aboutsummaryrefslogtreecommitdiffstats
diff options
context:
space:
mode:
authorNils Adermann <naderman@naderman.de>2006-04-27 14:20:43 +0000
committerNils Adermann <naderman@naderman.de>2006-04-27 14:20:43 +0000
commit5994371c0a098bf2f866b89ba0a8bba11f414a41 (patch)
tree78b0d65900b7f969702bac61142010766149ea80
parent5f503ddabf90693d0293d2c4fe91c9d965aa86de (diff)
downloadforums-5994371c0a098bf2f866b89ba0a8bba11f414a41.tar
forums-5994371c0a098bf2f866b89ba0a8bba11f414a41.tar.gz
forums-5994371c0a098bf2f866b89ba0a8bba11f414a41.tar.bz2
forums-5994371c0a098bf2f866b89ba0a8bba11f414a41.tar.xz
forums-5994371c0a098bf2f866b89ba0a8bba11f414a41.zip
- allow searches for multibyte characters in fulltext_native
- use preg_quote safely git-svn-id: file:///svn/phpbb/trunk@5853 89ea8834-ac86-4346-8a33-228a782c2dd0
Diffstat
-rw-r--r--phpBB/includes/functions_compress.php2
-rw-r--r--phpBB/includes/functions_messenger.php2
-rw-r--r--phpBB/includes/functions_privmsgs.php20
-rwxr-xr-xphpBB/includes/search/fulltext_native.php42
-rw-r--r--phpBB/search.php6
5 files changed, 43 insertions, 29 deletions
diff --git a/phpBB/includes/functions_compress.php b/phpBB/includes/functions_compress.php
index 700303fd7d..51fbdab6ae 100644
--- a/phpBB/includes/functions_compress.php
+++ b/phpBB/includes/functions_compress.php
@@ -23,7 +23,7 @@ class compress
$skip_files = explode(',', $skip_files);
// Remove rm prefix from src path
- $src_path = ($src_rm_prefix) ? preg_replace('#^(' . preg_quote($src_rm_prefix) . ')#', '', $src) : $src;
+ $src_path = ($src_rm_prefix) ? preg_replace('#^(' . preg_quote($src_rm_prefix, '#') . ')#', '', $src) : $src;
// Add src prefix
$src_path = ($src_add_prefix) ? ($src_add_prefix . ((substr($src_add_prefix, -1) != '/') ? '/' : '') . $src_path) : $src_path;
// Remove initial "/" if present
diff --git a/phpBB/includes/functions_messenger.php b/phpBB/includes/functions_messenger.php
index d6755857ab..8e20a418e2 100644
--- a/phpBB/includes/functions_messenger.php
+++ b/phpBB/includes/functions_messenger.php
@@ -1227,7 +1227,7 @@ function mail_encode($str, $encoding)
$str = chunk_split(base64_encode($str), $length, $spacer);
// remove trailing spacer and add start and end delimiters
- $str = preg_replace('#' . preg_quote($spacer) . '$#', '', $str);
+ $str = preg_replace('#' . preg_quote($spacer, '#') . '$#', '', $str);
return $start . $str . $end;
}
diff --git a/phpBB/includes/functions_privmsgs.php b/phpBB/includes/functions_privmsgs.php
index 6adfdf5605..d24db82b09 100644
--- a/phpBB/includes/functions_privmsgs.php
+++ b/phpBB/includes/functions_privmsgs.php
@@ -53,28 +53,28 @@ define('CHECK_TO', 5);
$global_privmsgs_rules = array(
CHECK_SUBJECT => array(
- RULE_IS_LIKE => array('check0' => 'message_subject', 'function' => 'preg_match("/" . preg_quote({STRING}) . "/i", {CHECK0})'),
- RULE_IS_NOT_LIKE => array('check0' => 'message_subject', 'function' => '!(preg_match("/" . preg_quote({STRING}) . "/i", {CHECK0}))'),
+ RULE_IS_LIKE => array('check0' => 'message_subject', 'function' => 'preg_match("/" . preg_quote({STRING}, "/") . "/i", {CHECK0})'),
+ RULE_IS_NOT_LIKE => array('check0' => 'message_subject', 'function' => '!(preg_match("/" . preg_quote({STRING}, "/") . "/i", {CHECK0}))'),
RULE_IS => array('check0' => 'message_subject', 'function' => '{CHECK0} == {STRING}'),
RULE_IS_NOT => array('check0' => 'message_subject', 'function' => '{CHECK0} != {STRING}'),
- RULE_BEGINS_WITH => array('check0' => 'message_subject', 'function' => 'preg_match("/^" . preg_quote({STRING}) . "/i", {CHECK0})'),
- RULE_ENDS_WITH => array('check0' => 'message_subject', 'function' => 'preg_match("/" . preg_quote({STRING}) . "$/i", {CHECK0})')),
+ RULE_BEGINS_WITH => array('check0' => 'message_subject', 'function' => 'preg_match("/^" . preg_quote({STRING}, "/") . "/i", {CHECK0})'),
+ RULE_ENDS_WITH => array('check0' => 'message_subject', 'function' => 'preg_match("/" . preg_quote({STRING}, "/") . "$/i", {CHECK0})')),
CHECK_SENDER => array(
- RULE_IS_LIKE => array('check0' => 'username', 'function' => 'preg_match("/" . preg_quote({STRING}) . "/i", {CHECK0})'),
- RULE_IS_NOT_LIKE => array('check0' => 'username', 'function' => '!(preg_match("/" . preg_quote({STRING}) . "/i", {CHECK0}))'),
+ RULE_IS_LIKE => array('check0' => 'username', 'function' => 'preg_match("/" . preg_quote({STRING}, "/") . "/i", {CHECK0})'),
+ RULE_IS_NOT_LIKE => array('check0' => 'username', 'function' => '!(preg_match("/" . preg_quote({STRING}, "/") . "/i", {CHECK0}))'),
RULE_IS => array('check0' => 'username', 'function' => '{CHECK0} == {STRING}'),
RULE_IS_NOT => array('check0' => 'username', 'function' => '{CHECK0} != {STRING}'),
- RULE_BEGINS_WITH => array('check0' => 'username', 'function' => 'preg_match("/^" . preg_quote({STRING}) . "/i", {CHECK0})'),
- RULE_ENDS_WITH => array('check0' => 'username', 'function' => 'preg_match("/" . preg_quote({STRING}) . "$/i", {CHECK0})'),
+ RULE_BEGINS_WITH => array('check0' => 'username', 'function' => 'preg_match("/^" . preg_quote({STRING}, "/") . "/i", {CHECK0})'),
+ RULE_ENDS_WITH => array('check0' => 'username', 'function' => 'preg_match("/" . preg_quote({STRING}, "/") . "$/i", {CHECK0})'),
RULE_IS_FRIEND => array('check0' => 'friend', 'function' => '{CHECK0} == 1'),
RULE_IS_FOE => array('check0' => 'foe', 'function' => '{CHECK0} == 1'),
RULE_IS_USER => array('check0' => 'author_id', 'function' => '{CHECK0} == {USER_ID}'),
RULE_IS_GROUP => array('check0' => 'author_in_group', 'function' => 'in_array({GROUP_ID}, {CHECK0})')),
CHECK_MESSAGE => array(
- RULE_IS_LIKE => array('check0' => 'message_text', 'function' => 'preg_match("/" . preg_quote({STRING}) . "/i", {CHECK0})'),
- RULE_IS_NOT_LIKE => array('check0' => 'message_text', 'function' => '!(preg_match("/" . preg_quote({STRING}) . "/i", {CHECK0}))'),
+ RULE_IS_LIKE => array('check0' => 'message_text', 'function' => 'preg_match("/" . preg_quote({STRING}, "/") . "/i", {CHECK0})'),
+ RULE_IS_NOT_LIKE => array('check0' => 'message_text', 'function' => '!(preg_match("/" . preg_quote({STRING}, "/") . "/i", {CHECK0}))'),
RULE_IS => array('check0' => 'message_text', 'function' => '{CHECK0} == {STRING}'),
RULE_IS_NOT => array('check0' => 'message_text', 'function' => '{CHECK0} != {STRING}')),
diff --git a/phpBB/includes/search/fulltext_native.php b/phpBB/includes/search/fulltext_native.php
index b69dc18061..48415ab409 100755
--- a/phpBB/includes/search/fulltext_native.php
+++ b/phpBB/includes/search/fulltext_native.php
@@ -50,8 +50,8 @@ class fulltext_native extends search_backend
{
global $db, $config;
- $drop_char_match = array('^', '$', ';', '#', '&', '(', ')', '<', '>', '`', '\'', '"', ',', '@', '_', '?', '%', '~', '.', '[', ']', '{', '}', ':', '\\', '/', '=', '!');
- $drop_char_replace = array(' ', ' ', ' ', ' ', ' ', ' ', ' ', ' ', ' ', '', '', ' ', ' ', ' ', '', ' ', ' ', ' ', ' ', ' ', ' ', ' ', ' ', ' ', '' , ' ', ' ', ' ');
+ $drop_char_match = array('^', '$', '(', ')', '<', '>', '`', '\'', '"', ',', '@', '_', '?', '%', '~', '.', '[', ']', '{', '}', ':', '\\', '/', '=', '!', "\n", "\r");
+ $drop_char_replace = array(' ', ' ', ' ', ' ', ' ', '', '', ' ', ' ', ' ', '', ' ', ' ', ' ', ' ', ' ', ' ', ' ', ' ', ' ', '' , ' ', ' ', ' ', ' ', ' ', ' ');
$this->get_ignore_words();
$this->get_synonyms();
@@ -65,15 +65,20 @@ class fulltext_native extends search_backend
}
$match = array();
- // New lines, carriage returns
- $match[] = "#[\n\r]+#";
// NCRs like &nbsp; etc.
- $match[] = '#(&amp;|&)[\#a-z0-9]+?;#i';
+ $match[] = '#(&amp;|&)[a-z0-9]+?;#i';
// Filter out as above
$keywords = preg_replace($match, ' ', strtolower(trim($keywords)));
+
+ // Filter out non alphabetical characters
$keywords = str_replace($drop_char_match, $drop_char_replace, $keywords);
+ // Filter out ; and # but not &#[0-9]+;
+ $keywords = preg_replace('#&\#([0-9]+);#', '<$1>', $keywords);
+ $keywords = str_replace(array(';', '&', '#'), ' ', $keywords);
+ $keywords = str_replace(array('<', '>'), array('&#', ';'), $keywords);
+
// Split words
$this->split_words = explode(' ', preg_replace('#\s+#', ' ', $keywords));
@@ -100,7 +105,7 @@ class fulltext_native extends search_backend
}
// check word length
- $clean_len = strlen(str_replace('*', '', $word));
+ $clean_len = $this->word_length($word);
if (($clean_len < $config['fulltext_native_min_chars']) || ($clean_len > $config['fulltext_native_max_chars']))
{
if ($prefixed)
@@ -152,6 +157,14 @@ class fulltext_native extends search_backend
}
/**
+ * Returns the string length but it counts multibyte characters as single characters and ignores "*"
+ */
+ function word_length($word)
+ {
+ return strlen(str_replace('*', '', preg_replace('#&\#[0-9]+;#', 'x', $word)));
+ }
+
+ /**
* Turns text into an array of words that can be stored in the word list table
*/
function split_message($text)
@@ -165,29 +178,30 @@ class fulltext_native extends search_backend
if (!is_array($drop_char_match))
{
- $drop_char_match = array('-', '^', '$', ';', '#', '&', '(', ')', '<', '>', '`', '\'', '"', '|', ',', '@', '_', '?', '%', '~', '.', '[', ']', '{', '}', ':', '\\', '/', '=', '\'', '!', '*', '+');
- $drop_char_replace = array(' ', ' ', ' ', ' ', ' ', ' ', ' ', ' ', ' ', ' ', '', '', ' ', ' ', ' ', ' ', '', ' ', ' ', ' ', ' ', ' ', ' ', ' ', ' ', ' ', '' , ' ', ' ', ' ', ' ', ' ', ' ');
+ $drop_char_match = array('-', '^', '$', '(', ')', '<', '>', '`', '\'', '"', '|', ',', '@', '_', '?', '%', '~', '.', '[', ']', '{', '}', ':', '\\', '/', '=', '\'', '!', '*', '+', "\n", "\r");
+ $drop_char_replace = array(' ', ' ', ' ', ' ', ' ', ' ', ' ', '', '', ' ', ' ', ' ', ' ', '', ' ', ' ', ' ', ' ', ' ', ' ', ' ', ' ', ' ', '' , ' ', ' ', ' ', ' ', ' ', ' ', ' ', ' ');
}
$match = array();
// Comments for hardcoded bbcode elements (urls, smilies, html)
$match[] = '#<!\-\- .* \-\->(.*?)<!\-\- .* \-\->#is';
- // New lines, carriage returns
- $match[] = "#[\n\r]+#";
// NCRs like &nbsp; etc.
- $match[] = '#(&amp;|&)[\#a-z0-9]+?;#i';
+ $match[] = '#(&amp;|&)[a-z0-9]+;#i';
// Do not index code
$match[] = '#\[code(?:=.*?)?(\:?[0-9a-z]{5,})\].*?\[\/code(\:?[0-9a-z]{5,})\]#is';
// BBcode
$match[] = '#\[\/?[a-z\*\+\-]+(?:=.*?)?(\:?[0-9a-z]{5,})\]#';
- // Filter out ; and # but not &#[0-9]+;
- //$match[] = '#(&\#[0-9]+;)|;|\#|&#';
$text = preg_replace($match, ' ', ' ' . strtolower(trim($text)) . ' ');
// Filter out non-alphabetical chars
$text = str_replace($drop_char_match, $drop_char_replace, $text);
+ // Filter out ; and # but not &#[0-9]+;
+ $text = preg_replace('#&\#([0-9]+);#', '<$1>', $text);
+ $text = str_replace(array(';', '&', '#'), ' ', $text);
+ $text = str_replace(array('<', '>'), array('&#', ';'), $text);
+
// Split words
$text = explode(' ', preg_replace('#\s+#', ' ', trim($text)));
@@ -206,7 +220,7 @@ class fulltext_native extends search_backend
for ($i = 0, $n = sizeof($text); $i < $n; $i++)
{
$text[$i] = trim($text[$i]);
- if (strlen($text[$i]) < $config['fulltext_native_min_chars'] || strlen($text[$i]) > $config['fulltext_native_max_chars'])
+ if ($this->word_length($text[$i]) < $config['fulltext_native_min_chars'] || $this->word_length($text[$i]) > $config['fulltext_native_max_chars'])
{
unset($text[$i]);
}
diff --git a/phpBB/search.php b/phpBB/search.php
index 80ad3dccbe..2095bc1e70 100644
--- a/phpBB/search.php
+++ b/phpBB/search.php
@@ -29,8 +29,8 @@ $topic_id = request_var('t', 0);
$view = request_var('view', '');
$submit = request_var('submit', false);
-$keywords = request_var('keywords', '');
-$add_keywords = request_var('add_keywords', '');
+$keywords = request_var('keywords', '', true);
+$add_keywords = request_var('add_keywords', '', true);
$author = request_var('author', '');
$show_results = ($topic_id) ? 'posts' : request_var('sr', 'posts');
$show_results = ($show_results == 'posts') ? 'posts' : 'topics';
@@ -657,7 +657,7 @@ if ($keywords || $author || $search_id || $submit)
if ($hilit)
{
- $row['post_text'] = preg_replace('#(?!<.*)(?<!\w)(' . preg_quote($hilit) . ')(?!\w|[^<>]*>)#i', '<span class="posthilit">$1</span>', $row['post_text']);
+ $row['post_text'] = preg_replace('#(?!<.*)(?<!\w)(' . preg_quote($hilit, '#') . ')(?!\w|[^<>]*>)#i', '<span class="posthilit">$1</span>', $row['post_text']);
}
$row['post_text'] = smiley_text($row['post_text']);
generated by cgit v1.2.1 (git 2.21.0) at 2025-04-01 10:50:18 +0000