Prevent zero values for idf measure in search algorithm (see contao#2172

)

Description
-----------

Our idf (inverse document frequency) measure is currently zero for terms that are present in every document. This results in divisions by zero and the ranking of other factors like term frequency and document length gets lost.

By adding one to the searchCount the idf value is always larger than zero.

Commits
-------

549da7c Prevent zero values for idf measure

core-bundle/src/Resources/contao/library/Contao/Search.php

@@ -357,7 +357,7 @@ public static function indexPage($arrData)
 					tl_search_index.pid,
 					SQRT(SUM(POW(
 						(1 + LOG(relevance)) * LOG((
-							" . (int) $objDatabase->query("SELECT COUNT(*) as count FROM tl_search")->count . "
+							" . (int) ($objDatabase->query("SELECT COUNT(*) as count FROM tl_search")->count + 1) . "
 						) / GREATEST(1, documentFrequency)),
 						2
 					))) as vectorLength
@@ -573,7 +573,7 @@ public static function searchFor($strKeywords, $blnOrSearch=false, $arrPid=array
 
 			if (isset($arrWildcards[$index]))
 			{
-				$strQuery .= "+ ((1+LOG(SUM(match$index * tl_search_index.relevance))) * POW(LOG(@searchCount / @wildcardCount$index), 2) / " . (\count($arrAllKeywords) - \count($arrExcludedMatches)) . ")";
+				$strQuery .= "+ ((1+LOG(SUM(match$index * tl_search_index.relevance))) * POW(LOG((@searchCount + 1) / @wildcardCount$index), 2) / " . (\count($arrAllKeywords) - \count($arrExcludedMatches)) . ")";
 			}
 			else
 			{
@@ -592,7 +592,7 @@ public static function searchFor($strKeywords, $blnOrSearch=false, $arrPid=array
 
 			if (isset($arrWildcards[$index]))
 			{
-				$strQuery .= " + POW(LOG(@searchCount / @wildcardCount$index) / " . (\count($arrAllKeywords) - \count($arrExcludedMatches)) . ", 2)";
+				$strQuery .= " + POW(LOG((@searchCount + 1) / @wildcardCount$index) / " . (\count($arrAllKeywords) - \count($arrExcludedMatches)) . ", 2)";
 			}
 			else
 			{
@@ -604,7 +604,7 @@ public static function searchFor($strKeywords, $blnOrSearch=false, $arrPid=array
 		$strQuery .= " FROM (SELECT id, term";
 
 		// Calculate inverse document frequency of every matching term
-		$strQuery .= ", LOG(@searchCount / GREATEST(1, documentFrequency)) AS idf";
+		$strQuery .= ", LOG((@searchCount + 1) / GREATEST(1, documentFrequency)) AS idf";
 
 		// Store the match of every keyword and wildcard in its own column match0, match1, ...
 		foreach ($arrAllKeywords as $index => $strKeywordExpression)