Merged r6534-6559 from 1.7 branch to trunk

git-svn-id: http://code.elgg.org/elgg/trunk@6840 36083f99-b078-4883-b0ff-0f9b5a30f544

2 files changed, 71 insertions, 45 deletions

diff --git a/mod/search/search_hooks.php b/mod/search/search_hooks.php
index 86a6b452f..516bb2098 100644
--- a/mod/search/search_hooks.php
+++ b/mod/search/search_hooks.php
@@ -24,18 +24,19 @@ function search_objects_hook($hook, $type, $value, $params) {
 	$params['joins'] = array($join);
 	$fields = array('title', 'description');
 
-	$where = search_get_where_sql('oe', $fields, $params);
+	$where = search_get_where_sql('oe', $fields, $params, FALSE);
 
 	$params['wheres'] = array($where);
-
-	$entities = elgg_get_entities($params);
 	$params['count'] = TRUE;
 	$count = elgg_get_entities($params);
-
+	
 	// no need to continue if nothing here.
 	if (!$count) {
 		return array('entities' => array(), 'count' => $count);
 	}
+	
+	$params['count'] = FALSE;
+	$entities = elgg_get_entities($params);
 
 	// add the volatile data for why these entities have been returned.
 	foreach ($entities as $entity) {
@@ -68,25 +69,28 @@ function search_groups_hook($hook, $type, $value, $params) {
 
 	$join = "JOIN {$CONFIG->dbprefix}groups_entity ge ON e.guid = ge.guid";
 	$params['joins'] = array($join);
+	
+	$fields = array('name', 'description');
+
+	// force into boolean mode because we've having problems with the
+	// "if > 50% match 0 sets are returns" problem.
+	$where = search_get_where_sql('ge', $fields, $params, FALSE);
 
-	$where = "(ge.guid = e.guid
-		AND (ge.name LIKE '%$query%'
-			OR ge.description LIKE '%$query%'
-			)
-		)";
 	$params['wheres'] = array($where);
 
 	// override subtype -- All groups should be returned regardless of subtype.
 	$params['subtype'] = ELGG_ENTITIES_ANY_VALUE;
 
-	$entities = elgg_get_entities($params);
 	$params['count'] = TRUE;
 	$count = elgg_get_entities($params);
-
+	
 	// no need to continue if nothing here.
 	if (!$count) {
 		return array('entities' => array(), 'count' => $count);
 	}
+	
+	$params['count'] = FALSE;
+	$entities = elgg_get_entities($params);
 
 	// add the volatile data for why these entities have been returned.
 	foreach ($entities as $entity) {
@@ -122,17 +126,20 @@ function search_users_hook($hook, $type, $value, $params) {
 	$join = "JOIN {$CONFIG->dbprefix}users_entity ue ON e.guid = ue.guid";
 	$params['joins'] = array($join);
 
-	$where = "(ue.guid = e.guid
-		AND (ue.username LIKE '%$query%'
-			OR ue.name LIKE '%$query%'
-			)
-		)";
+//	$where = "(ue.guid = e.guid
+//		AND (ue.username LIKE '%$query%'
+//			OR ue.name LIKE '%$query%'
+//			)
+//		)";
+
+	$fields = array('username', 'name');
+	$where = search_get_where_sql('ue', $fields, $params, FALSE);
+	
 	$params['wheres'] = array($where);
 
 	// override subtype -- All users should be returned regardless of subtype.
 	$params['subtype'] = ELGG_ENTITIES_ANY_VALUE;
 
-	$entities = elgg_get_entities($params);
 	$params['count'] = TRUE;
 	$count = elgg_get_entities($params);
 
@@ -140,6 +147,9 @@ function search_users_hook($hook, $type, $value, $params) {
 	if (!$count) {
 		return array('entities' => array(), 'count' => $count);
 	}
+	
+	$params['count'] = FALSE;
+	$entities = elgg_get_entities($params);
 
 	// add the volatile data for why these entities have been returned.
 	foreach ($entities as $entity) {
@@ -212,7 +222,6 @@ function search_tags_hook($hook, $type, $value, $params) {
 
 	$params['wheres'][] = "(msn.string IN ($tags_in) AND msv.string = '$query' AND $access)";
 
-	$entities = elgg_get_entities($params);
 	$params['count'] = TRUE;
 	$count = elgg_get_entities($params);
 
@@ -220,6 +229,9 @@ function search_tags_hook($hook, $type, $value, $params) {
 	if (!$count) {
 		return array('entities' => array(), 'count' => $count);
 	}
+	
+	$params['count'] = FALSE;
+	$entities = elgg_get_entities($params);
 
 	// add the volatile data for why these entities have been returned.
 	foreach ($entities as $entity) {
@@ -332,7 +344,7 @@ function search_comments_hook($hook, $type, $value, $params) {
 	$e_access = get_access_sql_suffix('e');
 	$a_access = get_access_sql_suffix('a');
 	// @todo this can probably be done through the api..
-	$q = "SELECT DISTINCT a.*, msv.string as comment FROM {$CONFIG->dbprefix}annotations a
+	$q = "SELECT count(DISTINCT a.id) as total FROM {$CONFIG->dbprefix}annotations a
 		JOIN {$CONFIG->dbprefix}metastrings msn ON a.name_id = msn.id
 		JOIN {$CONFIG->dbprefix}metastrings msv ON a.value_id = msv.id
 		JOIN {$CONFIG->dbprefix}entities e ON a.entity_guid = e.guid
@@ -341,13 +353,20 @@ function search_comments_hook($hook, $type, $value, $params) {
 			AND $e_access
 			AND $a_access
 			$container_and
-
-		LIMIT {$params['offset']}, {$params['limit']}
 		";
 
-	$comments = get_data($q);
-
-	$q = "SELECT count(DISTINCT a.id) as total FROM {$CONFIG->dbprefix}annotations a
+	if (!$result = get_data($q)) {
+		return FALSE;
+	}
+	
+	$count = $result[0]->total;
+	
+	// don't continue if nothing there...
+	if (!$count) {
+		return array ('entities' => array(), 'count' => 0);
+	}
+	
+	$q = "SELECT DISTINCT a.*, msv.string as comment FROM {$CONFIG->dbprefix}annotations a
 		JOIN {$CONFIG->dbprefix}metastrings msn ON a.name_id = msn.id
 		JOIN {$CONFIG->dbprefix}metastrings msv ON a.value_id = msv.id
 		JOIN {$CONFIG->dbprefix}entities e ON a.entity_guid = e.guid
@@ -356,14 +375,11 @@ function search_comments_hook($hook, $type, $value, $params) {
 			AND $e_access
 			AND $a_access
 			$container_and
-		";
 
-	$result = get_data($q);
-	$count = $result[0]->total;
+		LIMIT {$params['offset']}, {$params['limit']}
+		";
 
-	if (!is_array($comments)) {
-		return FALSE;
-	}
+	$comments = get_data($q);
 
 	// @todo if plugins are disabled causing subtypes
 	// to be invalid and there are comments on entities of those subtypes,
diff --git a/mod/search/start.php b/mod/search/start.php
index 781afb8ec..ae5e71c97 100644
--- a/mod/search/start.php
+++ b/mod/search/start.php
@@ -259,8 +259,12 @@ function search_highlight_words($words, $string) {
 	);
 
 	foreach ($words as $word) {
+		// remove any boolean mode operators
+		$word = preg_replace("/([\-\+~])([\w]+)/i", '$2', $word);
+		
 		// escape the delimiter and any other regexp special chars
 		$word = preg_quote($word, '/');
+		
 		$search = "/($word)/i";
 
 		// must replace with placeholders in case one of the search terms is
@@ -297,7 +301,9 @@ function search_remove_ignored_words($query, $format = 'array') {
 	global $CONFIG;
 
 	// don't worry about "s or boolean operators
-	$query = str_replace(array('"', '-', '+', '~'), '', stripslashes(strip_tags($query)));
+	//$query = str_replace(array('"', '-', '+', '~'), '', stripslashes(strip_tags($query)));
+	$query = stripslashes(strip_tags($query));
+	
 	$words = explode(' ', $query);
 
 	$min_chars = $CONFIG->search_info['min_chars'];
@@ -386,12 +392,8 @@ function search_get_where_sql($table, $fields, $params, $use_fulltext = TRUE) {
 			$fields[$i] = "$table.$field";
 		}
 	}
-
-	// if we're not using full text, rewrite the query for bool mode.
-	// exploiting a feature(ish) of bool mode where +-word is the same as -word
-	if (!$use_fulltext) {
-		$query = '+' . str_replace(' ', ' +', $query);
-	}
+	
+	$where = '';
 
 	// if query is shorter than the min for fts words
 	// it's likely a single acronym or similar
@@ -405,22 +407,30 @@ function search_get_where_sql($table, $fields, $params, $use_fulltext = TRUE) {
 		$likes_str = implode(' OR ', $likes);
 		$where = "($likes_str)";
 	} else {
-		// if using advanced or paired "s, switch into boolean mode
-		if (!$use_fulltext
-		|| (isset($params['advanced_search']) && $params['advanced_search'])
-		|| elgg_substr_count($query, '"') >= 2 ) {
+		// if we're not using full text, rewrite the query for bool mode.
+		// exploiting a feature(ish) of bool mode where +-word is the same as -word
+		if (!$use_fulltext) {
+			$query = '+' . str_replace(' ', ' +', $query);
+		}
+		
+		// if using advanced, boolean operators, or paired "s, switch into boolean mode
+		$booleans_used = preg_match("/([\-\+~])([\w]+)/i", $query);
+		$advanced_search = (isset($params['advanced_search']) && $params['advanced_search']);
+		$quotes_used = (elgg_substr_count($query, '"') >= 2); 
+		
+		if (!$use_fulltext || $booleans_used || $advanced_search || $quotes_used) {
 			$options = 'IN BOOLEAN MODE';
 		} else {
 			// natural language mode is default and this keyword isn't supported in < 5.1
 			//$options = 'IN NATURAL LANGUAGE MODE';
 			$options = '';
 		}
-
+		
 		// if short query, use query expansion.
 		// @todo doesn't seem to be working well.
-		if (elgg_strlen($query) < 5) {
-			//$options .= ' WITH QUERY EXPANSION';
-		}
+//		if (elgg_strlen($query) < 5) {
+//			$options .= ' WITH QUERY EXPANSION';
+//		}
 		$query = sanitise_string($query);
 
 		$fields_str = implode(',', $fields);