diff options
| author | Ludovic Arnaud <ludovic_arnaud@users.sourceforge.net> | 2006-07-15 17:01:59 +0000 |
|---|---|---|
| committer | Ludovic Arnaud <ludovic_arnaud@users.sourceforge.net> | 2006-07-15 17:01:59 +0000 |
| commit | 29d92430c5e364d63a5ec7db96d306cdf1c02e72 (patch) | |
| tree | 94b33eb4813a4d1a8046ec5ec86020210ca78bf9 /phpBB/includes/search | |
| parent | 7b8f0da356ea2fdc852a9ca719ed6909636d0d1c (diff) | |
| download | forums-29d92430c5e364d63a5ec7db96d306cdf1c02e72.tar forums-29d92430c5e364d63a5ec7db96d306cdf1c02e72.tar.gz forums-29d92430c5e364d63a5ec7db96d306cdf1c02e72.tar.bz2 forums-29d92430c5e364d63a5ec7db96d306cdf1c02e72.tar.xz forums-29d92430c5e364d63a5ec7db96d306cdf1c02e72.zip | |
Changed: moved functions that encode/decode NCRs from and to UTF-8 to utf_tools.php
git-svn-id: file:///svn/phpbb/trunk@6187 89ea8834-ac86-4346-8a33-228a782c2dd0
Diffstat (limited to 'phpBB/includes/search')
| -rw-r--r-- | phpBB/includes/search/fulltext_native_improved.php | 77 |
1 files changed, 9 insertions, 68 deletions
diff --git a/phpBB/includes/search/fulltext_native_improved.php b/phpBB/includes/search/fulltext_native_improved.php index dac964a64a..a01ccd64ed 100644 --- a/phpBB/includes/search/fulltext_native_improved.php +++ b/phpBB/includes/search/fulltext_native_improved.php @@ -47,10 +47,18 @@ class fulltext_native_improved extends search_backend $this->word_length = array('min' => $config['fulltext_native_min_chars'], 'max' => $config['fulltext_native_max_chars']); + /** + * Load the UTF tools + */ if (!class_exists('utf_normalizer')) { include($phpbb_root_path . 'includes/utf/utf_normalizer.' . $phpEx); } + if (!function_exists('utf8_strlen')) + { + include($phpbb_root_path . 'includes/utf/utf_tools.' . $phpEx); + } + $error = false; } @@ -865,14 +873,6 @@ class fulltext_native_improved extends search_backend $isset_min = $min - 1; /** - * Load the UTF tools - */ - if (!function_exists('utf8_strlen')) - { - include($phpbb_root_path . 'includes/utf/utf_tools.' . $phpEx); - } - - /** * Clean up the string, remove HTML tags, remove BBCodes */ $word = strtok($this->cleanup(preg_replace($match, ' ', strip_tags($text)), '', $user->lang['ENCODING']), ' '); @@ -1259,11 +1259,6 @@ class fulltext_native_improved extends search_backend $encoding = strtolower($encoding); if ($encoding != 'utf-8') { - if (!function_exists('utf8_recode')) - { - include($phpbb_root_path . 'includes/utf/utf_tools.' . $phpEx); - } - $text = utf8_recode($text, $encoding); } @@ -1277,7 +1272,7 @@ class fulltext_native_improved extends search_backend /** * Replace HTML entities and NCRs */ - $text = html_entity_decode($this->decode_ncr($text), ENT_QUOTES); + $text = html_entity_decode(utf8_decode_ncr($text), ENT_QUOTES); /** * Load the UTF-8 normalizer @@ -1482,60 +1477,6 @@ class fulltext_native_improved extends search_backend } /** - * Convert Numeric Character References to UTF-8 chars - * - * Notes: - * - we do not convert NCRs recursively, if you pass &#38; it will return & - * - we DO NOT check for the existence of the Unicode characters, therefore an entity - * may be converted to an inexistent codepoint - * - * @param string $text String to convert, encoded in UTF-8 (no normal form required) - * @return string UTF-8 string where NCRs have been replaced with the actual chars - */ - function decode_ncr($text) - { - /** - * @todo replace me with preg_replace_callback() or a loop - */ - return preg_replace( - '/&#([0-9]{1,6});/e', - "\$this->cp_to_utf(\$1)", - - preg_replace( - '/&#x([0-9A-F]{1,5});/ie', - "\$this->cp_to_utf(hexdec('\$1'))", - $text - ) - ); - } - - /** - * Convert a codepoint to a UTF-8 char - * - * @param integer $cp Unicode codepoint - * @return string UTF-8 string - */ - function cp_to_utf($cp) - { - if ($cp > 0xFFFF) - { - return chr(0xF0 | ($cp >> 18)) . chr(0x80 | (($cp >> 12) & 0x3F)) . chr(0x80 | (($cp >> 6) & 0x3F)) . chr(0x80 | ($cp & 0x3F)); - } - elseif ($cp > 0x7FF) - { - return chr(0xE0 | ($cp >> 12)) . chr(0x80 | (($cp >> 6) & 0x3F)) . chr(0x80 | ($cp & 0x3F)); - } - elseif ($cp > 0x7F) - { - return chr(0xC0 | ($cp >> 6)) . chr(0x80 | ($cp & 0x3F)); - } - else - { - return chr($cp); - } - } - - /** * Returns a list of options for the ACP to display */ function acp() |