improve parsing of emphasis

Parsedown.php

@@ -221,6 +221,8 @@ class Parsedown
 
 							$element['lines'] []= $line;
 
+							unset($element['interrupted']);
+
 							continue 2;
 						}
 					}
@@ -627,6 +629,81 @@ class Parsedown
 
 		$index = 0;
 
+		# inline link / inline image (recursive)
+
+		if (strpos($text, '](') !== FALSE and preg_match_all('/(!?)(\[((?:[^\[\]]|(?2))*)\])\((.*?)\)/', $text, $matches, PREG_SET_ORDER))
+		{
+			foreach ($matches as $matches)
+			{
+				$url = $matches[4];
+
+				strpos($url, '&') !== FALSE and $url = preg_replace('/&(?!#?\w+;)/', '&', $url);
+
+				if ($matches[1]) # image
+				{
+					$element = '<img alt="'.$matches[3].'" src="'.$url.'">';
+				}
+				else # link
+				{
+					$element_text = $this->parse_span_elements($matches[3]);
+
+					$element = '<a href="'.$url.'">'.$element_text.'</a>';
+				}
+
+				# ~
+
+				$code = "\x1A".'$'.$index;
+
+				$text = str_replace($matches[0], $code, $text);
+
+				$map[$code] = $element;
+
+				$index ++;
+			}
+		}
+
+		# reference link / reference image (recursive)
+
+		if ($this->reference_map and strpos($text, '[') !== FALSE and preg_match_all('/(!?)\[(.+?)\](?:\n?[ ]?\[(.*?)\])?/ms', $text, $matches, PREG_SET_ORDER))
+		{
+			foreach ($matches as $matches)
+			{
+				$link_definition = isset($matches[3]) && $matches[3]
+					? $matches[3]
+					: $matches[2]; # implicit
+
+				$link_definition = strtolower($link_definition);
+
+				if (isset($this->reference_map[$link_definition]))
+				{
+					$url = $this->reference_map[$link_definition];
+
+					strpos($url, '&') !== FALSE and $url = preg_replace('/&(?!#?\w+;)/', '&amp;', $url);
+
+					if ($matches[1]) # image
+					{
+						$element = '<img alt="'.$matches[2].'" src="'.$url.'">';
+					}
+					else # link
+					{
+						$element_text = $this->parse_span_elements($matches[2]);
+
+						$element = '<a href="'.$url.'">'.$element_text.'</a>';
+					}
+
+					# ~
+
+					$code = "\x1A".'$'.$index;
+
+					$text = str_replace($matches[0], $code, $text);
+
+					$map[$code] = $element;
+
+					$index ++;
+				}
+			}
+		}
+
 		# code span
 
 		if (strpos($text, '`') !== FALSE and preg_match_all('/`(.+?)`/', $text, $matches, PREG_SET_ORDER))
@@ -658,80 +735,7 @@ class Parsedown
 			}
 		}
 
-		# inline link or image
-
-		if (strpos($text, '](') !== FALSE and preg_match_all('/(!?)(\[((?:[^\[\]]|(?2))*)\])\((.*?)\)/', $text, $matches, PREG_SET_ORDER)) # inline
-		{
-			foreach ($matches as $matches)
-			{
-				$url = $matches[4];
-
-				strpos($url, '&') !== FALSE and $url = preg_replace('/&(?!#?\w+;)/', '&amp;', $url);
-
-				if ($matches[1]) # image
-				{
-					$element = '<img alt="'.$matches[3].'" src="'.$url.'">';
-				}
-				else
-				{
-					$element_text = $this->parse_span_elements($matches[3]);
-
-					$element = '<a href="'.$url.'">'.$element_text.'</a>';
-				}
-
-				# ~
-
-				$code = "\x1A".'$'.$index;
-
-				$text = str_replace($matches[0], $code, $text);
-
-				$map[$code] = $element;
-
-				$index ++;
-			}
-		}
-
-		# reference link or image
-
-		if ($this->reference_map and strpos($text, '[') !== FALSE and preg_match_all('/(!?)\[(.+?)\](?:\n?[ ]?\[(.*?)\])?/ms', $text, $matches, PREG_SET_ORDER))
-		{
-			foreach ($matches as $matches)
-			{
-				$link_definition = isset($matches[3]) && $matches[3]
-					? $matches[3]
-					: $matches[2]; # implicit
-
-				$link_definition = strtolower($link_definition);
-
-				if (isset($this->reference_map[$link_definition]))
-				{
-					$url = $this->reference_map[$link_definition];
-
-					strpos($url, '&') !== FALSE and $url = preg_replace('/&(?!#?\w+;)/', '&amp;', $url);
-
-					if ($matches[1]) # image
-					{
-						$element = '<img alt="'.$matches[2].'" src="'.$url.'">';
-					}
-					else # anchor
-					{
-						$element_text = $this->parse_span_elements($matches[2]);
-
-						$element = '<a href="'.$url.'">'.$element_text.'</a>';
-					}
-
-					# ~
-
-					$code = "\x1A".'$'.$index;
-
-					$text = str_replace($matches[0], $code, $text);
-
-					$map[$code] = $element;
-
-					$index ++;
-				}
-			}
-		}
+		# automatic link
 
 		if (strpos($text, '://') !== FALSE)
 		{
@@ -780,18 +784,15 @@ class Parsedown
 		if (strpos($text, '_') !== FALSE)
 		{
 			$text = preg_replace('/__(?=\S)([^_]+?)(?<=\S)__/s', '<strong>$1</strong>', $text, -1, $count);
-			$count or $text = preg_replace('/__(?=\S)(.+?)(?<=\S)__(?!_)/s', '<strong>$1</strong>', $text);
-
-			$text = preg_replace('/\b_(?=\S)(.+?)(?<=\S)_\b/s', '<em>$1</em>', $text);
+			$text = preg_replace('/(\b|_)_(?=\S)([^_]+?)(?<=\S)_(\b|_)/s', '$1<em>$2</em>$3', $text);
+			$text = preg_replace('/__(?=\S)([^_]+?)(?<=\S)__/s', '<strong>$1</strong>', $text, -1, $count);
 		}
 
 		if (strpos($text, '*') !== FALSE)
 		{
-			$text = preg_replace('/\*\*(?=\S)([^*]+?)(?<=\S)\*\*/s', '<strong>$1</strong>', $text, -1, $count);
-			$count or $text = preg_replace('/\*\*(?=\S)(.+?)(?<=\S)\*\*(?!\*)/s', '<strong>$1</strong>', $text);
-
-			$text = preg_replace('/\*(?=\S)([^*]+?)(?<=\S)\*/s', '<em>$1</em>', $text, -1, $count);
-			$count or $text = preg_replace('/\*(?=\S)(.+?)(?<=\S)\*(?!\*)/s', '<em>$1</em>', $text);
+			$text = preg_replace('/\*\*(?=\S)([^*]+?)(?<=\S)\*\*/s', '<strong>$1</strong>', $text);
+			$text = preg_replace('/\*(?=\S)([^*]+?)(?<=\S)\*/s', '<em>$1</em>', $text);
+			$text = preg_replace('/\*\*(?=\S)([^*]+?)(?<=\S)\*\*/s', '<strong>$1</strong>', $text);
 		}
 
 		$text = strtr($text, $map);

README.md

@@ -1,19 +1,19 @@
 ## Parsedown
 
-Better [Markdown][1] parser for PHP.
+Better [Markdown](http://daringfireball.net/projects/markdown/) parser for PHP.
 
 ***
 
-[home](http://parsedown.org/) · [demo](http://parsedown.org/demo) · [tests](http://parsedown.org/tests/)
+[demo](http://parsedown.org/demo) · [tests](http://parsedown.org/tests/)
 
 ***
 
-Features:
+### Features
 
 * [fast](http://parsedown.org/speed)
 * [consistent](http://parsedown.org/consistency)
-* [ GitHub Flavored ][2]
-* tested in PHP 5.2, 5.3, 5.4 and 5.5
+* [GitHub Flavored](https://help.github.com/articles/github-flavored-markdown)
+* [tested](https://travis-ci.org/erusev/parsedown) in PHP 5.2, 5.3, 5.4 and 5.5
 * friendly to international input
 
 ### Installation
@@ -29,6 +29,3 @@ $result = Parsedown::instance()->parse($text);
 
 echo $result; # prints: <p>Hello <strong>Parsedown</strong>!</p>
 ```
-
-[1]: http://daringfireball.net/projects/markdown/
-[2]: https://help.github.com/articles/github-flavored-markdown 

tests/data/em_strong.html

@@ -1,4 +1,6 @@
 <p><strong><em>em strong</em> strong</strong></p>
 <p><strong>strong <em>em strong</em></strong></p>
 <p><strong>strong <em>em strong</em> strong</strong></p>
-<p><strong>strong <em>em strong</em></strong></p>
+<p><strong><em>em strong</em> strong</strong></p>
+<p><strong>strong <em>em strong</em></strong></p>
+<p><strong>strong <em>em strong</em> strong</strong></p>

tests/data/em_strong.md

@@ -4,4 +4,8 @@ __strong _em strong___
 
 __strong _em strong_ strong__
 
-**strong *em strong***
+***em strong* strong**
+
+**strong *em strong***
+
+**strong *em strong* strong**

tests/data/emphasis.html

@@ -3,4 +3,5 @@
 emphasis</em></p>
 <p>_ this _ is not an emphasis, neither is _ this_, _this _, or _this*</p>
 <p>this_is_not_an_emphasis</p>
-<p>an empty emphasis __ ** is not an emphasis</p>
+<p>an empty emphasis __ ** is not an emphasis</p>
+<p>*mixed *<em>double and</em> single asterisk** spans</p>

tests/data/emphasis.md

@@ -7,4 +7,6 @@ _ this _ is not an emphasis, neither is _ this_, _this _, or _this*
 
 this_is_not_an_emphasis
 
-an empty emphasis __ ** is not an emphasis
+an empty emphasis __ ** is not an emphasis
+
+*mixed **double and* single asterisk** spans

tests/data/inline_link.html

@@ -1,2 +1,3 @@
 <p><a href="http://example.com">link</a></p>
+<p><a href="http://example.com"><code>link</code></a></p>
 <p><a href="http://example.com"><img alt="MD Logo" src="http://parsedown.org/md.png"></a></p>

tests/data/inline_link.md

@@ -1,3 +1,5 @@
 [link](http://example.com)
 
+[`link`](http://example.com)
+
 [![MD Logo](http://parsedown.org/md.png)](http://example.com)

tests/data/multiline_list_paragraph.html

@@ -0,0 +1,7 @@
+<ul>
+<li>
+<p>li</p>
+<p>line
+line</p>
+</li>
+</ul>

tests/data/multiline_list_paragraph.md

@@ -0,0 +1,4 @@
+- li
+
+  line
+  line

tests/data/special_characters.html

@@ -1,5 +1,4 @@
 <p>AT&amp;T has an ampersand in their name</p>
-<pre><code>Let's play some cards ♠ ♣ ♥ ♦</code></pre>
 <p>AT&amp;T is another way to write it</p>
 <p>this &amp; that</p>
 <p>4 &lt; 5 and 6 > 5</p>

tests/data/special_characters.md

@@ -1,7 +1,5 @@
 AT&T has an ampersand in their name
 
-    Let's play some cards ♠ ♣ ♥ ♦
-
 AT&T is another way to write it
 
 this & that

tests/data/text_reference.html

@@ -3,4 +3,5 @@
 <p>[one][404] with no definition</p>
 <p><a href="http://example.com">multiline
 one</a> defined on 2 lines</p>
-<p><a href="http://example.com">one</a> with an upper case label</p>
+<p><a href="http://example.com">one</a> with an upper case label</p>
+<p><a href="http://example.com"><code>link</code></a></p>

tests/data/text_reference.md

@@ -13,4 +13,6 @@ one][website] defined on 2 lines
 
 [one][label] with an upper case label
 
-[LABEL]: http://example.com
+[LABEL]: http://example.com
+
+[`link`][website]