public function testElement()
{
foreach ($this->html5Elements as $element) {
$this->assertGreaterThan(0, Elements::element($element));
}
$nonhtml5 = array('foo', 'bar', 'baz');
foreach ($nonhtml5 as $element) {
$this->assertFalse(Elements::element($element));
}
}
/**
* Process the start tag.
*
* @todo - XMLNS namespace handling (we need to parse, even if it's not valid)
* - XLink, MathML and SVG namespace handling
* - Omission rules: 8.1.2.4 Optional tags
*/
public function startTag($name, $attributes = array(), $selfClosing = false)
{
// fprintf(STDOUT, $name);
$lname = $this->normalizeTagName($name);
// Make sure we have an html element.
if (!$this->doc->documentElement && $name !== 'html' && !$this->frag) {
$this->startTag('html');
}
// Set quirks mode if we're at IM_INITIAL with no doctype.
if ($this->insertMode == static::IM_INITIAL) {
$this->quirks = true;
$this->parseError("No DOCTYPE specified.");
}
// SPECIAL TAG HANDLING:
// Spec says do this, and "don't ask."
if ($name == 'image') {
$name = 'img';
}
// Autoclose p tags where appropriate.
if ($this->insertMode >= static::IM_IN_BODY && Elements::isA($name, Elements::AUTOCLOSE_P)) {
$this->autoclose('p');
}
// Set insert mode:
switch ($name) {
case 'html':
$this->insertMode = static::IM_BEFORE_HEAD;
break;
case 'head':
if ($this->insertMode > static::IM_BEFORE_HEAD) {
$this->parseError("Unexpected head tag outside of head context.");
} else {
$this->insertMode = static::IM_IN_HEAD;
}
break;
case 'body':
$this->insertMode = static::IM_IN_BODY;
break;
case 'svg':
$this->insertMode = static::IM_IN_SVG;
break;
case 'math':
$this->insertMode = static::IM_IN_MATHML;
break;
case 'noscript':
if ($this->insertMode == static::IM_IN_HEAD) {
$this->insertMode = static::IM_IN_HEAD_NOSCRIPT;
}
break;
}
// Special case handling for SVG.
if ($this->insertMode == static::IM_IN_SVG) {
$lname = Elements::normalizeSvgElement($lname);
}
$pushes = 0;
// when we found a tag thats appears inside $nsRoots, we have to switch the defalut namespace
if (isset($this->nsRoots[$lname]) && $this->nsStack[0][''] !== $this->nsRoots[$lname]) {
array_unshift($this->nsStack, array('' => $this->nsRoots[$lname]) + $this->nsStack[0]);
$pushes++;
}
$needsWorkaround = false;
if (isset($this->options["xmlNamespaces"]) && $this->options["xmlNamespaces"]) {
// when xmlNamespaces is true a and we found a 'xmlns' or 'xmlns:*' attribute, we should add a new item to the $nsStack
foreach ($attributes as $aName => $aVal) {
if ($aName === 'xmlns') {
$needsWorkaround = $aVal;
array_unshift($this->nsStack, array('' => $aVal) + $this->nsStack[0]);
$pushes++;
} elseif ((($pos = strpos($aName, ':')) ? substr($aName, 0, $pos) : '') === 'xmlns') {
array_unshift($this->nsStack, array(substr($aName, $pos + 1) => $aVal) + $this->nsStack[0]);
$pushes++;
}
}
}
if ($this->onlyInline && Elements::isA($lname, Elements::BLOCK_TAG)) {
$this->autoclose($this->onlyInline);
$this->onlyInline = null;
}
try {
$prefix = ($pos = strpos($lname, ':')) ? substr($lname, 0, $pos) : '';
if ($needsWorkaround !== false) {
$xml = "<{$lname} xmlns=\"{$needsWorkaround}\" " . (strlen($prefix) && isset($this->nsStack[0][$prefix]) ? "xmlns:{$prefix}=\"" . $this->nsStack[0][$prefix] . "\"" : "") . "/>";
$frag = new \DOMDocument('1.0', 'UTF-8');
$frag->loadXML($xml);
$ele = $this->doc->importNode($frag->documentElement, true);
} else {
if (!isset($this->nsStack[0][$prefix]) || $prefix === "" && isset($this->options[self::OPT_DISABLE_HTML_NS]) && $this->options[self::OPT_DISABLE_HTML_NS]) {
$ele = $this->doc->createElement($lname);
} else {
$ele = $this->doc->createElementNS($this->nsStack[0][$prefix], $lname);
}
}
} catch (\DOMException $e) {
$this->parseError("Illegal tag name: <{$lname}>. Replaced with <invalid>.");
$ele = $this->doc->createElement('invalid');
}
if (Elements::isA($lname, Elements::BLOCK_ONLY_INLINE)) {
$this->onlyInline = $lname;
}
// When we add some namespacess, we have to track them. Later, when "endElement" is invoked, we have to remove them.
// When we are on a void tag, we do not need to care about namesapce nesting.
if ($pushes > 0 && !Elements::isA($name, Elements::VOID_TAG)) {
// PHP tends to free the memory used by DOM,
// to avoid spl_object_hash collisions whe have to avoid garbage collection of $ele storing it into $pushes
// see https://bugs.php.net/bug.php?id=67459
$this->pushes[spl_object_hash($ele)] = array($pushes, $ele);
// SEE https://github.com/facebook/hhvm/issues/2962
if (defined('HHVM_VERSION')) {
$ele->setAttribute('html5-php-fake-id-attribute', spl_object_hash($ele));
}
}
foreach ($attributes as $aName => $aVal) {
// xmlns attributes can't be set
if ($aName === 'xmlns') {
continue;
}
if ($this->insertMode == static::IM_IN_SVG) {
$aName = Elements::normalizeSvgAttribute($aName);
} elseif ($this->insertMode == static::IM_IN_MATHML) {
$aName = Elements::normalizeMathMlAttribute($aName);
}
try {
$prefix = ($pos = strpos($aName, ':')) ? substr($aName, 0, $pos) : false;
if ($prefix === 'xmlns') {
$ele->setAttributeNs(self::NAMESPACE_XMLNS, $aName, $aVal);
} elseif ($prefix !== false && isset($this->nsStack[0][$prefix])) {
$ele->setAttributeNs($this->nsStack[0][$prefix], $aName, $aVal);
} else {
$ele->setAttribute($aName, $aVal);
}
} catch (\DOMException $e) {
$this->parseError("Illegal attribute name for tag {$name}. Ignoring: {$aName}");
continue;
}
// This is necessary on a non-DTD schema, like HTML5.
if ($aName == 'id') {
$ele->setIdAttribute('id', true);
}
}
// Some elements have special processing rules. Handle those separately.
if ($this->rules->hasRules($name) && $this->frag !== $this->current) {
$this->current = $this->rules->evaluate($ele, $this->current);
} else {
$this->current->appendChild($ele);
// XXX: Need to handle self-closing tags and unary tags.
if (!Elements::isA($name, Elements::VOID_TAG)) {
$this->current = $ele;
}
}
// This is sort of a last-ditch attempt to correct for cases where no head/body
// elements are provided.
if ($this->insertMode <= static::IM_BEFORE_HEAD && $name != 'head' && $name != 'html') {
$this->insertMode = static::IM_IN_BODY;
}
// When we are on a void tag, we do not need to care about namesapce nesting,
// but we have to remove the namespaces pushed to $nsStack.
if ($pushes > 0 && Elements::isA($name, Elements::VOID_TAG)) {
// remove the namespaced definded by current node
for ($i = 0; $i < $pushes; $i++) {
array_shift($this->nsStack);
}
}
// Return the element mask, which the tokenizer can then use to set
// various processing rules.
return Elements::element($name);
}
