improve parsing of emphasis

**strong** and *em* and **strong** and *em*

Parsedown.php

@@ -631,9 +631,14 @@ class Parsedown
 
 	# ~
 
-	private $em_strong_regex = array(
+	private $strong_regex = array(
-		'*' => '/^[*](.*?)[*]{2}(.+?)[*]{2}(.*?)[*]/s',
+		'*' => '/^[*]{2}([^*]+?)[*]{2}(?![*])/s',
-		'_' => '/^_(.*?)__(.+?)__(.*?)_/s',
+		'_' => '/^__([^_]+?)__(?!_)/s',
+	);
+
+	private $em_regex = array(
+		'*' => '/^[*]([^*]+?)[*](?![*])/s',
+		'_' => '/^_([^_]+?)[_](?![_])\b/s',
 	);
 
 	private $strong_em_regex = array(
@@ -641,14 +646,9 @@ class Parsedown
 		'_' => '/^__(.*?)_(.+?)_(.*?)__/s',
 	);
 
-	private $strong_regex = array(
+	private $em_strong_regex = array(
-		'*' => '/^[*]{2}(.+?)[*]{2}/s',
+		'*' => '/^[*](.*?)[*]{2}(.+?)[*]{2}(.*?)[*]/s',
-		'_' => '/^__(.+?)__/s',
+		'_' => '/^_(.*?)__(.+?)__(.*?)_/s',
-	);
-
-	private $em_regex = array(
-		'*' => '/^[*](.+?)[*]/s',
-		'_' => '/^_(.+?)_\b/s',
 	);
 
 	private function parse_span_elements($text, $markers = array("  \n", '![', '&', '*', '<', '[', '_', '`', 'http', '~~'))
@@ -806,7 +806,28 @@ class Parsedown
 				case '*':
 				case '_':
 
-					if (preg_match($this->em_strong_regex[$closest_marker], $text, $matches))
+					if ($text[1] === $closest_marker and preg_match($this->strong_regex[$closest_marker], $text, $matches))
+					{
+						$matches[1] = $this->parse_span_elements($matches[1], $markers);
+
+						$markup .= '<strong>'.$matches[1].'</strong>';
+					}
+					elseif (preg_match($this->em_regex[$closest_marker], $text, $matches))
+					{
+						$matches[1] = $this->parse_span_elements($matches[1], $markers);
+
+						$markup .= '<em>'.$matches[1].'</em>';
+					}
+					elseif ($text[1] === $closest_marker and preg_match($this->strong_em_regex[$closest_marker], $text, $matches))
+					{
+						$matches[2] = $this->parse_span_elements($matches[2], $markers);
+
+						$matches[1] and $matches[1] = $this->parse_span_elements($matches[1], $markers);
+						$matches[3] and $matches[3] = $this->parse_span_elements($matches[3], $markers);
+
+						$markup .= '<strong>'.$matches[1].'<em>'.$matches[2].'</em>'.$matches[3].'</strong>';
+					}
+					elseif (preg_match($this->em_strong_regex[$closest_marker], $text, $matches))
 					{
 						$matches[2] = $this->parse_span_elements($matches[2], $markers);
 
@@ -815,27 +836,8 @@ class Parsedown
 
 						$markup .= '<em>'.$matches[1].'<strong>'.$matches[2].'</strong>'.$matches[3].'</em>';
 					}
-					elseif ($text[1] === $closest_marker)
-					{
-						if (preg_match($this->strong_em_regex[$closest_marker], $text, $matches))
-						{
-							$markup .= '<strong>'.$matches[1].'<em>'.$matches[2].'</em>'.$matches[3].'</strong>';
-						}
-						elseif (preg_match($this->strong_regex[$closest_marker], $text, $matches))
-						{
-							$matches[1] = $this->parse_span_elements($matches[1], $markers);
 
-							$markup .= '<strong>'.$matches[1].'</strong>';
+					if (isset($matches) and $matches)
-						}
-					}
-					elseif (preg_match($this->em_regex[$closest_marker], $text, $matches))
-					{
-						$element_text = $this->parse_span_elements($matches[1], $markers);
-
-						$markup .= '<em>'.$element_text.'</em>';
-					}
-
-					if ($matches)
 					{
 						$offset = strlen($matches[0]);
 					}

tests/data/em_strong.html

@@ -1,6 +1,8 @@
+<p><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>
+<p><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

@@ -1,9 +1,13 @@
+___em strong___
+
 ___em strong_ strong__
 
 __strong _em strong___
 
 __strong _em strong_ strong__
 
+***em strong***
+
 ***em strong* strong**
 
 **strong *em strong***

tests/data/emphasis.html

@@ -1,7 +1,8 @@
 <p><em>underscore</em>, <em>asterisk</em>, <em>one two</em>, <em>three four</em>, <em>a</em>, <em>b</em></p>
-<p><em>multiline
+<p><strong>strong</strong> and <em>em</em> and <strong>strong</strong> and <em>em</em></p>
-emphasis</em></p>
+<p><em>line
-<p>_ this _ is not an emphasis, neither is _ this_, _this _, or _this*</p>
+line
+line</em></p>
 <p>this_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

@@ -1,9 +1,10 @@
 _underscore_, *asterisk*, _one two_, *three four*, _a_, *b*
 
-_multiline
+**strong** and *em* and **strong** and *em*
-emphasis_
 
-_ this _ is not an emphasis, neither is _ this_, _this _, or _this*
+_line
+line
+line_
 
 this_is_not_an_emphasis
 

tests/data/strong_em.html

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

tests/data/strong_em.md

@@ -1,11 +1,11 @@
-***strong em*** 
-
 *em **strong em***
 
 ***strong em** em*
 
-___strong em___
+*em **strong em** em*
 
 _em __strong em___
 
 ___strong em__ em_
+
+_em __strong em__ em_