Index: codefilter.module =================================================================== RCS file: /cvs/drupal-contrib/contributions/modules/codefilter/codefilter.module,v retrieving revision 1.25 diff -u -r1.25 codefilter.module --- codefilter.module 3 Dec 2007 08:01:51 -0000 1.25 +++ codefilter.module 4 Dec 2007 14:43:58 -0000 @@ -30,6 +30,7 @@ $text = preg_replace('@@', '', str_replace('\"', '"', $text)); // Undo the escaping in the prepare step $text = decode_entities($text); + $text = str_replace("\xFEn\xFF", "\n", $text); // Trim leading and trailing linebreaks $text = trim($text, "\r\n"); // Highlight as PHP @@ -57,7 +58,7 @@ */ function codefilter_process_code($text) { // Undo linebreak escaping - $text = str_replace(' ', "\n", $text); + $text = str_replace("\xFEn\xFF", "\n", $text); // Inline or block level piece? $multiline = strpos($text, "\n") !== FALSE; // Note, pay attention to odd preg_replace-with-/e behaviour on slashes @@ -81,9 +82,98 @@ } function codefilter_escape($text) { - // Note, pay attention to odd preg_replace-with-/e behaviour on slashes // Protect newlines from linebreak filter - return str_replace(array("\r", "\n"), array('', ' '), check_plain(str_replace('\"', '"', $text))); + return str_replace(array("\r", "\n"), array('', "\xFEn\xFF"), check_plain($text)); +} + +function codefilter_prepare_find_block($text, $start_offset, $block_start, $block_end, &$chosen_start, &$chosen_end, &$chosen_starter, &$chosen_ender) { + $position = strpos($text, $block_start, $start_offset); + if ($position !== false) { + $position_end = strpos($text, $block_end, $position); + if ($position_end === false) { + $position = false; + } + else { + if ($position !== false && ($chosen_start === false || $position < $chosen_start)) { + $chosen_start = $position; + $chosen_end = $position_end; + $chosen_starter = $block_start; + $chosen_ender = $block_end; + } + } + } + return $position; +} + +/** + * A regex cannot easily find the end of a block as the terminating ?> may be within a string or multiline comment. + * Therefore, php blocks have to be looked at specially and an intelligent guess has to be made as to which ?> actually terminates + * the block. To do this, for each possible ending to a block, highlight_string is called to see if things after that ending are + * highlighted as non-php-code, and if so, then that must be the correct ending. Otherwise, we move onto the next possible ending + * and see what highlight_string thinks about ending there. + */ +function codefilter_prepare($text) { + $start_offset = 0; + + while(true) { + // Find the next php block, the next mini block and the next code block + $position_start = false; + $position_end = false; + $position_starter = ''; + $position_marker = 'php'; + + $position_php = codefilter_prepare_find_block($text, $start_offset, '', $position_start, $position_end, $position_starter, $position_ender); + $position_mini = codefilter_prepare_find_block($text, $start_offset, '<%', '%>', $position_start, $position_end, $position_starter, $position_ender); + $position_code = codefilter_prepare_find_block($text, $start_offset, '', '', $position_start, $position_end, $position_starter, $position_ender); + + if ($position_php === false && $position_mini === false && $position_code === false) + break; + if ($chosen_starter == '') + $position_marker = 'code'; + + $start_offset = $position_end; + + while($position_end) { + // See what highlight_string thinks about ending the block here + $to_highlight = substr($text, $position_start + strlen($position_starter), $position_end - $position_start - strlen($position_starter)); + $highlighted = highlight_string(' ?>', true); + /* + Iff the block to be highlighted is valid, then the first "?>" will be styled (?>) + and the second one will be unstyled (raw), so get the positions of the last styled and the last raw occurance of "?>" + and if the styled appears before the raw, then the block is valid. highlight_string has an odd bug where it can remove + single quotes (') from a piece of code when the code is invalid, so this is checked for. For example, this call: + highlight_string(" ?>"); + Will remove the single quote, style the penultimate "?>" and leave the final "?>" raw. + */ + $position_styled_close = strrpos($highlighted, '">?>'); + $position_raw_close = strrpos($highlighted, '?>'); + if ($position_styled_close !== false && $position_raw_close !== false && ($position_styled_close + 2) < $position_raw_close) { + if (substr_count($to_highlight, "'") == substr_count($highlighted, "'")) { + break; + } + } + + // Move onto the next possible ending + $position_end = strpos($text, $position_ender, $position_end + 1); + } + if ($position_end === false) { + // No more possible endings, but highlight_string hasn't been happy yet, so revert back to the first possible ending + $position_end = strpos($text, $position_ender, $position_start); + } + + // We've found the ending, so mark this block + $before_code = substr($text, 0, $position_start); + $position_start = $position_start + strlen($position_starter); + $the_code = substr($text, $position_start, $position_end - $position_start); + $position_end = $position_end + strlen($position_ender); + $after_code = substr($text, $position_end); + $text = $before_code ."\xFE". $position_marker . "\xFF". codefilter_escape($the_code) ."\xFE/". $position_marker ."\xFF"; + $start_offset = strlen($text); + $text = $text . $after_code; + } + + return $text; } /** @@ -100,8 +190,7 @@ case 'prepare': // Note: we use the bytes 0xFE and 0xFF to replace < > during the filtering process. // These bytes are not valid in UTF-8 data and thus least likely to cause problems. - $text = preg_replace('@(.+?)@se', "'\xFEcode\xFF'. codefilter_escape('\\1') .'\xFE/code\xFF'", $text); - $text = preg_replace('@[\[<](\?php|%)(.+?)(\?|%)[\]>]@se', "'\xFEphp\xFF'. codefilter_escape('\\2') .'\xFE/php\xFF'", $text); + $text = codefilter_prepare($text); return $text; case 'process':