Changed encoding to be a local attribute instead of a static attribute

Author: paquettg

src/PHPHtmlParser/Dom.php

@@ -5,6 +5,7 @@
 use Guzzle\Http\Client;
 use PHPHtmlParser\Dom\HtmlNode;
 use PHPHtmlParser\Dom\TextNode;
+use stringEncode\Encode;
 
 class Dom {
 
@@ -13,14 +14,7 @@ class Dom {
 	 *
 	 * @var string
 	 */
-	public static $charset = 'UTF-8';
-
-	/**
-	 * The charset that we expect the html to be in.
-	 *
-	 * @var string
-	 */
-	public static $expectedCharset = 'UTF-8';
+	protected $defaultCharset = 'UTF-8';
 
 	/**
 	 * Contains the root node of this dom tree.
@@ -57,12 +51,6 @@ class Dom {
 	 */
 	protected $size;
 
-	public function __construct()
-	{
-		self::$expectedCharset = 'UTF-8';
-		self::$charset = 'UTF-8';
-	}
-
 	/**
 	 * Returns the inner html of the root node.
 	 *
@@ -460,25 +448,35 @@ protected function parseTag()
 	 */
 	protected function detectCharset()
 	{
+		// set the default
+		$encode = new Encode;
+		$encode->from($this->defaultCharset);
+		$encode->to($this->defaultCharset);
+
 		$meta = $this->root->find('meta[http-equiv=Content-Type]', 0);
 		if (is_null($meta))
 		{
 			// could not find meta tag
+			$this->root->propagateEncoding($encode);
 			return false;
 		}
 		$content = $meta->content;
 		if (empty($content))
 		{
 			// could not find content
+			$this->root->propagateEncoding($encode);
 			return false;
 		}
 		$matches = [];
 		if (preg_match('/charset=(.+)/', $content, $matches))
 		{
-			static::$expectedCharset = trim($matches[1]);
+			$encode->from(trim($matches[1]));
+			$this->root->propagateEncoding($encode);
+			return true;
 		}
 
 		// no charset found
+		$this->root->propagateEncoding($encode);
 		return false;
 	}
 }

src/PHPHtmlParser/Dom/Node.php

@@ -2,6 +2,7 @@
 namespace PHPHtmlParser\Dom;
 
 use PHPHtmlParser\Selector;
+use stringEncode\Encode;
 
 /**
  * Dom node object.
@@ -39,10 +40,17 @@ class Node {
 	/**
 	 * The unique id of the class. Given by PHP.
 	 *
-	 * @string
+	 * @var string
 	 */
    	protected $id;
 
+   	/**
+   	 * The encoding class used to encode strings.
+   	 *
+   	 * @var mixed
+   	 */
+   	protected $encode;
+
 	public function __construct()
 	{
 		$this->id = spl_object_hash($this);
@@ -145,6 +153,23 @@ public function setParent(Node $parent)
         return $this;
     }
 
+	/** 
+	 * Sets the encoding class to this node and propagates it
+	 * to all its children.
+	 *
+	 * @param Encode $encode
+	 */
+    public function propagateEncoding(Encode $encode)
+    {
+    	$this->encode = $encode;
+    	$this->tag->setEncoding($encode);
+    	// check children
+    	foreach ($this->children as $id => $child)
+    	{
+    		$child['node']->propagateEncoding($encode);
+    	}
+    }
+
 	/**
 	 * Checks if this node has children.
 	 *

src/PHPHtmlParser/Dom/Tag.php

@@ -32,6 +32,13 @@ class Tag {
 	 */
 	protected $noise = '';
 
+	/**
+	 * The encoding class to... encode the tags
+	 *
+	 * @var mixed
+	 */
+	protected $encode = null;
+
 	public function __construct($name)
 	{
 		$this->name = $name;
@@ -78,6 +85,11 @@ public function isSelfClosing()
 		return $this->selfclosing;
 	}
 
+	public function setEncoding(Encode $encode)
+	{
+		$this->encode = $encode;
+	}
+
 	/**
 	 * Sets the noise for this tag (if any)
 	 *
@@ -146,13 +158,10 @@ public function getAttribute($key)
 			return null;
 		}
 		$value = $this->attr[$key]['value'];
-		if (is_string($value))
+		if (is_string($value) AND ! is_null($this->encode))
 		{
 			// convert charset
-			$encode = new Encode;
-			$encode->from(Dom::$expectedCharset);
-			$encode->to(Dom::$charset);
-			$this->attr[$key]['value'] = $encode->convert($value);
+			$this->attr[$key]['value'] = $this->encode->convert($value);
 		}
 
 		return $this->attr[$key];

src/PHPHtmlParser/Dom/TextNode.php

@@ -2,7 +2,6 @@
 namespace PHPHtmlParser\Dom;
 
 use PHPHtmlParser\Dom;
-use stringEncode\Encode;
 
 class TextNode extends Node {
 
@@ -43,10 +42,14 @@ public function __construct($text)
 	public function text()
 	{
 		// convert charset
-		$encode = new Encode;
-		$encode->from(Dom::$expectedCharset);
-		$encode->to(Dom::$charset);
-		$text = $encode->convert($this->text);
+		if ( ! is_null($this->encode))
+		{
+			$text = $this->encode->convert($this->text);
+		}
+		else
+		{
+			$text = $this->text;
+		}
 
 		return $text;
 	}

tests/DomTest.php

@@ -43,6 +43,13 @@ public function testLoadNoClosingTag()
 		$this->assertEquals('<div class="all"><p>Hey bro, <a href="google.com" data-quote="\"">click here</a></p></div><br />', $root->outerHtml);
 	}
 
+	public function testLoadWithFile()
+	{
+		$dom = new Dom;
+		$dom->loadFromFile('tests/small.html');
+		$this->assertEquals('VonBurgermeister', $dom->find('.post-user font', 0)->text);
+	}
+
 	public function testLoadFromFile()
 	{
 		$dom = new Dom;