improve tests

.travis.yml

@@ -4,4 +4,5 @@ php:
   - 5.5
   - 5.4
   - 5.3
-  - 5.2
+  - 5.2
+  - hhvm

Parsedown.php

@@ -59,9 +59,6 @@ class Parsedown
 
 	function parse($text)
 	{
-		# removes UTF-8 BOM and marker characters
-		$text = preg_replace('{^\xEF\xBB\xBF|\x1A}', '', $text);
-
 		# removes \r characters
 		$text = str_replace("\r\n", "\n", $text);
 		$text = str_replace("\r", "\n", $text);
@@ -90,7 +87,6 @@ class Parsedown
 
 		# ~
 
-		$text = preg_replace('/\n\s*\n/', "\n\n", $text);
 		$text = trim($text, "\n");
 
 		$lines = explode("\n", $text);
@@ -129,7 +125,7 @@ class Parsedown
 
 			switch ($element['type'])
 			{
-				case 'fenced_code_block':
+				case 'fenced block':
 
 					if ( ! isset($element['closed']))
 					{
@@ -149,16 +145,16 @@ class Parsedown
 
 					break;
 
-				case 'markup':
+				case 'block-level markup':
 
 					if ( ! isset($element['closed']))
 					{
-						if (preg_match('{<'.$element['subtype'].'>$}', $line)) # opening tag
+						if (preg_match('{<'.$element['root '].'>$}', $line)) # opening tag
 						{
 							$element['depth']++;
 						}
 
-						if (preg_match('{</'.$element['subtype'].'>$}', $line)) # closing tag
+						if (preg_match('{</'.$element['root '].'>$}', $line)) # closing tag
 						{
 							$element['depth'] > 0
 								? $element['depth']--
@@ -175,7 +171,9 @@ class Parsedown
 
 			# *
 
-			if ($line === '')
+			$deindented_line = ltrim($line);
+
+			if ($deindented_line === '')
 			{
 				$element['interrupted'] = true;
 
@@ -255,26 +253,17 @@ class Parsedown
 
 			# indentation sensitive types
 
-			$deindented_line = $line;
-
 			switch ($line[0])
 			{
 				case ' ':
 
-					# ~
-
-					$deindented_line = ltrim($line);
-
-					if ($deindented_line === '')
-					{
-						continue 2;
-					}
-
 					# code block
 
-					if (preg_match('/^[ ]{4}(.*)/', $line, $matches))
+					if (isset($line[3]) and $line[3] === ' ' and $line[2] === ' ' and $line[1] === ' ')
 					{
-						if ($element['type'] === 'code_block')
+						$code_line = substr($line, 4);
+
+						if ($element['type'] === 'code block')
 						{
 							if (isset($element['interrupted']))
 							{
@@ -283,15 +272,15 @@ class Parsedown
 								unset ($element['interrupted']);
 							}
 
-							$element['text'] .= "\n".$matches[1];
+							$element['text'] .= "\n".$code_line;
 						}
 						else
 						{
 							$elements []= $element;
 
 							$element = array(
-								'type' => 'code_block',
-								'text' => $matches[1],
+								'type' => 'code block',
+								'text' => $code_line,
 							);
 						}
 
@@ -304,15 +293,20 @@ class Parsedown
 
 					# atx heading (#)
 
-					if (preg_match('/^(#{1,6})[ ]*(.+?)[ ]*#*$/', $line, $matches))
+					if (isset($line[1]))
 					{
 						$elements []= $element;
 
-						$level = strlen($matches[1]);
+						$level = 1;
+
+						while (isset($line[$level]) and $line[$level] === '#')
+						{
+							$level++;
+						}
 
 						$element = array(
-							'type' => 'h.',
-							'text' => $matches[2],
+							'type' => 'heading',
+							'text' => trim($line, '# '),
 							'level' => $level,
 						);
 
@@ -322,27 +316,35 @@ class Parsedown
 					break;
 
 				case '-':
-
-					# setext heading (---)
-
-					if ($line[0] === '-' and $element['type'] === 'p' and ! isset($element['interrupted']) and preg_match('/^[-]+[ ]*$/', $line))
-					{
-						$element['type'] = 'h.';
-						$element['level'] = 2;
-
-						continue 2;
-					}
-
-					break;
-
 				case '=':
 
-					# setext heading (===)
+					# setext heading
 
-					if ($line[0] === '=' and $element['type'] === 'p' and ! isset($element['interrupted']) and preg_match('/^[=]+[ ]*$/', $line))
+					if ($element['type'] === 'paragraph' and isset($element['interrupted']) === false)
 					{
-						$element['type'] = 'h.';
-						$element['level'] = 1;
+						$is_heading = true;
+
+						$chopped_line = rtrim($line);
+
+						$i = 1;
+
+						while (isset($chopped_line[$i]))
+						{
+							if ($chopped_line[$i] !== $line[0])
+							{
+								$is_heading = false;
+
+								break;
+							}
+
+							$i++;
+						}
+
+						if ($is_heading)
+						{
+							$element['type'] = 'heading';
+							$element['level'] = $line[0] === '-' ? 2 : 1;
+						}
 
 						continue 2;
 					}
@@ -363,7 +365,7 @@ class Parsedown
 						$elements []= $element;
 
 						$element = array(
-							'type' => '',
+							'type' => 'self-closing tag',
 							'text' => $deindented_line,
 						);
 
@@ -377,9 +379,9 @@ class Parsedown
 						$elements []= $element;
 
 						$element = array(
-							'type' => 'markup',
-							'subtype' => strtolower($matches[1]),
+							'type' => 'block-level markup',
 							'text' => $deindented_line,
+							'root ' => strtolower($matches[1]),
 							'depth' => 0,
 						);
 
@@ -442,7 +444,7 @@ class Parsedown
 						$elements []= $element;
 
 						$element = array(
-							'type' => 'fenced_code_block',
+							'type' => 'fenced block',
 							'text' => '',
 							'fence' => $matches[1],
 						);
@@ -494,7 +496,7 @@ class Parsedown
 
 			# li
 
-			if ($deindented_line[0] <= '9' and $deindented_line >= '0' and preg_match('/^([ ]*)\d+[.][ ](.*)/', $line, $matches))
+			if ($deindented_line[0] <= '9' and $deindented_line[0] >= '0' and preg_match('/^([ ]*)\d+[.][ ](.*)/', $line, $matches))
 			{
 				$elements []= $element;
 
@@ -513,7 +515,7 @@ class Parsedown
 
 			# paragraph
 
-			if ($element['type'] === 'p')
+			if ($element['type'] === 'paragraph')
 			{
 				if (isset($element['interrupted']))
 				{
@@ -533,7 +535,7 @@ class Parsedown
 				$elements []= $element;
 
 				$element = array(
-					'type' => 'p',
+					'type' => 'paragraph',
 					'text' => $line,
 				);
 			}
@@ -553,7 +555,7 @@ class Parsedown
 		{
 			switch ($element['type'])
 			{
-				case 'p':
+				case 'paragraph':
 
 					$text = $this->parse_span_elements($element['text']);
 
@@ -583,8 +585,8 @@ class Parsedown
 
 					break;
 
-				case 'code_block':
-				case 'fenced_code_block':
+				case 'code block':
+				case 'fenced block':
 
 					$text = htmlspecialchars($element['text'], ENT_NOQUOTES, 'UTF-8');
 
@@ -598,7 +600,7 @@ class Parsedown
 
 					break;
 
-				case 'h.':
+				case 'heading':
 
 					$text = $this->parse_span_elements($element['text']);
 
@@ -634,7 +636,7 @@ class Parsedown
 
 					break;
 
-				case 'markup':
+				case 'block-level markup':
 
 					$markup .= $this->parse_span_elements($element['text'])."\n";
 

README.md

@@ -13,8 +13,8 @@ Better [Markdown](http://en.wikipedia.org/wiki/Markdown) parser for PHP.
 * [fast](http://parsedown.org/speed)
 * [consistent](http://parsedown.org/consistency)
 * [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
+* [tested](https://travis-ci.org/erusev/parsedown) in PHP 5.2, 5.3, 5.4 and 5.5 as well as in [HHVM](http://www.hhvm.com/)
 
 ### Installation
 

tests/data/atx_heading.html

@@ -4,4 +4,5 @@
 <h4>h4</h4>
 <h5>h5</h5>
 <h6>h6</h6>
-<h1>closed h1</h1>
+<h1>closed h1</h1>
+<p>#</p>

tests/data/atx_heading.md

@@ -10,4 +10,6 @@
 
 ###### h6
 
-# closed h1 #
+# closed h1 #
+
+#

tests/data/tab-indented_code_block.html

@@ -0,0 +1,6 @@
+<pre><code>&lt;?php
+
+$message = 'Hello World!'; 
+echo $message;
+
+echo "following a blank line";</code></pre>

tests/data/tab-indented_code_block.md

@@ -0,0 +1,6 @@
+	<?php
+	
+	$message = 'Hello World!'; 
+	echo $message;
+
+	echo "following a blank line";