Chris@0: <?php
Chris@0: namespace Masterminds\HTML5\Parser;
Chris@0: 
Chris@0: use Masterminds\HTML5\Elements;
Chris@0: 
Chris@0: /**
Chris@0:  * Create an HTML5 DOM tree from events.
Chris@0:  *
Chris@0:  * This attempts to create a DOM from events emitted by a parser. This
Chris@0:  * attempts (but does not guarantee) to up-convert older HTML documents
Chris@0:  * to HTML5. It does this by applying HTML5's rules, but it will not
Chris@0:  * change the architecture of the document itself.
Chris@0:  *
Chris@0:  * Many of the error correction and quirks features suggested in the specification
Chris@0:  * are implemented herein; however, not all of them are. Since we do not
Chris@0:  * assume a graphical user agent, no presentation-specific logic is conducted
Chris@0:  * during tree building.
Chris@0:  *
Chris@0:  * FIXME: The present tree builder does not exactly follow the state machine rules
Chris@0:  * for insert modes as outlined in the HTML5 spec. The processor needs to be
Chris@0:  * re-written to accomodate this. See, for example, the Go language HTML5
Chris@0:  * parser.
Chris@0:  */
Chris@0: class DOMTreeBuilder implements EventHandler
Chris@0: {
Chris@0:     /**
Chris@0:      * Defined in http://www.w3.org/TR/html51/infrastructure.html#html-namespace-0
Chris@0:      */
Chris@0:     const NAMESPACE_HTML = 'http://www.w3.org/1999/xhtml';
Chris@0: 
Chris@0:     const NAMESPACE_MATHML = 'http://www.w3.org/1998/Math/MathML';
Chris@0: 
Chris@0:     const NAMESPACE_SVG = 'http://www.w3.org/2000/svg';
Chris@0: 
Chris@0:     const NAMESPACE_XLINK = 'http://www.w3.org/1999/xlink';
Chris@0: 
Chris@0:     const NAMESPACE_XML = 'http://www.w3.org/XML/1998/namespace';
Chris@0: 
Chris@0:     const NAMESPACE_XMLNS = 'http://www.w3.org/2000/xmlns/';
Chris@0: 
Chris@0:     const OPT_DISABLE_HTML_NS = 'disable_html_ns';
Chris@0: 
Chris@0:     const OPT_TARGET_DOC = 'target_document';
Chris@0: 
Chris@0:     const OPT_IMPLICIT_NS = 'implicit_namespaces';
Chris@0: 
Chris@0:     /**
Chris@0:      * Holds the HTML5 element names that causes a namespace switch
Chris@0:      *
Chris@0:      * @var array
Chris@0:      */
Chris@0:     protected $nsRoots = array(
Chris@0:         'html' => self::NAMESPACE_HTML,
Chris@0:         'svg' => self::NAMESPACE_SVG,
Chris@0:         'math' => self::NAMESPACE_MATHML
Chris@0:     );
Chris@0: 
Chris@0:     /**
Chris@0:      * Holds the always available namespaces (which does not require the XMLNS declaration).
Chris@0:      *
Chris@0:      * @var array
Chris@0:      */
Chris@0:     protected $implicitNamespaces = array(
Chris@0:         'xml' => self::NAMESPACE_XML,
Chris@0:         'xmlns' => self::NAMESPACE_XMLNS,
Chris@0:         'xlink' => self::NAMESPACE_XLINK
Chris@0:     );
Chris@0: 
Chris@0:     /**
Chris@0:      * Holds a stack of currently active namespaces.
Chris@0:      *
Chris@0:      * @var array
Chris@0:      */
Chris@0:     protected $nsStack = array();
Chris@0: 
Chris@0:     /**
Chris@0:      * Holds the number of namespaces declared by a node.
Chris@0:      *
Chris@0:      * @var array
Chris@0:      */
Chris@0:     protected $pushes = array();
Chris@0: 
Chris@0:     /**
Chris@0:      * Defined in 8.2.5.
Chris@0:      */
Chris@0:     const IM_INITIAL = 0;
Chris@0: 
Chris@0:     const IM_BEFORE_HTML = 1;
Chris@0: 
Chris@0:     const IM_BEFORE_HEAD = 2;
Chris@0: 
Chris@0:     const IM_IN_HEAD = 3;
Chris@0: 
Chris@0:     const IM_IN_HEAD_NOSCRIPT = 4;
Chris@0: 
Chris@0:     const IM_AFTER_HEAD = 5;
Chris@0: 
Chris@0:     const IM_IN_BODY = 6;
Chris@0: 
Chris@0:     const IM_TEXT = 7;
Chris@0: 
Chris@0:     const IM_IN_TABLE = 8;
Chris@0: 
Chris@0:     const IM_IN_TABLE_TEXT = 9;
Chris@0: 
Chris@0:     const IM_IN_CAPTION = 10;
Chris@0: 
Chris@0:     const IM_IN_COLUMN_GROUP = 11;
Chris@0: 
Chris@0:     const IM_IN_TABLE_BODY = 12;
Chris@0: 
Chris@0:     const IM_IN_ROW = 13;
Chris@0: 
Chris@0:     const IM_IN_CELL = 14;
Chris@0: 
Chris@0:     const IM_IN_SELECT = 15;
Chris@0: 
Chris@0:     const IM_IN_SELECT_IN_TABLE = 16;
Chris@0: 
Chris@0:     const IM_AFTER_BODY = 17;
Chris@0: 
Chris@0:     const IM_IN_FRAMESET = 18;
Chris@0: 
Chris@0:     const IM_AFTER_FRAMESET = 19;
Chris@0: 
Chris@0:     const IM_AFTER_AFTER_BODY = 20;
Chris@0: 
Chris@0:     const IM_AFTER_AFTER_FRAMESET = 21;
Chris@0: 
Chris@0:     const IM_IN_SVG = 22;
Chris@0: 
Chris@0:     const IM_IN_MATHML = 23;
Chris@0: 
Chris@0:     protected $options = array();
Chris@0: 
Chris@0:     protected $stack = array();
Chris@0: 
Chris@0:     protected $current; // Pointer in the tag hierarchy.
Chris@0:     protected $doc;
Chris@0: 
Chris@0:     protected $frag;
Chris@0: 
Chris@0:     protected $processor;
Chris@0: 
Chris@0:     protected $insertMode = 0;
Chris@0: 
Chris@0:     /**
Chris@0:      * Track if we are in an element that allows only inline child nodes
Chris@0:      * @var string|null
Chris@0:      */
Chris@0:     protected $onlyInline;
Chris@0: 
Chris@0:     /**
Chris@0:      * Quirks mode is enabled by default.
Chris@0:      * Any document that is missing the
Chris@0:      * DT will be considered to be in quirks mode.
Chris@0:      */
Chris@0:     protected $quirks = true;
Chris@0: 
Chris@0:     protected $errors = array();
Chris@0: 
Chris@0:     public function __construct($isFragment = false, array $options = array())
Chris@0:     {
Chris@0:         $this->options = $options;
Chris@0: 
Chris@0:         if (isset($options[self::OPT_TARGET_DOC])) {
Chris@0:             $this->doc = $options[self::OPT_TARGET_DOC];
Chris@0:         } else {
Chris@0:             $impl = new \DOMImplementation();
Chris@0:             // XXX:
Chris@0:             // Create the doctype. For now, we are always creating HTML5
Chris@0:             // documents, and attempting to up-convert any older DTDs to HTML5.
Chris@0:             $dt = $impl->createDocumentType('html');
Chris@0:             // $this->doc = \DOMImplementation::createDocument(NULL, 'html', $dt);
Chris@0:             $this->doc = $impl->createDocument(null, null, $dt);
Chris@0:         }
Chris@0:         $this->errors = array();
Chris@0: 
Chris@0:         $this->current = $this->doc; // ->documentElement;
Chris@0: 
Chris@0:         // Create a rules engine for tags.
Chris@0:         $this->rules = new TreeBuildingRules($this->doc);
Chris@0: 
Chris@0:         $implicitNS = array();
Chris@0:         if (isset($this->options[self::OPT_IMPLICIT_NS])) {
Chris@0:             $implicitNS = $this->options[self::OPT_IMPLICIT_NS];
Chris@0:         } elseif (isset($this->options["implicitNamespaces"])) {
Chris@0:             $implicitNS = $this->options["implicitNamespaces"];
Chris@0:         }
Chris@0: 
Chris@0:         // Fill $nsStack with the defalut HTML5 namespaces, plus the "implicitNamespaces" array taken form $options
Chris@0:         array_unshift($this->nsStack, $implicitNS + array(
Chris@0:             '' => self::NAMESPACE_HTML
Chris@0:         ) + $this->implicitNamespaces);
Chris@0: 
Chris@0:         if ($isFragment) {
Chris@0:             $this->insertMode = static::IM_IN_BODY;
Chris@0:             $this->frag = $this->doc->createDocumentFragment();
Chris@0:             $this->current = $this->frag;
Chris@0:         }
Chris@0:     }
Chris@0: 
Chris@0:     /**
Chris@0:      * Get the document.
Chris@0:      */
Chris@0:     public function document()
Chris@0:     {
Chris@0:         return $this->doc;
Chris@0:     }
Chris@0: 
Chris@0:     /**
Chris@0:      * Get the DOM fragment for the body.
Chris@0:      *
Chris@0:      * This returns a DOMNodeList because a fragment may have zero or more
Chris@0:      * DOMNodes at its root.
Chris@0:      *
Chris@0:      * @see http://www.w3.org/TR/2012/CR-html5-20121217/syntax.html#concept-frag-parse-context
Chris@0:      *
Chris@0:      * @return \DOMFragmentDocumentFragment
Chris@0:      */
Chris@0:     public function fragment()
Chris@0:     {
Chris@0:         return $this->frag;
Chris@0:     }
Chris@0: 
Chris@0:     /**
Chris@0:      * Provide an instruction processor.
Chris@0:      *
Chris@0:      * This is used for handling Processor Instructions as they are
Chris@0:      * inserted. If omitted, PI's are inserted directly into the DOM tree.
Chris@0:      */
Chris@0:     public function setInstructionProcessor(\Masterminds\HTML5\InstructionProcessor $proc)
Chris@0:     {
Chris@0:         $this->processor = $proc;
Chris@0:     }
Chris@0: 
Chris@0:     public function doctype($name, $idType = 0, $id = null, $quirks = false)
Chris@0:     {
Chris@0:         // This is used solely for setting quirks mode. Currently we don't
Chris@0:         // try to preserve the inbound DT. We convert it to HTML5.
Chris@0:         $this->quirks = $quirks;
Chris@0: 
Chris@0:         if ($this->insertMode > static::IM_INITIAL) {
Chris@0:             $this->parseError("Illegal placement of DOCTYPE tag. Ignoring: " . $name);
Chris@0: 
Chris@0:             return;
Chris@0:         }
Chris@0: 
Chris@0:         $this->insertMode = static::IM_BEFORE_HTML;
Chris@0:     }
Chris@0: 
Chris@0:     /**
Chris@0:      * Process the start tag.
Chris@0:      *
Chris@0:      * @todo - XMLNS namespace handling (we need to parse, even if it's not valid)
Chris@0:      *       - XLink, MathML and SVG namespace handling
Chris@0:      *       - Omission rules: 8.1.2.4 Optional tags
Chris@0:      */
Chris@0:     public function startTag($name, $attributes = array(), $selfClosing = false)
Chris@0:     {
Chris@0:         // fprintf(STDOUT, $name);
Chris@0:         $lname = $this->normalizeTagName($name);
Chris@0: 
Chris@0:         // Make sure we have an html element.
Chris@0:         if (! $this->doc->documentElement && $name !== 'html' && ! $this->frag) {
Chris@0:             $this->startTag('html');
Chris@0:         }
Chris@0: 
Chris@0:         // Set quirks mode if we're at IM_INITIAL with no doctype.
Chris@0:         if ($this->insertMode == static::IM_INITIAL) {
Chris@0:             $this->quirks = true;
Chris@0:             $this->parseError("No DOCTYPE specified.");
Chris@0:         }
Chris@0: 
Chris@0:         // SPECIAL TAG HANDLING:
Chris@0:         // Spec says do this, and "don't ask."
Chris@0:         // find the spec where this is defined... looks problematic
Chris@0:         if ($name == 'image' && !($this->insertMode === static::IM_IN_SVG || $this->insertMode === static::IM_IN_MATHML)) {
Chris@0:             $name = 'img';
Chris@0:         }
Chris@0: 
Chris@0:         // Autoclose p tags where appropriate.
Chris@0:         if ($this->insertMode >= static::IM_IN_BODY && Elements::isA($name, Elements::AUTOCLOSE_P)) {
Chris@0:             $this->autoclose('p');
Chris@0:         }
Chris@0: 
Chris@0:         // Set insert mode:
Chris@0:         switch ($name) {
Chris@0:             case 'html':
Chris@0:                 $this->insertMode = static::IM_BEFORE_HEAD;
Chris@0:                 break;
Chris@0:             case 'head':
Chris@0:                 if ($this->insertMode > static::IM_BEFORE_HEAD) {
Chris@0:                     $this->parseError("Unexpected head tag outside of head context.");
Chris@0:                 } else {
Chris@0:                     $this->insertMode = static::IM_IN_HEAD;
Chris@0:                 }
Chris@0:                 break;
Chris@0:             case 'body':
Chris@0:                 $this->insertMode = static::IM_IN_BODY;
Chris@0:                 break;
Chris@0:             case 'svg':
Chris@0:                 $this->insertMode = static::IM_IN_SVG;
Chris@0:                 break;
Chris@0:             case 'math':
Chris@0:                 $this->insertMode = static::IM_IN_MATHML;
Chris@0:                 break;
Chris@0:             case 'noscript':
Chris@0:                 if ($this->insertMode == static::IM_IN_HEAD) {
Chris@0:                     $this->insertMode = static::IM_IN_HEAD_NOSCRIPT;
Chris@0:                 }
Chris@0:                 break;
Chris@0:         }
Chris@0: 
Chris@0:         // Special case handling for SVG.
Chris@0:         if ($this->insertMode == static::IM_IN_SVG) {
Chris@0:             $lname = Elements::normalizeSvgElement($lname);
Chris@0:         }
Chris@0: 
Chris@0:         $pushes = 0;
Chris@0:         // when we found a tag thats appears inside $nsRoots, we have to switch the defalut namespace
Chris@0:         if (isset($this->nsRoots[$lname]) && $this->nsStack[0][''] !== $this->nsRoots[$lname]) {
Chris@0:             array_unshift($this->nsStack, array(
Chris@0:                 '' => $this->nsRoots[$lname]
Chris@0:             ) + $this->nsStack[0]);
Chris@0:             $pushes ++;
Chris@0:         }
Chris@0:         $needsWorkaround = false;
Chris@0:         if (isset($this->options["xmlNamespaces"]) && $this->options["xmlNamespaces"]) {
Chris@0:             // when xmlNamespaces is true a and we found a 'xmlns' or 'xmlns:*' attribute, we should add a new item to the $nsStack
Chris@0:             foreach ($attributes as $aName => $aVal) {
Chris@0:                 if ($aName === 'xmlns') {
Chris@0:                     $needsWorkaround = $aVal;
Chris@0:                     array_unshift($this->nsStack, array(
Chris@0:                         '' => $aVal
Chris@0:                     ) + $this->nsStack[0]);
Chris@0:                     $pushes ++;
Chris@0:                 } elseif ((($pos = strpos($aName, ':')) ? substr($aName, 0, $pos) : '') === 'xmlns') {
Chris@0:                     array_unshift($this->nsStack, array(
Chris@0:                         substr($aName, $pos + 1) => $aVal
Chris@0:                     ) + $this->nsStack[0]);
Chris@0:                     $pushes ++;
Chris@0:                 }
Chris@0:             }
Chris@0:         }
Chris@0: 
Chris@0:         if ($this->onlyInline && Elements::isA($lname, Elements::BLOCK_TAG)) {
Chris@0:         	$this->autoclose($this->onlyInline);
Chris@0:         	$this->onlyInline = null;
Chris@0:         }
Chris@0: 
Chris@0:         try {
Chris@0:             $prefix = ($pos = strpos($lname, ':')) ? substr($lname, 0, $pos) : '';
Chris@0: 
Chris@0: 
Chris@0:             if ($needsWorkaround!==false) {
Chris@0: 
Chris@0:                 $xml = "<$lname xmlns=\"$needsWorkaround\" ".(strlen($prefix) && isset($this->nsStack[0][$prefix])?("xmlns:$prefix=\"".$this->nsStack[0][$prefix]."\""):"")."/>";
Chris@0: 
Chris@0:                 $frag = new \DOMDocument('1.0', 'UTF-8');
Chris@0:                 $frag->loadXML($xml);
Chris@0: 
Chris@0:                 $ele = $this->doc->importNode($frag->documentElement, true);
Chris@0: 
Chris@0:             } else {
Chris@0:                 if (!isset($this->nsStack[0][$prefix]) || ($prefix === "" && isset($this->options[self::OPT_DISABLE_HTML_NS]) && $this->options[self::OPT_DISABLE_HTML_NS])) {
Chris@0:                     $ele = $this->doc->createElement($lname);
Chris@0:                 } else {
Chris@0:                     $ele = $this->doc->createElementNS($this->nsStack[0][$prefix], $lname);
Chris@0:                 }
Chris@0:             }
Chris@0: 
Chris@0:         } catch (\DOMException $e) {
Chris@0:             $this->parseError("Illegal tag name: <$lname>. Replaced with <invalid>.");
Chris@0:             $ele = $this->doc->createElement('invalid');
Chris@0:         }
Chris@0: 
Chris@0:         if (Elements::isA($lname, Elements::BLOCK_ONLY_INLINE)) {
Chris@0:         	$this->onlyInline = $lname;
Chris@0:         }
Chris@0: 
Chris@0:         // When we add some namespacess, we have to track them. Later, when "endElement" is invoked, we have to remove them.
Chris@0:         // When we are on a void tag, we do not need to care about namesapce nesting.
Chris@0:         if ($pushes > 0 && !Elements::isA($name, Elements::VOID_TAG)) {
Chris@0:             // PHP tends to free the memory used by DOM,
Chris@0:             // to avoid spl_object_hash collisions whe have to avoid garbage collection of $ele storing it into $pushes
Chris@0:             // see https://bugs.php.net/bug.php?id=67459
Chris@0:             $this->pushes[spl_object_hash($ele)] = array($pushes, $ele);
Chris@0: 
Chris@0:             // SEE https://github.com/facebook/hhvm/issues/2962
Chris@0:             if (defined('HHVM_VERSION')) {
Chris@0:                 $ele->setAttribute('html5-php-fake-id-attribute', spl_object_hash($ele));
Chris@0:             }
Chris@0:         }
Chris@0: 
Chris@0:         foreach ($attributes as $aName => $aVal) {
Chris@0:             // xmlns attributes can't be set
Chris@0:             if ($aName === 'xmlns') {
Chris@0:                 continue;
Chris@0:             }
Chris@0: 
Chris@0:             if ($this->insertMode == static::IM_IN_SVG) {
Chris@0:                 $aName = Elements::normalizeSvgAttribute($aName);
Chris@0:             } elseif ($this->insertMode == static::IM_IN_MATHML) {
Chris@0:                 $aName = Elements::normalizeMathMlAttribute($aName);
Chris@0:             }
Chris@0: 
Chris@0:             try {
Chris@0:                 $prefix = ($pos = strpos($aName, ':')) ? substr($aName, 0, $pos) : false;
Chris@0: 
Chris@0:                 if ($prefix==='xmlns') {
Chris@0:                     $ele->setAttributeNs(self::NAMESPACE_XMLNS, $aName, $aVal);
Chris@0:                 } elseif ($prefix!==false && isset($this->nsStack[0][$prefix])) {
Chris@0:                     $ele->setAttributeNs($this->nsStack[0][$prefix], $aName, $aVal);
Chris@0:                 } else {
Chris@0:                     $ele->setAttribute($aName, $aVal);
Chris@0:                 }
Chris@0:             } catch (\DOMException $e) {
Chris@0:                 $this->parseError("Illegal attribute name for tag $name. Ignoring: $aName");
Chris@0:                 continue;
Chris@0:             }
Chris@0: 
Chris@0:             // This is necessary on a non-DTD schema, like HTML5.
Chris@0:             if ($aName == 'id') {
Chris@0:                 $ele->setIdAttribute('id', true);
Chris@0:             }
Chris@0:         }
Chris@0: 
Chris@0:         // Some elements have special processing rules. Handle those separately.
Chris@0:         if ($this->rules->hasRules($name) && $this->frag !== $this->current) {
Chris@0:             $this->current = $this->rules->evaluate($ele, $this->current);
Chris@0:         }         // Otherwise, it's a standard element.
Chris@0:         else {
Chris@0:             $this->current->appendChild($ele);
Chris@0: 
Chris@0:             // XXX: Need to handle self-closing tags and unary tags.
Chris@0:             if (! Elements::isA($name, Elements::VOID_TAG)) {
Chris@0:                 $this->current = $ele;
Chris@0:             }
Chris@0:         }
Chris@0: 
Chris@0:         // This is sort of a last-ditch attempt to correct for cases where no head/body
Chris@0:         // elements are provided.
Chris@0:         if ($this->insertMode <= static::IM_BEFORE_HEAD && $name != 'head' && $name != 'html') {
Chris@0:             $this->insertMode = static::IM_IN_BODY;
Chris@0:         }
Chris@0: 
Chris@0:         // When we are on a void tag, we do not need to care about namesapce nesting,
Chris@0:         // but we have to remove the namespaces pushed to $nsStack.
Chris@0:         if ($pushes > 0 && Elements::isA($name, Elements::VOID_TAG)) {
Chris@0:             // remove the namespaced definded by current node
Chris@0:             for ($i = 0; $i < $pushes; $i ++) {
Chris@0:                 array_shift($this->nsStack);
Chris@0:             }
Chris@0:         }
Chris@0:         // Return the element mask, which the tokenizer can then use to set
Chris@0:         // various processing rules.
Chris@0:         return Elements::element($name);
Chris@0:     }
Chris@0: 
Chris@0:     public function endTag($name)
Chris@0:     {
Chris@0:         $lname = $this->normalizeTagName($name);
Chris@0: 
Chris@0:         // Ignore closing tags for unary elements.
Chris@0:         if (Elements::isA($name, Elements::VOID_TAG)) {
Chris@0:             return;
Chris@0:         }
Chris@0: 
Chris@0:         if ($this->insertMode <= static::IM_BEFORE_HTML) {
Chris@0:             // 8.2.5.4.2
Chris@0:             if (in_array($name, array(
Chris@0:                 'html',
Chris@0:                 'br',
Chris@0:                 'head',
Chris@0:                 'title'
Chris@0:             ))) {
Chris@0:                 $this->startTag('html');
Chris@0:                 $this->endTag($name);
Chris@0:                 $this->insertMode = static::IM_BEFORE_HEAD;
Chris@0: 
Chris@0:                 return;
Chris@0:             }
Chris@0: 
Chris@0:             // Ignore the tag.
Chris@0:             $this->parseError("Illegal closing tag at global scope.");
Chris@0: 
Chris@0:             return;
Chris@0:         }
Chris@0: 
Chris@0:         // Special case handling for SVG.
Chris@0:         if ($this->insertMode == static::IM_IN_SVG) {
Chris@0:             $lname = Elements::normalizeSvgElement($lname);
Chris@0:         }
Chris@0: 
Chris@0:         // See https://github.com/facebook/hhvm/issues/2962
Chris@0:         if (defined('HHVM_VERSION') && ($cid = $this->current->getAttribute('html5-php-fake-id-attribute'))) {
Chris@0:             $this->current->removeAttribute('html5-php-fake-id-attribute');
Chris@0:         } else {
Chris@0:             $cid = spl_object_hash($this->current);
Chris@0:         }
Chris@0: 
Chris@0:         // XXX: Not sure whether we need this anymore.
Chris@0:         // if ($name != $lname) {
Chris@0:         // return $this->quirksTreeResolver($lname);
Chris@0:         // }
Chris@0: 
Chris@0:         // XXX: HTML has no parent. What do we do, though,
Chris@0:         // if this element appears in the wrong place?
Chris@0:         if ($lname == 'html') {
Chris@0:             return;
Chris@0:         }
Chris@0: 
Chris@0:         // remove the namespaced definded by current node
Chris@0:         if (isset($this->pushes[$cid])) {
Chris@0:             for ($i = 0; $i < $this->pushes[$cid][0]; $i ++) {
Chris@0:                 array_shift($this->nsStack);
Chris@0:             }
Chris@0:             unset($this->pushes[$cid]);
Chris@0:         }
Chris@0: 
Chris@0:         if (! $this->autoclose($lname)) {
Chris@0:             $this->parseError('Could not find closing tag for ' . $lname);
Chris@0:         }
Chris@0: 
Chris@0:         // switch ($this->insertMode) {
Chris@0:         switch ($lname) {
Chris@0:             case "head":
Chris@0:                 $this->insertMode = static::IM_AFTER_HEAD;
Chris@0:                 break;
Chris@0:             case "body":
Chris@0:                 $this->insertMode = static::IM_AFTER_BODY;
Chris@0:                 break;
Chris@0:             case "svg":
Chris@0:             case "mathml":
Chris@0:                 $this->insertMode = static::IM_IN_BODY;
Chris@0:                 break;
Chris@0:         }
Chris@0:     }
Chris@0: 
Chris@0:     public function comment($cdata)
Chris@0:     {
Chris@0:         // TODO: Need to handle case where comment appears outside of the HTML tag.
Chris@0:         $node = $this->doc->createComment($cdata);
Chris@0:         $this->current->appendChild($node);
Chris@0:     }
Chris@0: 
Chris@0:     public function text($data)
Chris@0:     {
Chris@0:         // XXX: Hmmm.... should we really be this strict?
Chris@0:         if ($this->insertMode < static::IM_IN_HEAD) {
Chris@0:             // Per '8.2.5.4.3 The "before head" insertion mode' the characters
Chris@0:             // " \t\n\r\f" should be ignored but no mention of a parse error. This is
Chris@0:             // practical as most documents contain these characters. Other text is not
Chris@0:             // expected here so recording a parse error is necessary.
Chris@0:             $dataTmp = trim($data, " \t\n\r\f");
Chris@0:             if (! empty($dataTmp)) {
Chris@0:                 // fprintf(STDOUT, "Unexpected insert mode: %d", $this->insertMode);
Chris@0:                 $this->parseError("Unexpected text. Ignoring: " . $dataTmp);
Chris@0:             }
Chris@0: 
Chris@0:             return;
Chris@0:         }
Chris@0:         // fprintf(STDOUT, "Appending text %s.", $data);
Chris@0:         $node = $this->doc->createTextNode($data);
Chris@0:         $this->current->appendChild($node);
Chris@0:     }
Chris@0: 
Chris@0:     public function eof()
Chris@0:     {
Chris@0:         // If the $current isn't the $root, do we need to do anything?
Chris@0:     }
Chris@0: 
Chris@0:     public function parseError($msg, $line = 0, $col = 0)
Chris@0:     {
Chris@0:         $this->errors[] = sprintf("Line %d, Col %d: %s", $line, $col, $msg);
Chris@0:     }
Chris@0: 
Chris@0:     public function getErrors()
Chris@0:     {
Chris@0:         return $this->errors;
Chris@0:     }
Chris@0: 
Chris@0:     public function cdata($data)
Chris@0:     {
Chris@0:         $node = $this->doc->createCDATASection($data);
Chris@0:         $this->current->appendChild($node);
Chris@0:     }
Chris@0: 
Chris@0:     public function processingInstruction($name, $data = null)
Chris@0:     {
Chris@0:         // XXX: Ignore initial XML declaration, per the spec.
Chris@0:         if ($this->insertMode == static::IM_INITIAL && 'xml' == strtolower($name)) {
Chris@0:             return;
Chris@0:         }
Chris@0: 
Chris@0:         // Important: The processor may modify the current DOM tree however
Chris@0:         // it sees fit.
Chris@0:         if (isset($this->processor)) {
Chris@0:             $res = $this->processor->process($this->current, $name, $data);
Chris@0:             if (! empty($res)) {
Chris@0:                 $this->current = $res;
Chris@0:             }
Chris@0: 
Chris@0:             return;
Chris@0:         }
Chris@0: 
Chris@0:         // Otherwise, this is just a dumb PI element.
Chris@0:         $node = $this->doc->createProcessingInstruction($name, $data);
Chris@0: 
Chris@0:         $this->current->appendChild($node);
Chris@0:     }
Chris@0: 
Chris@0:     // ==========================================================================
Chris@0:     // UTILITIES
Chris@0:     // ==========================================================================
Chris@0: 
Chris@0:     /**
Chris@0:      * Apply normalization rules to a tag name.
Chris@0:      *
Chris@0:      * See sections 2.9 and 8.1.2.
Chris@0:      *
Chris@0:      * @param string $name
Chris@0:      *            The tag name.
Chris@0:      * @return string The normalized tag name.
Chris@0:      */
Chris@0:     protected function normalizeTagName($name)
Chris@0:     {
Chris@0:         /*
Chris@0:          * 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); }
Chris@0:          */
Chris@0:         return $name;
Chris@0:     }
Chris@0: 
Chris@0:     protected function quirksTreeResolver($name)
Chris@0:     {
Chris@0:         throw new \Exception("Not implemented.");
Chris@0:     }
Chris@0: 
Chris@0:     /**
Chris@0:      * Automatically climb the tree and close the closest node with the matching $tag.
Chris@0:      */
Chris@0:     protected function autoclose($tag)
Chris@0:     {
Chris@0:         $working = $this->current;
Chris@0:         do {
Chris@0:             if ($working->nodeType != XML_ELEMENT_NODE) {
Chris@0:                 return false;
Chris@0:             }
Chris@0:             if ($working->tagName == $tag) {
Chris@0:                 $this->current = $working->parentNode;
Chris@0: 
Chris@0:                 return true;
Chris@0:             }
Chris@0:         } while ($working = $working->parentNode);
Chris@0:         return false;
Chris@0:     }
Chris@0: 
Chris@0:     /**
Chris@0:      * Checks if the given tagname is an ancestor of the present candidate.
Chris@0:      *
Chris@0:      * If $this->current or anything above $this->current matches the given tag
Chris@0:      * name, this returns true.
Chris@0:      */
Chris@0:     protected function isAncestor($tagname)
Chris@0:     {
Chris@0:         $candidate = $this->current;
Chris@0:         while ($candidate->nodeType === XML_ELEMENT_NODE) {
Chris@0:             if ($candidate->tagName == $tagname) {
Chris@0:                 return true;
Chris@0:             }
Chris@0:             $candidate = $candidate->parentNode;
Chris@0:         }
Chris@0: 
Chris@0:         return false;
Chris@0:     }
Chris@0: 
Chris@0:     /**
Chris@0:      * Returns true if the immediate parent element is of the given tagname.
Chris@0:      */
Chris@0:     protected function isParent($tagname)
Chris@0:     {
Chris@0:         return $this->current->tagName == $tagname;
Chris@0:     }
Chris@0: }