From 5a7caf65084369422fdbe4aaaae46e5de69cc594 Mon Sep 17 00:00:00 2001 From: Nicolas Grekas Date: Wed, 20 Nov 2013 13:47:31 +0100 Subject: [feature/patchwork-utf8] Normalize with intl, use patchwork/utf8 as fallback --- phpBB/phpbb/search/fulltext_native.php | 34 ++++++++++++++++------------------ 1 file changed, 16 insertions(+), 18 deletions(-) (limited to 'phpBB/phpbb/search/fulltext_native.php') diff --git a/phpBB/phpbb/search/fulltext_native.php b/phpBB/phpbb/search/fulltext_native.php index 34947b176d..ea239c0b36 100644 --- a/phpBB/phpbb/search/fulltext_native.php +++ b/phpBB/phpbb/search/fulltext_native.php @@ -18,6 +18,13 @@ namespace phpbb\search; */ class fulltext_native extends \phpbb\search\base { + const UTF8_HANGUL_FIRST = "\xEA\xB0\x80"; + const UTF8_HANGUL_LAST = "\xED\x9E\xA3"; + const UTF8_CJK_FIRST = "\xE4\xB8\x80"; + const UTF8_CJK_LAST = "\xE9\xBE\xBB"; + const UTF8_CJK_B_FIRST = "\xF0\xA0\x80\x80"; + const UTF8_CJK_B_LAST = "\xF0\xAA\x9B\x96"; + /** * Associative array holding index stats * @var array @@ -93,7 +100,7 @@ class fulltext_native extends \phpbb\search\base protected $user; /** - * Initialises the fulltext_native search backend with min/max word length and makes sure the UTF-8 normalizer is loaded + * Initialises the fulltext_native search backend with min/max word length * * @param boolean|string &$error is passed by reference and should either be set to false on success or an error message on failure */ @@ -110,10 +117,6 @@ class fulltext_native extends \phpbb\search\base /** * Load the UTF tools */ - if (!class_exists('utf_normalizer')) - { - include($this->phpbb_root_path . 'includes/utf/utf_normalizer.' . $this->php_ext); - } if (!function_exists('utf8_decode_ncr')) { include($this->phpbb_root_path . 'includes/utf/utf_tools.' . $this->php_ext); @@ -1175,9 +1178,9 @@ class fulltext_native extends \phpbb\search\base * Note: this could be optimized. If the codepoint is lower than Hangul's range * we know that it will also be lower than CJK ranges */ - if ((strncmp($word, UTF8_HANGUL_FIRST, 3) < 0 || strncmp($word, UTF8_HANGUL_LAST, 3) > 0) - && (strncmp($word, UTF8_CJK_FIRST, 3) < 0 || strncmp($word, UTF8_CJK_LAST, 3) > 0) - && (strncmp($word, UTF8_CJK_B_FIRST, 4) < 0 || strncmp($word, UTF8_CJK_B_LAST, 4) > 0)) + if ((strncmp($word, self::UTF8_HANGUL_FIRST, 3) < 0 || strncmp($word, self::UTF8_HANGUL_LAST, 3) > 0) + && (strncmp($word, self::UTF8_CJK_FIRST, 3) < 0 || strncmp($word, self::UTF8_CJK_LAST, 3) > 0) + && (strncmp($word, self::UTF8_CJK_B_FIRST, 4) < 0 || strncmp($word, self::UTF8_CJK_B_LAST, 4) > 0)) { $word = strtok(' '); continue; @@ -1544,8 +1547,6 @@ class fulltext_native extends \phpbb\search\base * @param string $allowed_chars String of special chars to allow * @param string $encoding Text encoding * @return string Cleaned up text, only alphanumeric chars are left - * - * @todo \normalizer::cleanup being able to be used? */ protected function cleanup($text, $allowed_chars = null, $encoding = 'utf-8') { @@ -1572,12 +1573,9 @@ class fulltext_native extends \phpbb\search\base $text = htmlspecialchars_decode(utf8_decode_ncr($text), ENT_QUOTES); /** - * Load the UTF-8 normalizer - * - * If we use it more widely, an instance of that class should be held in a - * a global variable instead + * Normalize to NFC */ - \utf_normalizer::nfc($text); + $text = \Normalizer::normalize($text); /** * The first thing we do is: @@ -1670,9 +1668,9 @@ class fulltext_native extends \phpbb\search\base $utf_char = substr($text, $pos, $utf_len); $pos += $utf_len; - if (($utf_char >= UTF8_HANGUL_FIRST && $utf_char <= UTF8_HANGUL_LAST) - || ($utf_char >= UTF8_CJK_FIRST && $utf_char <= UTF8_CJK_LAST) - || ($utf_char >= UTF8_CJK_B_FIRST && $utf_char <= UTF8_CJK_B_LAST)) + if (($utf_char >= self::UTF8_HANGUL_FIRST && $utf_char <= self::UTF8_HANGUL_LAST) + || ($utf_char >= self::UTF8_CJK_FIRST && $utf_char <= self::UTF8_CJK_LAST) + || ($utf_char >= self::UTF8_CJK_B_FIRST && $utf_char <= self::UTF8_CJK_B_LAST)) { /** * All characters within these ranges are valid -- cgit v1.2.1