Chris@0: <?php
Chris@0: /**
Chris@0:  * @file
Chris@0:  * The rules for generating output in the serializer.
Chris@0:  *
Chris@0:  * These output rules are likely to generate output similar to the document that
Chris@0:  * was parsed. It is not intended to output exactly the document that was parsed.
Chris@0:  */
Chris@17: 
Chris@0: namespace Masterminds\HTML5\Serializer;
Chris@0: 
Chris@0: use Masterminds\HTML5\Elements;
Chris@0: 
Chris@0: /**
Chris@0:  * Generate the output html5 based on element rules.
Chris@0:  */
Chris@17: class OutputRules implements RulesInterface
Chris@0: {
Chris@0:     /**
Chris@17:      * 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:     /**
Chris@17:      * Holds the HTML5 element names that causes a namespace switch.
Chris@0:      *
Chris@0:      * @var array
Chris@0:      */
Chris@0:     protected $implicitNamespaces = array(
Chris@0:         self::NAMESPACE_HTML,
Chris@0:         self::NAMESPACE_SVG,
Chris@0:         self::NAMESPACE_MATHML,
Chris@0:         self::NAMESPACE_XML,
Chris@0:         self::NAMESPACE_XMLNS,
Chris@0:     );
Chris@0: 
Chris@0:     const IM_IN_HTML = 1;
Chris@0: 
Chris@0:     const IM_IN_SVG = 2;
Chris@0: 
Chris@0:     const IM_IN_MATHML = 3;
Chris@0: 
Chris@0:     /**
Chris@17:      * Used as cache to detect if is available ENT_HTML5.
Chris@17:      *
Chris@17:      * @var bool
Chris@0:      */
Chris@0:     private $hasHTML5 = false;
Chris@0: 
Chris@0:     protected $traverser;
Chris@0: 
Chris@0:     protected $encode = false;
Chris@0: 
Chris@0:     protected $out;
Chris@0: 
Chris@0:     protected $outputMode;
Chris@0: 
Chris@0:     private $xpath;
Chris@0: 
Chris@0:     protected $nonBooleanAttributes = array(
Chris@0:         /*
Chris@0:         array(
Chris@0:             'nodeNamespace'=>'http://www.w3.org/1999/xhtml',
Chris@0:             'attrNamespace'=>'http://www.w3.org/1999/xhtml',
Chris@0: 
Chris@0:             'nodeName'=>'img', 'nodeName'=>array('img', 'a'),
Chris@0:             'attrName'=>'alt', 'attrName'=>array('title', 'alt'),
Chris@0:         ),
Chris@0:         */
Chris@0:         array(
Chris@0:             'nodeNamespace' => 'http://www.w3.org/1999/xhtml',
Chris@0:             'attrName' => array('href',
Chris@0:                 'hreflang',
Chris@0:                 'http-equiv',
Chris@0:                 'icon',
Chris@0:                 'id',
Chris@0:                 'keytype',
Chris@0:                 'kind',
Chris@0:                 'label',
Chris@0:                 'lang',
Chris@0:                 'language',
Chris@0:                 'list',
Chris@0:                 'maxlength',
Chris@0:                 'media',
Chris@0:                 'method',
Chris@0:                 'name',
Chris@0:                 'placeholder',
Chris@0:                 'rel',
Chris@0:                 'rows',
Chris@0:                 'rowspan',
Chris@0:                 'sandbox',
Chris@0:                 'spellcheck',
Chris@0:                 'scope',
Chris@0:                 'seamless',
Chris@0:                 'shape',
Chris@0:                 'size',
Chris@0:                 'sizes',
Chris@0:                 'span',
Chris@0:                 'src',
Chris@0:                 'srcdoc',
Chris@0:                 'srclang',
Chris@0:                 'srcset',
Chris@0:                 'start',
Chris@0:                 'step',
Chris@0:                 'style',
Chris@0:                 'summary',
Chris@0:                 'tabindex',
Chris@0:                 'target',
Chris@0:                 'title',
Chris@0:                 'type',
Chris@0:                 'value',
Chris@0:                 'width',
Chris@0:                 'border',
Chris@0:                 'charset',
Chris@0:                 'cite',
Chris@0:                 'class',
Chris@0:                 'code',
Chris@0:                 'codebase',
Chris@0:                 'color',
Chris@0:                 'cols',
Chris@0:                 'colspan',
Chris@0:                 'content',
Chris@0:                 'coords',
Chris@0:                 'data',
Chris@0:                 'datetime',
Chris@0:                 'default',
Chris@0:                 'dir',
Chris@0:                 'dirname',
Chris@0:                 'enctype',
Chris@0:                 'for',
Chris@0:                 'form',
Chris@0:                 'formaction',
Chris@0:                 'headers',
Chris@0:                 'height',
Chris@0:                 'accept',
Chris@0:                 'accept-charset',
Chris@0:                 'accesskey',
Chris@0:                 'action',
Chris@0:                 'align',
Chris@0:                 'alt',
Chris@0:                 'bgcolor',
Chris@0:             ),
Chris@0:         ),
Chris@0:         array(
Chris@0:             'nodeNamespace' => 'http://www.w3.org/1999/xhtml',
Chris@0:             'xpath' => 'starts-with(local-name(), \'data-\')',
Chris@0:         ),
Chris@0:     );
Chris@0: 
Chris@0:     const DOCTYPE = '<!DOCTYPE html>';
Chris@0: 
Chris@0:     public function __construct($output, $options = array())
Chris@0:     {
Chris@0:         if (isset($options['encode_entities'])) {
Chris@0:             $this->encode = $options['encode_entities'];
Chris@0:         }
Chris@0: 
Chris@0:         $this->outputMode = static::IM_IN_HTML;
Chris@0:         $this->out = $output;
Chris@0: 
Chris@0:         // If HHVM, see https://github.com/facebook/hhvm/issues/2727
Chris@0:         $this->hasHTML5 = defined('ENT_HTML5') && !defined('HHVM_VERSION');
Chris@0:     }
Chris@17: 
Chris@0:     public function addRule(array $rule)
Chris@0:     {
Chris@0:         $this->nonBooleanAttributes[] = $rule;
Chris@0:     }
Chris@0: 
Chris@17:     public function setTraverser(Traverser $traverser)
Chris@0:     {
Chris@0:         $this->traverser = $traverser;
Chris@0: 
Chris@0:         return $this;
Chris@0:     }
Chris@0: 
Chris@0:     public function document($dom)
Chris@0:     {
Chris@0:         $this->doctype();
Chris@0:         if ($dom->documentElement) {
Chris@0:             foreach ($dom->childNodes as $node) {
Chris@0:                 $this->traverser->node($node);
Chris@0:             }
Chris@0:             $this->nl();
Chris@0:         }
Chris@0:     }
Chris@0: 
Chris@0:     protected function doctype()
Chris@0:     {
Chris@0:         $this->wr(static::DOCTYPE);
Chris@0:         $this->nl();
Chris@0:     }
Chris@0: 
Chris@0:     public function element($ele)
Chris@0:     {
Chris@0:         $name = $ele->tagName;
Chris@0: 
Chris@0:         // Per spec:
Chris@0:         // If the element has a declared namespace in the HTML, MathML or
Chris@0:         // SVG namespaces, we use the lname instead of the tagName.
Chris@0:         if ($this->traverser->isLocalElement($ele)) {
Chris@0:             $name = $ele->localName;
Chris@0:         }
Chris@0: 
Chris@0:         // If we are in SVG or MathML there is special handling.
Chris@0:         // Using if/elseif instead of switch because it's faster in PHP.
Chris@17:         if ('svg' == $name) {
Chris@0:             $this->outputMode = static::IM_IN_SVG;
Chris@0:             $name = Elements::normalizeSvgElement($name);
Chris@17:         } elseif ('math' == $name) {
Chris@0:             $this->outputMode = static::IM_IN_MATHML;
Chris@0:         }
Chris@0: 
Chris@0:         $this->openTag($ele);
Chris@0:         if (Elements::isA($name, Elements::TEXT_RAW)) {
Chris@0:             foreach ($ele->childNodes as $child) {
Chris@0:                 if ($child instanceof \DOMCharacterData) {
Chris@0:                     $this->wr($child->data);
Chris@0:                 } elseif ($child instanceof \DOMElement) {
Chris@0:                     $this->element($child);
Chris@0:                 }
Chris@0:             }
Chris@0:         } else {
Chris@0:             // Handle children.
Chris@0:             if ($ele->hasChildNodes()) {
Chris@0:                 $this->traverser->children($ele->childNodes);
Chris@0:             }
Chris@0: 
Chris@0:             // Close out the SVG or MathML special handling.
Chris@17:             if ('svg' == $name || 'math' == $name) {
Chris@0:                 $this->outputMode = static::IM_IN_HTML;
Chris@0:             }
Chris@0:         }
Chris@0: 
Chris@0:         // If not unary, add a closing tag.
Chris@17:         if (!Elements::isA($name, Elements::VOID_TAG)) {
Chris@0:             $this->closeTag($ele);
Chris@0:         }
Chris@0:     }
Chris@0: 
Chris@0:     /**
Chris@0:      * Write a text node.
Chris@0:      *
Chris@17:      * @param \DOMText $ele The text node to write.
Chris@0:      */
Chris@0:     public function text($ele)
Chris@0:     {
Chris@0:         if (isset($ele->parentNode) && isset($ele->parentNode->tagName) && Elements::isA($ele->parentNode->localName, Elements::TEXT_RAW)) {
Chris@0:             $this->wr($ele->data);
Chris@17: 
Chris@0:             return;
Chris@0:         }
Chris@0: 
Chris@0:         // FIXME: This probably needs some flags set.
Chris@0:         $this->wr($this->enc($ele->data));
Chris@0:     }
Chris@0: 
Chris@0:     public function cdata($ele)
Chris@0:     {
Chris@0:         // This encodes CDATA.
Chris@0:         $this->wr($ele->ownerDocument->saveXML($ele));
Chris@0:     }
Chris@0: 
Chris@0:     public function comment($ele)
Chris@0:     {
Chris@0:         // These produce identical output.
Chris@0:         // $this->wr('<!--')->wr($ele->data)->wr('-->');
Chris@0:         $this->wr($ele->ownerDocument->saveXML($ele));
Chris@0:     }
Chris@0: 
Chris@0:     public function processorInstruction($ele)
Chris@0:     {
Chris@0:         $this->wr('<?')
Chris@0:             ->wr($ele->target)
Chris@0:             ->wr(' ')
Chris@0:             ->wr($ele->data)
Chris@0:             ->wr('?>');
Chris@0:     }
Chris@17: 
Chris@0:     /**
Chris@17:      * Write the namespace attributes.
Chris@0:      *
Chris@17:      * @param \DOMNode $ele The element being written.
Chris@0:      */
Chris@0:     protected function namespaceAttrs($ele)
Chris@0:     {
Chris@17:         if (!$this->xpath || $this->xpath->document !== $ele->ownerDocument) {
Chris@0:             $this->xpath = new \DOMXPath($ele->ownerDocument);
Chris@0:         }
Chris@0: 
Chris@17:         foreach ($this->xpath->query('namespace::*[not(.=../../namespace::*)]', $ele) as $nsNode) {
Chris@0:             if (!in_array($nsNode->nodeValue, $this->implicitNamespaces)) {
Chris@0:                 $this->wr(' ')->wr($nsNode->nodeName)->wr('="')->wr($nsNode->nodeValue)->wr('"');
Chris@0:             }
Chris@0:         }
Chris@0:     }
Chris@0: 
Chris@0:     /**
Chris@0:      * Write the opening tag.
Chris@0:      *
Chris@0:      * Tags for HTML, MathML, and SVG are in the local name. Otherwise, use the
Chris@0:      * qualified name (8.3).
Chris@0:      *
Chris@17:      * @param \DOMNode $ele The element being written.
Chris@0:      */
Chris@0:     protected function openTag($ele)
Chris@0:     {
Chris@0:         $this->wr('<')->wr($this->traverser->isLocalElement($ele) ? $ele->localName : $ele->tagName);
Chris@0: 
Chris@0:         $this->attrs($ele);
Chris@0:         $this->namespaceAttrs($ele);
Chris@0: 
Chris@0:         if ($this->outputMode == static::IM_IN_HTML) {
Chris@0:             $this->wr('>');
Chris@0:         }         // If we are not in html mode we are in SVG, MathML, or XML embedded content.
Chris@0:         else {
Chris@0:             if ($ele->hasChildNodes()) {
Chris@0:                 $this->wr('>');
Chris@0:             }             // If there are no children this is self closing.
Chris@0:             else {
Chris@0:                 $this->wr(' />');
Chris@0:             }
Chris@0:         }
Chris@0:     }
Chris@0: 
Chris@0:     protected function attrs($ele)
Chris@0:     {
Chris@0:         // FIXME: Needs support for xml, xmlns, xlink, and namespaced elements.
Chris@17:         if (!$ele->hasAttributes()) {
Chris@0:             return $this;
Chris@0:         }
Chris@0: 
Chris@0:         // TODO: Currently, this always writes name="value", and does not do
Chris@0:         // value-less attributes.
Chris@0:         $map = $ele->attributes;
Chris@0:         $len = $map->length;
Chris@17:         for ($i = 0; $i < $len; ++$i) {
Chris@0:             $node = $map->item($i);
Chris@0:             $val = $this->enc($node->value, true);
Chris@0: 
Chris@0:             // XXX: The spec says that we need to ensure that anything in
Chris@0:             // the XML, XMLNS, or XLink NS's should use the canonical
Chris@0:             // prefix. It seems that DOM does this for us already, but there
Chris@0:             // may be exceptions.
Chris@0:             $name = $node->nodeName;
Chris@0: 
Chris@0:             // Special handling for attributes in SVG and MathML.
Chris@0:             // Using if/elseif instead of switch because it's faster in PHP.
Chris@0:             if ($this->outputMode == static::IM_IN_SVG) {
Chris@0:                 $name = Elements::normalizeSvgAttribute($name);
Chris@0:             } elseif ($this->outputMode == static::IM_IN_MATHML) {
Chris@0:                 $name = Elements::normalizeMathMlAttribute($name);
Chris@0:             }
Chris@0: 
Chris@0:             $this->wr(' ')->wr($name);
Chris@0: 
Chris@17:             if ((isset($val) && '' !== $val) || $this->nonBooleanAttribute($node)) {
Chris@0:                 $this->wr('="')->wr($val)->wr('"');
Chris@0:             }
Chris@0:         }
Chris@0:     }
Chris@0: 
Chris@0:     protected function nonBooleanAttribute(\DOMAttr $attr)
Chris@0:     {
Chris@0:         $ele = $attr->ownerElement;
Chris@17:         foreach ($this->nonBooleanAttributes as $rule) {
Chris@17:             if (isset($rule['nodeNamespace']) && $rule['nodeNamespace'] !== $ele->namespaceURI) {
Chris@0:                 continue;
Chris@0:             }
Chris@17:             if (isset($rule['attNamespace']) && $rule['attNamespace'] !== $attr->namespaceURI) {
Chris@0:                 continue;
Chris@0:             }
Chris@17:             if (isset($rule['nodeName']) && !is_array($rule['nodeName']) && $rule['nodeName'] !== $ele->localName) {
Chris@0:                 continue;
Chris@0:             }
Chris@17:             if (isset($rule['nodeName']) && is_array($rule['nodeName']) && !in_array($ele->localName, $rule['nodeName'], true)) {
Chris@0:                 continue;
Chris@0:             }
Chris@17:             if (isset($rule['attrName']) && !is_array($rule['attrName']) && $rule['attrName'] !== $attr->localName) {
Chris@0:                 continue;
Chris@0:             }
Chris@17:             if (isset($rule['attrName']) && is_array($rule['attrName']) && !in_array($attr->localName, $rule['attrName'], true)) {
Chris@0:                 continue;
Chris@0:             }
Chris@17:             if (isset($rule['xpath'])) {
Chris@0:                 $xp = $this->getXPath($attr);
Chris@17:                 if (isset($rule['prefixes'])) {
Chris@17:                     foreach ($rule['prefixes'] as $nsPrefix => $ns) {
Chris@0:                         $xp->registerNamespace($nsPrefix, $ns);
Chris@0:                     }
Chris@0:                 }
Chris@17:                 if (!$xp->evaluate($rule['xpath'], $attr)) {
Chris@0:                     continue;
Chris@0:                 }
Chris@0:             }
Chris@0: 
Chris@0:             return true;
Chris@0:         }
Chris@0: 
Chris@0:         return false;
Chris@0:     }
Chris@0: 
Chris@17:     private function getXPath(\DOMNode $node)
Chris@17:     {
Chris@17:         if (!$this->xpath) {
Chris@0:             $this->xpath = new \DOMXPath($node->ownerDocument);
Chris@0:         }
Chris@17: 
Chris@0:         return $this->xpath;
Chris@0:     }
Chris@0: 
Chris@0:     /**
Chris@0:      * Write the closing tag.
Chris@0:      *
Chris@0:      * Tags for HTML, MathML, and SVG are in the local name. Otherwise, use the
Chris@0:      * qualified name (8.3).
Chris@0:      *
Chris@17:      * @param \DOMNode $ele The element being written.
Chris@0:      */
Chris@0:     protected function closeTag($ele)
Chris@0:     {
Chris@0:         if ($this->outputMode == static::IM_IN_HTML || $ele->hasChildNodes()) {
Chris@0:             $this->wr('</')->wr($this->traverser->isLocalElement($ele) ? $ele->localName : $ele->tagName)->wr('>');
Chris@0:         }
Chris@0:     }
Chris@0: 
Chris@0:     /**
Chris@0:      * Write to the output.
Chris@0:      *
Chris@17:      * @param string $text The string to put into the output
Chris@0:      *
Chris@17:      * @return $this
Chris@0:      */
Chris@0:     protected function wr($text)
Chris@0:     {
Chris@0:         fwrite($this->out, $text);
Chris@17: 
Chris@0:         return $this;
Chris@0:     }
Chris@0: 
Chris@0:     /**
Chris@0:      * Write a new line character.
Chris@0:      *
Chris@17:      * @return $this
Chris@0:      */
Chris@0:     protected function nl()
Chris@0:     {
Chris@0:         fwrite($this->out, PHP_EOL);
Chris@17: 
Chris@0:         return $this;
Chris@0:     }
Chris@0: 
Chris@0:     /**
Chris@0:      * Encode text.
Chris@0:      *
Chris@0:      * When encode is set to false, the default value, the text passed in is
Chris@0:      * escaped per section 8.3 of the html5 spec. For details on how text is
Chris@0:      * escaped see the escape() method.
Chris@0:      *
Chris@0:      * When encoding is set to true the text is converted to named character
Chris@0:      * references where appropriate. Section 8.1.4 Character references of the
Chris@0:      * html5 spec refers to using named character references. This is useful for
Chris@0:      * characters that can't otherwise legally be used in the text.
Chris@0:      *
Chris@0:      * The named character references are listed in section 8.5.
Chris@0:      *
Chris@0:      * @see http://www.w3.org/TR/2013/CR-html5-20130806/syntax.html#named-character-references True encoding will turn all named character references into their entities.
Chris@0:      *      This includes such characters as +.# and many other common ones. By default
Chris@0:      *      encoding here will just escape &'<>".
Chris@0:      *
Chris@0:      *      Note, PHP 5.4+ has better html5 encoding.
Chris@0:      *
Chris@0:      * @todo Use the Entities class in php 5.3 to have html5 entities.
Chris@0:      *
Chris@17:      * @param string $text      Text to encode.
Chris@17:      * @param bool   $attribute True if we are encoding an attrubute, false otherwise.
Chris@0:      *
Chris@0:      * @return string The encoded text.
Chris@0:      */
Chris@0:     protected function enc($text, $attribute = false)
Chris@0:     {
Chris@0:         // Escape the text rather than convert to named character references.
Chris@17:         if (!$this->encode) {
Chris@0:             return $this->escape($text, $attribute);
Chris@0:         }
Chris@0: 
Chris@0:         // If we are in PHP 5.4+ we can use the native html5 entity functionality to
Chris@0:         // convert the named character references.
Chris@0: 
Chris@0:         if ($this->hasHTML5) {
Chris@0:             return htmlentities($text, ENT_HTML5 | ENT_SUBSTITUTE | ENT_QUOTES, 'UTF-8', false);
Chris@0:         }         // If a version earlier than 5.4 html5 entities are not entirely handled.
Chris@0:         // This manually handles them.
Chris@0:         else {
Chris@17:             return strtr($text, HTML5Entities::$map);
Chris@0:         }
Chris@0:     }
Chris@0: 
Chris@0:     /**
Chris@0:      * Escape test.
Chris@0:      *
Chris@0:      * According to the html5 spec section 8.3 Serializing HTML fragments, text
Chris@0:      * within tags that are not style, script, xmp, iframe, noembed, and noframes
Chris@0:      * need to be properly escaped.
Chris@0:      *
Chris@0:      * The & should be converted to &amp;, no breaking space unicode characters
Chris@0:      * converted to &nbsp;, when in attribute mode the " should be converted to
Chris@0:      * &quot;, and when not in attribute mode the < and > should be converted to
Chris@0:      * &lt; and &gt;.
Chris@0:      *
Chris@0:      * @see http://www.w3.org/TR/2013/CR-html5-20130806/syntax.html#escapingString
Chris@0:      *
Chris@17:      * @param string $text      Text to escape.
Chris@17:      * @param bool   $attribute True if we are escaping an attrubute, false otherwise.
Chris@0:      */
Chris@0:     protected function escape($text, $attribute = false)
Chris@0:     {
Chris@0:         // Not using htmlspecialchars because, while it does escaping, it doesn't
Chris@0:         // match the requirements of section 8.5. For example, it doesn't handle
Chris@0:         // non-breaking spaces.
Chris@0:         if ($attribute) {
Chris@0:             $replace = array(
Chris@0:                 '"' => '&quot;',
Chris@0:                 '&' => '&amp;',
Chris@17:                 "\xc2\xa0" => '&nbsp;',
Chris@0:             );
Chris@0:         } else {
Chris@0:             $replace = array(
Chris@0:                 '<' => '&lt;',
Chris@0:                 '>' => '&gt;',
Chris@0:                 '&' => '&amp;',
Chris@17:                 "\xc2\xa0" => '&nbsp;',
Chris@0:             );
Chris@0:         }
Chris@0: 
Chris@0:         return strtr($text, $replace);
Chris@0:     }
Chris@0: }