Merge pull request #233 from JosephLees/feature/exact-match-limit

Flag to return all FuzzySearch results on exact match
teamtnt · Feb 16, 2023 · 161d336 · 161d336
2 parents ca00916 + fbde344
commit 161d336
Show file tree

Hide file tree

Showing 2 changed files with 29 additions and 4 deletions.
diff --git a/src/TNTSearch.php b/src/TNTSearch.php
@@ -24,6 +24,7 @@ class TNTSearch
     public $fuzzy_prefix_length  = 2;
     public $fuzzy_max_expansions = 50;
     public $fuzzy_distance       = 2;
+    public $fuzzy_no_limit       = false;
     protected $dbh               = null;
 
     /**
@@ -105,12 +106,13 @@ public function search($phrase, $numOfResults = 100)
         $dlWeight  = 0.5;
         $docScores = [];
         $count     = $this->totalDocumentsInCollection();
+        $noLimit   = $this->fuzzy_no_limit;
 
         foreach ($keywords as $index => $term) {
             $isLastKeyword = ($keywords->count() - 1) == $index;
             $df            = $this->totalMatchingDocuments($term, $isLastKeyword);
             $idf           = log($count / max(1, $df));
-            foreach ($this->getAllDocumentsForKeyword($term, false, $isLastKeyword) as $document) {
+            foreach ($this->getAllDocumentsForKeyword($term, $noLimit, $isLastKeyword) as $document) {
                 $docID = $document['doc_id'];
                 $tf    = $document['hit_count'];
                 $num   = ($tfWeight + 1) * $tf;
@@ -244,7 +246,7 @@ public function searchBoolean($phrase, $numOfResults = 100)
      */
     public function getAllDocumentsForKeyword($keyword, $noLimit = false, $isLastKeyword = false)
     {
-        $word = $this->getWordlistByKeyword($keyword, $isLastKeyword);
+        $word = $this->getWordlistByKeyword($keyword, $isLastKeyword, $noLimit);
         if (!isset($word[0])) {
             return new Collection([]);
         }
@@ -300,7 +302,7 @@ public function totalMatchingDocuments($keyword, $isLastWord = false)
      *
      * @return array
      */
-    public function getWordlistByKeyword($keyword, $isLastWord = false)
+    public function getWordlistByKeyword($keyword, $isLastWord = false, $noLimit = false)
     {
         $searchWordlist = "SELECT * FROM wordlist WHERE term like :keyword LIMIT 1";
         $stmtWord       = $this->index->prepare($searchWordlist);
@@ -315,7 +317,7 @@ public function getWordlistByKeyword($keyword, $isLastWord = false)
         $stmtWord->execute();
         $res = $stmtWord->fetchAll(PDO::FETCH_ASSOC);
 
-        if ($this->fuzziness && !isset($res[0])) {
+        if ($this->fuzziness && (!isset($res[0]) || $noLimit)) {
             return $this->fuzzySearch($keyword);
         }
         return $res;

diff --git a/tests/TNTSearchTest.php b/tests/TNTSearchTest.php
@@ -304,6 +304,29 @@ public function testFuzzySearchMultipleWordsFound()
         $this->assertContains(15, $res['ids']);
     }
 
+    public function testFuzzySearchOnExactMatchWithNoLimit()
+    {
+        $tnt = new TNTSearch();
+        $tnt->loadConfig($this->config);
+        $indexer                = $tnt->createIndex($this->indexName);
+        $indexer->disableOutput = true;
+        $indexer->query('SELECT id, title, article FROM articles;');
+        $indexer->run();
+
+        $tnt->selectIndex($this->indexName);
+        $index = $tnt->getIndex();
+
+        $index->insert(['id' => '14', 'title' => '199x', 'article' => 'Nineties with the x...']);
+        $index->insert(['id' => '15', 'title' => '199y', 'article' => 'Nineties with the y...']);
+        $tnt->fuzziness = true;
+        $res            = $tnt->search('199x');
+        $this->assertEquals([14], $res['ids']);
+
+        $tnt->fuzzy_no_limit    = true;
+        $res                    = $tnt->search('199x');
+        $this->assertEquals([14,15], $res['ids']);
+    }
+
     public function testIndexDoesNotExistException()
     {
         $this->expectException(IndexNotFoundException::class);