Added support for non-escaped quotes in attribute value

fixes paquettg#37

Author: paquettg

src/PHPHtmlParser/Content.php

@@ -34,23 +34,13 @@ class Content {
 	protected $slash = " />\r\n\t";
 	protected $attr  = ' >';
 
-	/**
-	 * Sets up the content block with its content.
-	 *
-	 * @param $content
-	 */
 	public function __construct($content)
 	{
 		$this->content = $content;
 		$this->size    = strlen($content);
 		$this->pos     = 0;
 	}
 
-	/**
-	 * Returns the current position of the parser.
-	 *
-	 * @return int
-	 */
 	public function getPosition()
 	{
 		return $this->pos;
@@ -163,19 +153,43 @@ public function copyUntil($string, $char = false, $escape = false)
 			$this->pos = $this->size;
 			return $return;
 		}
-		
+
 		if ($position == $this->pos)
 		{
 			// we are at the right place
 			return '';
 		}
-		
+
 		$return = substr($this->content, $this->pos, $position - $this->pos);
 		// set the new position
 		$this->pos = $position;
 		return $return;
 	}
 
+	/**
+	 * Copies the content until the string is found and return it
+	 * unless the 'unless' is found in the substring.
+	 *
+	 * @param string $string
+	 * @param string $unless
+	 * @return string
+	 */
+	public function copyUntilUnless($string, $unless)
+	{
+		$lastPos = $this->pos;
+		$this->fastForward(1);
+		$foundString = $this->copyUntil($string, true, true);
+
+		$position = strcspn($foundString, $unless);
+		if ($position == strlen($foundString))
+		{
+			return $string.$foundString;
+		}
+		// rewind changes and return nothing
+		$this->pos = $lastPos;
+		return '';
+	}
+
 	/**
 	 * Copies the content until it reaches the token string.,
 	 *
@@ -202,7 +216,7 @@ public function skip($string, $copy = false)
 	{
 		$len = strspn($this->content, $string, $this->pos);
 
-		// make it chain-able if they don't want a copy
+		// make it chainable if they don't want a copy
 		$return = $this;
 		if ($copy)
 		{

src/PHPHtmlParser/Dom.php

@@ -498,12 +498,12 @@ protected function parseTag()
 			$space = $this->content->skipByToken('blank', true);
 			if (empty($space))
 			{
-				break;
+				$this->content->fastForward(1);
+				continue;
 			}
 
 			$name = $this->content->copyByToken('equal', true);
-			if ($name == '/' OR
-			    empty($name))
+			if ($name == '/')
 			{
 				break;
 			}
@@ -525,14 +525,26 @@ protected function parseTag()
 					case '"':
 						$attr['doubleQuote'] = true;
 						$this->content->fastForward(1);
-						$attr['value'] = $this->content->copyUntil('"', false, true);
+						$string = $this->content->copyUntil('"', true, true);
+						do
+						{
+							$moreString = $this->content->copyUntilUnless('"', '=>');
+							$string .= $moreString;
+						} while( ! empty($moreString));
+						$attr['value'] = $string;
 						$this->content->fastForward(1);
 						$node->getTag()->$name = $attr;
 						break;
 					case "'":
 						$attr['doubleQuote'] = false;
 						$this->content->fastForward(1);
-						$attr['value'] = $this->content->copyUntil("'", false, true);
+						$string = $this->content->copyUntil("'", true, true);
+						do
+						{
+							$moreString = $this->content->copyUntilUnless("'", '=>');
+							$string .= $moreString;
+						} while( ! empty($moreString));
+						$attr['value'] = $string;
 						$this->content->fastForward(1);
 						$node->getTag()->$name = $attr;
 						break;

tests/DomTest.php

@@ -271,4 +271,18 @@ public function testScriptCleanerScriptTag()
 		<p>....</p>');
 		$this->assertEquals('....', $dom->getElementsByTag('p')[1]->innerHtml);
 	}
+
+	public function testMultipleDoubleQuotes()
+	{
+		$dom = new Dom;
+		$dom->load('<a title="This is a "test" of double quotes" href="http://www.example.com">Hello</a>');
+		$this->assertEquals('This is a "test" of double quotes', $dom->getElementsByTag('a')[0]->title);
+	}
+
+	public function testMultipleSingleQuotes()
+	{
+		$dom = new Dom;
+		$dom->load("<a title='Ain't this the best' href=\"http://www.example.com\">Hello</a>");
+		$this->assertEquals("Ain't this the best", $dom->getElementsByTag('a')[0]->title);
+	}
 }