parse link references as blocks to improve performance

Parsedown.php

@@ -75,18 +75,6 @@ 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);
@@ -217,56 +205,14 @@ class Parsedown
 			
 			# Quick Paragraph 
 			
-			if ($line[0] >= 'A' and $line['0'] !== '_')
+			if ($line[0] >= 'A' and $line[0] !== '_' and $line[0] !== '[')
 			{
 				goto paragraph; # trust me 
 			}
 			
-			# Setext Header (---) 
-			
-			if ($element['type'] === 'p' and ! isset($element['interrupted']) and preg_match('/^[-]+[ ]*$/', $line))
-			{
-				$element['type'] = 'h.';
-				$element['level'] = 2;
-				
-				continue;
-			}
-			
-			# Horizontal Rule  
-			
-			if (preg_match('/^[ ]{0,3}([-*_])([ ]{0,2}\1){2,}[ ]*$/', $line))
-			{
-				$elements []= $element;
-				
-				$element = array(
-					'type' => 'hr',
-				);
-				
-				continue;
-			}
-			
-			# List Item 
-			
-			if (preg_match('/^([ ]{0,3})(\d+[.]|[*+-])[ ](.*)/', $line, $matches))
-			{
-				$elements []= $element;
-				
-				$element = array(
-					'type' => 'li',
-					'ordered' => isset($matches[2][1]),
-					'indentation' => $matches[1],
-					'last' => true,
-					'lines' => array(
-						preg_replace('/^[ ]{0,4}/', '', $matches[3]),
-					),
-				);
-				
-				continue;
-			}
-			
 			# Code 
 			
-			if (preg_match('/^[ ]{4}(.*)/', $line, $matches))
+			if ($line[0] === ' ' and preg_match('/^[ ]{4}(.*)/', $line, $matches))
 			{
 				if ($element['type'] === 'code')
 				{
@@ -279,10 +225,20 @@ class Parsedown
 					$elements []= $element;
 					
 					$element = array(
-						'type' => 'code', 
+						'type' => 'code',
 						'text' => $matches[1],
 					);
 				}
+
+				continue;
+			}
+			
+			# Setext Header (---)  
+			
+			if ($line[0] === '-' and $element['type'] === 'p' and ! isset($element['interrupted']) and preg_match('/^[-]+[ ]*$/', $line))
+			{
+				$element['type'] = 'h.';
+				$element['level'] = 2;
 				
 				continue;
 			}
@@ -296,17 +252,43 @@ class Parsedown
 				$level = strlen($matches[1]);
 				
 				$element = array(
-					'type' => 'h.', 
-					'text' => $matches[2], 
+					'type' => 'h.',
+					'text' => $matches[2],
 					'level' => $level,
 				);
 				
 				continue;
 			}
 			
+			# Setext Header (===) 
+			
+			if ($line[0] === '=' and $element['type'] === 'p' and ! isset($element['interrupted']) and preg_match('/^[=]+[ ]*$/', $line))
+			{
+				$element['type'] = 'h.';
+				$element['level'] = 1;
+				
+				continue;
+			}
+			
+			# ~ 
+			
+			$pure_line = ltrim($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 
 			
-			if (preg_match('/^[ ]*>[ ]?(.*)/', $line, $matches))
+			if ($pure_line[0] === '>' and preg_match('/^>[ ]?(.*)/', $pure_line, $matches))
 			{
 				if ($element['type'] === 'blockquote')
 				{
@@ -334,45 +316,72 @@ class Parsedown
 				continue;
 			}
 			
-			# Setext Header (===) 
+			# HTML  
 			
-			if ($element['type'] === 'p' and ! isset($element['interrupted']) and preg_match('/^[=]+[ ]*$/', $line))
+			if ($pure_line[0] === '<')
 			{
-				$element['type'] = 'h.';
-				$element['level'] = 1;
-				
-				continue;
-			}
+				# Block-Level HTML <self-closing/>
 
-			# Block-Level HTML <self-closing/> 
+				if (preg_match('{^<.+?/>$}', $pure_line))
+				{
+					$elements []= $element;
+
+					$element = array(
+						'type' => '',
+						'text' => $pure_line,
+					);
+
+					continue;
+				}
+				
+				# Block-Level HTML <open>
+
+				if (preg_match('{^<(\w+)(?:[ ].*?)?>}', $pure_line, $matches))
+				{
+					$elements []= $element;
+
+					$element = array(
+						'type' => 'block',
+						'subtype' => strtolower($matches[1]),
+						'text' => $pure_line,
+						'depth' => 0,
+					);
+					
+					preg_match('{</'.$matches[1].'>\s*$}', $pure_line) and $element['closed'] = true;
+
+					continue;
+				}
+			}
 			
-			if (preg_match('{^<.+?/>$}', $line))
+			# Horizontal Rule   
+			
+			if (preg_match('/^([-*_])([ ]{0,2}\1){2,}[ ]*$/', $pure_line))
 			{
 				$elements []= $element;
 				
 				$element = array(
-					'type' => '',
-					'text' => $line,
+					'type' => 'hr',
 				);
 				
 				continue;
 			}
 			
-			# Block-Level HTML <open>
+			# List Item 
 			
-			if (preg_match('{^<(\w+)(?:[ ].*?)?>}', $line, $matches))
+			if (preg_match('/^([ ]*)(\d+[.]|[*+-])[ ](.*)/', $line, $matches))
 			{
 				$elements []= $element;
 				
 				$element = array(
-					'type' => 'block',
-					'subtype' => strtolower($matches[1]),
-					'text' => $line,
-					'depth' => 0,
+					'type' => 'li',
+					'ordered' => isset($matches[2][1]),
+					'indentation' => $matches[1],
+					'last' => true,
+					'lines' => array(
+						preg_replace('/^[ ]{0,4}/', '', $matches[3]),
+					),
 				);
 				
-				preg_match('{</'.$matches[1].'>\s*$}', $line) and $element['closed'] = true; 
-				
 				continue;
 			}
 			
@@ -550,15 +559,17 @@ class Parsedown
 		{
 			foreach ($matches as $matches)
 			{
+				$url = $this->escape_special_characters($matches[4]);
+				
 				if ($matches[1]) # image 
 				{
-					$element = '<img alt="'.$matches[3].'" src="'.$matches[4].'">';
+					$element = '<img alt="'.$matches[3].'" src="'.$url.'">';
 				}
-				else 
+				else
 				{
 					$element_text = $this->parse_inline_elements($matches[3]);
 					
-					$element = '<a href="'.$matches[4].'">'.$element_text.'</a>';
+					$element = '<a href="'.$url.'">'.$element_text.'</a>';
 				}
 				
 				# ~ 
@@ -588,6 +599,7 @@ class Parsedown
 				if (isset($this->reference_map[$link_definition]))
 				{
 					$url = $this->reference_map[$link_definition];
+					$url = $this->escape_special_characters($url);
 					
 					if ($matches[1]) # image 
 					{
@@ -613,13 +625,17 @@ class Parsedown
 			}
 		}
 		
+		# Automatic Links 
+		
 		if (strpos($text, '<') !== FALSE and preg_match_all('/<((https?|ftp|dict):[^\^\s]+?)>/i', $text, $matches, PREG_SET_ORDER))
 		{
 			foreach ($matches as $matches)
 			{
+				$url = $this->escape_special_characters($matches[1]);
+				
 				$element = '<a href=":href">:text</a>';
-				$element = str_replace(':text', $matches[1], $element);
-				$element = str_replace(':href', $matches[1], $element);
+				$element = str_replace(':text', $url, $element);
+				$element = str_replace(':href', $url, $element);
 				
 				# ~ 
 				
@@ -633,6 +649,12 @@ class Parsedown
 			}
 		}
 		
+		# ~ 
+		
+		$text = $this->escape_special_characters($text);
+
+		# ~ 
+		
 		if (strpos($text, '_') !== FALSE)
 		{
 			$text = preg_replace('/__(?=\S)(.+?)(?<=\S)__/', '<strong>$1</strong>', $text);
@@ -649,5 +671,13 @@ 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/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/sparse_list.html

@@ -1,14 +1,16 @@
-<p>Here's a list where items are separated by empty lines:</p>
-<ul>
-<li>
-<p>list item</p>
-</li>
-<li>another list item</li>
-</ul>
-<p>Here's an ordered one:</p>
-<ol>
-<li>
-<p>item one</p>
-</li>
-<li>item two</li>
-</ol>
+<p>Here's a sparse list:</p>
+<ul>
+<li>
+<p>list item</p>
+</li>
+<li>another list item</li>
+</ul>
+<p>Here's one with an indented list item:</p>
+<ul>
+<li>
+<p>li</p>
+<ul>
+<li>li</li>
+</ul>
+</li>
+</ul>

tests/data/sparse_list.md

@@ -1,11 +1,11 @@
-Here's a list where items are separated by empty lines:
-
-- list item
-
-- another list item
-
-Here's an ordered one:
-
-1. item one
-
-2. item two
+Here's a sparse list:
+
+- list item
+
+- another list item
+
+Here's one with an indented list item:
+
+- li
+
+    - li

tests/data/special_characters.html

@@ -0,0 +1,8 @@
+<p>AT&amp;T has an ampersand in their name.</p>
+<p>AT&amp;T is another way to write it.</p>
+<p>This &amp; that.</p>
+<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>
+<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

@@ -0,0 +1,17 @@
+AT&T has an ampersand in their name.
+
+AT&T is another way to write it.
+
+This & that.
+
+4 < 5 and 6 > 5.
+
+Here's a [link] [1] with an ampersand in the URL.
+
+Here's an inline [link](/script?foo=1&bar=2).
+
+[1]: http://example.com/?foo=1&bar=2
+
+---
+
+Based on [the original](http://daringfireball.net/projects/downloads/MarkdownTest_1.0.zip) test suite.