? geshifilter_process_0.txt Index: geshifilter.module =================================================================== RCS file: /cvs/drupal-contrib/contributions/modules/geshifilter/geshifilter.module,v retrieving revision 1.18.2.22 diff -u -u -p -r1.18.2.22 geshifilter.module --- geshifilter.module 14 Oct 2007 12:13:56 -0000 1.18.2.22 +++ geshifilter.module 14 Oct 2007 12:17:00 -0000 @@ -298,6 +298,9 @@ function geshifilter_filter($op, $delta case 'description': return t('Enables syntax highlighting of inline/block source code using the GeSHi engine'); + case 'prepare': + return _geshifilter_prepare($format, $text); + case 'process': return _geshifilter_process($format, $text); @@ -686,11 +689,7 @@ function geshifilter_admin_filter_confli $order_conflict_filters = array( 'filter/0' => array( // HTML filter 'order' => 'after', - 'description' => 'It is recommended to put %geshifilter after %cfilter to prevent loss of layout and highlighting.', - ), - 'filter/2' => array( // line break convertor - 'order' => 'before', - 'description' => 'It is suggested to put %geshifilter before %cfilter to prevent issues with inline source code.', + 'description' => 'It is required to put %geshifilter after %cfilter to prevent loss of layout and highlighting.', ), 'pearwiki_filter/0' => array( 'order' => 'after', @@ -720,32 +719,6 @@ function geshifilter_admin_filter_confli } } } - // check HTML filter allowed tags - if (isset($filters['filter/0']) && ($filters['filter/0']->weight <= $geshifilter->weight) - && (_geshifilter_brackets($format)==GESHIFILTER_BRACKETS_ANGLE || _geshifilter_brackets($format) == GESHIFILTER_BRACKETS_BOTH) - ) { - $cfilter = $filters['filter/0']; - $html_filter_allowed_tags = preg_split('/\s+|<|>/', variable_get("allowed_html_$format", '
    1. '), -1, PREG_SPLIT_NO_EMPTY); - // collect geshifilter tags - list($generic_code_tags, $language_tags, $tag_to_lang) = _geshifilter_get_tags($format); - $geshifilter_tags = array_merge($generic_code_tags, $language_tags); - foreach ($geshifilter_tags as $tag) { - if (!in_array($tag, $html_filter_allowed_tags)) { - $alerts[] = array( - l(t($input_format->name), "admin/settings/filters/$format"), - $cfilter->name, - t('%cfilter will remove %tag tags', array('%cfilter' => $cfilter->name, '%tag' => "<$tag>")), - t('Add tag %tag to !allowed_html_tags', array( - '%tag' => "<$tag>", - '!allowed_html_tags' => l(t('"%allowed_html_tags" of %cfilter', array( - '%allowed_html_tags' => t('Allowed HTML tags'), - '%cfilter' => $cfilter->name, - )), "admin/settings/filters/$format/configure", array(), drupal_get_destination(), NULL, FALSE, TRUE) - )), - ); - } - } - } } } if ($check_only) { @@ -912,20 +885,13 @@ function _geshifilter_get_tags($format) } /** - * geshifilter_filter callback for processing input text. + * geshifilter_filter callback for preparing input text. */ -function _geshifilter_process($format, $text) { - // load GeSHi library (if not already) - $geshi_library = _geshifilter_check_geshi_library(); - if (!$geshi_library['success']) { - drupal_set_message($geshi_library['message'], 'error'); - return $text; - } +function _geshifilter_prepare($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); $tags_string = implode('|', $tags); - // Pattern for matching "..." like stuff // Also matches "...$" where "$" refers to end of string, not end of // line (because PCRE_MULTILINE (modifier 'm') is not enabled), so matching @@ -941,7 +907,46 @@ function _geshifilter_process($format, $ $pattern = '#[<\[]('. $tags_string .')([^>\]]*)[>\]](.*?)([<\[]/\1\s*[>\]]|$)#s'; break; } - return preg_replace_callback($pattern, '_geshifilter_replace_callback', $text); + // replace the code container tag brackets + // and prepare the container content (newline and angle bracket protection) + $text = preg_replace_callback($pattern, '_geshifilter_prepare_callback', $text); + return $text; +} + +/** + * _geshifilter_prepare callback for preparing input text. + * Replaces the code tags brackets with 0xFA and 0xFB + * (e.g. "[code]" -> "\xFAcode\xFB"). + * Replaces newlines with " " and angle brackets with 0xFC and 0xFD + * to prevent messing up by other filters like the HTML filter + * and line break filter. + * The used bytes 0xFA to 0xFD are not valid UTF-8 data thus least likely to + * cause problems (based on the codefilter module). + */ +function _geshifilter_prepare_callback($match) { + return "\xFA". $match[1] . $match[2] ."\xFB" + . str_replace(array("\r", "\n", '<', '>'), array('', ' ', "\xFC", "\xFD"), $match[3]) + ."\xFA/". $match[1]. "\xFB"; +} + +/** + * geshifilter_filter callback for processing input text. + */ +function _geshifilter_process($format, $text) { + // load GeSHi library (if not already) + $geshi_library = _geshifilter_check_geshi_library(); + if (!$geshi_library['success']) { + drupal_set_message($geshi_library['message'], 'error'); + return $text; + } + // get the available tags + list($generic_code_tags, $language_tags, $tag_to_lang) = _geshifilter_get_tags($format); + $tags = array_merge($generic_code_tags, $language_tags); + $tags_string = implode('|', $tags); + // Pattern for matching the prepared "..." stuff + $pattern = '#\xFA('. $tags_string .')([^\xFB]*)\xFB(.*?)(\xFA/\1\xFB)#s'; + $text = preg_replace_callback($pattern, '_geshifilter_replace_callback', $text); + return $text; } /** @@ -957,6 +962,9 @@ function _geshifilter_replace_callback($ $tag_attributes = $match[2]; $source_code = $match[3]; + // Undo linebreak and angle bracket escaping from preparation phase + $source_code = str_replace(array(' ', "\xFC", "\xFD"), array("\n", '<', '>'), $source_code); + // get the possible tags and languages list($generic_code_tags, $language_tags, $tag_to_lang) = _geshifilter_get_tags($format); $enabled_languages = _geshifilter_get_enabled_languages(); @@ -1051,8 +1059,6 @@ function geshifilter_geshi_process($sour } // decode html entities, GeSHi will reencode these $source_code = decode_entities($source_code); - // remove newline (
      ) and paragraph (

      ) formatting, GeSHi will add these again - $source_code = preg_replace('@

      |

      |@', "", $source_code); // remove leading/trailing newlines $source_code = trim($source_code, "\n\r"); // create GeSHi object