| 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 |
public function setInstructionProcessor(InstructionProcessor $proc) |
| 236 |
{ |
| 237 |
$this->processor = $proc; |
| 238 |
} |
| 239 |
|
| 240 |
public function doctype($name, $idType = 0, $id = null, $quirks = false) |
| 241 |
{ |
| 242 |
// This is used solely for setting quirks mode. Currently we don't |
| 243 |
// try to preserve the inbound DT. We convert it to HTML5. |
| 244 |
$this->quirks = $quirks; |
| 245 |
|
| 246 |
if ($this->insertMode > static::IM_INITIAL) { |
| 247 |
$this->parseError('Illegal placement of DOCTYPE tag. Ignoring: ' . $name); |
| 248 |
|
| 249 |
return; |
| 250 |
} |
| 251 |
|
| 252 |
$this->insertMode = static::IM_BEFORE_HTML; |
| 253 |
} |
| 254 |
|
| 255 |
/** |
| 256 |
* Process the start tag. |
| 257 |
* |
| 258 |
* @todo - XMLNS namespace handling (we need to parse, even if it's not valid) |
| 259 |
* - XLink, MathML and SVG namespace handling |
| 260 |
* - Omission rules: 8.1.2.4 Optional tags |
| 261 |
* |
| 262 |
* @param string $name |
| 263 |
* @param array $attributes |
| 264 |
* @param bool $selfClosing |
| 265 |
* |
| 266 |
* @return int |
| 267 |
*/ |
| 268 |
public function startTag($name, $attributes = array(), $selfClosing = false) |
| 269 |
{ |
| 270 |
$lname = $this->normalizeTagName($name); |
| 271 |
|
| 272 |
// Make sure we have an html element. |
| 273 |
if (!$this->doc->documentElement && 'html' !== $name && !$this->frag) { |
| 274 |
$this->startTag('html'); |
| 275 |
} |
| 276 |
|
| 277 |
// Set quirks mode if we're at IM_INITIAL with no doctype. |
| 278 |
if ($this->insertMode === static::IM_INITIAL) { |
| 279 |
$this->quirks = true; |
| 280 |
$this->parseError('No DOCTYPE specified.'); |
| 281 |
} |
| 282 |
|
| 283 |
// SPECIAL TAG HANDLING: |
| 284 |
// Spec says do this, and "don't ask." |
| 285 |
// find the spec where this is defined... looks problematic |
| 286 |
if ('image' === $name && !($this->insertMode === static::IM_IN_SVG || $this->insertMode === static::IM_IN_MATHML)) { |
| 287 |
$name = 'img'; |
| 288 |
} |
| 289 |
|
| 290 |
// Autoclose p tags where appropriate. |
| 291 |
if ($this->insertMode >= static::IM_IN_BODY && Elements::isA($name, Elements::AUTOCLOSE_P)) { |
| 292 |
$this->autoclose('p'); |
| 293 |
} |
| 294 |
|
| 295 |
// Set insert mode: |
| 296 |
switch ($name) { |
| 297 |
case 'html': |
| 298 |
$this->insertMode = static::IM_BEFORE_HEAD; |
| 299 |
break; |
| 300 |
case 'head': |
| 301 |
if ($this->insertMode > static::IM_BEFORE_HEAD) { |
| 302 |
$this->parseError('Unexpected head tag outside of head context.'); |
| 303 |
} else { |
| 304 |
$this->insertMode = static::IM_IN_HEAD; |
| 305 |
} |
| 306 |
break; |
| 307 |
case 'body': |
| 308 |
$this->insertMode = static::IM_IN_BODY; |
| 309 |
break; |
| 310 |
case 'svg': |
| 311 |
$this->insertMode = static::IM_IN_SVG; |
| 312 |
break; |
| 313 |
case 'math': |
| 314 |
$this->insertMode = static::IM_IN_MATHML; |
| 315 |
break; |
| 316 |
case 'noscript': |
| 317 |
if ($this->insertMode === static::IM_IN_HEAD) { |
| 318 |
$this->insertMode = static::IM_IN_HEAD_NOSCRIPT; |
| 319 |
} |
| 320 |
break; |
| 321 |
} |
| 322 |
|
| 323 |
// Special case handling for SVG. |
| 324 |
if ($this->insertMode === static::IM_IN_SVG) { |
| 325 |
$lname = Elements::normalizeSvgElement($lname); |
| 326 |
} |
| 327 |
|
| 328 |
$pushes = 0; |
| 329 |
// when we found a tag thats appears inside $nsRoots, we have to switch the defalut namespace |
| 330 |
if (isset($this->nsRoots[$lname]) && $this->nsStack[0][''] !== $this->nsRoots[$lname]) { |
| 331 |
array_unshift($this->nsStack, array( |
| 332 |
'' => $this->nsRoots[$lname], |
| 333 |
) + $this->nsStack[0]); |
| 334 |
++$pushes; |
| 335 |
} |
| 336 |
$needsWorkaround = false; |
| 337 |
if (isset($this->options['xmlNamespaces']) && $this->options['xmlNamespaces']) { |
| 338 |
// when xmlNamespaces is true a and we found a 'xmlns' or 'xmlns:*' attribute, we should add a new item to the $nsStack |
| 339 |
foreach ($attributes as $aName => $aVal) { |
| 340 |
if ('xmlns' === $aName) { |
| 341 |
$needsWorkaround = $aVal; |
| 342 |
array_unshift($this->nsStack, array( |
| 343 |
'' => $aVal, |
| 344 |
) + $this->nsStack[0]); |
| 345 |
++$pushes; |
| 346 |
} elseif ('xmlns' === (($pos = strpos($aName, ':')) ? substr($aName, 0, $pos) : '')) { |
| 347 |
array_unshift($this->nsStack, array( |
| 348 |
substr($aName, $pos + 1) => $aVal, |
| 349 |
) + $this->nsStack[0]); |
| 350 |
++$pushes; |
| 351 |
} |
| 352 |
} |
| 353 |
} |
| 354 |
|
| 355 |
if ($this->onlyInline && Elements::isA($lname, Elements::BLOCK_TAG)) { |
| 356 |
$this->autoclose($this->onlyInline); |
| 357 |
$this->onlyInline = null; |
| 358 |
} |
| 359 |
|
| 360 |
// some elements as table related tags might have optional end tags that force us to auto close multiple tags |
| 361 |
// https://www.w3.org/TR/html401/struct/tables.html |
| 362 |
if ($this->current instanceof \DOMElement && isset(Elements::$optionalEndElementsParentsToClose[$lname])) { |
| 363 |
foreach (Elements::$optionalEndElementsParentsToClose[$lname] as $parentElName) { |
| 364 |
if ($this->current instanceof \DOMElement && $this->current->tagName === $parentElName) { |
| 365 |
$this->autoclose($parentElName); |
| 366 |
} |
| 367 |
} |
| 368 |
} |
| 369 |
|
| 370 |
try { |
| 371 |
$prefix = ($pos = strpos($lname, ':')) ? substr($lname, 0, $pos) : ''; |
| 372 |
|
| 373 |
if (false !== $needsWorkaround) { |
| 374 |
$xml = "<$lname xmlns=\"$needsWorkaround\" " . (strlen($prefix) && isset($this->nsStack[0][$prefix]) ? ("xmlns:$prefix=\"" . $this->nsStack[0][$prefix] . '"') : '') . '/>'; |
| 375 |
|
| 376 |
$frag = new \DOMDocument('1.0', 'UTF-8'); |
| 377 |
$frag->loadXML($xml); |
| 378 |
|
| 379 |
$ele = $this->doc->importNode($frag->documentElement, true); |
| 380 |
} else { |
| 381 |
if (!isset($this->nsStack[0][$prefix]) || ('' === $prefix && isset($this->options[self::OPT_DISABLE_HTML_NS]) && $this->options[self::OPT_DISABLE_HTML_NS])) { |
| 382 |
$ele = $this->doc->createElement($lname); |
| 383 |
} else { |
| 384 |
$ele = $this->doc->createElementNS($this->nsStack[0][$prefix], $lname); |
| 385 |
} |
| 386 |
} |
| 387 |
} catch (\DOMException $e) { |
| 388 |
$this->parseError("Illegal tag name: <$lname>. Replaced with <invalid>."); |
| 389 |
$ele = $this->doc->createElement('invalid'); |
| 390 |
} |
| 391 |
|
| 392 |
if (Elements::isA($lname, Elements::BLOCK_ONLY_INLINE)) { |
| 393 |
$this->onlyInline = $lname; |
| 394 |
} |
| 395 |
|
| 396 |
// When we add some namespacess, we have to track them. Later, when "endElement" is invoked, we have to remove them. |
| 397 |
// When we are on a void tag, we do not need to care about namesapce nesting. |
| 398 |
if ($pushes > 0 && !Elements::isA($name, Elements::VOID_TAG)) { |
| 399 |
// PHP tends to free the memory used by DOM, |
| 400 |
// to avoid spl_object_id collisions we have to avoid garbage collection of $ele storing it into $pushes |
| 401 |
// see https://bugs.php.net/bug.php?id=67459 |
| 402 |
$this->pushes[spl_object_id($ele)] = array($pushes, $ele); |
| 403 |
} |
| 404 |
|
| 405 |
foreach ($attributes as $aName => $aVal) { |
| 406 |
// xmlns attributes can't be set |
| 407 |
if ('xmlns' === $aName) { |
| 408 |
continue; |
| 409 |
} |
| 410 |
|
| 411 |
if ($this->insertMode === static::IM_IN_SVG) { |
| 412 |
$aName = Elements::normalizeSvgAttribute($aName); |
| 413 |
} elseif ($this->insertMode === static::IM_IN_MATHML) { |
| 414 |
$aName = Elements::normalizeMathMlAttribute($aName); |
| 415 |
} |
| 416 |
|
| 417 |
$aVal = (string) $aVal; |
| 418 |
|
| 419 |
try { |
| 420 |
$prefix = ($pos = strpos($aName, ':')) ? substr($aName, 0, $pos) : false; |
| 421 |
|
| 422 |
if ('xmlns' === $prefix) { |
| 423 |
$ele->setAttributeNS(self::NAMESPACE_XMLNS, $aName, $aVal); |
| 424 |
} elseif (false !== $prefix && isset($this->nsStack[0][$prefix])) { |
| 425 |
$ele->setAttributeNS($this->nsStack[0][$prefix], $aName, $aVal); |
| 426 |
} else { |
| 427 |
$ele->setAttribute($aName, $aVal); |
| 428 |
} |
| 429 |
} catch (\DOMException $e) { |
| 430 |
$this->parseError("Illegal attribute name for tag $name. Ignoring: $aName"); |
| 431 |
continue; |
| 432 |
} |
| 433 |
|
| 434 |
// This is necessary on a non-DTD schema, like HTML5. |
| 435 |
if ('id' === $aName) { |
| 436 |
$ele->setIdAttribute('id', true); |
| 437 |
} |
| 438 |
} |
| 439 |
|
| 440 |
if ($this->frag !== $this->current && $this->rules->hasRules($name)) { |
| 441 |
// Some elements have special processing rules. Handle those separately. |
| 442 |
$this->current = $this->rules->evaluate($ele, $this->current); |
| 443 |
} else { |
| 444 |
// Otherwise, it's a standard element. |
| 445 |
$this->current->appendChild($ele); |
| 446 |
|
| 447 |
if (!Elements::isA($name, Elements::VOID_TAG)) { |
| 448 |
$this->current = $ele; |
| 449 |
} |
| 450 |
|
| 451 |
// Self-closing tags should only be respected on foreign elements |
| 452 |
// (and are implied on void elements) |
| 453 |
// See: https://www.w3.org/TR/html5/syntax.html#start-tags |
| 454 |
if (Elements::isHtml5Element($name)) { |
| 455 |
$selfClosing = false; |
| 456 |
} |
| 457 |
} |
| 458 |
|
| 459 |
// This is sort of a last-ditch attempt to correct for cases where no head/body |
| 460 |
// elements are provided. |
| 461 |
if ($this->insertMode <= static::IM_BEFORE_HEAD && 'head' !== $name && 'html' !== $name) { |
| 462 |
$this->insertMode = static::IM_IN_BODY; |
| 463 |
} |
| 464 |
|
| 465 |
// When we are on a void tag, we do not need to care about namesapce nesting, |
| 466 |
// but we have to remove the namespaces pushed to $nsStack. |
| 467 |
if ($pushes > 0 && Elements::isA($name, Elements::VOID_TAG)) { |
| 468 |
// remove the namespaced definded by current node |
| 469 |
for ($i = 0; $i < $pushes; ++$i) { |
| 470 |
array_shift($this->nsStack); |
| 471 |
} |
| 472 |
} |
| 473 |
|
| 474 |
if ($selfClosing) { |
| 475 |
$this->endTag($name); |
| 476 |
} |
| 477 |
|
| 478 |
// Return the element mask, which the tokenizer can then use to set |
| 479 |
// various processing rules. |
| 480 |
return Elements::element($name); |
| 481 |
} |
| 482 |
|
| 483 |
public function endTag($name) |
| 484 |
{ |
| 485 |
$lname = $this->normalizeTagName($name); |
| 486 |
|
| 487 |
// Special case within 12.2.6.4.7: An end tag whose tag name is "br" should be treated as an opening tag |
| 488 |
if ('br' === $name) { |
| 489 |
$this->parseError('Closing tag encountered for void element br.'); |
| 490 |
|
| 491 |
$this->startTag('br'); |
| 492 |
} |
| 493 |
// Ignore closing tags for other unary elements. |
| 494 |
elseif (Elements::isA($name, Elements::VOID_TAG)) { |
| 495 |
return; |
| 496 |
} |
| 497 |
|
| 498 |
if ($this->insertMode <= static::IM_BEFORE_HTML) { |
| 499 |
// 8.2.5.4.2 |
| 500 |
if (in_array($name, array( |
| 501 |
'html', |
| 502 |
'br', |
| 503 |
'head', |
| 504 |
'title', |
| 505 |
))) { |
| 506 |
$this->startTag('html'); |
| 507 |
$this->endTag($name); |
| 508 |
$this->insertMode = static::IM_BEFORE_HEAD; |
| 509 |
|
| 510 |
return; |
| 511 |
} |
| 512 |
|
| 513 |
// Ignore the tag. |
| 514 |
$this->parseError('Illegal closing tag at global scope.'); |
| 515 |
|
| 516 |
return; |
| 517 |
} |
| 518 |
|
| 519 |
// Special case handling for SVG. |
| 520 |
if ($this->insertMode === static::IM_IN_SVG) { |
| 521 |
$lname = Elements::normalizeSvgElement($lname); |
| 522 |
} |
| 523 |
|
| 524 |
$cid = spl_object_id($this->current); |
| 525 |
|
| 526 |
// XXX: HTML has no parent. What do we do, though, |
| 527 |
// if this element appears in the wrong place? |
| 528 |
if ('html' === $lname) { |
| 529 |
return; |
| 530 |
} |
| 531 |
|
| 532 |
// remove the namespaced definded by current node |
| 533 |
if (isset($this->pushes[$cid])) { |
| 534 |
for ($i = 0; $i < $this->pushes[$cid][0]; ++$i) { |
| 535 |
array_shift($this->nsStack); |
| 536 |
} |
| 537 |
unset($this->pushes[$cid]); |
| 538 |
} |
| 539 |
|
| 540 |
if (!$this->autoclose($lname)) { |
| 541 |
$this->parseError('Could not find closing tag for ' . $lname); |
| 542 |
} |
| 543 |
|
| 544 |
switch ($lname) { |
| 545 |
case 'head': |
| 546 |
$this->insertMode = static::IM_AFTER_HEAD; |
| 547 |
break; |
| 548 |
case 'body': |
| 549 |
$this->insertMode = static::IM_AFTER_BODY; |
| 550 |
break; |
| 551 |
case 'svg': |
| 552 |
case 'mathml': |
| 553 |
$this->insertMode = static::IM_IN_BODY; |
| 554 |
break; |
| 555 |
} |
| 556 |
} |
| 557 |
|
| 558 |
public function comment($cdata) |
| 559 |
{ |
| 560 |
// TODO: Need to handle case where comment appears outside of the HTML tag. |
| 561 |
$node = $this->doc->createComment($cdata); |
| 562 |
$this->current->appendChild($node); |
| 563 |
} |
| 564 |
|
| 565 |
public function text($data) |
| 566 |
{ |
| 567 |
// XXX: Hmmm.... should we really be this strict? |
| 568 |
if ($this->insertMode < static::IM_IN_HEAD) { |
| 569 |
// Per '8.2.5.4.3 The "before head" insertion mode' the characters |
| 570 |
// " \t\n\r\f" should be ignored but no mention of a parse error. This is |
| 571 |
// practical as most documents contain these characters. Other text is not |
| 572 |
// expected here so recording a parse error is necessary. |
| 573 |
$dataTmp = trim($data, " \t\n\r\f"); |
| 574 |
if (!empty($dataTmp)) { |
| 575 |
// fprintf(STDOUT, "Unexpected insert mode: %d", $this->insertMode); |
| 576 |
$this->parseError('Unexpected text. Ignoring: ' . $dataTmp); |
| 577 |
} |
| 578 |
|
| 579 |
return; |
| 580 |
} |
| 581 |
// fprintf(STDOUT, "Appending text %s.", $data); |
| 582 |
$node = $this->doc->createTextNode($data); |
| 583 |
$this->current->appendChild($node); |
| 584 |
} |
| 585 |
|
| 586 |
public function eof() |
| 587 |
{ |
| 588 |
// If the $current isn't the $root, do we need to do anything? |
| 589 |
} |
| 590 |
|
| 591 |
public function parseError($msg, $line = 0, $col = 0) |
| 592 |
{ |
| 593 |
$this->errors[] = sprintf('Line %d, Col %d: %s', $line, $col, $msg); |
| 594 |
} |
| 595 |
|
| 596 |
public function getErrors() |
| 597 |
{ |
| 598 |
return $this->errors; |
| 599 |
} |
| 600 |
|
| 601 |
public function cdata($data) |
| 602 |
{ |
| 603 |
$node = $this->doc->createCDATASection($data); |
| 604 |
$this->current->appendChild($node); |
| 605 |
} |
| 606 |
|
| 607 |
public function processingInstruction($name, $data = null) |
| 608 |
{ |
| 609 |
// XXX: Ignore initial XML declaration, per the spec. |
| 610 |
if ($this->insertMode === static::IM_INITIAL && 'xml' === strtolower($name)) { |
| 611 |
return; |
| 612 |
} |
| 613 |
|
| 614 |
// Important: The processor may modify the current DOM tree however it sees fit. |
| 615 |
if ($this->processor instanceof InstructionProcessor) { |
| 616 |
$res = $this->processor->process($this->current, $name, $data); |
| 617 |
if (!empty($res)) { |
| 618 |
$this->current = $res; |
| 619 |
} |
| 620 |
|
| 621 |
return; |
| 622 |
} |
| 623 |
|
| 624 |
// Otherwise, this is just a dumb PI element. |
| 625 |
$node = $this->doc->createProcessingInstruction($name, $data); |
| 626 |
|
| 627 |
$this->current->appendChild($node); |
| 628 |
} |
| 629 |
|
| 630 |
// ========================================================================== |
| 631 |
// UTILITIES |
| 632 |
// ========================================================================== |
| 633 |
|
| 634 |
/** |
| 635 |
* Apply normalization rules to a tag name. |
| 636 |
* See sections 2.9 and 8.1.2. |
| 637 |
* |
| 638 |
* @param string $tagName |
| 639 |
* |
| 640 |
* @return string The normalized tag name. |
| 641 |
*/ |
| 642 |
protected function normalizeTagName($tagName) |
| 643 |
{ |
| 644 |
/* |
| 645 |
* 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); } |
| 646 |
*/ |
| 647 |
return $tagName; |
| 648 |
} |
| 649 |
|
| 650 |
protected function quirksTreeResolver($name) |
| 651 |
{ |
| 652 |
throw new \Exception('Not implemented.'); |
| 653 |
} |
| 654 |
|
| 655 |
/** |
| 656 |
* Automatically climb the tree and close the closest node with the matching $tag. |
| 657 |
* |
| 658 |
* @param string $tagName |
| 659 |
* |
| 660 |
* @return bool |
| 661 |
*/ |
| 662 |
protected function autoclose($tagName) |
| 663 |
{ |
| 664 |
$working = $this->current; |
| 665 |
do { |
| 666 |
if (XML_ELEMENT_NODE !== $working->nodeType) { |
| 667 |
return false; |
| 668 |
} |
| 669 |
if ($working->tagName === $tagName) { |
| 670 |
$this->current = $working->parentNode; |
| 671 |
|
| 672 |
return true; |
| 673 |
} |
| 674 |
} while ($working = $working->parentNode); |
| 675 |
|
| 676 |
return false; |
| 677 |
} |
| 678 |
|
| 679 |
/** |
| 680 |
* Checks if the given tagname is an ancestor of the present candidate. |
| 681 |
* |
| 682 |
* If $this->current or anything above $this->current matches the given tag |
| 683 |
* name, this returns true. |
| 684 |
* |
| 685 |
* @param string $tagName |
| 686 |
* |
| 687 |
* @return bool |
| 688 |
*/ |
| 689 |
protected function isAncestor($tagName) |
| 690 |
{ |
| 691 |
$candidate = $this->current; |
| 692 |
while (XML_ELEMENT_NODE === $candidate->nodeType) { |
| 693 |
if ($candidate->tagName === $tagName) { |
| 694 |
return true; |
| 695 |
} |
| 696 |
$candidate = $candidate->parentNode; |
| 697 |
} |
| 698 |
|
| 699 |
return false; |
| 700 |
} |
| 701 |
|
| 702 |
/** |
| 703 |
* Returns true if the immediate parent element is of the given tagname. |
| 704 |
* |
| 705 |
* @param string $tagName |
| 706 |
* |
| 707 |
* @return bool |
| 708 |
*/ |
| 709 |
protected function isParent($tagName) |
| 710 |
{ |
| 711 |
return $this->current->tagName === $tagName; |
| 712 |
} |
| 713 |
} |
| 714 |
|
