Menu
▾
▴
[Phpbb-checkins] CVS: phpBB2/develop search_fill.php,1.7,1.8
|
From: Paul S. O. <ps...@us...> - 2002-01-25 01:31:45
|
Update of /cvsroot/phpbb/phpBB2/develop
In directory usw-pr-cvs1:/tmp/cvs-serv2755/develop
Modified Files:
search_fill.php
Log Message:
Added title support and updated for changes in includes/search ... title slows it down, guess it needs some work if possible ...
Index: search_fill.php
===================================================================
RCS file: /cvsroot/phpbb/phpBB2/develop/search_fill.php,v
retrieving revision 1.7
retrieving revision 1.8
diff -C2 -r1.7 -r1.8
*** search_fill.php 2002/01/03 11:06:22 1.7
--- search_fill.php 2002/01/25 01:31:42 1.8
***************
*** 2,6 ****
set_time_limit(0);
! $common_percent = 40; // Percentage of posts in which a word has to appear to be marked as common
$phpbb_root_path = "../";
--- 2,6 ----
set_time_limit(0);
! $common_percent = 0.4; // Percentage of posts in which a word has to appear to be marked as common
$phpbb_root_path = "../";
***************
*** 22,28 ****
// This needs fixing! Shouldn't be hardcoded to English files!
! $stopword_array = file($phpbb_root_path . "language/lang_english/search_stopwords.txt");
$synonym_array = file($phpbb_root_path . "language/lang_english/search_synonyms.txt");
!
for ($j = 0; $j < count($stopword_array); $j++)
{
--- 22,28 ----
// This needs fixing! Shouldn't be hardcoded to English files!
! //$stopword_array = file($phpbb_root_path . "language/lang_english/search_stopwords.txt");
$synonym_array = file($phpbb_root_path . "language/lang_english/search_synonyms.txt");
! /*
for ($j = 0; $j < count($stopword_array); $j++)
{
***************
*** 30,47 ****
$search[] = "/\b" . phpbb_preg_quote($filter_word, "/") . "\b/is";
$replace[] = '';
- }
-
- for ($j = 0; $j < count($synonym_list); $j++)
- {
- list($replace_synonym, $match_synonym) = split(" ", trim(strtolower($synonym_list[$j])));
- $search[] = "/\b" . phpbb_preg_quote(trim($match_synonym), "/") . "\b/is";
- $replace[] = " " . trim($replace_synonym) . " ";
}
//
// Fetch a batch of posts_text entries
//
! $sql = "
! SELECT count(*) as total, max(post_id) as max_post_id
FROM ". POSTS_TEXT_TABLE;
if(!$result = $db->sql_query($sql))
--- 30,40 ----
$search[] = "/\b" . phpbb_preg_quote($filter_word, "/") . "\b/is";
$replace[] = '';
}
+ */
//
// Fetch a batch of posts_text entries
//
! $sql = "SELECT count(*) as total, max(post_id) as max_post_id
FROM ". POSTS_TEXT_TABLE;
if(!$result = $db->sql_query($sql))
***************
*** 64,69 ****
$batchcount++;
! $sql = "SELECT *
! FROM " . POSTS_TEXT_TABLE ."
WHERE post_id
BETWEEN $batchstart
--- 57,62 ----
$batchcount++;
! $sql = "SELECT *
! FROM " . POSTS_TEXT_TABLE . "
WHERE post_id
BETWEEN $batchstart
***************
*** 93,196 ****
$post_id = $rowset[$post_nr]['post_id'];
! $data = $rowset[$post_nr]['post_text']; // Raw data
! $text = clean_words($data, $search, $replace); // Cleaned up post
! $matches = split_words($text);
! $num_matches = count($matches);
! if($num_matches < 1)
! {
! // Skip this post if no words where found
! continue;
! }
!
! $word = array();
! $word_count = array();
! $sql_in = "";
! $phrase_string = $text;
!
! // For all words in the posting
! $sql_insert = '';
! $sql_select = '';
! for($j = 0; $j < $num_matches; $j++)
{
! $this_word = strtolower(trim($matches[$j]));
! if($this_word != '')
{
! $word_count[$this_word]++;
! $comma = ($sql_insert != '')? ', ': '';
!
! $sql_insert .= "$comma('" .$this_word. "')";
! $sql_select .= "$comma'" .$this_word. "'";
}
- }
- if($sql_insert == '')
- {
- die("no words found");
- }
-
- $sql = 'INSERT IGNORE INTO '.SEARCH_WORD_TABLE."
- (word_text)
- VALUES $sql_insert";
- if( !$result = $db->sql_query($sql) )
- {
- $error = $db->sql_error();
- die("Couldn't INSERT words :: " . $sql . " :: " . $error['message']);
- }
-
- // Get the word_id's out of the DB (to see if they are already there)
- $sql = "SELECT word_id, word_text
- FROM ".SEARCH_WORD_TABLE."
- WHERE word_text IN ($sql_select)
- GROUP BY word_text";
- $result = $db->sql_query($sql);
- if( !$result )
- {
- $error = $db->sql_error();
- die("Couldn't select words :: " . $sql . " :: " . $error['message']);
- }
- if( $word_check_count = $db->sql_numrows($result) )
- {
- $selected_words = $db->sql_fetchrowset($result);
- }
- else
- {
- print "Couldn't do sql_numrows<br>\n";
- }
- $db->sql_freeresult($result);
-
- $sql_insert = '';
- while(list($junk, $row) = each($selected_words))
- {
- $comma = ($sql_insert != '')? ', ': '';
- $sql_insert .= "$comma($post_id, ".$row['word_id'].", 0)";
- }
-
- $sql = "INSERT INTO ".SEARCH_MATCH_TABLE."
- (post_id, word_id, title_match)
- VALUES
- $sql_insert
- ";
- $result = $db->sql_query($sql);
- if( !$result )
- {
- $error = $db->sql_error();
- die("Couldn't insert new word match :: " . $sql . " :: " . $error['message']);
- }
! /*
! //$phrase_string = preg_replace("/\b" . phpbb_preg_quote($word[$j], "/") . "\b/is", $word_id, $phrase_string);
! $phrase_string = trim(preg_replace("/ {2,}/s", " ", str_replace(array("*", "'"), " ", $phrase_string)));
! $sql = "INSERT INTO phpbb_search_phrasematch (post_id, phrase_list)
! VALUES ($post_id, '$phrase_string')";
! $result = $db->sql_query($sql);
! if( !$result )
! {
! $error = $db->sql_error();
! die("Couldn't insert new phrase match :: " . $sql . " :: " . $error['message']);
! }
! */
! } // All posts
// $sql = "UNLOCK TABLES";
// $result = $db->sql_query($sql);
--- 86,177 ----
$post_id = $rowset[$post_nr]['post_id'];
! $text = clean_words("post", $rowset[$post_nr]['post_text'], $synonym_array); // Cleaned up post
! $text_title = clean_words("post", $rowset[$post_nr]['post_subject'], $synonym_array);
! $matches = array();
! $matches['text'] = split_words($text);
! $matches['title'] = split_words($text_title);
!
! while( list($match_type, $match_ary) = @each($matches) )
{
! $title_match = ( $match_type == 'title' ) ? 1 : 0;
!
! $num_matches = count($match_ary);
!
! if ( $num_matches < 1 )
{
! // Skip this post if no words where found
! continue;
}
! // For all words in the posting
! $sql_in = "";
! $sql_insert = '';
! $sql_select = '';
+ $word = array();
+ $word_count = array();
+
+ for($j = 0; $j < $num_matches; $j++)
+ {
+ $this_word = strtolower(trim($match_ary[$j]));
+ if ( $this_word != '' )
+ {
+ $word_count[$this_word] = ( isset($word_count[$this_word]) ) ? $word_count[$this_word] + 1 : 0;
+ $comma = ($sql_insert != '')? ', ': '';
+
+ $sql_insert .= "$comma('" . $this_word . "')";
+ $sql_select .= "$comma'" . $this_word . "'";
+ }
+ }
+
+ if ( $sql_insert == '' )
+ {
+ die("no words found");
+ }
+
+ $sql = 'INSERT IGNORE INTO ' . SEARCH_WORD_TABLE . "
+ (word_text)
+ VALUES $sql_insert";
+ if ( !$result = $db->sql_query($sql) )
+ {
+ $error = $db->sql_error();
+ die("Couldn't INSERT words :: " . $sql . " :: " . $error['message']);
+ }
+
+ // Get the word_id's out of the DB (to see if they are already there)
+ $sql = "SELECT word_id, word_text
+ FROM " . SEARCH_WORD_TABLE . "
+ WHERE word_text IN ($sql_select)
+ GROUP BY word_text";
+ $result = $db->sql_query($sql);
+ if ( !$result )
+ {
+ $error = $db->sql_error();
+ die("Couldn't select words :: " . $sql . " :: " . $error['message']);
+ }
+
+ $sql_insert = array();
+ while( $row = $db->sql_fetchrow($result) )
+ {
+ $sql_insert[] = "($post_id, " . $row['word_id'] . ", $title_match)";
+ }
+
+ $db->sql_freeresult($result);
+
+ $sql = "INSERT INTO " . SEARCH_MATCH_TABLE . "
+ (post_id, word_id, title_match)
+ VALUES " . implode(", ", $sql_insert);
+ $result = $db->sql_query($sql);
+ if ( !$result )
+ {
+ $error = $db->sql_error();
+ die("Couldn't insert new word match :: " . $sql . " :: " . $error['message']);
+ }
+
+ } // All posts
+ }
+
// $sql = "UNLOCK TABLES";
// $result = $db->sql_query($sql);
***************
*** 209,215 ****
print "<br>Removing common words (words that appear in more than $common_percent of the posts)<br>\n";
flush();
! print "Removed ". remove_common_global($common_percent, 1) ." words that where too common.<br>";
}
-
}
--- 190,195 ----
print "<br>Removing common words (words that appear in more than $common_percent of the posts)<br>\n";
flush();
! print "Removed ". remove_common("global", $common_percent) ." words that where too common.<br>";
}
}
|
