Seperated the Traverser from the Output generation.

The Traverser now simply walks through a document. The OutputRules convert the nodes into output html. The rules is a configurable options. By default OutputRules will generate html close to the html5 that was parsed. Alternate rule implementation (e.g., minify rules, pretty spacing rules) can be set as the default or on an individual case.
author: Matt Farina <[email protected]> 2013-05-27 17:38:02 -0400
committer: Matt Farina <[email protected]> 2013-05-27 17:38:02 -0400
commit: 0e7cd49d390160603563c17c76dea82eba1824b9 (patch)
tree: beae415abb00ddcaa025d5b93aff793c5dfd8b07 /src
parent: 3f7e489e0eab9b34d6c707d13726eaa195bef2eb (diff)
4 files changed, 269 insertions, 249 deletions
diff --git a/src/HTML5.php b/src/HTML5.php
index 8d95921..47c646c 100644
--- a/src/HTML5.php
+++ b/src/HTML5.php
@@ -18,6 +18,11 @@ use HTML5\Serializer\Serializer;
  */
 class HTML5 {
 
+  public static $options = array(
+    'encode' => FALSE,
+    'rules' => '\HTML5\Serializer\OutputRules',
+  );
+
   /**
    * Load and parse an HTML file.
    *
@@ -86,14 +91,17 @@ class HTML5 {
    * @param string $file
    *   The filename to be written.
    * @param array $options
-   *   Configuration options when serialing the DOM. These include:
-   *   - format: a bool value to specify if formatting (e.g. add indentation)
-   *     should be used on the output. Defaults to TRUE.
+   *   Configuration options when serializing the DOM. These include:
+   *   - rules: The class with the serializer writing rules. Defaults to
+   *     \HTML5\Serializer\OutputRules. The standard rules are representative of the
+   *     original document. This can be replaced by alternatives that can
+   *     minify or make other alterations.
    *   - encode: Text written to the output is escaped by default and not all
    *     entities are encoded. If this is set to TRUE all entities will be encoded.
    *     Defaults to FALSE.
    */
   public static function save($dom, $file, $options = array()) {
+    $options = $options + self::options();
     $serializer = new \HTML5\Serializer\Serializer($dom, $options);
     return $serializer->save($file);
   }
@@ -104,9 +112,11 @@ class HTML5 {
    * @param mixed $dom
    *   The DOM to be serialized.
    * @param array $options
-   *   Configuration options when serialing the DOM. These include:
-   *   - format: a bool value to specify if formatting (e.g. add indentation)
-   *     should be used on the output. Defaults to TRUE.
+   *   Configuration options when serializing the DOM. These include:
+   *   - rules: The class with the serializer writing rules. Defaults to
+   *     \HTML5\Serializer\OutputRules. The standard rules are representative of the
+   *     original document. This can be replaced by alternatives that can
+   *     minify or make other alterations.
    *   - encode: Text written to the output is escaped by default and not all
    *     entities are encoded. If this is set to TRUE all entities will be encoded.
    *     Defaults to FALSE.
@@ -115,6 +125,7 @@ class HTML5 {
    *   A HTML5 documented generated from the DOM.
    */
   public static function saveHTML($dom, $options = array()) {
+    $options = $options + self::options();
     $serializer = new \HTML5\Serializer\Serializer($dom, $options);
     return $serializer->saveHTML();
   }
@@ -135,4 +146,26 @@ class HTML5 {
     return $events->document();
   }
 
+  /**
+   * Get the default options.
+   *
+   * @return array
+   *   The default options.
+   */
+  public static function options() {
+    return self::$options;
+  }
+
+  /**
+   * Set a default option.
+   *
+   * @param string $name
+   *   The option name.
+   * @param mixed $value
+   *   The option value.
+   */
+  public static function setOption($name, $value) {
+    self::$options[$name] = $value;
+  }
+
 }
diff --git a/src/HTML5/Serializer/OutputRules.php b/src/HTML5/Serializer/OutputRules.php
new file mode 100644
index 0000000..5780d61
--- /dev/null
+++ b/src/HTML5/Serializer/OutputRules.php
@@ -0,0 +1,203 @@
+<?php
+namespace HTML5\Serializer;
+
+use \HTML5\Elements;
+
+class OutputRules {
+
+  protected $traverser;
+  protected $encode = FALSE;
+  protected $out;
+
+  const DOCTYPE = '<!DOCTYPE html>';
+
+  public function __construct($traverser, $output, $options = array()) {
+    $this->traverser = $traverser;
+
+    if (isset($options['encode'])) {
+      $this->encode = $options['encode'];
+    }
+
+    $this->out = $output;
+
+  }
+
+  public function document($dom) {
+    $this->doctype();
+    $this->traverser->node($dom->documentElement);
+    $this->nl();
+  }
+
+  protected function doctype() {
+    $this->wr(self::DOCTYPE);
+    $this->nl();
+  }
+
+  public function element($ele) {
+    $name = $ele->tagName;
+
+    // Per spec:
+    // If the element has a declared namespace in the HTML, MathML or
+    // SVG namespaces, we use the lname instead of the tagName.
+    if ($this->traverser->isLocalElement($ele)) {
+      $name = $ele->localName;
+    }
+
+    $this->openTag($ele);
+
+    // Handle children.
+    if ($ele->hasChildNodes()) {
+      $this->traverser->children($ele->childNodes);
+    }
+
+    // If not unary, add a closing tag.
+    if (!Elements::isA($name, Elements::VOID_TAG)) {
+      $this->closeTag($ele);
+    }
+  }
+
+  /**
+   * Write a text node.
+   *
+   * @param \DOMText $ele 
+   *   The text node to write.
+   */
+  public function text($ele) {
+    if (isset($ele->parentNode) && Elements::isA($ele->parentNode->tagName, Elements::TEXT_RAW)) {
+      $this->wr($ele->wholeText);
+      return;
+    }
+
+    // FIXME: This probably needs some flags set.
+    $this->wr($this->enc($ele->wholeText));
+
+  }
+
+  public function cdata($ele) {
+    $this->wr('<![CDATA[')->wr($ele->wholeText)->wr(']]>');
+  }
+
+  public function comment($ele) {
+    $this->wr('<!--')->wr($ele->data)->wr('-->');
+  }
+
+  public function processorInstruction($ele) {
+    $this->wr('<?')->wr($ele->target)->wr(' ')->wr($ele->data)->wr(' ?>');
+  }
+
+  /**
+   * Write the opening tag.
+   *
+   * Tags for HTML, MathML, and SVG are in the local name. Otherwise, use the
+   * qualified name (8.3).
+   * 
+   * @param \DOMNode $ele
+   *   The element being written.
+   */
+  protected function openTag($ele) {
+    // FIXME: Needs support for SVG, MathML, and namespaced XML.
+    $this->wr('<')->wr($ele->tagName);
+    $this->attrs($ele);
+    $this->wr('>');
+  }
+
+  protected function attrs($ele) {
+    // FIXME: Needs support for xml, xmlns, xlink, and namespaced elements.
+    if (!$ele->hasAttributes()) {
+      return $this;
+    }
+
+    // TODO: Currently, this always writes name="value", and does not do
+    // value-less attributes.
+    $map = $ele->attributes;
+    $len = $map->length;
+    for ($i = 0; $i < $len; ++$i) {
+      $node = $map->item($i);
+      $val = $this->enc($node->value);
+
+      // XXX: The spec says that we need to ensure that anything in
+      // the XML, XMLNS, or XLink NS's should use the canonical
+      // prefix. It seems that DOM does this for us already, but there
+      // may be exceptions.
+      $this->wr(' ')->wr($node->name)->wr('="')->wr($val)->wr('"');
+    }
+  }
+
+  /**
+   * Write the closing tag.
+   * 
+   * Tags for HTML, MathML, and SVG are in the local name. Otherwise, use the
+   * qualified name (8.3).
+   *
+   * @param \DOMNode $ele
+   *   The element being written.
+   */
+  protected function closeTag($ele) {
+    // FIXME: Needs support for SVG, MathML, and namespaced XML.
+    $this->wr('</')->wr($ele->tagName)->wr('>');
+  }
+
+  /**
+   * Write to the output.
+   *
+   * @param string $text
+   *   The string to put into the output.
+   *
+   * @return HTML5\Serializer\Traverser
+   *   $this so it can be used in chaining.
+   */
+  protected function wr($text) {
+    fwrite($this->out, $text);
+    return $this;
+  }
+
+  /**
+   * Write a new line character.
+   *
+   * @return HTML5\Serializer\Traverser
+   *   $this so it can be used in chaining.
+   */
+  protected function nl() {
+    fwrite($this->out, PHP_EOL);
+    return $this;
+  }
+
+  /**
+   * Encode text.
+   *
+   * True encoding will turn all named character references into their entities.
+   * This includes such characters as +.# and many other common ones. By default
+   * encoding here will just escape &'<>".
+   *
+   * Note, PHP 5.4+ has better html5 encoding.
+   *
+   * @todo Use the Entities class in php 5.3 to have html5 entities.
+   *
+   * @param string $text
+   *   text to encode.
+   *
+   * @return string
+   *   The encoded text.
+   */
+  protected function enc($text) {
+    $flags = ENT_QUOTES;
+
+    // Escape rather than encode all entities.
+    if (!$this->encode) {
+      return htmlspecialchars($text, $flags, 'UTF-8');
+    }
+
+    // If we are in PHP 5.4+ we can use the native html5 entity functionality.
+    if (defined('ENT_HTML5')) {
+      $flags = ENT_HTML5 | ENT_SUBSTITUTE | ENT_QUOTES;
+      $ret = htmlentities($text, $flags, 'UTF-8', FALSE);
+    }
+    // If a version earlier than 5.4 html5 entities are not entirely handled.
+    // This manually handles them.
+    else {
+      $ret = strtr($text, \HTML5\Serializer\HTML5Entities::$map);
+    }
+    return $ret;
+  }
+
+}
+\ No newline at end of file
diff --git a/src/HTML5/Serializer/Serializer.php b/src/HTML5/Serializer/Serializer.php
index 3f9187f..7053df5 100644
--- a/src/HTML5/Serializer/Serializer.php
+++ b/src/HTML5/Serializer/Serializer.php
@@ -15,8 +15,7 @@ namespace HTML5\Serializer;
  */
 class Serializer {
   protected $dom;
-  protected $pretty = TRUE;
-  protected $encode = FALSE;
+  protected $options = array();
 
   /**
    * Create a serializer.
@@ -36,14 +35,7 @@ class Serializer {
    */
   public function __construct($dom, $options = array()) {
     $this->dom = $dom;
-    
-    if (isset($options['format']) && is_bool($options['format'])) {
-      $this->pretty = $options['format'];
-    }
-
-    if (isset($options['encode']) && is_bool($options['encode'])) {
-      $this->encode = $options['encode'];
-    }
+    $this->options = $options;
   }
 
   /**
@@ -63,9 +55,7 @@ class Serializer {
     else {
       $file = fopen($filename, 'w');
     }
-    $trav = new Traverser($this->dom, $file);
-    $trav->formatOutput($this->pretty);
-    $trav->encodeOutput($this->encode);
+    $trav = new Traverser($this->dom, $file, $this->options);
 
     $trav->walk();
 
diff --git a/src/HTML5/Serializer/Traverser.php b/src/HTML5/Serializer/Traverser.php
index 68dea82..bd9d1ce 100644
--- a/src/HTML5/Serializer/Traverser.php
+++ b/src/HTML5/Serializer/Traverser.php
@@ -1,8 +1,6 @@
 <?php
 namespace HTML5\Serializer;
 
-use \HTML5\Elements;
-
 /**
  * Traverser for walking a DOM tree.
  *
@@ -22,11 +20,10 @@ class Traverser {
   );
 
   protected $dom;
-  protected $out;
-  protected $pretty = TRUE;
+  protected $options;
   protected $encode = FALSE;
-
-  const DOCTYPE = '<!DOCTYPE html>';
+  protected $rules;
+  protected $out;
 
   /**
    * Create a traverser.
@@ -36,44 +33,19 @@ class Traverser {
    * @param resource $out
    *   A stream that allows writing. The traverser will output into this 
    *   stream.
+   * @param array $options
+   *   An array or options for the traverser as key/value pairs. These include:
+   *   - encode: A bool to specify if full encding should happen for all named
+   *     charachter references. Defaults to FALSE which escapes &'<>".
+   *   - rules: The path to the class handling the output rules.
    */
-  public function __construct($dom, $out) {
+  public function __construct($dom, $out, $options = array()) {
     $this->dom = $dom;
     $this->out = $out;
-  }
+    $this->options = $options;
 
-  /**
-   * Determine whether output should be formatted.
-   *
-   * IMPORTANT: Neither option will GUARANTEE that the spacing of the output
-   * will exactly match the spacing of an origin document. The HTML5 specification
-   * does not require any such behavior.
-   *
-   * Semantically (according to the HTML5 spec's definition), either flag
-   * will produce an identical document. (Insignificant 
-   * whitespace does not impact semantics).
-   *
-   * @param boolean $useFormatting
-   *   If TRUE (default) output will be formatted. If FALSE,
-   *   the little or no formatting is done.
-   */
-  public function formatOutput($useFormatting = TRUE) {
-    $this->pretty = $useFormatting;
-  }
-
-  /**
-   * Set whether encoding should encode all html5 entities.
-   *
-   * True encoding will turn all named character references into their entities.
-   * This includes such characters as +.# and many other common ones. By default
-   * encoding here will just escape &'<>". which is what most users expect.
-   * 
-   * @param  bool $encode
-   *   Whether to encode all html5 entities. Defaults to FALSE where only
-   *   &'<>". are escaped.
-   */
-  public function encodeOutput($encode = FALSE) {
-    $this->encode = $encode;
+    $rulesClass = $this->options['rules'];
+    $this->rules = new $rulesClass($this, $out, $this->options);
   }
 
   /**
@@ -85,8 +57,7 @@ class Traverser {
   public function walk() {
     // If DOMDocument, start with the DOCTYPE and travers.
     if ($this->dom instanceof \DOMDocument) {
-      $this->doctype();
-      $this->document($this->dom);
+      $this->rules->document($this->dom);
     }
     // If NodeList, loop
     elseif ($this->dom instanceof \DOMNodeList) {
@@ -100,40 +71,30 @@ class Traverser {
     return $this->out;
   }
 
-  protected function doctype() {
-    $this->wr(self::DOCTYPE);
-    $this->nl();
-  }
-
-  protected function document($node) {
-    $this->node($node->documentElement);
-    $this->nl();
-  }
-
   /**
    * Process a node in the DOM.
    *
    * @param mixed $node
    *   A node implementing \DOMNode.
    */
-  protected function node($node) {
+  public function node($node) {
     // A listing of types is at http://php.net/manual/en/dom.constants.php
     switch ($node->nodeType) {
       case XML_ELEMENT_NODE:
-        $this->element($node);
+        $this->rules->element($node);
         break;
       case XML_TEXT_NODE:
-        $this->text($node);
+        $this->rules->text($node);
         break;
       case XML_CDATA_SECTION_NODE:
-        $this->cdata($node);
+        $this->rules->cdata($node);
         break;
       // FIXME: It appears that the parser doesn't do PI's.
       case XML_PI_NODE:
-        $this->processorInstruction($ele);
+        $this->rules->processorInstruction($ele);
         break;
       case XML_COMMENT_NODE:
-        $this->comment($node);
+        $this->rules->comment($node);
         break;
       // Currently we don't support embedding DTDs.
       default:
@@ -142,186 +103,19 @@ class Traverser {
     }
   }
 
-  protected function element($ele) {
-    $name = $ele->tagName;
-    $block = $this->pretty && Elements::isA($name, Elements::BLOCK_TAG);
-
-    // Per spec:
-    // If the element has a declared namespace in the HTML, MathML or
-    // SVG namespaces, we use the lname instead of the tagName.
-    if ($this->isLocalElement($ele)) {
-      $name = $ele->localName;
-    }
-
-    // TODO: Really need to fix the spacing.
-    // Add a newline for a block element.
-    if ($block) $this->nl();
-
-    $this->openTag($ele);
-
-    // Handle children.
-    if ($ele->hasChildNodes()) {
-      $this->children($ele->childNodes);
-    }
-
-    // If not unary, add a closing tag.
-    if (!Elements::isA($name, Elements::VOID_TAG)) {
-      $this->closeTag($ele);
-      if ($block) $this->nl();
-    }
-  }
-
   /**
-   * Write a text node.
+   * Walk through all the nodes on a node list.
    *
-   * @param \DOMText $ele 
-   *   The text node to write.
+   * @param \DOMNodeList $nl
+   *   A list of child elements to walk through.
    */
-  protected function text($ele) {
-    if (isset($ele->parentNode) && Elements::isA($ele->parentNode->tagName, Elements::TEXT_RAW)) {
-      $this->wr($ele->wholeText);
-      return;
-    }
-
-    // FIXME: This probably needs some flags set.
-    $this->wr($this->enc($ele->wholeText));
-
-  }
-
-  protected function cdata($ele) {
-    $this->wr('<![CDATA[')->wr($ele->wholeText)->wr(']]>');
-  }
-
-  protected function comment($ele) {
-    $this->wr('<!--')->wr($ele->data)->wr('-->');
-  }
-
-  protected function processorInstruction($ele) {
-    $this->wr('<?')->wr($ele->target)->wr(' ')->wr($ele->data)->wr(' ?>');
-  }
-
-  protected function children($nl) {
+  public function children($nl) {
     foreach ($nl as $node) {
       $this->node($node);
     }
   }
 
   /**
-   * Write the opening tag.
-   *
-   * Tags for HTML, MathML, and SVG are in the local name. Otherwise, use the
-   * qualified name (8.3).
-   * 
-   * @param \DOMNode $ele
-   *   The element being written.
-   */
-  protected function openTag($ele) {
-    // FIXME: Needs support for SVG, MathML, and namespaced XML.
-    $this->wr('<')->wr($ele->tagName);
-    $this->attrs($ele);
-    $this->wr('>');
-  }
-
-  protected function attrs($ele) {
-    // FIXME: Needs support for xml, xmlns, xlink, and namespaced elements.
-    if (!$ele->hasAttributes()) {
-      return $this;
-    }
-
-    // TODO: Currently, this always writes name="value", and does not do
-    // value-less attributes.
-    $map = $ele->attributes;
-    $len = $map->length;
-    for ($i = 0; $i < $len; ++$i) {
-      $node = $map->item($i);
-      $val = $this->enc($node->value);
-
-      // XXX: The spec says that we need to ensure that anything in
-      // the XML, XMLNS, or XLink NS's should use the canonical
-      // prefix. It seems that DOM does this for us already, but there
-      // may be exceptions.
-      $this->wr(' ')->wr($node->name)->wr('="')->wr($val)->wr('"');
-    }
-  }
-
-  /**
-   * Write the closing tag.
-   * 
-   * Tags for HTML, MathML, and SVG are in the local name. Otherwise, use the
-   * qualified name (8.3).
-   *
-   * @param \DOMNode $ele
-   *   The element being written.
-   */
-  protected function closeTag($ele) {
-    // FIXME: Needs support for SVG, MathML, and namespaced XML.
-    $this->wr('</')->wr($ele->tagName)->wr('>');
-  }
-
-  /**
-   * Write to the output.
-   *
-   * @param string $text
-   *   The string to put into the output.
-   *
-   * @return HTML5\Serializer\Traverser
-   *   $this so it can be used in chaining.
-   */
-  protected function wr($text) {
-    fwrite($this->out, $text);
-    return $this;
-  }
-
-  /**
-   * Write a new line character.
-   *
-   * @return HTML5\Serializer\Traverser
-   *   $this so it can be used in chaining.
-   */
-  protected function nl() {
-    fwrite($this->out, PHP_EOL);
-    return $this;
-  }
-
-  /**
-   * Encode text.
-   *
-   * True encoding will turn all named character references into their entities.
-   * This includes such characters as +.# and many other common ones. By default
-   * encoding here will just escape &'<>".
-   *
-   * Note, PHP 5.4+ has better html5 encoding.
-   *
-   * @todo Use the Entities class in php 5.3 to have html5 entities.
-   *
-   * @param string $text
-   *   text to encode.
-   *
-   * @return string
-   *   The encoded text.
-   */
-  protected function enc($text) {
-    $flags = ENT_QUOTES;
-
-    // Escape rather than encode all entities.
-    if (!$this->encode) {
-      return htmlspecialchars($text, $flags, 'UTF-8');
-    }
-
-    // If we are in PHP 5.4+ we can use the native html5 entity functionality.
-    if (defined('ENT_HTML5')) {
-      $flags = ENT_HTML5 | ENT_SUBSTITUTE | ENT_QUOTES;
-      $ret = htmlentities($text, $flags, 'UTF-8', FALSE);
-    }
-    // If a version earlier than 5.4 html5 entities are not entirely handled.
-    // This manually handles them.
-    else {
-      $ret = strtr($text, \HTML5\Serializer\HTML5Entities::$map);
-    }
-    return $ret;
-  }
-
-  /**
    * Is an element local?
    *
    * @param mixed $ele
@@ -330,7 +124,7 @@ class Traverser {
    * @return bool
    *   True if local and false otherwise.
    */
-  protected function isLocalElement($ele) {
+  public function isLocalElement($ele) {
     $uri = $ele->namespaceURI;
     if (empty($uri)) {
       return FALSE;
author	Matt Farina <[email protected]>	2013-05-27 17:38:02 -0400
committer	Matt Farina <[email protected]>	2013-05-27 17:38:02 -0400
commit	0e7cd49d390160603563c17c76dea82eba1824b9 (patch)
tree	beae415abb00ddcaa025d5b93aff793c5dfd8b07 /src
parent	3f7e489e0eab9b34d6c707d13726eaa195bef2eb (diff)