1
0
mirror of https://github.com/erusev/parsedown.git synced 2023-08-10 21:13:06 +03:00

Compare commits

..

9 Commits
0.4.3 ... 0.4.6

64 changed files with 491 additions and 625 deletions

View File

@ -3,4 +3,9 @@ language: php
php: php:
- 5.5 - 5.5
- 5.4 - 5.4
- 5.3 - 5.3
- 5.2
matrix:
allow_failures:
- php: 5.2

View File

@ -1,172 +1,172 @@
<?php <?php
# #
# #
# Parsedown # Parsedown
# http://parsedown.org # http://parsedown.org
# #
# (c) Emanuil Rusev # (c) Emanuil Rusev
# http://erusev.com # http://erusev.com
# #
# For the full license information, please view the LICENSE file that was # For the full license information, please view the LICENSE file that was
# distributed with this source code. # distributed with this source code.
# #
# #
class Parsedown class Parsedown
{ {
# #
# Multiton (http://en.wikipedia.org/wiki/Multiton_pattern) # Multiton (http://en.wikipedia.org/wiki/Multiton_pattern)
# #
static function instance($name = 'default') static function instance($name = 'default')
{ {
if (isset(self::$instances[$name])) if (isset(self::$instances[$name]))
return self::$instances[$name]; return self::$instances[$name];
$instance = new Parsedown(); $instance = new Parsedown();
self::$instances[$name] = $instance; self::$instances[$name] = $instance;
return $instance; return $instance;
} }
private static $instances = array(); private static $instances = array();
# #
# Fields # Fields
# #
private $reference_map = array(); private $reference_map = array();
private $escape_sequence_map = array(); private $escape_sequence_map = array();
# #
# Public Methods # Public Methods
# #
function parse($text) function parse($text)
{ {
# Removes UTF-8 BOM and marker characters. # Removes UTF-8 BOM and marker characters.
$text = preg_replace('{^\xEF\xBB\xBF|\x1A}', '', $text); $text = preg_replace('{^\xEF\xBB\xBF|\x1A}', '', $text);
# Removes \r characters. # Removes \r characters.
$text = str_replace("\r\n", "\n", $text); $text = str_replace("\r\n", "\n", $text);
$text = str_replace("\r", "\n", $text); $text = str_replace("\r", "\n", $text);
# Replaces tabs with spaces. # Replaces tabs with spaces.
$text = str_replace("\t", ' ', $text); $text = str_replace("\t", ' ', $text);
# Encodes escape sequences. # Encodes escape sequences.
if (strpos($text, '\\') !== FALSE) if (strpos($text, '\\') !== FALSE)
{ {
$escape_sequences = array('\\\\', '\`', '\*', '\_', '\{', '\}', '\[', '\]', '\(', '\)', '\>', '\#', '\+', '\-', '\.', '\!'); $escape_sequences = array('\\\\', '\`', '\*', '\_', '\{', '\}', '\[', '\]', '\(', '\)', '\>', '\#', '\+', '\-', '\.', '\!');
foreach ($escape_sequences as $index => $escape_sequence) foreach ($escape_sequences as $index => $escape_sequence)
{ {
if (strpos($text, $escape_sequence) !== FALSE) if (strpos($text, $escape_sequence) !== FALSE)
{ {
$code = "\x1A".'\\'.$index; $code = "\x1A".'\\'.$index;
$text = str_replace($escape_sequence, $code, $text); $text = str_replace($escape_sequence, $code, $text);
$this->escape_sequence_map[$code] = $escape_sequence; $this->escape_sequence_map[$code] = $escape_sequence;
} }
} }
} }
# ~ # ~
$text = preg_replace('/\n\s*\n/', "\n\n", $text); $text = preg_replace('/\n\s*\n/', "\n\n", $text);
$text = trim($text, "\n "); $text = trim($text, "\n");
$lines = explode("\n", $text); $lines = explode("\n", $text);
$text = $this->parse_block_elements($lines); $text = $this->parse_block_elements($lines);
# Decodes escape sequences (leaves out backslashes). # Decodes escape sequences (leaves out backslashes).
foreach ($this->escape_sequence_map as $code => $escape_sequence) foreach ($this->escape_sequence_map as $code => $escape_sequence)
{ {
$text = str_replace($code, $escape_sequence[1], $text); $text = str_replace($code, $escape_sequence[1], $text);
} }
$text = rtrim($text, "\n"); $text = rtrim($text, "\n");
return $text; return $text;
} }
# #
# Private Methods # Private Methods
# #
private function parse_block_elements(array $lines, $context = '') private function parse_block_elements(array $lines, $context = '')
{ {
$elements = array(); $elements = array();
$element = array( $element = array(
'type' => '', 'type' => '',
); );
foreach ($lines as $line) foreach ($lines as $line)
{ {
# Block-Level HTML # Block-Level HTML
if ($element['type'] === 'block' and ! isset($element['closed'])) if ($element['type'] === 'block' and ! isset($element['closed']))
{ {
if (preg_match('{<'.$element['subtype'].'>$}', $line)) # <open> if (preg_match('{<'.$element['subtype'].'>$}', $line)) # <open>
{ {
$element['depth']++; $element['depth']++;
} }
if (preg_match('{</'.$element['subtype'].'>$}', $line)) # </close> if (preg_match('{</'.$element['subtype'].'>$}', $line)) # </close>
{ {
$element['depth'] > 0 $element['depth'] > 0
? $element['depth']-- ? $element['depth']--
: $element['closed'] = true; : $element['closed'] = true;
} }
$element['text'] .= "\n".$line; $element['text'] .= "\n".$line;
continue; continue;
} }
# Empty # Empty
if ($line === '') if ($line === '')
{ {
$element['interrupted'] = true; $element['interrupted'] = true;
continue; continue;
} }
# Lazy Blockquote # Lazy Blockquote
if ($element['type'] === 'blockquote' and ! isset($element['interrupted'])) if ($element['type'] === 'blockquote' and ! isset($element['interrupted']))
{ {
$line = preg_replace('/^[ ]*>[ ]?/', '', $line); $line = preg_replace('/^[ ]*>[ ]?/', '', $line);
$element['lines'] []= $line; $element['lines'] []= $line;
continue; continue;
} }
# Lazy List Item # Lazy List Item
if ($element['type'] === 'li') if ($element['type'] === 'li')
{ {
if (preg_match('/^([ ]{0,3})(\d+[.]|[*+-])[ ](.*)/', $line, $matches)) if (preg_match('/^([ ]{0,3})(\d+[.]|[*+-])[ ](.*)/', $line, $matches))
{ {
if ($element['indentation'] !== $matches[1]) if ($element['indentation'] !== $matches[1])
{ {
$element['lines'] []= $line; $element['lines'] []= $line;
} }
else else
{ {
unset($element['last']); unset($element['last']);
$elements []= $element; $elements []= $element;
$element = array( $element = array(
'type' => 'li', 'type' => 'li',
'indentation' => $matches[1], 'indentation' => $matches[1],
@ -176,59 +176,64 @@ class Parsedown
), ),
); );
} }
continue; continue;
} }
if (isset($element['interrupted'])) if (isset($element['interrupted']))
{ {
if ($line[0] === ' ') if ($line[0] === ' ')
{ {
$element['lines'] []= ''; $element['lines'] []= '';
$line = preg_replace('/^[ ]{0,4}/', '', $line);; $line = preg_replace('/^[ ]{0,4}/', '', $line);;
$element['lines'] []= $line; $element['lines'] []= $line;
continue; continue;
} }
} }
else else
{ {
$line = preg_replace('/^[ ]{0,4}/', '', $line);; $line = preg_replace('/^[ ]{0,4}/', '', $line);;
$element['lines'] []= $line; $element['lines'] []= $line;
continue; continue;
} }
} }
# Quick Paragraph # Quick Paragraph
if ($line[0] >= 'A' and $line[0] !== '_' and $line[0] !== '[') if ($line[0] >= 'a' or $line[0] >= 'A' and $line[0] <= 'Z')
{ {
goto paragraph; # trust me goto paragraph;
} }
# Code # Code Block
if ($line[0] === ' ' and preg_match('/^[ ]{4}(.*)/', $line, $matches)) if ($line[0] === ' ' and preg_match('/^[ ]{4}(.*)/', $line, $matches))
{ {
if (trim($line) === '')
{
continue;
}
if ($element['type'] === 'code') if ($element['type'] === 'code')
{ {
if (isset($element['interrupted'])) if (isset($element['interrupted']))
{ {
$element['text'] .= "\n"; $element['text'] .= "\n";
unset ($element['interrupted']); unset ($element['interrupted']);
} }
$element['text'] .= "\n".$matches[1]; $element['text'] .= "\n".$matches[1];
} }
else else
{ {
$elements []= $element; $elements []= $element;
$element = array( $element = array(
'type' => 'code', 'type' => 'code',
'text' => $matches[1], 'text' => $matches[1],
@ -237,62 +242,67 @@ class Parsedown
continue; continue;
} }
# Setext Header (---) # Setext Header (---)
if ($line[0] === '-' and $element['type'] === 'p' and ! isset($element['interrupted']) and preg_match('/^[-]+[ ]*$/', $line)) if ($line[0] === '-' and $element['type'] === 'p' and ! isset($element['interrupted']) and preg_match('/^[-]+[ ]*$/', $line))
{ {
$element['type'] = 'h.'; $element['type'] = 'h.';
$element['level'] = 2; $element['level'] = 2;
continue; continue;
} }
# Atx Header (#) # Atx Header (#)
if ($line[0] === '#' and preg_match('/^(#{1,6})[ ]*(.+?)[ ]*#*$/', $line, $matches)) if ($line[0] === '#' and preg_match('/^(#{1,6})[ ]*(.+?)[ ]*#*$/', $line, $matches))
{ {
$elements []= $element; $elements []= $element;
$level = strlen($matches[1]); $level = strlen($matches[1]);
$element = array( $element = array(
'type' => 'h.', 'type' => 'h.',
'text' => $matches[2], 'text' => $matches[2],
'level' => $level, 'level' => $level,
); );
continue; continue;
} }
# Setext Header (===) # Setext Header (===)
if ($line[0] === '=' and $element['type'] === 'p' and ! isset($element['interrupted']) and preg_match('/^[=]+[ ]*$/', $line)) if ($line[0] === '=' and $element['type'] === 'p' and ! isset($element['interrupted']) and preg_match('/^[=]+[ ]*$/', $line))
{ {
$element['type'] = 'h.'; $element['type'] = 'h.';
$element['level'] = 1; $element['level'] = 1;
continue; continue;
} }
# ~ # ~
$pure_line = ltrim($line); $pure_line = $line[0] !== ' ' ? $line : ltrim($line);
# Link Reference if ($pure_line === '')
{
continue;
}
# Link Reference
if ($pure_line[0] === '[' and preg_match('/^\[(.+?)\]:[ ]*([^ ]+)/', $pure_line, $matches)) if ($pure_line[0] === '[' and preg_match('/^\[(.+?)\]:[ ]*([^ ]+)/', $pure_line, $matches))
{ {
$label = $matches[1]; $label = strtolower($matches[1]);
$url = trim($matches[2], '<>'); $url = trim($matches[2], '<>');
$this->reference_map[$label] = $url; $this->reference_map[$label] = $url;
continue; continue;
} }
# Blockquote # Blockquote
if ($pure_line[0] === '>' and preg_match('/^>[ ]?(.*)/', $pure_line, $matches)) if ($pure_line[0] === '>' and preg_match('/^>[ ]?(.*)/', $pure_line, $matches))
{ {
if ($element['type'] === 'blockquote') if ($element['type'] === 'blockquote')
@ -300,16 +310,16 @@ class Parsedown
if (isset($element['interrupted'])) if (isset($element['interrupted']))
{ {
$element['lines'] []= ''; $element['lines'] []= '';
unset($element['interrupted']); unset($element['interrupted']);
} }
$element['lines'] []= $matches[1]; $element['lines'] []= $matches[1];
} }
else else
{ {
$elements []= $element; $elements []= $element;
$element = array( $element = array(
'type' => 'blockquote', 'type' => 'blockquote',
'lines' => array( 'lines' => array(
@ -317,12 +327,12 @@ class Parsedown
), ),
); );
} }
continue; continue;
} }
# HTML # HTML
if ($pure_line[0] === '<') if ($pure_line[0] === '<')
{ {
# Block-Level HTML <self-closing/> # Block-Level HTML <self-closing/>
@ -338,7 +348,7 @@ class Parsedown
continue; continue;
} }
# Block-Level HTML <open> # Block-Level HTML <open>
if (preg_match('{^<(\w+)(?:[ ].*?)?>}', $pure_line, $matches)) if (preg_match('{^<(\w+)(?:[ ].*?)?>}', $pure_line, $matches))
@ -351,32 +361,32 @@ class Parsedown
'text' => $pure_line, 'text' => $pure_line,
'depth' => 0, 'depth' => 0,
); );
preg_match('{</'.$matches[1].'>\s*$}', $pure_line) and $element['closed'] = true; preg_match('{</'.$matches[1].'>\s*$}', $pure_line) and $element['closed'] = true;
continue; continue;
} }
} }
# Horizontal Rule # Horizontal Rule
if (preg_match('/^([-*_])([ ]{0,2}\1){2,}[ ]*$/', $pure_line)) if (preg_match('/^([-*_])([ ]{0,2}\1){2,}[ ]*$/', $pure_line))
{ {
$elements []= $element; $elements []= $element;
$element = array( $element = array(
'type' => 'hr', 'type' => 'hr',
); );
continue; continue;
} }
# List Item # List Item
if (preg_match('/^([ ]*)(\d+[.]|[*+-])[ ](.*)/', $line, $matches)) if (preg_match('/^([ ]*)(\d+[.]|[*+-])[ ](.*)/', $line, $matches))
{ {
$elements []= $element; $elements []= $element;
$element = array( $element = array(
'type' => 'li', 'type' => 'li',
'ordered' => isset($matches[2][1]), 'ordered' => isset($matches[2][1]),
@ -386,22 +396,22 @@ class Parsedown
preg_replace('/^[ ]{0,4}/', '', $matches[3]), preg_replace('/^[ ]{0,4}/', '', $matches[3]),
), ),
); );
continue; continue;
} }
# ~ # ~
paragraph: paragraph:
if ($element['type'] === 'p') if ($element['type'] === 'p')
{ {
if (isset($element['interrupted'])) if (isset($element['interrupted']))
{ {
$elements []= $element; $elements []= $element;
$element['text'] = $line; $element['text'] = $line;
unset($element['interrupted']); unset($element['interrupted']);
} }
else else
@ -412,215 +422,215 @@ class Parsedown
else else
{ {
$elements []= $element; $elements []= $element;
$element = array( $element = array(
'type' => 'p', 'type' => 'p',
'text' => $line, 'text' => $line,
); );
} }
} }
$elements []= $element; $elements []= $element;
array_shift($elements); array_shift($elements);
# #
# ~ # ~
# #
$markup = ''; $markup = '';
foreach ($elements as $index => $element) foreach ($elements as $index => $element)
{ {
switch ($element['type']) switch ($element['type'])
{ {
case 'li': case 'li':
if (isset($element['ordered'])) # first if (isset($element['ordered'])) # first
{ {
$list_type = $element['ordered'] ? 'ol' : 'ul'; $list_type = $element['ordered'] ? 'ol' : 'ul';
$markup .= '<'.$list_type.'>'."\n"; $markup .= '<'.$list_type.'>'."\n";
} }
if (isset($element['interrupted']) and ! isset($element['last'])) if (isset($element['interrupted']) and ! isset($element['last']))
{ {
$element['lines'] []= ''; $element['lines'] []= '';
} }
$text = $this->parse_block_elements($element['lines'], 'li'); $text = $this->parse_block_elements($element['lines'], 'li');
$markup .= '<li>'.$text.'</li>'."\n"; $markup .= '<li>'.$text.'</li>'."\n";
isset($element['last']) and $markup .= '</'.$list_type.'>'."\n"; isset($element['last']) and $markup .= '</'.$list_type.'>'."\n";
break; break;
case 'p': case 'p':
$text = $this->parse_inline_elements($element['text']); $text = $this->parse_inline_elements($element['text']);
$text = preg_replace('/[ ]{2}\n/', '<br />'."\n", $text); $text = preg_replace('/[ ]{2}\n/', '<br />'."\n", $text);
if ($context === 'li' and $index === 0) if ($context === 'li' and $index === 0)
{ {
if (isset($element['interrupted'])) if (isset($element['interrupted']))
{ {
$markup .= "\n".'<p>'.$text.'</p>'."\n"; $markup .= "\n".'<p>'.$text.'</p>'."\n";
} }
else else
{ {
$markup .= $text; $markup .= $text;
} }
} }
else else
{ {
$markup .= '<p>'.$text.'</p>'."\n"; $markup .= '<p>'.$text.'</p>'."\n";
} }
break; break;
case 'code': case 'code':
$text = htmlentities($element['text'], ENT_NOQUOTES); $text = htmlentities($element['text'], ENT_NOQUOTES);
strpos($text, "\x1A\\") !== FALSE and $text = strtr($text, $this->escape_sequence_map); strpos($text, "\x1A\\") !== FALSE and $text = strtr($text, $this->escape_sequence_map);
$markup .= '<pre><code>'.$text.'</code></pre>'."\n"; $markup .= '<pre><code>'.$text.'</code></pre>'."\n";
break; break;
case 'blockquote': case 'blockquote':
$text = $this->parse_block_elements($element['lines']); $text = $this->parse_block_elements($element['lines']);
$markup .= '<blockquote>'."\n".$text.'</blockquote>'."\n"; $markup .= '<blockquote>'."\n".$text.'</blockquote>'."\n";
break; break;
case 'h.': case 'h.':
$text = $this->parse_inline_elements($element['text']); $text = $this->parse_inline_elements($element['text']);
$markup .= '<h'.$element['level'].'>'.$text.'</h'.$element['level'].'>'."\n"; $markup .= '<h'.$element['level'].'>'.$text.'</h'.$element['level'].'>'."\n";
break; break;
case 'hr': case 'hr':
$markup .= '<hr />'."\n"; $markup .= '<hr />'."\n";
break; break;
default: default:
$markup .= $element['text']."\n"; $markup .= $element['text']."\n";
} }
} }
return $markup; return $markup;
} }
private function parse_inline_elements($text) private function parse_inline_elements($text)
{ {
$map = array(); $map = array();
$index = 0; $index = 0;
# Code Span # Code Span
if (strpos($text, '`') !== FALSE and preg_match_all('/`(.+?)`/', $text, $matches, PREG_SET_ORDER)) if (strpos($text, '`') !== FALSE and preg_match_all('/`(.+?)`/', $text, $matches, PREG_SET_ORDER))
{ {
foreach ($matches as $matches) foreach ($matches as $matches)
{ {
$element_text = $matches[1]; $element_text = $matches[1];
$element_text = htmlentities($element_text, ENT_NOQUOTES); $element_text = htmlentities($element_text, ENT_NOQUOTES);
# Decodes escape sequences. # Decodes escape sequences.
$this->escape_sequence_map $this->escape_sequence_map
and strpos($element_text, "\x1A") !== FALSE and strpos($element_text, "\x1A") !== FALSE
and $element_text = strtr($element_text, $this->escape_sequence_map); and $element_text = strtr($element_text, $this->escape_sequence_map);
# Composes element. # Composes element.
$element = '<code>'.$element_text.'</code>'; $element = '<code>'.$element_text.'</code>';
# Encodes element. # Encodes element.
$code = "\x1A".'$'.$index; $code = "\x1A".'$'.$index;
$text = str_replace($matches[0], $code, $text); $text = str_replace($matches[0], $code, $text);
$map[$code] = $element; $map[$code] = $element;
$index ++; $index ++;
} }
} }
# Inline Link / Image # Inline Link / Image
if (strpos($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 $matches) foreach ($matches as $matches)
{ {
$url = $matches[4]; $url = $matches[4];
strpos($url, '&') !== FALSE and $url = preg_replace('/&(?!#?\w+;)/', '&amp;', $url); strpos($url, '&') !== FALSE and $url = preg_replace('/&(?!#?\w+;)/', '&amp;', $url);
if ($matches[1]) # image if ($matches[1]) # image
{ {
$element = '<img alt="'.$matches[3].'" src="'.$url.'">'; $element = '<img alt="'.$matches[3].'" src="'.$url.'">';
} }
else else
{ {
$element_text = $this->parse_inline_elements($matches[3]); $element_text = $this->parse_inline_elements($matches[3]);
$element = '<a href="'.$url.'">'.$element_text.'</a>'; $element = '<a href="'.$url.'">'.$element_text.'</a>';
} }
# ~ # ~
$code = "\x1A".'$'.$index; $code = "\x1A".'$'.$index;
$text = str_replace($matches[0], $code, $text); $text = str_replace($matches[0], $code, $text);
$map[$code] = $element; $map[$code] = $element;
$index ++; $index ++;
} }
} }
# Reference(d) Link / Image # Reference(d) Link / Image
if ($this->reference_map and strpos($text, '[') !== FALSE and preg_match_all('/(!?)\[(.+?)\](?:\n?[ ]?\[(.*?)\])?/ms', $text, $matches, PREG_SET_ORDER)) if ($this->reference_map and strpos($text, '[') !== FALSE and preg_match_all('/(!?)\[(.+?)\](?:\n?[ ]?\[(.*?)\])?/ms', $text, $matches, PREG_SET_ORDER))
{ {
foreach ($matches as $matches) foreach ($matches as $matches)
{ {
$link_definition = isset($matches[3]) && $matches[3] $link_definition = isset($matches[3]) && $matches[3]
? $matches[3] ? $matches[3]
: $matches[2]; # implicit : $matches[2]; # implicit
$link_definition = strtolower($link_definition); $link_definition = strtolower($link_definition);
if (isset($this->reference_map[$link_definition])) if (isset($this->reference_map[$link_definition]))
{ {
$url = $this->reference_map[$link_definition]; $url = $this->reference_map[$link_definition];
strpos($url, '&') !== FALSE and $url = preg_replace('/&(?!#?\w+;)/', '&amp;', $url); strpos($url, '&') !== FALSE and $url = preg_replace('/&(?!#?\w+;)/', '&amp;', $url);
if ($matches[1]) # image if ($matches[1]) # image
{ {
$element = '<img alt="'.$matches[2].'" src="'.$url.'">'; $element = '<img alt="'.$matches[2].'" src="'.$url.'">';
} }
else # anchor else # anchor
{ {
$element_text = $this->parse_inline_elements($matches[2]); $element_text = $this->parse_inline_elements($matches[2]);
$element = '<a href="'.$url.'">'.$element_text.'</a>'; $element = '<a href="'.$url.'">'.$element_text.'</a>';
} }
# ~ # ~
$code = "\x1A".'$'.$index; $code = "\x1A".'$'.$index;
@ -632,54 +642,54 @@ class Parsedown
} }
} }
} }
# Automatic Links # Automatic Links
if (strpos($text, '<') !== FALSE and preg_match_all('/<((https?|ftp|dict):[^\^\s]+?)>/i', $text, $matches, PREG_SET_ORDER)) if (strpos($text, '<') !== FALSE and preg_match_all('/<((https?|ftp|dict):[^\^\s]+?)>/i', $text, $matches, PREG_SET_ORDER))
{ {
foreach ($matches as $matches) foreach ($matches as $matches)
{ {
$url = $matches[1]; $url = $matches[1];
strpos($url, '&') !== FALSE and $url = preg_replace('/&(?!#?\w+;)/', '&amp;', $url); strpos($url, '&') !== FALSE and $url = preg_replace('/&(?!#?\w+;)/', '&amp;', $url);
$element = '<a href=":href">:text</a>'; $element = '<a href=":href">:text</a>';
$element = str_replace(':text', $url, $element); $element = str_replace(':text', $url, $element);
$element = str_replace(':href', $url, $element); $element = str_replace(':href', $url, $element);
# ~ # ~
$code = "\x1A".'$'.$index; $code = "\x1A".'$'.$index;
$text = str_replace($matches[0], $code, $text); $text = str_replace($matches[0], $code, $text);
$map[$code] = $element; $map[$code] = $element;
$index ++; $index ++;
} }
} }
# ~ # ~
strpos($text, '&') !== FALSE and $text = preg_replace('/&(?!#?\w+;)/', '&amp;', $text); strpos($text, '&') !== FALSE and $text = preg_replace('/&(?!#?\w+;)/', '&amp;', $text);
strpos($text, '<') !== FALSE and $text = preg_replace('/<(?!\/?\w.*?>)/', '&lt;', $text); strpos($text, '<') !== FALSE and $text = preg_replace('/<(?!\/?\w.*?>)/', '&lt;', $text);
# ~ # ~
if (strpos($text, '_') !== FALSE) if (strpos($text, '_') !== FALSE)
{ {
$text = preg_replace('/__(?=\S)(.+?)(?<=\S)__/', '<strong>$1</strong>', $text); $text = preg_replace('/__(?=\S)(.+?)(?<=\S)__/', '<strong>$1</strong>', $text);
$text = preg_replace('/_(?=\S)(.+?)(?<=\S)_/', '<em>$1</em>', $text); $text = preg_replace('/_(?=\S)(.+?)(?<=\S)_/', '<em>$1</em>', $text);
} }
if (strpos($text, '*') !== FALSE) if (strpos($text, '*') !== FALSE)
{ {
$text = preg_replace('/\*\*(?=\S)(.+?)(?<=\S)\*\*/', '<strong>$1</strong>', $text); $text = preg_replace('/\*\*(?=\S)(.+?)(?<=\S)\*\*/', '<strong>$1</strong>', $text);
$text = preg_replace('/\*(?=\S)(.+?)(?<=\S)\*/', '<em>$1</em>', $text); $text = preg_replace('/\*(?=\S)(.+?)(?<=\S)\*/', '<em>$1</em>', $text);
} }
$text = strtr($text, $map); $text = strtr($text, $map);
return $text; return $text;
} }
} }

View File

@ -1,6 +1,6 @@
<h1>h1</h1> <h1>h1</h1>
<h2>h2</h2> <h2>h2</h2>
<h3>h3</h3> <h3>h3</h3>
<h4>h4</h4> <h4>h4</h4>
<h5>h5</h5> <h5>h5</h5>
<h6>h6</h6> <h6>h6</h6>

View File

@ -1,11 +1,11 @@
# h1 # # h1 #
## h2 ## ## h2 ##
### h3 ### ### h3 ###
#### h4 #### #### h4 ####
##### h5 ##### ##### h5 #####
###### h6 ###### ###### h6 ######

View File

@ -1,8 +1,4 @@
<p>Here's a regular blockquote:</p> <p>Here's a blockquote:</p>
<blockquote>
<p>blockquote</p>
</blockquote>
<p>Here's one with no space after the ">":</p>
<blockquote> <blockquote>
<p>blockquote</p> <p>blockquote</p>
</blockquote> </blockquote>

View File

@ -1,11 +1,7 @@
Here's a regular blockquote: Here's a blockquote:
> blockquote > blockquote
Here's one with no space after the ">":
>blockquote
Here's one on multiple lines: Here's one on multiple lines:
> line 1 > line 1

View File

@ -1,16 +1,16 @@
<p>Here's one with multiple paragraphs:</p> <p>Here's one with multiple paragraphs:</p>
<blockquote> <blockquote>
<p>This is line one.</p> <p>This is line one.</p>
<p>This is line two.</p> <p>This is line two.</p>
</blockquote> </blockquote>
<p>Here's one with multiple types of blocks:</p> <p>Here's one with multiple types of blocks:</p>
<blockquote> <blockquote>
<p>This is a quoted paragraph.</p> <p>This is a quoted paragraph.</p>
<ul> <ul>
<li>This is a list item of a quoted list.</li> <li>This is a list item of a quoted list.</li>
<li>This is another list item.</li> <li>This is another list item.</li>
</ul> </ul>
<blockquote> <blockquote>
<p>This is a nested quote block.</p> <p>This is a nested quote block.</p>
</blockquote> </blockquote>
</blockquote> </blockquote>

View File

@ -1,14 +1,14 @@
Here's one with multiple paragraphs: Here's one with multiple paragraphs:
> This is line one. > This is line one.
> >
> This is line two. > This is line two.
Here's one with multiple types of blocks: Here's one with multiple types of blocks:
> This is a quoted paragraph. > This is a quoted paragraph.
> >
> - This is a list item of a quoted list. > - This is a list item of a quoted list.
> - This is another list item. > - This is another list item.
> >
> > This is a nested quote block. > > This is a nested quote block.

View File

@ -0,0 +1,11 @@
<p>Here's a lazy blockquote:</p>
<blockquote>
<p>line
line</p>
</blockquote>
<p>Here's one with multiple lines:</p>
<blockquote>
<p>line
line
line</p>
</blockquote>

View File

@ -0,0 +1,10 @@
Here's a lazy blockquote:
> line
line
Here's one with multiple lines:
> line
line
line

View File

@ -0,0 +1,12 @@
<p>Here's a blockquote with no space after the ">":</p>
<blockquote>
<p>blockquote</p>
</blockquote>
<p>Here's a blockquote with leading space:</p>
<blockquote>
<p>blockquote</p>
</blockquote>
<p>Here's a blockquote on the next line:</p>
<blockquote>
<p>blockquote</p>
</blockquote>

View File

@ -0,0 +1,10 @@
Here's a blockquote with no space after the ">":
>blockquote
Here's a blockquote with leading space:
> blockquote
Here's a blockquote on the next line:
> blockquote

View File

@ -2,7 +2,4 @@
<pre><code>&lt;?php <pre><code>&lt;?php
$message = 'Hello World!'; $message = 'Hello World!';
echo $message;</code></pre> echo $message;</code></pre>
<p>Here's one that holds a list:</p>
<pre><code>- list item
- another list item</code></pre>

View File

@ -3,10 +3,4 @@ Here's a code block:
<?php <?php
$message = 'Hello World!'; $message = 'Hello World!';
echo $message; echo $message;
Here's one that holds a list:
- list item
- another list item

View File

@ -1,24 +0,0 @@
<p>Here's a regular list:</p>
<ul>
<li>list item</li>
<li>another list item</li>
<li>3rd list item</li>
</ul>
<p>Here's one with white space around items:</p>
<ul>
<li>list item </li>
<li>another list item </li>
</ul>
<p>Here's one with too much space before items:</p>
<pre><code>- list item
- another list item</code></pre>
<p>Here's one with no space after markers:</p>
<p>-list item
-another list item</p>
<p>Here's one where items contain line breaks:</p>
<ul>
<li>list
item</li>
<li>another
list item</li>
</ul>

View File

@ -1,27 +0,0 @@
Here's a regular list:
- list item
- another list item
- 3rd list item
Here's one with white space around items:
- list item
- another list item
Here's one with too much space before items:
- list item
- another list item
Here's one with no space after markers:
-list item
-another list item
Here's one where items contain line breaks:
- list
item
- another
list item

View File

@ -1,7 +1,7 @@
<p>Here's <em>an emphasis</em>.</p> <p>Here's <em>an emphasis</em>.</p>
<p>A short emphasis <em>a</em> <em>b</em> .</p> <p>A short one <em>a</em> <em>b</em> .</p>
<p>Here's <strong>a strong one</strong>. </p> <p>Here's <strong>a strong one</strong>. </p>
<p>Here's <em>an emphasis that uses underscores</em>. </p> <p>Here's <em>one that uses underscores</em>. </p>
<p>Here's <strong>a strong emphasis that uses underscores</strong>.</p> <p>Here's <strong>a strong one that uses underscores</strong>.</p>
<p>This is not _ an emphasis _ neither is * this * neither is _ this_ neither is _this _.</p> <p>This is not _ one _ neither is * this * neither is _ this_ neither is _this _.</p>
<p>Empty emphasis ** is not __ an emphasis.</p> <p>An empty emphasis ** is not __ an emphasis.</p>

View File

@ -1,13 +1,13 @@
Here's *an emphasis*. Here's *an emphasis*.
A short emphasis _a_ *b* . A short one _a_ *b* .
Here's **a strong one**. Here's **a strong one**.
Here's _an emphasis that uses underscores_. Here's _one that uses underscores_.
Here's __a strong emphasis that uses underscores__. Here's __a strong one that uses underscores__.
This is not _ an emphasis _ neither is * this * neither is _ this_ neither is _this _. This is not _ one _ neither is * this * neither is _ this_ neither is _this _.
Empty emphasis ** is not __ an emphasis. An empty emphasis ** is not __ an emphasis.

View File

@ -1,6 +1,2 @@
<p>Here's an <em>emphasis</em> and here's an escaped *emphasis*. Here are also an escaped `code span`, escaped [inline link](http://example.com).</p> <p>Here's an <em>emphasis</em> and here's an escaped *emphasis*. Here are also an escaped `code span`, an escaped [inline link](http://example.com) and an escaped <code>\*emphasis\*</code> inside of a code span.</p>
<p>Here's <code>an escaped \*emphasis\* inside of a code span</code>.</p> <pre><code>An escaped \*emphasis\* inside of a code block.</code></pre>
<p>Here's one inside of a code block:</p>
<pre><code>An escaped \*emphasis\*.</code></pre>
<p>Finally, an escaped reference:</p>
<p>[1]: http://example.com</p>

View File

@ -1,11 +1,3 @@
Here's an *emphasis* and here's an escaped \*emphasis\*. Here are also an escaped \`code span\`, escaped \[inline link](http://example.com). Here's an *emphasis* and here's an escaped \*emphasis\*. Here are also an escaped \`code span\`, an escaped \[inline link](http://example.com) and an escaped `\*emphasis\*` inside of a code span.
Here's `an escaped \*emphasis\* inside of a code span`. An escaped \*emphasis\* inside of a code block.
Here's one inside of a code block:
An escaped \*emphasis\*.
Finally, an escaped reference:
\[1]: http://example.com

View File

@ -2,15 +2,9 @@
<hr /> <hr />
<hr /> <hr />
<hr /> <hr />
<hr />
<pre><code>---</code></pre>
<hr />
<hr />
<hr />
<hr />
<pre><code>- - -</code></pre>
<p>Asterisks:</p> <p>Asterisks:</p>
<hr /> <hr />
<p>Underscores:</p> <p>Underscores:</p>
<hr /> <hr />
<p>Based on <a href="http://daringfireball.net/projects/downloads/MarkdownTest_1.0.zip">the original</a> test suite.</p> <p>On the next line:</p>
<hr />

View File

@ -2,24 +2,10 @@ Dashes:
--- ---
---
---
---
---
- - - - - -
- - -
- - -
- - - - - -
- - -
Asterisks: Asterisks:
*** ***
@ -28,4 +14,5 @@ Underscores:
___ ___
Based on [the original](http://daringfireball.net/projects/downloads/MarkdownTest_1.0.zip) test suite. On the next line:
___

View File

@ -1,11 +1,11 @@
<p>Self-closing tag:</p> <p>A self-closing tag:</p>
<hr/> <hr/>
<p>Self-closing tag with attributes:</p> <p>One with attributes:</p>
<hr style="background: #eaa" /> <hr style="background: #eaa" />
<p>Bare element:</p> <p>A bare element:</p>
<div>content</div> <div>content</div>
<p>Element with attributes:</p> <p>One with attributes:</p>
<a href="http://parsedown.org">link</a> <a href="http://example.com">link</a>
<p>Nested elements:</p> <p>Nested elements:</p>
<div> <div>
parent parent

View File

@ -1,18 +1,18 @@
Self-closing tag: A self-closing tag:
<hr/> <hr/>
Self-closing tag with attributes: One with attributes:
<hr style="background: #eaa" /> <hr style="background: #eaa" />
Bare element: A bare element:
<div>content</div> <div>content</div>
Element with attributes: One with attributes:
<a href="http://parsedown.org">link</a> <a href="http://example.com">link</a>
Nested elements: Nested elements:

View File

@ -1,2 +1,2 @@
<p>Here's a <a href="http://parsedown.org">link</a>.</p> <p>Here's a <a href="http://example.com">link</a>.</p>
<p>Here's an image link: <a href="http://daringfireball.net/projects/markdown/"><img alt="MD Logo" src="http://parsedown.org/md.png"></a>.</p> <p>Here's one that is based on an image: <a href="http://daringfireball.net/projects/markdown/"><img alt="MD Logo" src="http://parsedown.org/md.png"></a>.</p>

View File

@ -1,3 +1,3 @@
Here's a [link](http://parsedown.org). Here's a [link](http://example.com).
Here's an image link: [![MD Logo](http://parsedown.org/md.png)](http://daringfireball.net/projects/markdown/). Here's one that is based on an image: [![MD Logo](http://parsedown.org/md.png)](http://daringfireball.net/projects/markdown/).

View File

@ -1,4 +0,0 @@
<blockquote>
<p>line 1
line 2</p>
</blockquote>

View File

@ -1,2 +0,0 @@
> line 1
line 2

5
tests/data/list.html Normal file
View File

@ -0,0 +1,5 @@
<p>Here's a list:</p>
<ul>
<li>li</li>
<li>li</li>
</ul>

4
tests/data/list.md Normal file
View File

@ -0,0 +1,4 @@
Here's a list:
- li
- li

View File

@ -6,5 +6,4 @@ Here's a compound list:
- This is another list item. - This is another list item.
> This is a quote block that belongs to it. > This is a quote block that belongs to it.

View File

@ -1,4 +1,4 @@
<ul> <ul>
<li>li <li>li
more text</li> more text</li>
</ul> </ul>

View File

@ -1,2 +1,2 @@
- li - li
more text more text

View File

@ -2,7 +2,6 @@
<ol> <ol>
<li>one</li> <li>one</li>
<li>two</li> <li>two</li>
<li>three</li>
</ol> </ol>
<p>Here's one with repeating numbers:</p> <p>Here's one with repeating numbers:</p>
<ol> <ol>
@ -12,5 +11,4 @@
<p>Here's one with large numbers:</p> <p>Here's one with large numbers:</p>
<ol> <ol>
<li>one</li> <li>one</li>
<li>two</li>
</ol> </ol>

View File

@ -2,7 +2,6 @@ Here's a regular ordered list:
1. one 1. one
2. two 2. two
3. three
Here's one with repeating numbers: Here's one with repeating numbers:
@ -11,6 +10,4 @@ Here's one with repeating numbers:
Here's one with large numbers: Here's one with large numbers:
123. one 123. one
123. two

View File

@ -1,16 +1,16 @@
<p>Here's a sparse list:</p> <p>Here's a sparse list:</p>
<ul> <ul>
<li> <li>
<p>list item</p> <p>list item</p>
</li> </li>
<li>another list item</li> <li>another list item</li>
</ul> </ul>
<p>Here's one with an indented list item:</p> <p>Here's one with an indented list item:</p>
<ul> <ul>
<li> <li>
<p>li</p> <p>li</p>
<ul> <ul>
<li>li</li> <li>li</li>
</ul> </ul>
</li> </li>
</ul> </ul>

View File

@ -1,11 +1,11 @@
Here's a sparse list: Here's a sparse list:
- list item - list item
- another list item - another list item
Here's one with an indented list item: Here's one with an indented list item:
- li - li
- li - li

View File

@ -0,0 +1,11 @@
<p>Here's an unordered list:</p>
<ul>
<li>li</li>
<li>li</li>
</ul>
<p>Here's one with mixed markers:</p>
<ul>
<li>li</li>
<li>li</li>
<li>li</li>
</ul>

View File

@ -0,0 +1,10 @@
Here's an unordered list:
- li
- li
Here's one with mixed markers:
- li
+ li
* li

View File

@ -0,0 +1,5 @@
<p>Here's one with white space around items:</p>
<ul>
<li>li </li>
<li>li </li>
</ul>

View File

@ -0,0 +1,4 @@
Here's one with white space around items:
- li
- li

View File

@ -1,4 +0,0 @@
<p>Here's a paragraph.</p>
<blockquote>
<p>a block quote that belongs to it.</p>
</blockquote>

View File

@ -1,2 +0,0 @@
Here's a paragraph.
> a block quote that belongs to it.

View File

@ -1,5 +0,0 @@
<p>Here's a list that's "inside" a paragraph:</p>
<ul>
<li>list item</li>
<li>another list item</li>
</ul>

View File

@ -1,4 +0,0 @@
Here's a list that's "inside" a paragraph:
- list item
- another list item

View File

@ -1,20 +0,0 @@
<p>Here's a regular quote block:</p>
<blockquote>
<p>Some quoted text.
Here goes some more.</p>
</blockquote>
<p>Here's one with space before lines:</p>
<blockquote>
<p>Some quoted text.
Here goes some more.</p>
</blockquote>
<p>Here's one with no space after >:</p>
<blockquote>
<p>Some quoted text.
Here goes some more.</p>
</blockquote>
<p>Here's one with no > on the second line:</p>
<blockquote>
<p>Some quoted text.
Here goes some more.</p>
</blockquote>

View File

@ -1,19 +0,0 @@
Here's a regular quote block:
> Some quoted text.
> Here goes some more.
Here's one with space before lines:
> Some quoted text.
> Here goes some more.
Here's one with no space after >:
>Some quoted text.
>Here goes some more.
Here's one with no > on the second line:
> Some quoted text.
Here goes some more.

View File

@ -1,11 +1,7 @@
<p>Here's a <a href="http://parsedown.org">reference link</a>.</p> <p>Here's a <a href="http://example.com">reference link</a>.</p>
<p>Here's <a href="http://parsedown.org">one</a> on the next line.</p> <p>Here's <a href="http://example.com">one</a> with a semantic name.</p>
<p>Here's <a href="http://parsedown.org/tests/">one</a> with a different URL.</p> <p>Here's <a href="http://example.com">one</a> with an upper case label definition.</p>
<p>Here's <a href="http://parsedown.org">one</a> with a semantic name.</p> <p>Here's <a href="http://example.com">one</a> with definition name on the next line.</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 [one][404] with no definition.</p>
<p>Here's an image: <img alt="Markdown Logo" src="/md.png"></p> <p>Here's a <a href="http://example.com">multiline
<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> one</a> defined on 2 lines.</p>

View File

@ -1,32 +1,19 @@
Here's a [reference link][1]. Here's a [reference link][1].
[1]: http://parsedown.org [1]: http://example.com
Here's [one][2] on the next line.
[2]: http://parsedown.org
Here's [one][3] with a different URL.
[3]: http://parsedown.org/tests/
Here's [one][website] with a semantic name. Here's [one][website] with a semantic name.
[website]: http://parsedown.org [website]: http://example.com
Here's [one][case] with an upper case label definition.
[CASE]: http://example.com
Here's [one] Here's [one]
[website] with definition name on the next line. [website] with definition name on the next line.
Here's [one][404] with no definition. Here's [one][404] with no definition.
Here's an image: ![Markdown Logo][image]
[image]: /md.png
Here's an [implicit one].
[implicit one]: http://google.com
Here's an [implicit one][] with an empty link definition.
Here's a [multiline Here's a [multiline
one][website] defined on 2 lines. one][website] defined on 2 lines.

View File

@ -0,0 +1 @@
<p>Here's an image: <img alt="Markdown Logo" src="/md.png"></p>

View File

@ -0,0 +1,3 @@
Here's an image: ![Markdown Logo][image]
[image]: /md.png

View File

@ -0,0 +1,2 @@
<p>Here's an <a href="http://example.com">implicit</a> reference link.</p>
<p>Here's an <a href="http://example.com">implicit</a> one with an empty link definition.</p>

View File

@ -0,0 +1,5 @@
Here's an [implicit] reference link.
[implicit]: http://example.com
Here's an [implicit][] one with an empty link definition.

View File

@ -0,0 +1 @@
<p>Here's a <a href="http://example.com">reference link</a> with a definition on the next line.</p>

View File

@ -0,0 +1,2 @@
Here's a [reference link][2] with a definition on the next line.
[2]: http://example.com

View File

@ -1,6 +0,0 @@
<h1>Heading 1</h1>
<h2>Heading 2</h2>
<h2>Block Heading</h2>
<p>This is the rest of the block.</p>
<h1>Single "="</h1>
<h2>Single "-"</h2>

View File

@ -1,16 +0,0 @@
Heading 1
=========
Heading 2
---------
Block Heading
-------------
This is the rest of the block.
Single "="
=
Single "-"
-

View File

@ -2,8 +2,6 @@
<p>AT&amp;T is another way to write it.</p> <p>AT&amp;T is another way to write it.</p>
<p>This &amp; that.</p> <p>This &amp; that.</p>
<p>4 &lt; 5 and 6 > 5.</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 autolink <a href="http://example.com/autolink?a=1&amp;b=2">http://example.com/autolink?a=1&amp;b=2</a></p>
<p>Here's an inline <a href="/script?foo=1&amp;bar=2">link</a>.</p> <p>Here's an inline <a href="/script?a=1&amp;b=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> <p>Here's a reference <a href="http://example.com/?a=1&amp;b=2">link</a> with an ampersand in the URL.</p>
<hr />
<p>Based on <a href="http://daringfireball.net/projects/downloads/MarkdownTest_1.0.zip">the original</a> test suite.</p>

View File

@ -1,19 +1,15 @@
AT&T has an ampersand in their name. AT&T has an ampersand in their name.
AT&T is another way to write it. AT&amp;T is another way to write it.
This & that. This & that.
4 < 5 and 6 > 5. 4 < 5 and 6 > 5.
Here's a [link] [1] with an ampersand in the URL. Here's an autolink <http://example.com/autolink?a=1&b=2>
Here's an inline [link](/script?foo=1&bar=2). Here's an inline [link](/script?a=1&b=2).
[1]: http://example.com/?foo=1&bar=2 Here's a reference [link] [1] with an ampersand in the URL.
<http://example.com/autolink?a=1&b=2> [1]: http://example.com/?a=1&b=2
---
Based on [the original](http://daringfireball.net/projects/downloads/MarkdownTest_1.0.zip) test suite.

View File

@ -1,20 +0,0 @@
<p>Here's a regular unordered list:</p>
<ul>
<li>list item</li>
<li>another list item</li>
<li>3rd list item</li>
</ul>
<p>Here's one with a variety of markers:</p>
<ul>
<li>hyphen</li>
<li>plus</li>
<li>asterisk</li>
</ul>
<p>Here's one with white space around items:</p>
<ul>
<li>list item </li>
<li>another list item </li>
</ul>
<p>Here's one with no space after markers:</p>
<p>-list item
-another list item</p>

View File

@ -1,21 +0,0 @@
Here's a regular unordered list:
- list item
- another list item
- 3rd list item
Here's one with a variety of markers:
- hyphen
+ plus
* asterisk
Here's one with white space around items:
- list item
- another list item
Here's one with no space after markers:
-list item
-another list item

View File

@ -0,0 +1 @@
<pre><code>This text starts with a line that consists of 4 spaces and it ends with one. This is a code block to make sure that leading spaces don't get trimmed.</code></pre>

5
tests/data/whitespace.md Normal file
View File

@ -0,0 +1,5 @@
This text starts with a line that consists of 4 spaces and it ends with one. This is a code block to make sure that leading spaces don't get trimmed.