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

1331 lines
38 KiB
PHP
Raw Normal View History

2013-11-09 01:40:00 +04:00
<?php
2013-07-11 00:22:16 +04:00
2013-11-09 01:40:00 +04:00
#
#
2013-07-11 00:22:16 +04:00
# Parsedown
# http://parsedown.org
2013-11-09 01:40:00 +04:00
#
# (c) Emanuil Rusev
2013-07-11 00:22:16 +04:00
# http://erusev.com
2013-11-09 01:40:00 +04:00
#
2014-01-24 03:28:03 +04:00
# For the full license information, view the LICENSE file that was distributed
# with this source code.
2013-11-09 01:40:00 +04:00
#
#
2013-07-11 00:22:16 +04:00
class Parsedown
{
2014-02-21 04:23:17 +04:00
#
# Philosophy
2014-02-21 04:23:17 +04:00
#
# Markdown is intended to be easy-to-read by humans - those of us who read
# line by line, left to right, top to bottom. In order to take advantage of
# this, Parsedown tries to read in a similar way. It breaks texts into
# lines, it iterates through them and it looks at how they start and relate
# to each other.
#
# Setters
#
2014-01-25 20:47:44 +04:00
# Enables GFM line breaks.
2014-01-24 03:28:03 +04:00
2014-02-24 03:38:58 +04:00
function setBreaksEnabled($breaksEnabled)
{
2014-02-24 03:38:58 +04:00
$this->breaksEnabled = $breaksEnabled;
return $this;
}
2014-02-24 03:38:58 +04:00
private $breaksEnabled = false;
2013-11-09 01:40:00 +04:00
2014-01-29 04:14:59 +04:00
#
# Methods
#
2013-11-09 01:40:00 +04:00
function parse($text)
{
2014-01-24 02:41:45 +04:00
# standardize line breaks
$text = str_replace("\r\n", "\n", $text);
$text = str_replace("\r", "\n", $text);
2013-11-09 01:40:00 +04:00
2014-01-24 02:41:45 +04:00
# replace tabs with spaces
$text = str_replace("\t", ' ', $text);
2013-11-09 01:40:00 +04:00
2014-01-24 02:41:45 +04:00
# remove surrounding line breaks
$text = trim($text, "\n");
2013-11-09 01:40:00 +04:00
2014-01-24 02:41:45 +04:00
# split text into lines
$lines = explode("\n", $text);
2013-11-09 01:40:00 +04:00
# iterate through lines to identify blocks
2014-02-24 03:38:58 +04:00
$blocks = $this->findBlocks($lines);
2013-11-09 01:40:00 +04:00
# iterate through blocks to build markup
$markup = $this->compile($blocks);
2013-11-09 01:40:00 +04:00
# trim line breaks
$markup = trim($markup, "\n");
return $markup;
}
2013-11-09 01:40:00 +04:00
#
2014-01-25 20:47:44 +04:00
# Private
2013-11-09 01:40:00 +04:00
2014-02-24 03:38:58 +04:00
private function findBlocks(array $lines, $blockContext = null)
{
$block = null;
$context = null;
2014-02-24 03:38:58 +04:00
$contextData = null;
foreach ($lines as $line)
{
2014-02-24 03:38:58 +04:00
$indentedLine = $line;
2013-11-09 01:40:00 +04:00
$indentation = 0;
while(isset($line[$indentation]) and $line[$indentation] === ' ')
{
$indentation++;
}
if ($indentation > 0)
{
$line = ltrim($line);
}
2013-11-09 01:40:00 +04:00
# ~
2013-11-09 01:40:00 +04:00
switch ($context)
{
case null:
2013-11-09 01:40:00 +04:00
2014-02-24 03:38:58 +04:00
$contextData = null;
if ($line === '')
{
continue 2;
}
break;
2013-11-09 01:40:00 +04:00
# ~~~ javascript
# var message = 'Hello!';
case 'fenced code':
if ($line === '')
{
$block['content'][0]['content'] .= "\n";
2013-11-09 01:40:00 +04:00
continue 2;
}
2013-11-09 01:40:00 +04:00
2014-02-24 03:38:58 +04:00
if (preg_match('/^[ ]*'.$contextData['marker'].'{3,}[ ]*$/', $line))
{
$context = null;
}
else
{
if ($block['content'][0]['content'])
{
$block['content'][0]['content'] .= "\n";
}
2014-02-24 03:38:58 +04:00
$string = htmlspecialchars($indentedLine, ENT_NOQUOTES, 'UTF-8');
$block['content'][0]['content'] .= $string;
}
continue 2;
case 'markup':
2014-02-24 03:38:58 +04:00
if (stripos($line, $contextData['start']) !== false) # opening tag
{
2014-02-24 03:38:58 +04:00
$contextData['depth']++;
}
2014-02-24 03:38:58 +04:00
if (stripos($line, $contextData['end']) !== false) # closing tag
{
2014-02-24 03:38:58 +04:00
if ($contextData['depth'] > 0)
{
2014-02-24 03:38:58 +04:00
$contextData['depth']--;
}
else
{
$context = null;
}
}
2014-02-24 03:38:58 +04:00
$block['content'] .= "\n".$indentedLine;
continue 2;
case 'li':
if ($line === '')
{
2014-02-24 03:38:58 +04:00
$contextData['interrupted'] = true;
continue 2;
}
2014-02-24 03:38:58 +04:00
if ($contextData['indentation'] === $indentation and preg_match('/^'.$contextData['marker'].'[ ]+(.*)/', $line, $matches))
{
2014-02-24 03:38:58 +04:00
if (isset($contextData['interrupted']))
{
2014-02-24 03:38:58 +04:00
$nestedBlock['content'] []= '';
2014-02-24 03:38:58 +04:00
unset($contextData['interrupted']);
}
2014-02-24 03:38:58 +04:00
unset($nestedBlock);
2014-02-24 03:38:58 +04:00
$nestedBlock = array(
'name' => 'li',
2014-02-24 02:55:34 +04:00
'content type' => 'lines',
'content' => array(
$matches[1],
),
);
2014-02-24 03:38:58 +04:00
$block['content'] []= & $nestedBlock;
continue 2;
}
2013-11-09 01:40:00 +04:00
2014-02-24 03:38:58 +04:00
if (empty($contextData['interrupted']))
{
$value = $line;
2014-02-24 03:38:58 +04:00
if ($indentation > $contextData['baseline'])
{
2014-02-24 03:38:58 +04:00
$value = str_repeat(' ', $indentation - $contextData['baseline']) . $value;
}
2013-11-09 01:40:00 +04:00
2014-02-24 03:38:58 +04:00
$nestedBlock['content'] []= $value;
continue 2;
}
if ($indentation > 0)
{
2014-02-24 03:38:58 +04:00
$nestedBlock['content'] []= '';
$value = $line;
2014-02-24 03:38:58 +04:00
if ($indentation > $contextData['baseline'])
{
2014-02-24 03:38:58 +04:00
$value = str_repeat(' ', $indentation - $contextData['baseline']) . $value;
}
2013-11-09 01:40:00 +04:00
2014-02-24 03:38:58 +04:00
$nestedBlock['content'] []= $value;
2013-11-09 01:40:00 +04:00
2014-02-24 03:38:58 +04:00
unset($contextData['interrupted']);
continue 2;
}
2013-11-09 01:40:00 +04:00
$context = null;
break;
2013-11-09 01:40:00 +04:00
case 'quote':
2013-11-09 01:40:00 +04:00
if ($line === '')
{
2014-02-24 03:38:58 +04:00
$contextData['interrupted'] = true;
continue 2;
}
2013-11-10 00:23:56 +04:00
if (preg_match('/^>[ ]?(.*)/', $line, $matches))
{
$block['content'] []= $matches[1];
continue 2;
}
2014-02-24 03:38:58 +04:00
if (empty($contextData['interrupted']))
{
$block['content'] []= $line;
2014-01-17 01:43:12 +04:00
continue 2;
}
$context = null;
2013-11-09 01:40:00 +04:00
break;
case 'code':
if ($line === '')
{
2014-02-24 03:38:58 +04:00
$contextData['interrupted'] = true;
continue 2;
}
if ($indentation >= 4)
{
2014-02-24 03:38:58 +04:00
if (isset($contextData['interrupted']))
{
$block['content'][0]['content'] .= "\n";
2014-02-24 03:38:58 +04:00
unset($contextData['interrupted']);
}
$block['content'][0]['content'] .= "\n";
$string = htmlspecialchars($line, ENT_NOQUOTES, 'UTF-8');
$string = str_repeat(' ', $indentation - 4) . $string;
$block['content'][0]['content'] .= $string;
continue 2;
}
2013-11-09 01:40:00 +04:00
$context = null;
break;
2013-11-09 01:40:00 +04:00
2014-02-23 20:55:34 +04:00
case 'table':
if ($line === '')
{
$context = null;
continue 2;
}
if ($line[0] === '|')
{
2014-02-24 03:38:58 +04:00
$nestedBlocks = array();
2014-02-23 20:55:34 +04:00
$substring = preg_replace('/^[|][ ]*/', '', $line);
$substring = preg_replace('/[|]?[ ]*$/', '', $substring);
$parts = explode('|', $substring);
foreach ($parts as $index => $part)
{
$substring = trim($part);
2014-02-24 03:38:58 +04:00
$nestedBlock = array(
2014-02-23 20:55:34 +04:00
'name' => 'td',
2014-02-24 02:55:34 +04:00
'content type' => 'line',
2014-02-23 20:55:34 +04:00
'content' => $substring,
);
2014-02-24 03:38:58 +04:00
if (isset($contextData['alignments'][$index]))
2014-02-23 20:55:34 +04:00
{
2014-02-24 03:38:58 +04:00
$nestedBlock['attributes'] = array(
'align' => $contextData['alignments'][$index],
2014-02-23 20:55:34 +04:00
);
}
2014-02-24 03:38:58 +04:00
$nestedBlocks []= $nestedBlock;
2014-02-23 20:55:34 +04:00
}
2014-02-24 03:38:58 +04:00
$nestedBlock = array(
2014-02-23 20:55:34 +04:00
'name' => 'tr',
'content type' => 'blocks',
2014-02-24 03:38:58 +04:00
'content' => $nestedBlocks,
2014-02-23 20:55:34 +04:00
);
2014-02-24 03:38:58 +04:00
$block['content'][1]['content'] []= $nestedBlock;
2014-02-23 20:55:34 +04:00
continue 2;
}
$context = null;
break;
case 'paragraph':
if ($line === '')
{
$block['name'] = 'p'; # dense li
$context = null;
2013-11-09 01:40:00 +04:00
continue 2;
}
if ($line[0] === '=' and chop($line, '=') === '')
{
$block['name'] = 'h1';
2013-11-09 01:40:00 +04:00
$context = null;
2014-01-17 02:36:11 +04:00
continue 2;
}
2013-11-09 01:40:00 +04:00
if ($line[0] === '-' and chop($line, '-') === '')
{
$block['name'] = 'h2';
$context = null;
2013-11-09 01:40:00 +04:00
continue 2;
}
2013-11-09 01:40:00 +04:00
2014-02-23 20:55:34 +04:00
if ($line[0] === '|' and $block['content'][0] === '|' and chop($line, ' -:|') === '')
{
$values = array();
$substring = trim($line, ' |');
$parts = explode('|', $substring);
foreach ($parts as $part)
{
$substring = trim($part);
$value = null;
if ($substring[0] === ':')
{
$value = 'left';
}
if (substr($substring, -1) === ':')
{
$value = $value === 'left' ? 'center' : 'right';
}
$values []= $value;
}
# ~
2014-02-24 03:38:58 +04:00
$nestedBlocks = array();
2014-02-23 20:55:34 +04:00
$substring = preg_replace('/^[|][ ]*/', '', $block['content']);
$substring = preg_replace('/[|]?[ ]*$/', '', $substring);
$parts = explode('|', $substring);
foreach ($parts as $index => $part)
{
$substring = trim($part);
2014-02-24 03:38:58 +04:00
$nestedBlock = array(
2014-02-23 20:55:34 +04:00
'name' => 'th',
2014-02-24 02:55:34 +04:00
'content type' => 'line',
2014-02-23 20:55:34 +04:00
'content' => $substring,
);
if (isset($values[$index]))
{
$value = $values[$index];
2014-02-24 03:38:58 +04:00
$nestedBlock['attributes'] = array(
2014-02-23 20:55:34 +04:00
'align' => $value,
);
}
2014-02-24 03:38:58 +04:00
$nestedBlocks []= $nestedBlock;
2014-02-23 20:55:34 +04:00
}
# ~
$block = array(
'name' => 'table',
'content type' => 'blocks',
'content' => array(),
);
$block['content'] []= array(
'name' => 'thead',
'content type' => 'blocks',
'content' => array(),
);
$block['content'] []= array(
'name' => 'tbody',
'content type' => 'blocks',
'content' => array(),
);
$block['content'][0]['content'] []= array(
'name' => 'tr',
'content type' => 'blocks',
'content' => array(),
);
2014-02-24 03:38:58 +04:00
$block['content'][0]['content'][0]['content'] = $nestedBlocks;
2014-02-23 20:55:34 +04:00
# ~
$context = 'table';
2014-02-24 03:38:58 +04:00
$contextData = array(
2014-02-23 20:55:34 +04:00
'alignments' => $values,
);
# ~
continue 2;
}
break;
2013-11-09 01:40:00 +04:00
default:
throw new Exception('Unrecognized context - '.$context);
}
if ($indentation >= 4)
{
$blocks []= $block;
$string = htmlspecialchars($line, ENT_NOQUOTES, 'UTF-8');
$string = str_repeat(' ', $indentation - 4) . $string;
$block = array(
'name' => 'pre',
'content type' => 'blocks',
'content' => array(
array(
'name' => 'code',
2014-02-24 02:55:34 +04:00
'content type' => null,
'content' => $string,
),
),
);
2013-11-09 01:40:00 +04:00
$context = 'code';
continue;
}
switch ($line[0])
{
case '#':
2013-11-09 01:40:00 +04:00
if (isset($line[1]))
{
$blocks []= $block;
2013-11-09 01:40:00 +04:00
$level = 1;
2013-11-09 01:40:00 +04:00
while (isset($line[$level]) and $line[$level] === '#')
{
$level++;
}
$string = trim($line, '# ');
2014-02-24 03:38:58 +04:00
$string = $this->parseLine($string);
$block = array(
'name' => 'h'.$level,
2014-02-24 02:55:34 +04:00
'content type' => 'line',
'content' => $string,
);
$context = null;
2013-11-09 01:40:00 +04:00
continue 2;
}
break;
2013-11-10 00:23:56 +04:00
case '<':
2013-11-09 01:40:00 +04:00
$position = strpos($line, '>');
2013-11-09 01:40:00 +04:00
2014-01-27 02:10:24 +04:00
if ($position > 1)
{
$substring = substr($line, 1, $position - 1);
2013-11-09 01:40:00 +04:00
2014-01-27 02:10:24 +04:00
$substring = chop($substring);
if (substr($substring, -1) === '/')
{
2014-02-24 03:38:58 +04:00
$isClosing = true;
2013-11-09 01:40:00 +04:00
2014-01-27 02:10:24 +04:00
$substring = substr($substring, 0, -1);
}
2014-01-27 02:10:24 +04:00
$position = strpos($substring, ' ');
if ($position)
{
2014-01-27 02:10:24 +04:00
$name = substr($substring, 0, $position);
}
else
{
$name = $substring;
}
$name = strtolower($name);
if ($name[0] == 'h' and strpos('r123456', $name[1]) !== false) # hr, h1, h2, ...
{
if ($name == 'hr')
{
2014-02-24 03:38:58 +04:00
$isClosing = true;
}
}
elseif ( ! ctype_alpha($name))
{
break;
}
2014-02-24 03:38:58 +04:00
if (in_array($name, self::$textLevelElements))
{
break;
}
2014-01-26 15:47:56 +04:00
$blocks []= $block;
$block = array(
'name' => null,
2014-02-24 02:55:34 +04:00
'content type' => null,
2014-02-24 03:38:58 +04:00
'content' => $indentedLine,
);
2014-02-24 03:38:58 +04:00
if (isset($isClosing))
{
2014-02-24 03:38:58 +04:00
unset($isClosing);
continue 2;
}
$context = 'markup';
2014-02-24 03:38:58 +04:00
$contextData = array(
'start' => '<'.$name.'>',
'end' => '</'.$name.'>',
'depth' => 0,
);
2013-11-09 01:40:00 +04:00
2014-02-24 03:38:58 +04:00
if (stripos($line, $contextData['end']) !== false)
{
$context = null;
}
continue 2;
}
2013-11-09 01:40:00 +04:00
break;
2013-11-09 01:40:00 +04:00
case '>':
2013-11-09 01:40:00 +04:00
if (preg_match('/^>[ ]?(.*)/', $line, $matches))
{
2014-01-26 15:47:56 +04:00
$blocks []= $block;
2013-11-09 01:40:00 +04:00
2014-01-26 15:47:56 +04:00
$block = array(
'name' => 'blockquote',
2014-02-24 02:55:34 +04:00
'content type' => 'lines',
'content' => array(
$matches[1],
),
);
$context = 'quote';
2014-02-24 03:38:58 +04:00
$contextData = array();
continue 2;
}
break;
case '[':
2013-11-09 01:40:00 +04:00
$position = strpos($line, ']:');
2014-02-05 16:03:23 +04:00
if ($position)
{
2014-02-05 16:03:23 +04:00
$reference = array();
2013-11-02 23:42:55 +04:00
$label = substr($line, 1, $position - 1);
$label = strtolower($label);
2014-02-05 16:03:23 +04:00
$substring = substr($line, $position + 2);
2014-02-05 16:03:23 +04:00
$substring = trim($substring);
2014-02-06 04:36:11 +04:00
if ($substring === '')
{
break;
}
2014-02-05 16:03:23 +04:00
if ($substring[0] === '<')
{
$position = strpos($substring, '>');
if ($position === false)
{
break;
}
$reference['link'] = substr($substring, 1, $position - 1);
2014-02-05 16:03:23 +04:00
$substring = substr($substring, $position + 1);
}
else
{
$position = strpos($substring, ' ');
if ($position === false)
{
$reference['link'] = $substring;
2013-12-26 23:53:48 +04:00
2014-02-05 16:03:23 +04:00
$substring = false;
}
else
{
$reference['link'] = substr($substring, 0, $position);
2014-02-05 16:03:23 +04:00
$substring = substr($substring, $position + 1);
}
}
if ($substring !== false)
{
2014-02-06 04:36:22 +04:00
if ($substring[0] !== '"' and $substring[0] !== "'" and $substring[0] !== '(')
2014-02-05 16:03:23 +04:00
{
2014-02-06 04:36:22 +04:00
break;
}
2014-02-05 16:03:23 +04:00
2014-02-24 03:38:58 +04:00
$lastChar = substr($substring, -1);
2014-02-05 16:03:23 +04:00
2014-02-24 03:38:58 +04:00
if ($lastChar !== '"' and $lastChar !== "'" and $lastChar !== ')')
2014-02-06 04:36:22 +04:00
{
break;
2014-02-05 16:03:23 +04:00
}
2014-02-06 04:36:22 +04:00
$reference['title'] = substr($substring, 1, -1);
}
2013-11-09 01:40:00 +04:00
2014-02-24 03:38:58 +04:00
$this->referenceMap[$label] = $reference;
2014-02-05 16:03:23 +04:00
continue 2;
}
break;
2013-11-09 01:40:00 +04:00
case '`':
case '~':
if (preg_match('/^([`]{3,}|[~]{3,})[ ]*(\w+)?[ ]*$/', $line, $matches))
{
2014-01-26 15:47:56 +04:00
$blocks []= $block;
2014-01-26 15:47:56 +04:00
$block = array(
'name' => 'pre',
'content type' => 'blocks',
'content' => array(
array(
'name' => 'code',
2014-02-24 02:55:34 +04:00
'content type' => null,
'content' => '',
),
),
);
if (isset($matches[2]))
{
$block['content'][0]['attributes'] = array(
'class' => 'language-'.$matches[2],
);
}
$context = 'fenced code';
2014-02-24 03:38:58 +04:00
$contextData = array(
'marker' => $matches[1][0],
);
continue 2;
}
break;
case '-':
case '*':
case '_':
2013-11-09 01:40:00 +04:00
if (preg_match('/^([-*_])([ ]{0,2}\1){2,}[ ]*$/', $line))
{
2014-01-26 15:47:56 +04:00
$blocks []= $block;
2013-11-09 01:40:00 +04:00
2014-01-26 15:47:56 +04:00
$block = array(
'name' => 'hr',
'content' => null,
);
continue 2;
}
}
switch (true)
{
case $line[0] <= '-' and preg_match('/^([*+-][ ]+)(.*)/', $line, $matches):
case $line[0] <= '9' and preg_match('/^([0-9]+[.][ ]+)(.*)/', $line, $matches):
$blocks []= $block;
2014-02-03 00:27:22 +04:00
$name = $line[0] >= '0' ? 'ol' : 'ul';
$block = array(
'name' => $name,
'content type' => 'blocks',
'content' => array(),
);
2014-02-24 03:38:58 +04:00
unset($nestedBlock);
2013-11-09 01:40:00 +04:00
2014-02-24 03:38:58 +04:00
$nestedBlock = array(
'name' => 'li',
2014-02-24 02:55:34 +04:00
'content type' => 'lines',
'content' => array(
$matches[2],
),
);
2013-11-09 01:40:00 +04:00
2014-02-24 03:38:58 +04:00
$block['content'] []= & $nestedBlock;
2013-11-09 01:40:00 +04:00
$baseline = $indentation + strlen($matches[1]);
2014-02-03 00:27:22 +04:00
$marker = $line[0] >= '0' ? '[0-9]+[.]' : '[*+-]';
2013-11-09 01:40:00 +04:00
$context = 'li';
2014-02-24 03:38:58 +04:00
$contextData = array(
'indentation' => $indentation,
'baseline' => $baseline,
'marker' => $marker,
'lines' => array(
$matches[2],
),
);
continue 2;
}
2013-11-09 01:40:00 +04:00
if ($context === 'paragraph')
{
$block['content'] .= "\n".$line;
2013-11-09 01:40:00 +04:00
continue;
}
else
{
2014-01-26 15:47:56 +04:00
$blocks []= $block;
2013-11-09 01:40:00 +04:00
2014-01-26 15:47:56 +04:00
$block = array(
'name' => 'p',
2014-02-24 02:55:34 +04:00
'content type' => 'line',
'content' => $line,
);
2013-11-09 01:40:00 +04:00
2014-02-24 03:38:58 +04:00
if ($blockContext === 'li' and empty($blocks[1]))
{
$block['name'] = null;
}
2013-11-09 01:40:00 +04:00
$context = 'paragraph';
}
}
2013-11-09 01:40:00 +04:00
2014-02-24 03:38:58 +04:00
if ($blockContext === 'li' and $block['name'] === null)
{
return $block['content'];
}
2013-11-17 00:04:26 +04:00
$blocks []= $block;
2013-11-09 01:40:00 +04:00
unset($blocks[0]);
2013-11-09 01:40:00 +04:00
return $blocks;
}
2013-11-09 01:40:00 +04:00
2014-02-21 04:22:31 +04:00
private function compile(array $blocks)
{
$markup = '';
foreach ($blocks as $block)
{
$markup .= "\n";
if (isset($block['name']))
{
$markup .= '<'.$block['name'];
if (isset($block['attributes']))
{
foreach ($block['attributes'] as $name => $value)
{
$markup .= ' '.$name.'="'.$value.'"';
}
}
if ($block['content'] === null)
{
$markup .= ' />';
continue;
}
else
{
$markup .= '>';
}
}
switch ($block['content type'])
{
2014-02-24 02:55:34 +04:00
case null:
2014-02-21 04:22:31 +04:00
$markup .= $block['content'];
break;
2014-02-24 02:55:34 +04:00
case 'line':
2014-02-21 04:22:31 +04:00
2014-02-24 03:38:58 +04:00
$markup .= $this->parseLine($block['content']);
2014-02-21 04:22:31 +04:00
break;
2014-02-24 02:55:34 +04:00
case 'lines':
2014-02-21 04:22:31 +04:00
2014-02-24 03:38:58 +04:00
$result = $this->findBlocks($block['content'], $block['name']);
2014-02-21 04:22:31 +04:00
if (is_string($result)) # dense li
{
2014-02-24 03:38:58 +04:00
$markup .= $this->parseLine($result);
2014-02-21 04:22:31 +04:00
break;
}
$markup .= $this->compile($result);
break;
case 'blocks':
$markup .= $this->compile($block['content']);
break;
}
if (isset($block['name']))
{
$markup .= '</'.$block['name'].'>';
}
}
$markup .= "\n";
return $markup;
}
2014-02-24 03:38:58 +04:00
private function parseLine($text, $markers = array(" \n", '![', '&', '*', '<', '[', '\\', '_', '`', 'http', '~~'))
{
if (isset($text[1]) === false or $markers === array())
{
return $text;
}
2013-11-09 01:40:00 +04:00
# ~
2013-11-09 01:40:00 +04:00
$markup = '';
2013-11-09 01:40:00 +04:00
while ($markers)
{
2014-02-24 03:38:58 +04:00
$closestMarker = null;
$closestMarkerIndex = 0;
$closestMarkerPosition = null;
2013-11-09 01:40:00 +04:00
foreach ($markers as $index => $marker)
{
2014-02-24 03:38:58 +04:00
$markerPosition = strpos($text, $marker);
2013-11-09 01:40:00 +04:00
2014-02-24 03:38:58 +04:00
if ($markerPosition === false)
{
unset($markers[$index]);
continue;
}
2013-11-09 01:40:00 +04:00
2014-02-24 03:38:58 +04:00
if ($closestMarker === null or $markerPosition < $closestMarkerPosition)
{
2014-02-24 03:38:58 +04:00
$closestMarker = $marker;
$closestMarkerIndex = $index;
$closestMarkerPosition = $markerPosition;
}
}
2013-11-09 01:40:00 +04:00
# ~
2014-02-24 03:38:58 +04:00
if ($closestMarker === null or isset($text[$closestMarkerPosition + 1]) === false)
{
$markup .= $text;
break;
}
else
{
2014-02-24 03:38:58 +04:00
$markup .= substr($text, 0, $closestMarkerPosition);
}
2014-02-24 03:38:58 +04:00
$text = substr($text, $closestMarkerPosition);
# ~
2013-11-09 01:40:00 +04:00
2014-02-24 03:38:58 +04:00
unset($markers[$closestMarkerIndex]);
2013-11-09 01:40:00 +04:00
# ~
2013-11-09 01:40:00 +04:00
2014-02-24 03:38:58 +04:00
switch ($closestMarker)
{
case " \n":
2014-01-20 11:26:25 +04:00
$markup .= '<br />'."\n";
2014-01-20 11:26:25 +04:00
$offset = 3;
2014-01-20 11:26:25 +04:00
break;
2014-01-20 11:26:25 +04:00
case '![':
case '[':
if (strpos($text, ']') and preg_match('/\[((?:[^][]|(?R))*)\]/', $text, $matches))
{
$element = array(
'!' => $text[0] === '!',
'text' => $matches[1],
);
$offset = strlen($matches[0]);
if ($element['!'])
{
$offset++;
}
2014-02-24 03:38:58 +04:00
$remainingText = substr($text, $offset);
2014-02-24 03:38:58 +04:00
if ($remainingText[0] === '(' and preg_match('/\([ ]*(.*?)(?:[ ]+[\'"](.+?)[\'"])?[ ]*\)/', $remainingText, $matches))
{
$element['link'] = $matches[1];
2013-07-11 00:22:16 +04:00
if (isset($matches[2]))
{
$element['title'] = $matches[2];
}
2013-12-26 23:53:48 +04:00
$offset += strlen($matches[0]);
}
2014-02-24 03:38:58 +04:00
elseif ($this->referenceMap)
{
$reference = $element['text'];
2014-02-24 03:38:58 +04:00
if (preg_match('/^\s*\[(.*?)\]/', $remainingText, $matches))
{
2014-02-24 01:43:18 +04:00
$reference = $matches[1] === '' ? $element['text'] : $matches[1];
$offset += strlen($matches[0]);
}
$reference = strtolower($reference);
2014-02-24 03:38:58 +04:00
if (isset($this->referenceMap[$reference]))
{
2014-02-24 03:38:58 +04:00
$element['link'] = $this->referenceMap[$reference]['link'];
2013-12-26 23:53:48 +04:00
2014-02-24 03:38:58 +04:00
if (isset($this->referenceMap[$reference]['title']))
{
2014-02-24 03:38:58 +04:00
$element['title'] = $this->referenceMap[$reference]['title'];
}
}
else
{
unset($element);
}
}
else
{
unset($element);
}
}
if (isset($element))
{
$element['link'] = str_replace('&', '&amp;', $element['link']);
$element['link'] = str_replace('<', '&lt;', $element['link']);
if ($element['!'])
{
$markup .= '<img alt="'.$element['text'].'" src="'.$element['link'].'"';
2014-01-21 00:19:23 +04:00
if (isset($element['title']))
{
$markup .= ' title="'.$element['title'].'"';
}
2014-01-21 00:19:23 +04:00
$markup .= ' />';
}
else
{
2014-02-24 03:38:58 +04:00
$element['text'] = $this->parseLine($element['text'], $markers);
$markup .= '<a href="'.$element['link'].'"';
2014-01-21 00:49:10 +04:00
if (isset($element['title']))
{
$markup .= ' title="'.$element['title'].'"';
}
2014-01-21 00:49:10 +04:00
$markup .= '>'.$element['text'].'</a>';
}
unset($element);
}
else
{
2014-02-24 03:38:58 +04:00
$markup .= $closestMarker;
2013-07-11 00:22:16 +04:00
2014-02-24 03:38:58 +04:00
$offset = $closestMarker === '![' ? 2 : 1;
}
break;
case '&':
if (preg_match('/^&#?\w+;/', $text, $matches))
{
$markup .= $matches[0];
$offset = strlen($matches[0]);
}
else
{
$markup .= '&amp;';
$offset = 1;
}
break;
case '*':
case '_':
2014-02-24 03:38:58 +04:00
if ($text[1] === $closestMarker and preg_match(self::$strongRegex[$closestMarker], $text, $matches))
{
2014-02-24 03:38:58 +04:00
$markers[] = $closestMarker;
$matches[1] = $this->parseLine($matches[1], $markers);
$markup .= '<strong>'.$matches[1].'</strong>';
}
2014-02-24 03:38:58 +04:00
elseif (preg_match(self::$emRegex[$closestMarker], $text, $matches))
{
2014-02-24 03:38:58 +04:00
$markers[] = $closestMarker;
$matches[1] = $this->parseLine($matches[1], $markers);
$markup .= '<em>'.$matches[1].'</em>';
}
if (isset($matches) and $matches)
{
$offset = strlen($matches[0]);
}
else
{
2014-02-24 03:38:58 +04:00
$markup .= $closestMarker;
$offset = 1;
}
break;
case '<':
if (strpos($text, '>') !== false)
{
2014-01-30 03:05:05 +04:00
if ($text[1] === 'h' and preg_match('/^<(https?:[\/]{2}[^\s]+?)>/i', $text, $matches))
{
2014-02-24 03:38:58 +04:00
$elementUrl = $matches[1];
$elementUrl = str_replace('&', '&amp;', $elementUrl);
$elementUrl = str_replace('<', '&lt;', $elementUrl);
2013-12-06 02:29:51 +04:00
2014-02-24 03:38:58 +04:00
$markup .= '<a href="'.$elementUrl.'">'.$elementUrl.'</a>';
$offset = strlen($matches[0]);
}
elseif (strpos($text, '@') > 1 and preg_match('/<(\S+?@\S+?)>/', $text, $matches))
{
$markup .= '<a href="mailto:'.$matches[1].'">'.$matches[1].'</a>';
2014-01-21 16:23:04 +04:00
$offset = strlen($matches[0]);
}
elseif (preg_match('/^<\/?\w.*?>/', $text, $matches))
{
$markup .= $matches[0];
$offset = strlen($matches[0]);
}
else
{
$markup .= '&lt;';
$offset = 1;
}
}
else
{
$markup .= '&lt;';
$offset = 1;
}
break;
case '\\':
2014-01-20 00:49:43 +04:00
2014-02-24 03:38:58 +04:00
if (in_array($text[1], self::$specialCharacters))
{
$markup .= $text[1];
2014-01-20 00:49:43 +04:00
$offset = 2;
}
else
{
$markup .= '\\';
2014-01-20 00:49:43 +04:00
$offset = 1;
}
2014-01-20 00:49:43 +04:00
break;
2014-01-20 00:49:43 +04:00
case '`':
2014-02-06 16:16:14 +04:00
if (preg_match('/^(`+)[ ]*(.+?)[ ]*(?<!`)\1(?!`)/', $text, $matches))
{
2014-02-24 03:38:58 +04:00
$elementText = $matches[2];
$elementText = htmlspecialchars($elementText, ENT_NOQUOTES, 'UTF-8');
2013-11-09 01:40:00 +04:00
2014-02-24 03:38:58 +04:00
$markup .= '<code>'.$elementText.'</code>';
$offset = strlen($matches[0]);
}
else
{
$markup .= '`';
2013-11-09 01:40:00 +04:00
$offset = 1;
}
2013-11-09 01:40:00 +04:00
break;
2013-11-09 01:40:00 +04:00
case 'http':
2013-11-09 01:40:00 +04:00
2014-02-06 02:10:18 +04:00
if (preg_match('/^https?:[\/]{2}[^\s]+\b\/*/ui', $text, $matches))
{
2014-02-24 03:38:58 +04:00
$elementUrl = $matches[0];
$elementUrl = str_replace('&', '&amp;', $elementUrl);
$elementUrl = str_replace('<', '&lt;', $elementUrl);
2013-11-09 01:40:00 +04:00
2014-02-24 03:38:58 +04:00
$markup .= '<a href="'.$elementUrl.'">'.$elementUrl.'</a>';
2013-11-09 01:40:00 +04:00
$offset = strlen($matches[0]);
}
else
{
$markup .= 'http';
2013-11-22 02:20:45 +04:00
$offset = 4;
}
2013-11-22 02:20:45 +04:00
break;
2013-11-09 01:40:00 +04:00
case '~~':
2013-11-09 01:40:00 +04:00
if (preg_match('/^~~(?=\S)(.+?)(?<=\S)~~/', $text, $matches))
{
2014-02-24 03:38:58 +04:00
$matches[1] = $this->parseLine($matches[1], $markers);
2013-11-09 01:40:00 +04:00
$markup .= '<del>'.$matches[1].'</del>';
2013-11-09 01:40:00 +04:00
$offset = strlen($matches[0]);
}
else
{
$markup .= '~~';
2013-11-21 15:39:00 +04:00
$offset = 2;
}
2013-11-09 01:40:00 +04:00
break;
}
if (isset($offset))
{
$text = substr($text, $offset);
}
2013-11-09 01:40:00 +04:00
2014-02-24 03:38:58 +04:00
$markers[$closestMarkerIndex] = $closestMarker;
}
2013-11-09 01:40:00 +04:00
return $markup;
}
2014-01-18 17:10:24 +04:00
2014-02-21 04:22:31 +04:00
#
# Static
static function instance($name = 'default')
{
if (isset(self::$instances[$name]))
{
return self::$instances[$name];
}
$instance = new Parsedown();
self::$instances[$name] = $instance;
return $instance;
}
private static $instances = array();
#
2014-01-26 21:53:24 +04:00
# Fields
#
2014-02-24 03:38:58 +04:00
private $referenceMap = array();
2014-01-26 21:53:24 +04:00
#
# Read-only
2014-01-18 17:10:24 +04:00
2014-02-24 03:38:58 +04:00
private static $strongRegex = array(
'*' => '/^[*]{2}((?:[^*]|[*][^*]*[*])+?)[*]{2}(?![*])/s',
'_' => '/^__((?:[^_]|_[^_]*_)+?)__(?!_)/us',
);
2014-01-18 17:10:24 +04:00
2014-02-24 03:38:58 +04:00
private static $emRegex = array(
'*' => '/^[*]((?:[^*]|[*][*][^*]+?[*][*])+?)[*](?![*])/s',
'_' => '/^_((?:[^_]|__[^_]*__)+?)_(?!_)\b/us',
);
2014-01-27 02:58:18 +04:00
2014-02-24 03:38:58 +04:00
private static $specialCharacters = array(
2014-01-27 02:58:18 +04:00
'\\', '`', '*', '_', '{', '}', '[', ']', '(', ')', '>', '#', '+', '-', '.', '!',
);
2014-02-24 03:38:58 +04:00
private static $textLevelElements = array(
2014-01-27 02:58:18 +04:00
'a', 'br', 'bdo', 'abbr', 'blink', 'nextid', 'acronym', 'basefont',
'b', 'em', 'big', 'cite', 'small', 'spacer', 'listing',
'i', 'rp', 'sub', 'code', 'strike', 'marquee',
'q', 'rt', 'sup', 'font', 'strong',
's', 'tt', 'var', 'mark',
'u', 'xm', 'wbr', 'nobr',
'ruby',
'span',
'time',
);
}