diff options
| -rw-r--r-- | system/helpers/text_helper.php | 17 | ||||
| -rw-r--r-- | user_guide/changelog.html | 1 |
2 files changed, 13 insertions, 5 deletions
diff --git a/system/helpers/text_helper.php b/system/helpers/text_helper.php index 6e61f776a..e79a2419d 100644 --- a/system/helpers/text_helper.php +++ b/system/helpers/text_helper.php @@ -230,21 +230,28 @@ if ( ! function_exists('word_censor')) { return $str; } + + $str = ' '.$str.' '; + + // \w, \b and a few others do not match on a unicode character + // set for performance reasons. As a result words like über + // will not match on a word boundary. Instead, we'll assume that + // a bad word will be bookeneded by any of these characters. + $delim = '[-_\'\"`(){}<>\[\]|!?@#%&,.:;^~*+=\/ 0-9\n\r\t]'; - $str = ' '.$str.' '; foreach ($censored as $badword) { if ($replacement != '') { - $str = preg_replace("/\b(".str_replace('\*', '\w*?', preg_quote($badword, '/')).")\b/i", $replacement, $str); + $str = preg_replace("/({$delim})(".str_replace('\*', '\w*?', preg_quote($badword, '/')).")({$delim})/i", "\\1{$replacement}\\3", $str); } else { - $str = preg_replace("/\b(".str_replace('\*', '\w*?', preg_quote($badword, '/')).")\b/ie", "str_repeat('#', strlen('\\1'))", $str); + $str = preg_replace("/({$delim})(".str_replace('\*', '\w*?', preg_quote($badword, '/')).")({$delim})/ie", "'\\1'.str_repeat('#', strlen('\\2')).'\\3'", $str); } } - - return trim($str); + + return trim($str); } } diff --git a/user_guide/changelog.html b/user_guide/changelog.html index dbe4c6653..501529658 100644 --- a/user_guide/changelog.html +++ b/user_guide/changelog.html @@ -106,6 +106,7 @@ SVN Revision: </p> <li>Fixed a bug with non-breaking space cleanup in Typography::auto_typography().</li> <li>Fixed a bug in database escaping where a compound statement (ie: SUM()) wasn't handled correctly with database prefixes.</li> <li>Fixed a bug when an opening quote is preceded by a paragraph tag and immediately followed by another tag.</li> + <li>Fixed a bug in the Text Helper affecting some locales where word_censor() would not work on words beginning or ending with an accented character.</li> <li>Tweaked Typography::auto_typography() for some edge-cases.</li> </ul> |