simplify regex for inline link in attempt to resolve #23

Parsedown.php

@@ -75,22 +75,10 @@ class Parsedown
 			}
 		}
 		
-		# Extracts link references.
-		
-		if (preg_match_all('/^[ ]{0,3}\[(.+)\][ ]?:[ ]*\n?[ ]*(.+)$/m', $text, $matches, PREG_SET_ORDER))
-		{
-			foreach ($matches as $matches)
-			{
-				$this->reference_map[strtolower($matches[1])] = $matches[2];
-				
-				$text = str_replace($matches[0], '', $text);
-			}
-		}
-		
 		# ~ 
 		
 		$text = preg_replace('/\n\s*\n/', "\n\n", $text);
-		$text = trim($text, "\n");
+		$text = trim($text, "\n ");
 		
 		$lines = explode("\n", $text);
 		
@@ -217,7 +205,7 @@ class Parsedown
 			
 			# Quick Paragraph 
 			
-			if ($line[0] >= 'A' and $line[0] !== '_')
+			if ($line[0] >= 'a' or $line[0] >= 'A' and $line[0] <= 'Z')
 			{
 				goto paragraph; # trust me 
 			}
@@ -228,7 +216,12 @@ class Parsedown
 			{
 				if ($element['type'] === 'code')
 				{
-					isset($element['interrupted']) and $element['text'] .= "\n";
+					if (isset($element['interrupted']))
+					{
+						$element['text'] .= "\n";
+						
+						unset ($element['interrupted']);
+					}
 					
 					$element['text'] .= "\n".$matches[1];
 				}
@@ -282,9 +275,21 @@ class Parsedown
 				continue;
 			}
 			
-			# ~  
+			# ~ 
 			
-			$pure_line = ltrim($line);
+			$pure_line = $line[0] !== ' ' ? $line : ltrim($line);
+			
+			# Link Reference 
+			
+			if ($pure_line[0] === '[' and preg_match('/^\[(.+?)\]:[ ]*([^ ]+)/', $pure_line, $matches))
+			{
+				$label = $matches[1];
+				$url = trim($matches[2], '<>');
+				
+				$this->reference_map[$label] = $url;
+				
+				continue;
+			}
 			
 			# Blockquote 
 			
@@ -555,11 +560,13 @@ class Parsedown
 		
 		# Inline Link / Image 
 		
-		if (strpos($text, '](') !== FALSE and preg_match_all('/(!?)(\[((?:[^][]+|(?2))*)\])\((.*?)\)/', $text, $matches, PREG_SET_ORDER)) # inline 
+		if (strpos($text, '](') !== FALSE and preg_match_all('/(!?)(\[((?:[^\[\]]|(?2))*)\])\((.*?)\)/', $text, $matches, PREG_SET_ORDER)) # inline 
 		{
 			foreach ($matches as $matches)
 			{
-				$url = $this->escape_special_characters($matches[4]);
+				$url = $matches[4];
+				
+				strpos($url, '&') !== FALSE and $url = preg_replace('/&(?!#?\w+;)/', '&amp;', $url);
 				
 				if ($matches[1]) # image 
 				{
@@ -599,7 +606,8 @@ class Parsedown
 				if (isset($this->reference_map[$link_definition]))
 				{
 					$url = $this->reference_map[$link_definition];
-					$url = $this->escape_special_characters($url);
+					
+					strpos($url, '&') !== FALSE and $url = preg_replace('/&(?!#?\w+;)/', '&amp;', $url);
 					
 					if ($matches[1]) # image 
 					{
@@ -631,7 +639,9 @@ class Parsedown
 		{
 			foreach ($matches as $matches)
 			{
-				$url = $this->escape_special_characters($matches[1]);
+				$url = $matches[1];
+				
+				strpos($url, '&') !== FALSE and $url = preg_replace('/&(?!#?\w+;)/', '&amp;', $url);
 				
 				$element = '<a href=":href">:text</a>';
 				$element = str_replace(':text', $url, $element);
@@ -651,8 +661,9 @@ class Parsedown
 		
 		# ~ 
 		
-		$text = $this->escape_special_characters($text);
-
+		strpos($text, '&') !== FALSE and $text = preg_replace('/&(?!#?\w+;)/', '&amp;', $text);
+		strpos($text, '<') !== FALSE and $text = preg_replace('/<(?!\/?\w.*?>)/', '&lt;', $text);
+		
 		# ~ 
 		
 		if (strpos($text, '_') !== FALSE)
@@ -671,13 +682,4 @@ class Parsedown
 		
 		return $text;
 	}
-	
-	private function escape_special_characters($text)
-	{
-		strpos($text, '&') !== FALSE and $text = preg_replace('/&(?!#?\w+;)/', '&amp;', $text);
-		
-		$text = str_replace('<', '&lt;', $text);
-		
-		return $text;
-	}
 }

tests/data/code_block.html

@@ -1,9 +1,8 @@
-<p>Here's a regular code block:</p>
-<pre><code>&lt;?php
-
-echo 'Hello World!'; 
-
-?&gt;</code></pre>
-<p>Here's one that holds a list:</p>
-<pre><code>- list item
+<p>Here's a code block:</p>
+<pre><code>&lt;?php
+
+$message = 'Hello World!'; 
+echo $message;</code></pre>
+<p>Here's one that holds a list:</p>
+<pre><code>- list item
 - another list item</code></pre>

tests/data/code_block.md

@@ -1,13 +1,12 @@
-Here's a regular code block:
-
-    <?php
-    
-    echo 'Hello World!'; 
-	
-	?>
-
-Here's one that holds a list:
-
-    - list item
-    - another list item
-
+Here's a code block:
+
+    <?php
+    
+    $message = 'Hello World!'; 
+    echo $message;
+
+Here's one that holds a list:
+
+    - list item
+    - another list item
+

tests/data/reference_link.html

@@ -1,14 +1,11 @@
 <p>Here's a <a href="http://parsedown.org">reference link</a>.</p>
-<p>Here's <a href="http://parsedown.org">one</a> with an alternative syntax.</p>
 <p>Here's <a href="http://parsedown.org">one</a> on the next line.</p>
-<p>Here's <a href="http://parsedown.org">one</a> on 2 lines.</p>
 <p>Here's <a href="http://parsedown.org/tests/">one</a> with a different URL.</p>
 <p>Here's <a href="http://parsedown.org">one</a> with a semantic name.</p>
 <p>Here's <a href="http://parsedown.org">one</a> with definition name on the next line.</p>
 <p>Here's [one][404] with no definition.</p>
 <p>Here's an image: <img alt="Markdown Logo" src="/md.png"></p>
 <p>Here's an <a href="http://google.com">implicit one</a>.</p>
-<p>Here's an <a href="http://google.com">implicit one</a>.</p>
 <p>Here's an <a href="http://google.com">implicit one</a> with an empty link definition.</p>
 <p>Here's a <a href="http://parsedown.org">multiline
 one</a> defined on 2 lines.</p>

tests/data/reference_link.md

@@ -2,21 +2,12 @@ Here's a [reference link][1].
 
 [1]: http://parsedown.org
 
-Here's [one] [2] with an alternative syntax.
+Here's [one][2] on the next line.
+[2]: http://parsedown.org
 
-[2] :http://parsedown.org
+Here's [one][3] with a different URL.
 
-Here's [one][3] on the next line.
-[3]: http://parsedown.org
-
-Here's [one][4] on 2 lines.
-
-[4]:
-http://parsedown.org
-
-Here's [one][5] with a different URL.
-
-[5]: http://parsedown.org/tests/
+[3]: http://parsedown.org/tests/
 
 Here's [one][website] with a semantic name.
 
@@ -33,8 +24,6 @@ Here's an image: ![Markdown Logo][image]
 
 Here's an [implicit one].
 
-Here's an [implicit one].
-
 [implicit one]: http://google.com
 
 Here's an [implicit one][] with an empty link definition.

tests/data/span_level_html.html

@@ -0,0 +1 @@
+<p>Here's an <b>important</b> <a href=''>link</a>.</p>

tests/data/span_level_html.md

@@ -0,0 +1 @@
+Here's an <b>important</b> <a href=''>link</a>.

tests/data/special_characters.html

@@ -4,5 +4,6 @@
 <p>4 &lt; 5 and 6 > 5.</p>
 <p>Here's a <a href="http://example.com/?foo=1&amp;bar=2">link</a> with an ampersand in the URL.</p>
 <p>Here's an inline <a href="/script?foo=1&amp;bar=2">link</a>.</p>
+<p><a href="http://example.com/autolink?a=1&amp;b=2">http://example.com/autolink?a=1&amp;b=2</a></p>
 <hr />
 <p>Based on <a href="http://daringfireball.net/projects/downloads/MarkdownTest_1.0.zip">the original</a> test suite.</p>

tests/data/special_characters.md

@@ -1,6 +1,6 @@
 AT&T has an ampersand in their name.
 
-AT&T is another way to write it.
+AT&T is another way to write it.
 
 This & that.
 
@@ -12,6 +12,8 @@ Here's an inline [link](/script?foo=1&bar=2).
 
 [1]: http://example.com/?foo=1&bar=2
 
+<http://example.com/autolink?a=1&b=2>
+
 ---
 
 Based on [the original](http://daringfireball.net/projects/downloads/MarkdownTest_1.0.zip) test suite.