[TikiWiki-commits] [Git][tikiwiki/tiki][27.x] [FIX] Search: fix morelikethis for MySQL and manticore

Victor Emanouilov pushed to branch 27.x at Tiki Wiki CMS Groupware / Tiki


Commits:
81f9751f by Josaphat Imani at 2026-01-15T10:00:02+02:00
[FIX] Search: fix morelikethis for MySQL and manticore
---
* Autoload classes

* Added MoreLikeThis tests for MySQL and Manticore

* [FIX] Search: fix morelikethis for mySQL and manticore

See merge request tikiwiki/tiki!6767

- - - - -


8 changed files:

- lib/core/Search/Manticore/Index.php
- lib/core/Search/Manticore/PdoClient.php
- lib/core/Search/Manticore/QueryBuilder.php
- lib/core/Search/MySql/QueryBuilder.php
- + lib/test/core/Search/AbstractMoreLikeThis.php
- lib/test/core/Search/Elastic/MoreLikeThisTest.php
- + lib/test/core/Search/Manticore/MoreLikeThisTest.php
- + lib/test/core/Search/MySql/MoreLikeThisTest.php


Changes:

=====================================
lib/core/Search/Manticore/Index.php
=====================================
@@ -421,6 +421,11 @@ class Index implements \Search_Index_Interface, \Search_Index_QueryRepository
         return $search;
     }
 
+    public function getIndexTableName()
+    {
+        return $this->index;
+    }
+
     public function find(\Search_Query_Interface $query, $resultStart, $resultCount)
     {
         $builder = new QueryBuilder($this);


=====================================
lib/core/Search/Manticore/PdoClient.php
=====================================
@@ -352,7 +352,7 @@ class PdoClient
     public function document($index, $type, $id): ResultSet
     {
         $stmt = $this->prepareAndExecute("SELECT * FROM $index WHERE object_type = :object_type AND object_id = :object_id", ['object_type' => $type, 'object_id' => $id]);
-        return new ResultSet($stmt->fetch());
+        return new ResultSet($stmt->fetch(), 1, 0, 1);
     }
 
     public function optimize($index)


=====================================
lib/core/Search/Manticore/QueryBuilder.php
=====================================
@@ -104,6 +104,12 @@ class QueryBuilder
         try {
             if (! $node instanceof NotX && count($fields) == 1 && $this->isFullText($node)) {
                 $query = $this->fieldBuilder->build($node, $this->factory);
+                if ($node instanceof MoreLikeThis) {
+                    $type = $node->getObjectType();
+                    $object = $node->getObjectId();
+                    $query = $this->getDocumentContent($type, $object);
+                }
+
                 if (preg_match('/^[\d\.]+$/', $query)) {
                     // version strings should be phrases
                     $query = '"' . $query . '"';
@@ -194,6 +200,13 @@ class QueryBuilder
                     $this->select[$key] = $query;
                     return "$key = 0";
                 }
+                $query = preg_replace_callback(
+                    '/\s(AND|OR)\s/',
+                    function ($matches) {
+                        return $matches[1] === 'AND' ? ' OR ' : ' AND ';
+                    },
+                    $query
+                );
                 $query = str_replace(' = ', ' <> ', $query);
                 $query = str_replace(' < ', ' >= ', $query);
                 $query = str_replace(' > ', ' <= ', $query);
@@ -390,4 +403,14 @@ class QueryBuilder
         // single quotes require only one slash escape
         return addcslashes($qs, "'");
     }
+
+    private function getDocumentContent($type, $object)
+    {
+        $doc = $this->pdo_client->document($this->index->getIndexTableName(), $type, $object, 'contents');
+        if (! empty($doc['contents'])) {
+            return $doc['contents'];
+        }
+
+        return '';
+    }
 }


=====================================
lib/core/Search/MySql/QueryBuilder.php
=====================================
@@ -72,7 +72,13 @@ class Search_MySql_QueryBuilder
                 // $query contains the token string to compare against $fields[0] in the unified search table
                 // $fields[0] can be i.e  'allowed_users', 'allowed_groups'
                 $query = $this->fieldBuilder->build($node, $this->factory);
-                $str = $this->db->qstr($query);
+                if ($node instanceof MoreLikeThis) {
+                    $type = $node->getObjectType();
+                    $object = $node->getObjectId();
+                    $str = $node->getContent() ?: $this->getDocumentContent($type, $object);
+                } else {
+                    $str = $this->db->qstr($query);
+                }
                 $this->requireIndex($fields[0], 'fulltext', $node->getWeight());
                 $type = $this->fieldBuilder->isInverted()
                     ? 'NOT MATCH'
@@ -173,4 +179,15 @@ class Search_MySql_QueryBuilder
     {
         return $node->getValue($this->factory)->getValue();
     }
+
+    private function getDocumentContent($type, $object)
+    {
+        $results = $this->table->fetchAllIndex(['contents'], ['object_type' => $type, 'object_id' => $object]);
+
+        if (! empty($results[0]['contents'])) {
+            return $this->db->qstr($results[0]['contents']);
+        }
+
+        return '';
+    }
 }


=====================================
lib/test/core/Search/AbstractMoreLikeThis.php
=====================================
@@ -0,0 +1,93 @@
+<?php
+
+namespace Search;
+
+use PHPUnit\Framework\TestCase;
+use Search_ContentSource_Static;
+use Search_Indexer;
+use Search_Query;
+
+abstract class AbstractMoreLikeThis extends TestCase
+{
+    protected $index;
+
+    abstract protected function getIndex();
+
+    protected function setUp(): void
+    {
+        $this->index = $this->getIndex();
+        $this->index->destroy();
+
+        $this->populate($this->index);
+    }
+
+    protected function tearDown(): void
+    {
+        if ($this->index) {
+            $this->index->destroy();
+        }
+    }
+
+    public function populate($index)
+    {
+        $data = [
+            'X' => [
+                'wiki_content' => 'this does not work',
+            ],
+        ];
+
+        $words = ['hello', 'world', 'some', 'random', 'content', 'populated', 'through', 'automatic', 'sampling'];
+
+        // Generate 50 documents with random words (in a stable way)
+        foreach (range(1, 50) as $doc) {
+            $parts = [];
+            foreach ($words as $key => $word) {
+                if ($doc % ($key + 2) === 0) {
+                    $parts[] = $word;
+                    $parts[] = $word;
+                    $parts[] = $word;
+                }
+            }
+
+            $data[$doc] = [
+                'object_type' => 'wiki page',
+                'object_id' => $doc,
+                'wiki_content' => implode(' ', $parts),
+            ];
+        }
+
+        $source = new Search_ContentSource_Static(
+            $data,
+            [
+                'object_type' => 'identifier',
+                'object_id' => 'identifier',
+                'wiki_content' => 'plaintext',
+            ]
+        );
+
+        $indexer = new Search_Indexer($index);
+        $indexer->addContentSource('wiki page', $source);
+
+        $indexer->rebuild();
+    }
+
+    public function testObtainSimilarDocument()
+    {
+        $query = new Search_Query();
+        $query->filterSimilar('wiki page', 12);
+
+        $results = $query->search($this->index);
+
+        $this->assertGreaterThan(0, count($results));
+    }
+
+    public function testDocumentTooDifferent()
+    {
+        $query = new Search_Query();
+        $query->filterSimilar('wiki page', 'X');
+
+        $results = $query->search($this->index);
+
+        $this->assertCount(0, $results);
+    }
+}


=====================================
lib/test/core/Search/Elastic/MoreLikeThisTest.php
=====================================
@@ -1,10 +1,13 @@
 <?php
 
-class Search_Elastic_MoreLikeThisTest extends PHPUnit\Framework\TestCase
-{
-    private $index;
+namespace Search\Elastic;
+
+use Search_Elastic_Connection;
+use Search_Elastic_Index;
 
-    protected function setUp(): void
+class Search_Elastic_MoreLikeThisTest extends \Search\AbstractMoreLikeThis
+{
+    protected function getIndex()
     {
         $elasticSearchHost = empty(getenv('ELASTICSEARCH_HOST')) ? 'localhost' : getenv('ELASTICSEARCH_HOST');
         $connection = Search_Elastic_Connection::build('http://' . $elasticSearchHost . ':9200');
@@ -15,79 +18,6 @@ class Search_Elastic_MoreLikeThisTest extends PHPUnit\Framework\TestCase
             $this->markTestSkipped('Elasticsearch needs to be available on ' . $elasticSearchHost . ':9200 for the test to run.');
         }
 
-        $this->index = new Search_Elastic_Index($connection, 'test_index');
-        $this->index->destroy();
-
-        $this->populate($this->index);
-    }
-
-    protected function tearDown(): void
-    {
-        if ($this->index) {
-            $this->index->destroy();
-        }
-    }
-
-    public function populate($index)
-    {
-        $data = [
-            'X' => [
-                'wiki_content' => 'this does not work',
-            ],
-        ];
-
-        $words = ['hello', 'world', 'some', 'random', 'content', 'populated', 'through', 'automatic', 'sampling'];
-
-        // Generate 50 documents with random words (in a stable way)
-        foreach (range(1, 50) as $doc) {
-            $parts = [];
-            foreach ($words as $key => $word) {
-                if ($doc % ($key + 2) === 0) {
-                    $parts[] = $word;
-                    $parts[] = $word;
-                    $parts[] = $word;
-                }
-            }
-
-            $data[$doc] = [
-                'object_type' => 'wiki page',
-                'object_id' => $doc,
-                'wiki_content' => implode(' ', $parts),
-            ];
-        }
-
-        $source = new Search_ContentSource_Static(
-            $data,
-            [
-                'object_type' => 'identifier',
-                'object_id' => 'identifier',
-                'wiki_content' => 'plaintext',
-            ]
-        );
-
-        $indexer = new Search_Indexer($index);
-        $indexer->addContentSource('wiki page', $source);
-
-        $indexer->rebuild();
-    }
-
-    public function testObtainSimilarDocument()
-    {
-        $query = new Search_Query();
-        $query->filterSimilar('wiki page', 12);
-
-        $results = $query->search($this->index);
-
-        $this->assertGreaterThan(0, count($results));
-    }
-
-    public function testDocumentTooDifferent()
-    {
-        $query = new Search_Query();
-        $query->filterSimilar('wiki page', 'X');
-
-        $results = $query->search($this->index);
-
-        $this->assertCount(0, $results);
+        return new Search_Elastic_Index($connection, 'test_index');
     }
 }


=====================================
lib/test/core/Search/Manticore/MoreLikeThisTest.php
=====================================
@@ -0,0 +1,8 @@
+<?php
+
+namespace Search\Manticore;
+
+class MoreLikeThisTest extends \Search\AbstractMoreLikeThis
+{
+    use IndexBuilder;
+}


=====================================
lib/test/core/Search/MySql/MoreLikeThisTest.php
=====================================
@@ -0,0 +1,14 @@
+<?php
+
+namespace Search\MySql;
+
+use Search_MySql_Index;
+use TikiDb;
+
+class MoreLikeThisTest extends \Search\AbstractMoreLikeThis
+{
+    protected function getIndex()
+    {
+        return new Search_MySql_Index(TikiDb::get(), 'test_index');
+    }
+}



View it on GitLab: https://gitlab.com/tikiwiki/tiki/-/commit/81f9751f6e304c57ca5317078df5c5fde7666307

-- 
View it on GitLab: https://gitlab.com/tikiwiki/tiki/-/commit/81f9751f6e304c57ca5317078df5c5fde7666307
You're receiving this email because of your account on gitlab.com.

_______________________________________________
TikiWiki-cvs mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/tikiwiki-cvs