#37 When searching Solr, matches with a score of 0 are now ignored by…

… default

Setting a field's boost factor to 0 (via the "search.simple" config option) will cause documents with matches just in that field to get a score of 0

src/Result/Base.php

@@ -38,8 +38,10 @@
 use Opus\Search\Log;
 use RuntimeException;
 
+use function array_filter;
 use function array_key_exists;
 use function array_map;
+use function array_values;
 use function count;
 use function ctype_digit;
 use function intval;
@@ -192,9 +194,11 @@ public function getFacet($fieldName)
      * Retrieves set of matching and locally existing documents returned in
      * response to some search query.
      *
+     * @param bool $ignoreZeroScoreMatches ignore any matches with score 0.0
+     *        (true) or not (false); defaults to true
      * @return ResultMatch[]
      */
-    public function getReturnedMatches()
+    public function getReturnedMatches($ignoreZeroScoreMatches = true)
     {
         if ($this->data['matches'] === null) {
             return [];
@@ -208,7 +212,10 @@ public function getReturnedMatches()
         foreach ($this->data['matches'] as $match) {
             try {
                 $match->getDocument();
-                $matches[] = $match;
+                $ignoreMatch = $ignoreZeroScoreMatches === true && $match->getScore() === 0.0;
+                if ($ignoreMatch !== true) {
+                    $matches[] = $match;
+                }
             } catch (DocumentException $e) {
                 Log::get()->warn('skipping matching but locally missing document #' . $match->getId());
             }
@@ -223,18 +230,22 @@ public function getReturnedMatches()
      *
      * @note If query was requesting to retrieve non-qualified matches this set
      *       might include IDs of documents that doesn't exist locally anymore.
+     * @param bool $ignoreZeroScoreMatches ignore any matches with score 0.0
+     *        (true) or not (false); defaults to true
      * @return int[]
      */
-    public function getReturnedMatchingIds()
+    public function getReturnedMatchingIds($ignoreZeroScoreMatches = true)
     {
         if ($this->data['matches'] === null) {
             return [];
         }
 
-        return array_map(function ($match) {
-            /** @var ResultMatch $match */
-            return $match->getId();
+        $matchingIds = array_map(function (ResultMatch $match) use ($ignoreZeroScoreMatches) {
+            $ignoreMatch = $ignoreZeroScoreMatches === true && $match->getScore() === 0.0;
+            return $ignoreMatch !== true ? $match->getId() : null;
         }, $this->data['matches']);
+
+        return array_values(array_filter($matchingIds));
     }
 
     /**
@@ -247,7 +258,7 @@ public function getReturnedMatchingIds()
      *       has changed in that it's returning set of Opus_Document instances
      *       rather than set of Opus_Search_Util_Result instances.
      * @note The wording is less specific in that all information in response to
-     *       search query may considered results of search. Thus this new API
+     *       search query may be considered results of search. Thus this new API
      *       prefers "matches" over "results".
      */
     public function getResults()

test/Solr/Solarium/AdapterSearchingTest.php

@@ -453,7 +453,7 @@ public function testWeightedAndSearchWithZeroedBoost()
         $query->setUnion(false); // use AND as default query operator
 
         $result  = $search->customSearch($query);
-        $matches = $result->getReturnedMatches();
+        $matches = $result->getReturnedMatches(false);
 
         // expect only docA & docB to get found (which both contain the full query string in one of their fields)
         $this->assertEquals(2, count($matches));
@@ -486,7 +486,7 @@ public function testWeightedOrSearchWithZeroedBoost()
         $query->setUnion(true); // use OR as default query operator
 
         $result  = $search->customSearch($query);
-        $matches = $result->getReturnedMatches();
+        $matches = $result->getReturnedMatches(false);
 
         // expect all documents to get found since all of them contain at least one query term in one of their fields
         $this->assertEquals(3, count($matches));
@@ -504,6 +504,35 @@ public function testWeightedOrSearchWithZeroedBoost()
         $this->assertTrue($matches[2]->getScore() === 0.0);
     }
 
+    /**
+     * Test that a weighted `AND` search with a field's boost factor set to 0 will
+     * by default ignore any matches with a score of 0.
+     */
+    public function testWeightedAndSearchLeavingOutZeroScoredMatches()
+    {
+        $docA = $this->createDocument('weightedTestDocA'); // full query string only occurs in abstract
+        $docB = $this->createDocument('weightedTestDocB'); // full query string only occurs in title
+        $docE = $this->createDocument('weightedTestDocE'); // title & abstract contain one query term each
+        $this->indexDocuments([$docA, $docB, $docE]);
+
+        $search = Service::selectSearchingService(null, 'solr');
+        $query  = $this->queryWithSearchString($search, 'test document');
+
+        $query->setWeightedSearch(true);
+        $query->setWeightedFields(['abstract' => 0, 'title' => 1.0]);
+        $query->setUnion(false); // use AND as default query operator
+
+        $result  = $search->customSearch($query);
+        $matches = $result->getReturnedMatches(); // by default, ignores any matches with score 0.0
+
+        // expect only docB to get found, since docA only contains the full query string in the (ignored) abstract
+        $this->assertEquals(1, count($matches));
+
+        // expect just docB (contains full query string in title) with a score greater than 0
+        $this->assertEquals($docB->getId(), $matches[0]->getDocument()->getId());
+        $this->assertTrue($matches[0]->getScore() > 0.0);
+    }
+
     /**
      * Test that a "weighted" search with undefined weights (i.e. no field-specific boost factors defined at all)
      * will cause Solr to fall back to its standard query parser (which by default only searches the title field).