1
0
mirror of https://github.com/phpbb/phpbb.git synced 2025-06-08 07:27:17 +02:00

Latest variants ... not finished yet ...

git-svn-id: file:///svn/phpbb/trunk@1423 89ea8834-ac86-4346-8a33-228a782c2dd0
This commit is contained in:
Paul S. Owen 2001-11-23 01:04:31 +00:00
parent 6f02536adf
commit ed88c19320
2 changed files with 188 additions and 293 deletions

View File

@ -112,6 +112,8 @@ function remove_common($percent, $word_id_list = array())
$row = $db->sql_fetchrow($result); $row = $db->sql_fetchrow($result);
if( $row['total_posts'] > 100 )
{
$common_threshold = floor($row['total_posts'] * $percent); $common_threshold = floor($row['total_posts'] * $percent);
$sql = "SELECT word_id $sql = "SELECT word_id
@ -125,7 +127,7 @@ function remove_common($percent, $word_id_list = array())
message_die(GENERAL_ERROR, "Couldn't obtain common word list", "", __LINE__, __FILE__, $sql); message_die(GENERAL_ERROR, "Couldn't obtain common word list", "", __LINE__, __FILE__, $sql);
} }
if( $post_count = $db->sql_numrows($result) ) if( $word_count = $db->sql_numrows($result) )
{ {
$common_word_id_list = array(); $common_word_id_list = array();
while( $row = $db->sql_fetchrow($result) ) while( $row = $db->sql_fetchrow($result) )
@ -140,7 +142,7 @@ function remove_common($percent, $word_id_list = array())
$common_word_id_list = implode(", ", $common_word_id_list); $common_word_id_list = implode(", ", $common_word_id_list);
$sql = "UPDATE " . SEARCH_WORD_TABLE . " $sql = "UPDATE " . SEARCH_WORD_TABLE . "
SET word_common = 1 SET word_common = " . TRUE . "
WHERE word_id IN ($common_word_id_list)"; WHERE word_id IN ($common_word_id_list)";
$result = $db->sql_query($sql); $result = $db->sql_query($sql);
if( !$result ) if( !$result )
@ -161,120 +163,94 @@ function remove_common($percent, $word_id_list = array())
return 0; return 0;
} }
} }
else
return $words_removed;
}
function remove_old_words($post_id)
{ {
global $db, $phpbb_root_path, $board_config, $lang; return 0;
$stopword_array = @file($phpbb_root_path . "language/lang_" . $board_config['default_lang'] . "/search_stopwords.txt");
$synonym_array = @file($phpbb_root_path . "language/lang_" . $board_config['default_lang'] . "/search_synonyms.txt");
$sql = "SELECT post_text
FROM " . POSTS_TEXT_TABLE . "
WHERE post_id = $post_id";
if( $result = $db->sql_query($sql) )
{
$row = $db->sql_fetchrow($result);
$search_text = clean_words($row['post_text'], $stopword_array, $synonym_array);
$search_matches = split_words($search_text);
if( count($search_matches) )
{
$word = array();
$word_count = array();
$phrase_string = $text;
$sql_in = "";
for ($j = 0; $j < count($search_matches); $j++)
{
$this_word = strtolower(trim($search_matches[$j]));
if( empty($word_count[$this_word]) )
{
$word_count[$this_word] = 1;
}
$new_word = true;
for($k = 0; $k < count($word); $k++)
{
if( $this_word == $word[$k] )
{
$new_word = false;
$word_count[$this_word]++;
} }
} }
else
if( $new_word )
{ {
$word[] = $this_word; return 0;
}
} }
for($j = 0; $j < count($word); $j++) return $word_count;
{
if( $word[$j] )
{
if( $sql_in != "" )
{
$sql_in .= ", ";
}
$sql_in .= "'" . $word[$j] . "'";
}
} }
$sql = "SELECT word_id, word_text function remove_unmatched_words()
FROM " . SEARCH_WORD_TABLE . " {
WHERE word_text IN ($sql_in)"; global $db;
switch(SQL_LAYER)
{
case 'postgresql':
$sql = "DELETE FROM " . SEARCH_WORD_TABLE . "
WHERE word_id NOT IN (
SELECT word_id
FROM " . SEARCH_MATCH_TABLE . ")
GROUP BY word_id";
$result = $db->sql_query($sql); $result = $db->sql_query($sql);
if( !$result ) if( !$result )
{ {
message_die(GENERAL_ERROR, "Couldn't select words", "", __LINE__, __FILE__, $sql); message_die(GENERAL_ERROR, "Couldn't delete old words from word table", __LINE__, __FILE__, $sql);
} }
if( $word_check_count = $db->sql_numrows($result) ) $unmatched_count = $db->sql_affectedrows();
{
$check_words = $db->sql_fetchrowset($result);
$word_id_sql = ""; break;
for($i = 0; $i < count($check_words); $i++ )
case 'oracle':
$sql = "DELETE FROM " . SEARCH_WORD_TABLE . "
WHERE word_id IN (
SELECT w.word_id
FROM " . SEARCH_WORD_TABLE . " w, " . SEARCH_MATCH_TABLE . " m
WHERE w.word_id = m.word_id(+)
AND m.word_id IS NULL)";
$result = $db->sql_query($sql);
if( !$result )
{ {
if( $word_id_sql != "" ) message_die(GENERAL_ERROR, "Couldn't delete old words from word table", __LINE__, __FILE__, $sql);
{
$word_id_sql .= ", ";
}
$word_id_sql .= $check_words[$i]['word_id'];
} }
$sql = "SELECT word_id, COUNT(post_id) AS post_occur_count $unmatched_count = $db->sql_affectedrows();
FROM " . SEARCH_MATCH_TABLE . "
WHERE word_id IN ($word_id_sql) break;
GROUP BY word_id";
if( !$result = $db->sql_query($sql) ) case 'mssql':
case 'msaccess':
$sql = "DELETE FROM " . SEARCH_WORD_TABLE . "
WHERE word_id IN (
SELECT w.word_id
FROM " . SEARCH_WORD_TABLE . " w
LEFT JOIN " . SEARCH_MATCH_TABLE . " m ON m.word_id = w.word_id
WHERE m.word_id IS NULL)";
$result = $db->sql_query($sql);
if( !$result )
{ {
message_die(GENERAL_ERROR, "Couldn't obtain search word sums", "", __LINE__, __FILE__, $sql); message_die(GENERAL_ERROR, "Couldn't delete old words from word table", __LINE__, __FILE__, $sql);
} }
if( $post_count = $db->sql_numrows($result) ) $unmatched_count = $db->sql_affectedrows();
{
$rowset = $db->sql_fetchrowset($result);
$word_id_sql = ""; break;
for($i = 0; $i < $post_count; $i++)
case 'mysql':
case 'mysql4':
$sql = "SELECT w.word_id
FROM " . SEARCH_WORD_TABLE . " w
LEFT JOIN " . SEARCH_MATCH_TABLE . " m ON m.word_id = w.word_id
WHERE m.word_id IS NULL";
if( $result = $db->sql_query($sql) )
{ {
if( $rowset[$i]['post_occur_count'] == 1 ) if( $unmatched_count = $db->sql_numrows($result) )
{ {
if( $word_id_sql != "" ) $rowset = array();
while( $row = $db->sql_fetchrow($result) )
{ {
$word_id_sql .= ", "; $rowset[] = $row['word_id'];
}
$word_id_sql .= $rowset[$i]['word_id'];
}
} }
$word_id_sql = implode(", ", $rowset);
if( $word_id_sql ) if( $word_id_sql )
{ {
$sql = "DELETE FROM " . SEARCH_WORD_TABLE . " $sql = "DELETE FROM " . SEARCH_WORD_TABLE . "
@ -285,24 +261,21 @@ function remove_old_words($post_id)
message_die(GENERAL_ERROR, "Couldn't delete word list entry", "", __LINE__, __FILE__, $sql); message_die(GENERAL_ERROR, "Couldn't delete word list entry", "", __LINE__, __FILE__, $sql);
} }
} }
} else
$sql = "DELETE FROM " . SEARCH_MATCH_TABLE . "
WHERE post_id = $post_id";
$result = $db->sql_query($sql);
if( !$result )
{ {
message_die(GENERAL_ERROR, "Couldn't delete word match entry for this post", "", __LINE__, __FILE__, $sql); return 0;
}
}
} }
} }
else else
{ {
message_die(GENERAL_ERROR, "Couldn't obtain post text", "", __LINE__, __FILE__, $sql); return 0;
}
} }
return; break;
}
return $unmatched_count;
} }
function add_search_words($post_id, $post_text, $post_title = "") function add_search_words($post_id, $post_text, $post_title = "")
@ -1658,11 +1631,19 @@ if( ( $submit || $confirm || $mode == "delete" ) && !$error )
if( $delete || $mode == "delete" ) if( $delete || $mode == "delete" )
{ {
remove_old_words($post_id); $sql = "DELETE FROM " . SEARCH_MATCH_TABLE . "
WHERE post_id = $post_id";
$result = $db->sql_query($sql, BEGIN_TRANSACTION);
if( !$result )
{
message_die(GENERAL_ERROR, "Couldn't delete word match entry for this post", "", __LINE__, __FILE__, $sql);
}
remove_unmatched_words();
$sql = "DELETE FROM " . POSTS_TEXT_TABLE . " $sql = "DELETE FROM " . POSTS_TEXT_TABLE . "
WHERE post_id = $post_id"; WHERE post_id = $post_id";
if($db->sql_query($sql, BEGIN_TRANSACTION)) if( $db->sql_query($sql) )
{ {
$sql = "DELETE FROM " . POSTS_TABLE . " $sql = "DELETE FROM " . POSTS_TABLE . "
WHERE post_id = $post_id"; WHERE post_id = $post_id";
@ -1882,12 +1863,18 @@ if( ( $submit || $confirm || $mode == "delete" ) && !$error )
} }
} }
remove_old_words($post_id); $sql = "DELETE FROM " . SEARCH_MATCH_TABLE . "
WHERE post_id = $post_id";
$result = $db->sql_query($sql, BEGIN_TRANSACTION);
if( !$result )
{
message_die(GENERAL_ERROR, "Couldn't delete word match entry for this post", "", __LINE__, __FILE__, $sql);
}
$sql = "UPDATE " . POSTS_TABLE . " $sql = "UPDATE " . POSTS_TABLE . "
SET bbcode_uid = '$bbcode_uid', enable_bbcode = $bbcode_on, enable_html = $html_on, enable_smilies = $smilies_on, enable_sig = $attach_sig" . $edited_sql . " SET bbcode_uid = '$bbcode_uid', enable_bbcode = $bbcode_on, enable_html = $html_on, enable_smilies = $smilies_on, enable_sig = $attach_sig" . $edited_sql . "
WHERE post_id = $post_id"; WHERE post_id = $post_id";
if($db->sql_query($sql, BEGIN_TRANSACTION)) if($db->sql_query($sql))
{ {
$sql = "UPDATE " . POSTS_TEXT_TABLE . " $sql = "UPDATE " . POSTS_TEXT_TABLE . "
SET post_text = '$post_message', post_subject = '$post_subject' SET post_text = '$post_message', post_subject = '$post_subject'
@ -1898,6 +1885,7 @@ if( ( $submit || $confirm || $mode == "delete" ) && !$error )
if( $db->sql_query($sql) ) if( $db->sql_query($sql) )
{ {
add_search_words($post_id, stripslashes($post_message)); add_search_words($post_id, stripslashes($post_message));
remove_unmatched_words();
// //
// Update topics table here // Update topics table here
@ -2020,8 +2008,8 @@ if( ( $submit || $confirm || $mode == "delete" ) && !$error )
} }
else else
{ {
remove_old_words($post_id);
add_search_words($post_id, stripslashes($post_message)); add_search_words($post_id, stripslashes($post_message));
remove_unmatched_words();
if( $db->sql_query($sql, END_TRANSACTION) ) if( $db->sql_query($sql, END_TRANSACTION) )
{ {

View File

@ -88,74 +88,6 @@ function split_words(&$entry)
return $split_entries[1]; return $split_entries[1];
} }
function arrayintersect($array1, $array2)
{
//
// make sure arguments are actually arrays
//
if( !is_array($array1) || !is_array($array2) )
{
return false;
}
if( sizeof($array2) > sizeof($array1) )
{
$temp = $array1;
$array1 = $array2;
$array2 = $temp;
unset($temp);
}
sort($array1);
rsort($array2);
// the greatest element in array2
$eg2 = $array2[0];
for($i = 0; $i < sizeof($array1); $i++)
{
$e1 = $array1[$i];
for($j = 0; $j < sizeof($array2); $j++)
{
$e2 = $array2[$j];
if( $e1 > $e2 )
{
//
// we have passed the match
//
if( $e2 == $eg2 )
{
break(2);
}
else
{
break;
}
}
else if( $e1 == $e2 )
{
//
// we found a match
//
$retArray[] = $e1;
break;
}
}
}
return $retArray;
}
function inarray($needle, $haystack)
{
for($i = 0; $i < count($haystack) && $haystack[$i] != $needle; $i++);
return ( $i != count($haystack) );
}
// //
// End of functions defns // End of functions defns
// ---------------------- // ----------------------
@ -352,7 +284,9 @@ else if( $query_keywords != "" || $query_author != "" || $search_id )
$split_search = array(); $split_search = array();
$split_search = split_words($cleaned_search); $split_search = split_words($cleaned_search);
$word_count = 0;
$word_match = array(); $word_match = array();
$result_list = array();
$current_match_type = "and"; $current_match_type = "and";
for($i = 0; $i < count($split_search); $i++) for($i = 0; $i < count($split_search); $i++)
@ -376,20 +310,7 @@ else if( $query_keywords != "" || $query_author != "" || $search_id )
$current_match_type = "and"; $current_match_type = "and";
} }
$word_match[$current_match_type][] = $split_search[$i]; $match_word = str_replace("*", "%", $split_search[$i]);
}
}
@reset($word_match);
$word_count = 0;
$result_list = array();
while( list($match_type, $match_word_list) = each($word_match) )
{
for($i = 0; $i < count($match_word_list); $i++ )
{
$match_word = str_replace("*", "%", $match_word_list[$i]);
$sql = "SELECT m.post_id $sql = "SELECT m.post_id
FROM " . SEARCH_WORD_TABLE . " w, " . SEARCH_MATCH_TABLE . " m FROM " . SEARCH_WORD_TABLE . " w, " . SEARCH_MATCH_TABLE . " m
@ -399,49 +320,30 @@ else if( $query_keywords != "" || $query_author != "" || $search_id )
$result = $db->sql_query($sql); $result = $db->sql_query($sql);
if( !$result ) if( !$result )
{ {
message_die(GENERAL_ERROR, "Couldn't matched posts", "", __LINE__, __FILE__, $sql); message_die(GENERAL_ERROR, "Couldn't obtain matched posts list", "", __LINE__, __FILE__, $sql);
} }
$row = array(); $row = array();
while( $temp_row = $db->sql_fetchrow($result) ) while( $temp_row = $db->sql_fetchrow($result) )
{ {
$row['' . $temp_row['post_id'] . ''] = 1; $row['' . $temp_row['post_id'] . ''] = 1;
}
@reset($row);
while( list($post_id, $match_count) = each($row) )
{
if( !$word_count ) if( !$word_count )
{ {
$result_list['' . $post_id . ''] = $match_count; $result_list['' . $temp_row['post_id'] . ''] = 1;
} }
else if( $match_type == "and" ) else if( $current_match_type == "or" )
{ {
$result_list['' . $post_id . ''] = ( $result_list['' . $post_id . ''] ) ? $result_list['' . $post_id . ''] + intval($match_count) : 0; $result_list['' . $temp_row['post_id'] . ''] = 1;
} }
else if( $match_type == "or" ) else if( $current_match_type == "not" )
{ {
if( $result_list['' . $post_id . ''] ) $result_list['' . $temp_row['post_id'] . ''] = 0;
{
$result_list['' . $post_id . ''] += intval($match_count);
}
else
{
$result_list['' . $post_id . ''] = 0;
$result_list['' . $post_id . ''] += intval($match_count);
}
}
else if( $match_type == "not" )
{
$result_list['' . $post_id . ''] = 0;
} }
} }
if( $match_type == "and" && $word_count ) if( $current_match_type == "and" && $word_count )
{ {
@reset($row);
@reset($result_list); @reset($result_list);
while( list($post_id, $match_count) = each($result_list) ) while( list($post_id, $match_count) = each($result_list) )
@ -494,6 +396,10 @@ else if( $query_keywords != "" || $query_author != "" || $search_id )
} }
else else
{ {
switch(SQL_LAYER)
{
case 'mysql':
case 'mysql4':
$sql = "SELECT topic_id $sql = "SELECT topic_id
FROM " . POSTS_TABLE . " FROM " . POSTS_TABLE . "
WHERE post_id IN ($sql_post_id_in) WHERE post_id IN ($sql_post_id_in)
@ -516,6 +422,16 @@ else if( $query_keywords != "" || $query_author != "" || $search_id )
$search_sql .= "t.topic_id IN ($sql_post_id_in) "; $search_sql .= "t.topic_id IN ($sql_post_id_in) ";
break;
default:
$search_sql .= "t.topic_id IN (
SELECT topic_id
FROM " . POSTS_TABLE . "
WHERE post_id IN ($sql_post_id_in)
GROUP BY topic_id )";
break;
}
} }
} }
// //
@ -611,11 +527,7 @@ else if( $query_keywords != "" || $query_author != "" || $search_id )
} }
$total_match_count = $db->sql_numrows($result); $total_match_count = $db->sql_numrows($result);
/* if( $total_match_count > 500 )
{
message_die(GENERAL_MESSAGE, $lang['Too_many_results']);//"Your search returned too many matches, refine your search criteria and try again";
}
*/
$searchset = $db->sql_fetchrowset($result); $searchset = $db->sql_fetchrowset($result);
$db->sql_freeresult($result); $db->sql_freeresult($result);
@ -868,10 +780,7 @@ else if( $query_keywords != "" || $query_author != "" || $search_id )
{ {
if($return_chars != 0 ) if($return_chars != 0 )
{ {
if($return_chars != -1) $bbcode_uid = $searchset[$i]['bbcode_uid'];
{
$message = (strlen($message) > $return_chars) ? substr($message, 0, $return_chars) . " ..." : $message;
}
// //
// If the board has HTML off but the post has HTML // If the board has HTML off but the post has HTML
@ -879,14 +788,13 @@ else if( $query_keywords != "" || $query_author != "" || $search_id )
// //
if( $return_chars != -1 ) if( $return_chars != -1 )
{ {
$message = preg_replace("#<([\/]?.*?)>#is", "&lt;\\1&gt;", $message); $message = (strlen($message) > $return_chars) ? substr($message, 0, $return_chars) . " ..." : $message;
$message = preg_replace("/[img\:[0-9a-z\:]+\].*?\[\/img\:[0-9a-z\:]+\]/si", "", $message); $message = strip_tags($message);
$message = preg_replace("/[\/[a-z\*]+\:[0-9a-z\:]+\]/si", "", $message); $message = preg_replace("/\[.*?:$bbcode_uid:?.*?\]/si", "", $message);
$message = preg_replace("/\[url\]|\[\/url\]/si", "", $message);
} }
else else
{ {
$bbcode_uid = $searchset[$i]['bbcode_uid'];
$user_sig = $searchset[$i]['user_sig']; $user_sig = $searchset[$i]['user_sig'];
$user_sig_bbcode_uid = $searchset[$i]['user_sig_bbcode_uid']; $user_sig_bbcode_uid = $searchset[$i]['user_sig_bbcode_uid'];
@ -965,7 +873,6 @@ else if( $query_keywords != "" || $query_author != "" || $search_id )
{ {
$message = ""; $message = "";
if( count($orig_word) ) if( count($orig_word) )
{ {
$topic_title = preg_replace($orig_word, $replacement_word, $searchset[$i]['topic_title']); $topic_title = preg_replace($orig_word, $replacement_word, $searchset[$i]['topic_title']);