, we have to remove them. |
| 399 | // When we are on a void tag, we do not need to care about namesapce nesting. |
| 400 | if ($pushes > 0 && !Elements::isA($name, Elements::VOID_TAG)) { |
| 401 | // PHP tends to free the memory used by DOM, |
| 402 | // to avoid spl_object_hash collisions whe have to avoid garbage collection of $ele storing it into $pushes |
| 403 | // see https://bugs.php.net/bug.php?id=67459 |
| 404 | $this->pushes[spl_object_hash($ele)] = array($pushes, $ele); |
| 405 | } |
| 406 | |
| 407 | foreach ($attributes as $aName => $aVal) { |
| 408 | // xmlns attributes can't be set |
| 409 | if ('xmlns' === $aName) { |
| 410 | continue; |
| 411 | } |
| 412 | |
| 413 | if ($this->insertMode === static::IM_IN_SVG) { |
| 414 | $aName = Elements::normalizeSvgAttribute($aName); |
| 415 | } elseif ($this->insertMode === static::IM_IN_MATHML) { |
| 416 | $aName = Elements::normalizeMathMlAttribute($aName); |
| 417 | } |
| 418 | |
| 419 | $aVal = (string) $aVal; |
| 420 | |
| 421 | try { |
| 422 | $prefix = ($pos = strpos($aName, ':')) ? substr($aName, 0, $pos) : false; |
| 423 | |
| 424 | if ('xmlns' === $prefix) { |
| 425 | $ele->setAttributeNS(self::NAMESPACE_XMLNS, $aName, $aVal); |
| 426 | } elseif (false !== $prefix && isset($this->nsStack[0][$prefix])) { |
| 427 | $ele->setAttributeNS($this->nsStack[0][$prefix], $aName, $aVal); |
| 428 | } else { |
| 429 | $ele->setAttribute($aName, $aVal); |
| 430 | } |
| 431 | } catch (\DOMException $e) { |
| 432 | $this->parseError("Illegal attribute name for tag $name. Ignoring: $aName"); |
| 433 | continue; |
| 434 | } |
| 435 | |
| 436 | // This is necessary on a non-DTD schema, like HTML5. |
| 437 | if ('id' === $aName) { |
| 438 | $ele->setIdAttribute('id', true); |
| 439 | } |
| 440 | } |
| 441 | |
| 442 | if ($this->frag !== $this->current && $this->rules->hasRules($name)) { |
| 443 | // Some elements have special processing rules. Handle those separately. |
| 444 | $this->current = $this->rules->evaluate($ele, $this->current); |
| 445 | } else { |
| 446 | // Otherwise, it's a standard element. |
| 447 | $this->current->appendChild($ele); |
| 448 | |
| 449 | if (!Elements::isA($name, Elements::VOID_TAG)) { |
| 450 | $this->current = $ele; |
| 451 | } |
| 452 | |
| 453 | // Self-closing tags should only be respected on foreign elements |
| 454 | // (and are implied on void elements) |
| 455 | // See: https://www.w3.org/TR/html5/syntax.html#start-tags |
| 456 | if (Elements::isHtml5Element($name)) { |
| 457 | $selfClosing = false; |
| 458 | } |
| 459 | } |
| 460 | |
| 461 | // This is sort of a last-ditch attempt to correct for cases where no head/body |
| 462 | // elements are provided. |
| 463 | if ($this->insertMode <= static::IM_BEFORE_HEAD && 'head' !== $name && 'html' !== $name) { |
| 464 | $this->insertMode = static::IM_IN_BODY; |
| 465 | } |
| 466 | |
| 467 | // When we are on a void tag, we do not need to care about namesapce nesting, |
| 468 | // but we have to remove the namespaces pushed to $nsStack. |
| 469 | if ($pushes > 0 && Elements::isA($name, Elements::VOID_TAG)) { |
| 470 | // remove the namespaced definded by current node |
| 471 | for ($i = 0; $i < $pushes; ++$i) { |
| 472 | array_shift($this->nsStack); |
| 473 | } |
| 474 | } |
| 475 | |
| 476 | if ($selfClosing) { |
| 477 | $this->endTag($name); |
| 478 | } |
| 479 | |
| 480 | // Return the element mask, which the tokenizer can then use to set |
| 481 | // various processing rules. |
| 482 | return Elements::element($name); |
| 483 | } |
| 484 | |
| 485 | public function endTag($name) |
| 486 | { |
| 487 | $lname = $this->normalizeTagName($name); |
| 488 | |
| 489 | // Special case within 12.2.6.4.7: An end tag whose tag name is "br" should be treated as an opening tag |
| 490 | if ('br' === $name) { |
| 491 | $this->parseError('Closing tag encountered for void element br.'); |
| 492 | |
| 493 | $this->startTag('br'); |
| 494 | } |
| 495 | // Ignore closing tags for other unary elements. |
| 496 | elseif (Elements::isA($name, Elements::VOID_TAG)) { |
| 497 | return; |
| 498 | } |
| 499 | |
| 500 | if ($this->insertMode <= static::IM_BEFORE_HTML) { |
| 501 | // 8.2.5.4.2 |
| 502 | if (in_array($name, array( |
| 503 | 'html', |
| 504 | 'br', |
| 505 | 'head', |
| 506 | 'title', |
| 507 | ))) { |
| 508 | $this->startTag('html'); |
| 509 | $this->endTag($name); |
| 510 | $this->insertMode = static::IM_BEFORE_HEAD; |
| 511 | |
| 512 | return; |
| 513 | } |
| 514 | |
| 515 | // Ignore the tag. |
| 516 | $this->parseError('Illegal closing tag at global scope.'); |
| 517 | |
| 518 | return; |
| 519 | } |
| 520 | |
| 521 | // Special case handling for SVG. |
| 522 | if ($this->insertMode === static::IM_IN_SVG) { |
| 523 | $lname = Elements::normalizeSvgElement($lname); |
| 524 | } |
| 525 | |
| 526 | $cid = spl_object_hash($this->current); |
| 527 | |
| 528 | // XXX: HTML has no parent. What do we do, though, |
| 529 | // if this element appears in the wrong place? |
| 530 | if ('html' === $lname) { |
| 531 | return; |
| 532 | } |
| 533 | |
| 534 | // remove the namespaced definded by current node |
| 535 | if (isset($this->pushes[$cid])) { |
| 536 | for ($i = 0; $i < $this->pushes[$cid][0]; ++$i) { |
| 537 | array_shift($this->nsStack); |
| 538 | } |
| 539 | unset($this->pushes[$cid]); |
| 540 | } |
| 541 | |
| 542 | if (!$this->autoclose($lname)) { |
| 543 | $this->parseError('Could not find closing tag for ' . $lname); |
| 544 | } |
| 545 | |
| 546 | switch ($lname) { |
| 547 | case 'head': |
| 548 | $this->insertMode = static::IM_AFTER_HEAD; |
| 549 | break; |
| 550 | case 'body': |
| 551 | $this->insertMode = static::IM_AFTER_BODY; |
| 552 | break; |
| 553 | case 'svg': |
| 554 | case 'mathml': |
| 555 | $this->insertMode = static::IM_IN_BODY; |
| 556 | break; |
| 557 | } |
| 558 | } |
| 559 | |
| 560 | public function comment($cdata) |
| 561 | { |
| 562 | // TODO: Need to handle case where comment appears outside of the HTML tag. |
| 563 | $node = $this->doc->createComment($cdata); |
| 564 | $this->current->appendChild($node); |
| 565 | } |
| 566 | |
| 567 | public function text($data) |
| 568 | { |
| 569 | // XXX: Hmmm.... should we really be this strict? |
| 570 | if ($this->insertMode < static::IM_IN_HEAD) { |
| 571 | // Per '8.2.5.4.3 The "before head" insertion mode' the characters |
| 572 | // " \t\n\r\f" should be ignored but no mention of a parse error. This is |
| 573 | // practical as most documents contain these characters. Other text is not |
| 574 | // expected here so recording a parse error is necessary. |
| 575 | $dataTmp = trim($data, " \t\n\r\f"); |
| 576 | if (!empty($dataTmp)) { |
| 577 | // fprintf(STDOUT, "Unexpected insert mode: %d", $this->insertMode); |
| 578 | $this->parseError('Unexpected text. Ignoring: ' . $dataTmp); |
| 579 | } |
| 580 | |
| 581 | return; |
| 582 | } |
| 583 | // fprintf(STDOUT, "Appending text %s.", $data); |
| 584 | $node = $this->doc->createTextNode($data); |
| 585 | $this->current->appendChild($node); |
| 586 | } |
| 587 | |
| 588 | public function eof() |
| 589 | { |
| 590 | // If the $current isn't the $root, do we need to do anything? |
| 591 | } |
| 592 | |
| 593 | public function parseError($msg, $line = 0, $col = 0) |
| 594 | { |
| 595 | $this->errors[] = sprintf('Line %d, Col %d: %s', $line, $col, $msg); |
| 596 | } |
| 597 | |
| 598 | public function getErrors() |
| 599 | { |
| 600 | return $this->errors; |
| 601 | } |
| 602 | |
| 603 | public function cdata($data) |
| 604 | { |
| 605 | $node = $this->doc->createCDATASection($data); |
| 606 | $this->current->appendChild($node); |
| 607 | } |
| 608 | |
| 609 | public function processingInstruction($name, $data = null) |
| 610 | { |
| 611 | // XXX: Ignore initial XML declaration, per the spec. |
| 612 | if ($this->insertMode === static::IM_INITIAL && 'xml' === strtolower($name)) { |
| 613 | return; |
| 614 | } |
| 615 | |
| 616 | // Important: The processor may modify the current DOM tree however it sees fit. |
| 617 | if ($this->processor instanceof InstructionProcessor) { |
| 618 | $res = $this->processor->process($this->current, $name, $data); |
| 619 | if (!empty($res)) { |
| 620 | $this->current = $res; |
| 621 | } |
| 622 | |
| 623 | return; |
| 624 | } |
| 625 | |
| 626 | // Otherwise, this is just a dumb PI element. |
| 627 | $node = $this->doc->createProcessingInstruction($name, $data); |
| 628 | |
| 629 | $this->current->appendChild($node); |
| 630 | } |
| 631 | |
| 632 | // ========================================================================== |
| 633 | // UTILITIES |
| 634 | // ========================================================================== |
| 635 | |
| 636 | /** |
| 637 | * Apply normalization rules to a tag name. |
| 638 | * See sections 2.9 and 8.1.2. |
| 639 | * |
| 640 | * @param string $tagName |
| 641 | * |
| 642 | * @return string The normalized tag name. |
| 643 | */ |
| 644 | protected function normalizeTagName($tagName) |
| 645 | { |
| 646 | /* |
| 647 | * Section 2.9 suggests that we should not do this. if (strpos($name, ':') !== false) { // We know from the grammar that there must be at least one other // char besides :, since : is not a legal tag start. $parts = explode(':', $name); return array_pop($parts); } |
| 648 | */ |
| 649 | return $tagName; |
| 650 | } |
| 651 | |
| 652 | protected function quirksTreeResolver($name) |
| 653 | { |
| 654 | throw new \Exception('Not implemented.'); |
| 655 | } |
| 656 | |
| 657 | /** |
| 658 | * Automatically climb the tree and close the closest node with the matching $tag. |
| 659 | * |
| 660 | * @param string $tagName |
| 661 | * |
| 662 | * @return bool |
| 663 | */ |
| 664 | protected function autoclose($tagName) |
| 665 | { |
| 666 | $working = $this->current; |
| 667 | do { |
| 668 | if (XML_ELEMENT_NODE !== $working->nodeType) { |
| 669 | return false; |
| 670 | } |
| 671 | if ($working->tagName === $tagName) { |
| 672 | $this->current = $working->parentNode; |
| 673 | |
| 674 | return true; |
| 675 | } |
| 676 | } while ($working = $working->parentNode); |
| 677 | |
| 678 | return false; |
| 679 | } |
| 680 | |
| 681 | /** |
| 682 | * Checks if the given tagname is an ancestor of the present candidate. |
| 683 | * |
| 684 | * If $this->current or anything above $this->current matches the given tag |
| 685 | * name, this returns true. |
| 686 | * |
| 687 | * @param string $tagName |
| 688 | * |
| 689 | * @return bool |
| 690 | */ |
| 691 | protected function isAncestor($tagName) |
| 692 | { |
| 693 | $candidate = $this->current; |
| 694 | while (XML_ELEMENT_NODE === $candidate->nodeType) { |
| 695 | if ($candidate->tagName === $tagName) { |
| 696 | return true; |
| 697 | } |
| 698 | $candidate = $candidate->parentNode; |
| 699 | } |
| 700 | |
| 701 | return false; |
| 702 | } |
| 703 | |
| 704 | /** |
| 705 | * Returns true if the immediate parent element is of the given tagname. |
| 706 | * |
| 707 | * @param string $tagName |
| 708 | * |
| 709 | * @return bool |
| 710 | */ |
| 711 | protected function isParent($tagName) |
| 712 | { |
| 713 | return $this->current->tagName === $tagName; |
| 714 | } |
| 715 | } |
| 716 |