', '\#', '\+', '\-', '\.', '\!'); foreach ($escape_sequences as $index => $escape_sequence) { if (strpos($text, $escape_sequence) !== FALSE) { $code = "\x1A".'\\'.$index.';'; $text = str_replace($escape_sequence, $code, $text); $this->escape_sequence_map[$code] = $escape_sequence; } } } # ~ $text = preg_replace('/\n\s*\n/', "\n\n", $text); $text = trim($text, "\n"); $lines = explode("\n", $text); $text = $this->parse_block_elements($lines); # decodes escape sequences foreach ($this->escape_sequence_map as $code => $escape_sequence) { $text = str_replace($code, $escape_sequence[1], $text); } $text = rtrim($text, "\n"); return $text; } # # Private Methods # private function parse_block_elements(array $lines, $context = '') { $elements = array(); $element = array( 'type' => '', ); foreach ($lines as $line) { # markup (open) if ($element['type'] === 'markup' and ! isset($element['closed'])) { if (preg_match('{<'.$element['subtype'].'>$}', $line)) # { $element['depth']++; } if (preg_match('{$}', $line)) # { $element['depth'] > 0 ? $element['depth']-- : $element['closed'] = true; } $element['text'] .= "\n".$line; continue; } # * if ($line === '') { $element['interrupted'] = true; continue; } # blockquote (existing) if ($element['type'] === 'blockquote' and ! isset($element['interrupted'])) { $line = preg_replace('/^[ ]*>[ ]?/', '', $line); $element['lines'] []= $line; continue; } # list (existing) if ($element['type'] === 'li') { if (preg_match('/^([ ]{0,3})(\d+[.]|[*+-])[ ](.*)/', $line, $matches)) { if ($element['indentation'] !== $matches[1]) { $element['lines'] []= $line; } else { unset($element['last']); $elements []= $element; $element = array( 'type' => 'li', 'indentation' => $matches[1], 'last' => true, 'lines' => array( preg_replace('/^[ ]{0,4}/', '', $matches[3]), ), ); } continue; } if (isset($element['interrupted'])) { if ($line[0] === ' ') { $element['lines'] []= ''; $line = preg_replace('/^[ ]{0,4}/', '', $line); $element['lines'] []= $line; continue; } } else { $line = preg_replace('/^[ ]{0,4}/', '', $line); $element['lines'] []= $line; continue; } } # paragraph if ($line[0] >= 'a' or $line[0] >= 'A' and $line[0] <= 'Z') { goto paragraph; } # code block if ($line[0] === ' ' and preg_match('/^[ ]{4}(.*)/', $line, $matches)) { if (trim($line) === '') { continue; } if ($element['type'] === 'code') { if (isset($element['interrupted'])) { $element['text'] .= "\n"; unset ($element['interrupted']); } $element['text'] .= "\n".$matches[1]; } else { $elements []= $element; $element = array( 'type' => 'code', 'text' => $matches[1], ); } continue; } # setext heading (---) if ($line[0] === '-' and $element['type'] === 'p' and ! isset($element['interrupted']) and preg_match('/^[-]+[ ]*$/', $line)) { $element['type'] = 'h.'; $element['level'] = 2; continue; } # atx heading (#) if ($line[0] === '#' and preg_match('/^(#{1,6})[ ]*(.+?)[ ]*#*$/', $line, $matches)) { $elements []= $element; $level = strlen($matches[1]); $element = array( 'type' => 'h.', 'text' => $matches[2], 'level' => $level, ); continue; } # setext heading (===) if ($line[0] === '=' and $element['type'] === 'p' and ! isset($element['interrupted']) and preg_match('/^[=]+[ ]*$/', $line)) { $element['type'] = 'h.'; $element['level'] = 1; continue; } # ~ $pure_line = $line[0] !== ' ' ? $line : ltrim($line); if ($pure_line === '') { continue; } # reference if ($pure_line[0] === '[' and preg_match('/^\[(.+?)\]:[ ]*([^ ]+)/', $pure_line, $matches)) { $label = strtolower($matches[1]); $url = trim($matches[2], '<>'); $this->reference_map[$label] = $url; continue; } # blockquote if ($pure_line[0] === '>' and preg_match('/^>[ ]?(.*)/', $pure_line, $matches)) { if ($element['type'] === 'blockquote') { if (isset($element['interrupted'])) { $element['lines'] []= ''; unset($element['interrupted']); } $element['lines'] []= $matches[1]; } else { $elements []= $element; $element = array( 'type' => 'blockquote', 'lines' => array( $matches[1], ), ); } continue; } # markup if ($pure_line[0] === '<') { # Block-Level HTML if (preg_match('{^<.+?/>$}', $pure_line)) { $elements []= $element; $element = array( 'type' => '', 'text' => $pure_line, ); continue; } # Block-Level HTML if (preg_match('{^<(\w+)(?:[ ].*?)?>}', $pure_line, $matches)) { $elements []= $element; $element = array( 'type' => 'markup', 'subtype' => strtolower($matches[1]), 'text' => $pure_line, 'depth' => 0, ); preg_match('{\s*$}', $pure_line) and $element['closed'] = true; continue; } } # horizontal rule if (preg_match('/^([-*_])([ ]{0,2}\1){2,}[ ]*$/', $pure_line)) { $elements []= $element; $element = array( 'type' => 'hr', ); continue; } # list item if (preg_match('/^([ ]*)(\d+[.]|[*+-])[ ](.*)/', $line, $matches)) { $elements []= $element; $element = array( 'type' => 'li', 'ordered' => isset($matches[2][1]), 'indentation' => $matches[1], 'last' => true, 'lines' => array( preg_replace('/^[ ]{0,4}/', '', $matches[3]), ), ); continue; } # ~ paragraph: if ($element['type'] === 'p') { if (isset($element['interrupted'])) { $elements []= $element; $element['text'] = $line; unset($element['interrupted']); } else { $element['text'] .= "\n".$line; } } else { $elements []= $element; $element = array( 'type' => 'p', 'text' => $line, ); } } $elements []= $element; array_shift($elements); # # ~ # $markup = ''; foreach ($elements as $index => $element) { switch ($element['type']) { case 'p': $text = $this->parse_span_elements($element['text']); $text = preg_replace('/[ ]{2}\n/', '
'."\n", $text); if ($context === 'li' and $index === 0) { if (isset($element['interrupted'])) { $markup .= "\n".'

'.$text.'

'."\n"; } else { $markup .= $text; } } else { $markup .= '

'.$text.'

'."\n"; } break; case 'blockquote': $text = $this->parse_block_elements($element['lines']); $markup .= '
'."\n".$text.'
'."\n"; break; case 'code': $text = htmlentities($element['text'], ENT_NOQUOTES); strpos($text, "\x1A\\") !== FALSE and $text = strtr($text, $this->escape_sequence_map); $markup .= '
'.$text.'
'."\n"; break; case 'h.': $text = $this->parse_span_elements($element['text']); $markup .= ''.$text.''."\n"; break; case 'hr': $markup .= '
'."\n"; break; case 'li': if (isset($element['ordered'])) # first { $list_type = $element['ordered'] ? 'ol' : 'ul'; $markup .= '<'.$list_type.'>'."\n"; } if (isset($element['interrupted']) and ! isset($element['last'])) { $element['lines'] []= ''; } $text = $this->parse_block_elements($element['lines'], 'li'); $markup .= '
  • '.$text.'
  • '."\n"; isset($element['last']) and $markup .= ''."\n"; break; default: $markup .= $element['text']."\n"; } } return $markup; } private function parse_span_elements($text) { $map = array(); $index = 0; # code span if (strpos($text, '`') !== FALSE and preg_match_all('/`(.+?)`/', $text, $matches, PREG_SET_ORDER)) { foreach ($matches as $matches) { $element_text = $matches[1]; $element_text = htmlentities($element_text, ENT_NOQUOTES); # Decodes escape sequences. $this->escape_sequence_map and strpos($element_text, "\x1A") !== FALSE and $element_text = strtr($element_text, $this->escape_sequence_map); # Composes element. $element = ''.$element_text.''; # Encodes element. $code = "\x1A".'$'.$index; $text = str_replace($matches[0], $code, $text); $map[$code] = $element; $index ++; } } # inline link or image if (strpos($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+;)/', '&', $url); if ($matches[1]) # image { $element = ''.$matches[3].''; } else { $element_text = $this->parse_span_elements($matches[3]); $element = ''.$element_text.''; } # ~ $code = "\x1A".'$'.$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+;)/', '&', $url); if ($matches[1]) # image { $element = ''.$matches[2].''; } else # anchor { $element_text = $this->parse_span_elements($matches[2]); $element = ''.$element_text.''; } # ~ $code = "\x1A".'$'.$index; $text = str_replace($matches[0], $code, $text); $map[$code] = $element; $index ++; } } } # automatic link if (strpos($text, '<') !== FALSE and preg_match_all('/<((https?|ftp|dict):[^\^\s]+?)>/i', $text, $matches, PREG_SET_ORDER)) { foreach ($matches as $matches) { $url = $matches[1]; strpos($url, '&') !== FALSE and $url = preg_replace('/&(?!#?\w+;)/', '&', $url); $element = ':text'; $element = str_replace(':text', $url, $element); $element = str_replace(':href', $url, $element); # ~ $code = "\x1A".'$'.$index; $text = str_replace($matches[0], $code, $text); $map[$code] = $element; $index ++; } } # ~ strpos($text, '&') !== FALSE and $text = preg_replace('/&(?!#?\w+;)/', '&', $text); strpos($text, '<') !== FALSE and $text = preg_replace('/<(?!\/?\w.*?>)/', '<', $text); # ~ if (strpos($text, '_') !== FALSE) { $text = preg_replace('/__(?=\S)(.+?)(?<=\S)__(?!_)/s', '$1', $text); $text = preg_replace('/_(?=\S)(.+?)(?<=\S)_/s', '$1', $text); } if (strpos($text, '*') !== FALSE) { $text = preg_replace('/\*\*(?=\S)(.+?)(?<=\S)\*\*(?!\*)/s', '$1', $text); $text = preg_replace('/\*(?=\S)(.+?)(?<=\S)\*/s', '$1', $text); } $text = strtr($text, $map); return $text; } }