geshifilter.pages.inc in GeSHi Filter for syntax highlighting 7

Same filename and directory in other branches
File

geshifilter.pages.inc
View source
<?php

/**
 * @file
 * Functions that handle the actual GeSHi processing.
 *
 * The actual GeSHi filter stuff: parsing the text to filter,
 * configure the GeSHi parser, feed the code to the GeSHi parser and put the
 * result back in the text.
 */
require_once drupal_get_path('module', 'geshifilter') . '/geshifilter.inc';

/**
 * Helper function for parsing the attributes of GeSHi code tags.
 * to get the settings for language, line numbers, etc.
 *
 * @param $attributes string with the attributes.
 * @param $format the concerning text format.
 * @return array of settings with fields 'language', 'line_numbering', 'linenumbers_start' and 'title'.
 */
function _geshifilter_parse_attributes($attributes, $format) {

  // Initial values.
  $lang = NULL;
  $line_numbering = NULL;
  $linenumbers_start = NULL;
  $title = NULL;

  // Get the possible tags and languages.
  list($generic_code_tags, $language_tags, $tag_to_lang) = _geshifilter_get_tags($format);
  $language_attributes = _geshifilter_whitespace_explode(GESHIFILTER_ATTRIBUTES_LANGUAGE);
  $attributes_preg_string = implode('|', array_merge($language_attributes, array(
    GESHIFILTER_ATTRIBUTE_LINE_NUMBERING,
    GESHIFILTER_ATTRIBUTE_LINE_NUMBERING_START,
    GESHIFILTER_ATTRIBUTE_FANCY_N,
    GESHIFILTER_ATTRIBUTE_TITLE,
  )));
  $enabled_languages = _geshifilter_get_enabled_languages();

  // Parse $attributes to an array $attribute_matches with:
  // $attribute_matches[0][xx] .... fully matched string, e.g. 'language="python"'
  // $attribute_matches[1][xx] .... param name, e.g. 'language'
  // $attribute_matches[2][xx] .... param value, e.g. 'python'
  preg_match_all('#(' . $attributes_preg_string . ')="?([^"]*)"?#', $attributes, $attribute_matches);
  foreach ($attribute_matches[1] as $a_key => $att_name) {

    // get attribute value
    $att_value = $attribute_matches[2][$a_key];

    // Check for the language attributes.
    $class_to_lang = str_replace('language-', '', $att_value);
    if (in_array($att_name, $language_attributes)) {

      // Try first to map the attribute value to geshi language code.
      if (in_array($att_value, $language_tags)) {
        $att_value = $tag_to_lang[$att_value];
      }

      // Set language if extracted language is an enabled language.
      if (array_key_exists($att_value, $enabled_languages)) {
        $lang = $att_value;
      }
      elseif ($att_name == 'class' && array_key_exists($class_to_lang, $enabled_languages)) {
        $lang = $class_to_lang;
      }
    }
    elseif ($att_name == GESHIFILTER_ATTRIBUTE_LINE_NUMBERING) {
      switch (strtolower($att_value)) {
        case "off":
          $line_numbering = 0;
          break;
        case "normal":
          $line_numbering = 1;
          break;
        case "fancy":
          $line_numbering = 5;
          break;
      }
    }
    elseif ($att_name == GESHIFILTER_ATTRIBUTE_FANCY_N) {
      $att_value = (int) $att_value;
      if ($att_value >= 2) {
        $line_numbering = $att_value;
      }
    }
    elseif ($att_name == GESHIFILTER_ATTRIBUTE_LINE_NUMBERING_START) {
      if ($line_numbering < 1) {
        $line_numbering = 1;
      }
      $linenumbers_start = (int) $att_value;
    }
    elseif ($att_name == GESHIFILTER_ATTRIBUTE_TITLE) {
      $title = $att_value;
    }
  }

  // Return parsed results.
  return array(
    'language' => $lang,
    'line_numbering' => $line_numbering,
    'linenumbers_start' => $linenumbers_start,
    'title' => $title,
  );
}

/**
 * geshifilter_filter callback for preparing input text.
 */
function _geshifilter_prepare($format, $text) {

  // get the available tags
  list($generic_code_tags, $language_tags, $tag_to_lang) = _geshifilter_get_tags($format);
  $tags = array_merge($generic_code_tags, $language_tags);

  // escape special (regular expression) characters in tags (for tags like 'c++' and 'c#')
  $tags = preg_replace('#(\\+|\\#)#', '\\\\$1', $tags);
  $tags_string = implode('|', $tags);

  // Pattern for matching "<code>...</code>" like stuff
  // Also matches "<code>...$"  where "$" refers to end of string, not end of
  // line (because PCRE_MULTILINE (modifier 'm') is not enabled), so matching
  // still works when teaser view trims inside the source code.
  // Replace the code container tag brackets
  // and prepare the container content (newline and angle bracket protection).
  // @todo: make sure that these replacements can be done in series.
  $tag_styles = array_filter(_geshifilter_tag_styles($format));
  if (in_array(GESHIFILTER_BRACKETS_ANGLE, $tag_styles)) {

    // Prepare <foo>..</foo> blocks.
    $pattern = '#(<)(' . $tags_string . ')((\\s+[^>]*)*)(>)(.*?)(</\\2\\s*>|$)#s';
    $text = preg_replace_callback($pattern, function ($match) use ($format) {
      return _geshifilter_prepare_callback($match, $format);
    }, $text);
  }
  if (in_array(GESHIFILTER_BRACKETS_SQUARE, $tag_styles)) {

    // Prepare [foo]..[/foo] blocks.
    $pattern = '#((?<!\\[)\\[)(' . $tags_string . ')((\\s+[^\\]]*)*)(\\])(.*?)((?<!\\[)\\[/\\2\\s*\\]|$)#s';
    $text = preg_replace_callback($pattern, function ($match) use ($format) {
      return _geshifilter_prepare_callback($match, $format);
    }, $text);
  }
  if (in_array(GESHIFILTER_BRACKETS_DOUBLESQUARE, $tag_styles)) {

    // Prepare [[foo]]..[[/foo]] blocks.
    $pattern = '#(\\[\\[)(' . $tags_string . ')((\\s+[^\\]]*)*)(\\]\\])(.*?)(\\[\\[/\\2\\s*\\]\\]|$)#s';
    $text = preg_replace_callback($pattern, function ($match) use ($format) {
      return _geshifilter_prepare_callback($match, $format);
    }, $text);
  }
  if (in_array(GESHIFILTER_BRACKETS_PHPBLOCK, $tag_styles)) {

    // Prepare < ?php ... ? > blocks.
    $pattern = '#[\\[<](\\?php|\\?PHP|%)(.+?)((\\?|%)[\\]>]|$)#s';
    $text = preg_replace_callback($pattern, function ($match) use ($format) {
      return _geshifilter_prepare_php_callback($match, $format);
    }, $text);
  }
  return $text;
}

/**
 * _geshifilter_prepare callback for preparing input text.
 * Replaces the code tags brackets with geshifilter specific ones to prevent
 * possible messing up by other filters, e.g.
 *   '[python]foo[/python]' to '[geshifilter-python]foo[/geshifilter-python]'.
 * Replaces newlines with "&#10;" to prevent issues with the line break filter
 * Escapes the tricky characters like angle brackets with check_plain() to
 * prevent messing up by other filters like the HTML filter.
 */
function _geshifilter_prepare_callback($match, $format) {

  // $match[0]: complete matched string
  // $match[1]: opening bracket ('<' or '[')
  // $match[2]: tag
  // $match[3] and $match[4]: attributes
  // $match[5]: closing bracket
  // $match[6]: source code
  // $match[7]: closing tag
  $tag_name = $match[2];
  $tag_attributes = $match[3];
  $content = $match[6];

  // Decode HTML entities if option has been set
  if (geshifilter_decode_entities($format)) {
    $content = html_entity_decode($content, ENT_QUOTES);
  }

  // get the default highlighting mode
  $lang = variable_get('geshifilter_default_highlighting', GESHIFILTER_DEFAULT_PLAINTEXT);
  if ($lang == GESHIFILTER_DEFAULT_DONOTHING) {

    // If the default highlighting mode is GESHIFILTER_DEFAULT_DONOTHING
    // and there is no language set (with language tag or language attribute),
    // we should not do any escaping in this prepare phase,
    // so that other filters can do their thing.
    $enabled_languages = _geshifilter_get_enabled_languages();

    // Usage of language tag?
    list($generic_code_tags, $language_tags, $tag_to_lang) = _geshifilter_get_tags($format);
    if (isset($tag_to_lang[$tag_name]) && isset($enabled_languages[$tag_to_lang[$tag_name]])) {
      $lang = $tag_to_lang[$tag_name];
    }
    else {

      // Get additional settings from the tag attributes.
      $settings = _geshifilter_parse_attributes($tag_attributes, $format);
      if ($settings['language'] && isset($enabled_languages[$settings['language']])) {
        $lang = $settings['language'];
      }
    }

    // If no language was set: prevent escaping and return original string
    if ($lang == GESHIFILTER_DEFAULT_DONOTHING) {
      return $match[0];
    }
  }

  // return escaped code block
  return '[geshifilter-' . $tag_name . $tag_attributes . ']' . str_replace(array(
    "\r",
    "\n",
  ), array(
    '',
    '&#10;',
  ), check_plain($content)) . '[/geshifilter-' . $tag_name . ']';
}

/**
 * _geshifilter_prepare callback for < ?php ... ? > blocks.
 */
function _geshifilter_prepare_php_callback($match, $format) {

  // Decode HTML entities if option has been set
  if (geshifilter_decode_entities($format)) {
    $match[2] = html_entity_decode($match[2], ENT_QUOTES);
  }
  return '[geshifilter-questionmarkphp]' . str_replace(array(
    "\r",
    "\n",
  ), array(
    '',
    '&#10;',
  ), check_plain($match[2])) . '[/geshifilter-questionmarkphp]';
}

/**
 * geshifilter_filter callback for processing input text.
 */
function _geshifilter_process($format, $text) {

  // load GeSHi library (if not already)
  $geshi_library = libraries_load('geshi');
  if (!$geshi_library['loaded']) {
    drupal_set_message($geshi_library['error message'], 'error');
    return $text;
  }

  // get the available tags
  list($generic_code_tags, $language_tags, $tag_to_lang) = _geshifilter_get_tags($format);
  if (in_array(GESHIFILTER_BRACKETS_PHPBLOCK, array_filter(_geshifilter_tag_styles($format)))) {
    $language_tags[] = 'questionmarkphp';
    $tag_to_lang['questionmarkphp'] = 'php';
  }
  $tags = array_merge($generic_code_tags, $language_tags);

  // escape special (regular expression) characters in tags (for tags like 'c++' and 'c#')
  $tags = preg_replace('#(\\+|\\#)#', '\\\\$1', $tags);
  $tags_string = implode('|', $tags);

  // Pattern for matching the prepared "<code>...</code>" stuff
  $pattern = '#\\[geshifilter-(' . $tags_string . ')([^\\]]*)\\](.*?)(\\[/geshifilter-\\1\\])#s';
  $text = preg_replace_callback($pattern, function ($match) use ($format) {
    return _geshifilter_replace_callback($match, $format);
  }, $text);
  return $text;
}

/**
 * preg_replace_callback callback.
 */
function _geshifilter_replace_callback($match, $format) {

  // $match[0]: complete matched string
  // $match[1]: tag name
  // $match[2]: tag attributes
  // $match[3]: tag content
  $complete_match = $match[0];
  $tag_name = $match[1];
  $tag_attributes = $match[2];
  $source_code = $match[3];

  // Undo linebreak and escaping from preparation phase.
  $source_code = decode_entities($source_code);

  // Initialize to default settings.
  $lang = variable_get('geshifilter_default_highlighting', GESHIFILTER_DEFAULT_PLAINTEXT);
  $line_numbering = variable_get('geshifilter_default_line_numbering', GESHIFILTER_LINE_NUMBERS_DEFAULT_NONE);
  $linenumbers_start = 1;
  $title = NULL;

  // Determine language based on tag name if possible.
  list($generic_code_tags, $language_tags, $tag_to_lang) = _geshifilter_get_tags($format);
  if (in_array(GESHIFILTER_BRACKETS_PHPBLOCK, array_filter(_geshifilter_tag_styles($format)))) {
    $language_tags[] = 'questionmarkphp';
    $tag_to_lang['questionmarkphp'] = 'php';
  }
  if (isset($tag_to_lang[$tag_name])) {
    $lang = $tag_to_lang[$tag_name];
  }

  // Get additional settings from the tag attributes.
  $settings = _geshifilter_parse_attributes($tag_attributes, $format);
  if (isset($settings['language'])) {
    $lang = $settings['language'];
  }
  if (isset($settings['line_numbering'])) {
    $line_numbering = $settings['line_numbering'];
  }
  if (isset($settings['linenumbers_start'])) {
    $linenumbers_start = $settings['linenumbers_start'];
  }
  if (isset($settings['title'])) {
    $title = $settings['title'];
  }
  if ($lang == GESHIFILTER_DEFAULT_DONOTHING) {

    // Do nothing, and return the original.
    return $complete_match;
  }
  if ($lang == GESHIFILTER_DEFAULT_PLAINTEXT) {

    // Use plain text 'highlighting'
    $lang = 'text';
  }
  $inline_mode = strpos($source_code, "\n") === FALSE;

  // process and return
  return geshifilter_process_sourcecode($source_code, $lang, $line_numbering, $linenumbers_start, $inline_mode, $title);
}

/**
 * Helper function for overriding some GeSHi defaults.
 */
function _geshifilter_override_geshi_defaults(&$geshi, $langcode) {

  // override the some default GeSHi styles (e.g. GeSHi uses Courier by default, which is ugly)
  $geshi
    ->set_line_style('font-family: monospace; font-weight: normal;', 'font-family: monospace; font-weight: bold; font-style: italic;');
  $geshi
    ->set_code_style('font-family: monospace; font-weight: normal; font-style: normal');

  // overall class needed for CSS
  $geshi
    ->set_overall_class('geshifilter-' . $langcode);

  // set keyword linking
  $geshi
    ->enable_keyword_links(variable_get('geshifilter_enable_keyword_urls', TRUE));
}

/**
 * General geshifilter processing function for a chunk of source code.
 */
function geshifilter_process_sourcecode($source_code, $lang, $line_numbering = 0, $linenumbers_start = 1, $inline_mode = FALSE, $title = NULL) {

  // process
  if ($lang == 'php' && variable_get('geshifilter_use_highlight_string_for_php', FALSE)) {
    return geshifilter_highlight_string_process($source_code, $inline_mode);
  }
  else {

    // process with GeSHi
    return geshifilter_geshi_process($source_code, $lang, $line_numbering, $linenumbers_start, $inline_mode, $title);
  }
}

/**
 * geshifilter wrapper for GeSHi processing.
 */
function geshifilter_geshi_process($source_code, $lang, $line_numbering = 0, $linenumbers_start = 1, $inline_mode = FALSE, $title = NULL) {

  // load GeSHi library (if not already)
  $geshi_library = libraries_load('geshi');
  if (!$geshi_library['loaded']) {
    drupal_set_message($geshi_library['error message'], 'error');
    return $source_code;
  }

  // Check for a cached version of this source code and return it if available.
  // @todo: Use a dedicated table instead of using cache_filter? If so,
  // also take care of the flushing in _geshifilter_clear_filter_cache().
  $cache_id = "geshifilter:{$lang}:{$line_numbering}:{$linenumbers_start}:{$inline_mode}" . md5($title . $source_code);
  if ($cached = cache_get($cache_id, 'cache_filter')) {
    return $cached->data;
  }

  // remove leading/trailing newlines
  $source_code = trim($source_code, "\n\r");

  // create GeSHi object
  $geshi = _geshifilter_geshi_factory($source_code, $lang);

  // CSS mode
  $ccs_mode = variable_get('geshifilter_css_mode', GESHIFILTER_CSS_INLINE);
  if ($ccs_mode == GESHIFILTER_CSS_CLASSES_AUTOMATIC || $ccs_mode == GESHIFILTER_CSS_CLASSES_ONLY) {
    $geshi
      ->enable_classes(TRUE);
  }
  _geshifilter_override_geshi_defaults($geshi, $lang);

  // some more GeSHi settings and parsing
  if ($inline_mode) {

    // inline source code mode
    $geshi
      ->set_header_type(GESHI_HEADER_NONE);

    // To make highlighting work we have to manually set a class on the code
    // element we will wrap the code in.
    // To counter a change between GeSHi version 1.0.7.22 and 1.0.8 (svn
    // commit 1610), we use both the language and overall_class for the class,
    // to mimic the 1.0.8 behavior, which is backward compatible.
    // $language and $overall_class are protected with $geshi, with no get functions,
    // recreate them manually.
    $overall_class = 'geshifilter-' . $lang;
    $code_class = "{$lang} {$overall_class}";
    $source_code = '<span class="geshifilter"' . (isset($title) ? ' title="' . check_plain($title) . '"' : '') . '><code class="' . $code_class . '">' . $geshi
      ->parse_code() . '</code></span>';
  }
  else {

    // block source code mode
    $geshi
      ->set_header_type((int) variable_get('geshifilter_code_container', GESHI_HEADER_PRE));
    if ($line_numbering == 1) {
      $geshi
        ->enable_line_numbers(GESHI_NORMAL_LINE_NUMBERS);
      $geshi
        ->start_line_numbers_at($linenumbers_start);
    }
    elseif ($line_numbering >= 2) {
      $geshi
        ->enable_line_numbers(GESHI_FANCY_LINE_NUMBERS, $line_numbering);
      $geshi
        ->start_line_numbers_at($linenumbers_start);
    }
    if (isset($title)) {
      $source_code = '<div class="geshifilter-title">' . check_plain($title) . '</div>';
    }
    else {
      $source_code = '';
    }
    $source_code .= '<div class="geshifilter">' . $geshi
      ->parse_code() . '</div>';
  }

  // Store in cache with a minimum expiration time of 1 day.
  cache_set($cache_id, $source_code, 'cache_filter', time() + 60 * 60 * 24);
  return $source_code;
}

/**
 * geshifilter wrapper for highlight_string() processing of PHP.
 */
function geshifilter_highlight_string_process($source_code, $inline_mode) {

  // Make sure that the source code starts with < ?php and ends with ? >
  $text = trim($source_code);
  if (substr($text, 0, 5) != '<?php') {
    $source_code = '<?php' . $source_code;
  }
  if (substr($text, -2) != '?>') {
    $source_code = $source_code . '?>';
  }

  // Use the right container
  $container = $inline_mode ? 'span' : 'div';

  // Process with highlight_string()
  $text = '<' . $container . ' class="codeblock geshifilter">' . highlight_string($source_code, TRUE) . '</' . $container . '>';

  // Remove newlines (added by highlight_string()) to avoid issues with the linebreak filter
  $text = str_replace("\n", '', $text);
  return $text;
}
Functions

Name	Description
geshifilter_geshi_process	geshifilter wrapper for GeSHi processing.
geshifilter_highlight_string_process	geshifilter wrapper for highlight_string() processing of PHP.
geshifilter_process_sourcecode	General geshifilter processing function for a chunk of source code.
_geshifilter_override_geshi_defaults	Helper function for overriding some GeSHi defaults.
_geshifilter_parse_attributes	Helper function for parsing the attributes of GeSHi code tags. to get the settings for language, line numbers, etc.
_geshifilter_prepare	geshifilter_filter callback for preparing input text.
_geshifilter_prepare_callback	_geshifilter_prepare callback for preparing input text. Replaces the code tags brackets with geshifilter specific ones to prevent possible messing up by other filters, e.g. '[python]foo[/python]' to…
_geshifilter_prepare_php_callback	_geshifilter_prepare callback for < ?php ... ? > blocks.
_geshifilter_process	geshifilter_filter callback for processing input text.
_geshifilter_replace_callback	preg_replace_callback callback.
You are here

geshifilter.pages.inc in GeSHi Filter for syntax highlighting 7

File

Functions

API Navigation
