diff options
| author | Christian Schneider | 2012-03-28 13:00:39 +0000 |
|---|---|---|
| committer | Christian Schneider | 2012-03-28 13:00:39 +0000 |
| commit | b1c0b4946572027c8de564730a89ec584c830bf3 (patch) | |
| tree | 81859c280f43a5cdb31ddede7300641fa530e0f8 /it_html.class | |
| parent | 49d2a5ce1b6ad201f051263db7c3a1f5ad6a39ab (diff) | |
| download | itools-b1c0b4946572027c8de564730a89ec584c830bf3.tar.gz itools-b1c0b4946572027c8de564730a89ec584c830bf3.tar.bz2 itools-b1c0b4946572027c8de564730a89ec584c830bf3.zip | |
Added it::any2utf8, fixed it::replace fast path to add u modified, added error reporting for invalid utf-8 input to it::match and it::replace
Diffstat (limited to 'it_html.class')
| -rw-r--r-- | it_html.class | 4 |
1 files changed, 3 insertions, 1 deletions
diff --git a/it_html.class b/it_html.class index 71ac965..f20c4be 100644 --- a/it_html.class +++ b/it_html.class @@ -431,9 +431,11 @@ function _strip_tags($html) function sanitize($html) { $result = ""; + $charset = $GLOBALS['it_html']->p['charset'] ? $GLOBALS['it_html']->p['charset'] : 'iso-8859-1'; + if ($charset == "utf-8") + $html = it::any2utf8($html); $html = it::replace(array('[\0\s]+' => " "), $html); # \s also matches \r and \n $urlpattern = 'https?://[^">]+'; - $charset = $GLOBALS['it_html']->p['charset'] ? $GLOBALS['it_html']->p['charset'] : 'iso-8859-1'; if ($tag = it::match("(.*)<(div|p|i|b)\b[^>]*>(.*?)</\\2>(.*)", $html)) { |