Query regex

src/Database/Adapter.php

@@ -1442,4 +1442,38 @@ public function enableAlterLocks(bool $enable): self
 
         return $this;
     }
+
+    /**
+     * Does the adapter support trigram index?
+     *
+     * @return bool
+     */
+    abstract public function getSupportForTrigramIndex(): bool;
+
+    /**
+     * Is PCRE regex supported?
+     * PCRE (Perl Compatible Regular Expressions) supports \b for word boundaries
+     *
+     * @return bool
+     */
+    abstract public function getSupportForPCRERegex(): bool;
+
+    /**
+     * Is POSIX regex supported?
+     * POSIX regex uses \y for word boundaries instead of \b
+     *
+     * @return bool
+     */
+    abstract public function getSupportForPOSIXRegex(): bool;
+
+    /**
+     * Is regex supported at all?
+     * Returns true if either PCRE or POSIX regex is supported
+     *
+     * @return bool
+     */
+    public function getSupportForRegex(): bool
+    {
+        return $this->getSupportForPCRERegex() || $this->getSupportForPOSIXRegex();
+    }
 }

src/Database/Adapter/MariaDB.php

@@ -2230,4 +2230,19 @@ public function getSupportForAlterLocks(): bool
     {
         return true;
     }
+
+    public function getSupportForTrigramIndex(): bool
+    {
+        return false;
+    }
+
+    public function getSupportForPCRERegex(): bool
+    {
+        return true;
+    }
+
+    public function getSupportForPOSIXRegex(): bool
+    {
+        return false;
+    }
 }

src/Database/Adapter/Mongo.php

@@ -2469,7 +2469,8 @@ protected function getQueryOperator(string $operator): string
             Query::TYPE_STARTS_WITH,
             Query::TYPE_NOT_STARTS_WITH,
             Query::TYPE_ENDS_WITH,
-            Query::TYPE_NOT_ENDS_WITH => '$regex',
+            Query::TYPE_NOT_ENDS_WITH,
+            Query::TYPE_REGEX => '$regex',
             Query::TYPE_OR => '$or',
             Query::TYPE_AND => '$and',
             default => throw new DatabaseException('Unknown operator:' . $operator . '. Must be one of ' . Query::TYPE_EQUAL . ', ' . Query::TYPE_NOT_EQUAL . ', ' . Query::TYPE_LESSER . ', ' . Query::TYPE_LESSER_EQUAL . ', ' . Query::TYPE_GREATER . ', ' . Query::TYPE_GREATER_EQUAL . ', ' . Query::TYPE_IS_NULL . ', ' . Query::TYPE_IS_NOT_NULL . ', ' . Query::TYPE_BETWEEN . ', ' . Query::TYPE_NOT_BETWEEN . ', ' . Query::TYPE_STARTS_WITH . ', ' . Query::TYPE_NOT_STARTS_WITH . ', ' . Query::TYPE_ENDS_WITH . ', ' . Query::TYPE_NOT_ENDS_WITH . ', ' . Query::TYPE_CONTAINS . ', ' . Query::TYPE_NOT_CONTAINS . ', ' . Query::TYPE_SEARCH . ', ' . Query::TYPE_NOT_SEARCH . ', ' . Query::TYPE_SELECT),
@@ -2740,6 +2741,26 @@ public function getSupportForGetConnectionId(): bool
         return false;
     }
 
+    /**
+     * Is PCRE regex supported?
+     *
+     * @return bool
+     */
+    public function getSupportForPCRERegex(): bool
+    {
+        return true;
+    }
+
+    /**
+     * Is POSIX regex supported?
+     *
+     * @return bool
+     */
+    public function getSupportForPOSIXRegex(): bool
+    {
+        return false;
+    }
+
     /**
      * Is cache fallback supported?
      *
@@ -3221,4 +3242,9 @@ public function getSupportForAlterLocks(): bool
     {
         return false;
     }
+
+    public function getSupportForTrigramIndex(): bool
+    {
+        return false;
+    }
 }

src/Database/Adapter/Pool.php

@@ -365,6 +365,21 @@ public function getSupportForFulltextWildcardIndex(): bool
         return $this->delegate(__FUNCTION__, \func_get_args());
     }
 
+    public function getSupportForPCRERegex(): bool
+    {
+        return $this->delegate(__FUNCTION__, \func_get_args());
+    }
+
+    public function getSupportForPOSIXRegex(): bool
+    {
+        return $this->delegate(__FUNCTION__, \func_get_args());
+    }
+
+    public function getSupportForTrigramIndex(): bool
+    {
+        return $this->delegate(__FUNCTION__, \func_get_args());
+    }
+
     public function getSupportForCasting(): bool
     {
         return $this->delegate(__FUNCTION__, \func_get_args());

src/Database/Adapter/Postgres.php

@@ -154,6 +154,7 @@ public function create(string $name): bool
         // Enable extensions
         $this->getPDO()->prepare('CREATE EXTENSION IF NOT EXISTS postgis')->execute();
         $this->getPDO()->prepare('CREATE EXTENSION IF NOT EXISTS vector')->execute();
+        $this->getPDO()->prepare('CREATE EXTENSION IF NOT EXISTS pg_trgm')->execute();
 
         $collation = "
             CREATE COLLATION IF NOT EXISTS utf8_ci_ai (
@@ -899,9 +900,10 @@ public function createIndex(string $collection, string $id, string $type, array
             Database::INDEX_SPATIAL,
             Database::INDEX_HNSW_EUCLIDEAN,
             Database::INDEX_HNSW_COSINE,
-            Database::INDEX_HNSW_DOT => 'INDEX',
+            Database::INDEX_HNSW_DOT,
+            Database::INDEX_OBJECT,
+            Database::INDEX_TRIGRAM => 'INDEX',
             Database::INDEX_UNIQUE => 'UNIQUE INDEX',
-            Database::INDEX_OBJECT => 'INDEX',
             default => throw new DatabaseException('Unknown index type: ' . $type . '. Must be one of ' . Database::INDEX_KEY . ', ' . Database::INDEX_UNIQUE . ', ' . Database::INDEX_FULLTEXT . ', ' . Database::INDEX_SPATIAL . ', ' . Database::INDEX_OBJECT . ', ' . Database::INDEX_HNSW_EUCLIDEAN . ', ' . Database::INDEX_HNSW_COSINE . ', ' . Database::INDEX_HNSW_DOT),
         };
 
@@ -922,6 +924,11 @@ public function createIndex(string $collection, string $id, string $type, array
             Database::INDEX_HNSW_COSINE => " USING HNSW ({$attributes} vector_cosine_ops)",
             Database::INDEX_HNSW_DOT => " USING HNSW ({$attributes} vector_ip_ops)",
             Database::INDEX_OBJECT => " USING GIN ({$attributes})",
+            Database::INDEX_TRIGRAM =>
+                " USING GIN (" . implode(', ', array_map(
+                    fn ($a) => "$a gin_trgm_ops",
+                    array_map('trim', explode(',', $attributes))
+                )) . ")",
             default => " ({$attributes})",
         };
 
@@ -2112,6 +2119,21 @@ public function getSupportForVectors(): bool
         return true;
     }
 
+    public function getSupportForPCRERegex(): bool
+    {
+        return false;
+    }
+
+    public function getSupportForPOSIXRegex(): bool
+    {
+        return true;
+    }
+
+    public function getSupportForTrigramIndex(): bool
+    {
+        return true;
+    }
+
     /**
      * @return string
      */
@@ -2120,6 +2142,14 @@ public function getLikeOperator(): string
         return 'ILIKE';
     }
 
+    /**
+     * @return string
+     */
+    public function getRegexOperator(): string
+    {
+        return '~';
+    }
+
     protected function processException(PDOException $e): \Exception
     {
         // Timeout

src/Database/Adapter/SQL.php

@@ -1794,6 +1794,8 @@ protected function getSQLOperator(string $method): string
             case Query::TYPE_NOT_ENDS_WITH:
             case Query::TYPE_NOT_CONTAINS:
                 return $this->getLikeOperator();
+            case Query::TYPE_REGEX:
+                return $this->getRegexOperator();
             case Query::TYPE_VECTOR_DOT:
             case Query::TYPE_VECTOR_COSINE:
             case Query::TYPE_VECTOR_EUCLIDEAN:
@@ -2284,6 +2286,14 @@ public function getLikeOperator(): string
         return 'LIKE';
     }
 
+    /**
+     * @return string
+     */
+    public function getRegexOperator(): string
+    {
+        return 'REGEXP';
+    }
+
     public function getInternalIndexesKeys(): array
     {
         return [];

src/Database/Adapter/SQLite.php

@@ -1876,4 +1876,26 @@ public function getSupportForAlterLocks(): bool
     {
         return false;
     }
+
+    /**
+     * Is PCRE regex supported?
+     * SQLite does not have native REGEXP support - it requires compile-time option or user-defined function
+     *
+     * @return bool
+     */
+    public function getSupportForPCRERegex(): bool
+    {
+        return false;
+    }
+
+    /**
+     * Is POSIX regex supported?
+     * SQLite does not have native REGEXP support - it requires compile-time option or user-defined function
+     *
+     * @return bool
+     */
+    public function getSupportForPOSIXRegex(): bool
+    {
+        return false;
+    }
 }

src/Database/Database.php

@@ -85,6 +85,7 @@ class Database
     public const INDEX_HNSW_EUCLIDEAN = 'hnsw_euclidean';
     public const INDEX_HNSW_COSINE = 'hnsw_cosine';
     public const INDEX_HNSW_DOT = 'hnsw_dot';
+    public const INDEX_TRIGRAM = 'trigram';
 
     // Max limits
     public const MAX_INT = 2147483647;
@@ -1641,6 +1642,7 @@ public function createCollection(string $id, array $attributes = [], array $inde
                 $this->adapter->getSupportForMultipleFulltextIndexes(),
                 $this->adapter->getSupportForIdenticalIndexes(),
                 $this->adapter->getSupportForObject(),
+                $this->adapter->getSupportForTrigramIndex(),
             );
             foreach ($indexes as $index) {
                 if (!$validator->isValid($index)) {
@@ -2785,7 +2787,8 @@ public function updateAttribute(string $collection, string $id, ?string $type =
                         $this->adapter->getSupportForAttributes(),
                         $this->adapter->getSupportForMultipleFulltextIndexes(),
                         $this->adapter->getSupportForIdenticalIndexes(),
-                        $this->adapter->getSupportForObject()
+                        $this->adapter->getSupportForObject(),
+                        $this->adapter->getSupportForTrigramIndex()
                     );
 
                     foreach ($indexes as $index) {
@@ -3665,8 +3668,14 @@ public function createIndex(string $collection, string $id, string $type, array
                 }
                 break;
 
+            case self::INDEX_TRIGRAM:
+                if (!$this->adapter->getSupportForTrigramIndex()) {
+                    throw new DatabaseException('Trigram indexes are not supported');
+                }
+                break;
+
             default:
-                throw new DatabaseException('Unknown index type: ' . $type . '. Must be one of ' . Database::INDEX_KEY . ', ' . Database::INDEX_UNIQUE . ', ' . Database::INDEX_FULLTEXT . ', ' . Database::INDEX_SPATIAL . ', ' . Database::INDEX_OBJECT . ', ' . Database::INDEX_HNSW_EUCLIDEAN . ', ' . Database::INDEX_HNSW_COSINE . ', ' . Database::INDEX_HNSW_DOT);
+                throw new DatabaseException('Unknown index type: ' . $type . '. Must be one of ' . Database::INDEX_KEY . ', ' . Database::INDEX_UNIQUE . ', ' . Database::INDEX_FULLTEXT . ', ' . Database::INDEX_SPATIAL . ', ' . Database::INDEX_OBJECT . ', ' . Database::INDEX_HNSW_EUCLIDEAN . ', ' . Database::INDEX_HNSW_COSINE . ', ' . Database::INDEX_HNSW_DOT . ', '.Database::INDEX_TRIGRAM);
         }
 
         /** @var array<Document> $collectionAttributes */
@@ -3722,6 +3731,7 @@ public function createIndex(string $collection, string $id, string $type, array
                 $this->adapter->getSupportForMultipleFulltextIndexes(),
                 $this->adapter->getSupportForIdenticalIndexes(),
                 $this->adapter->getSupportForObject(),
+                $this->adapter->getSupportForTrigramIndex(),
             );
             if (!$validator->isValid($index)) {
                 throw new IndexException($validator->getDescription());

src/Database/Query.php

@@ -26,6 +26,7 @@ class Query
     public const TYPE_NOT_STARTS_WITH = 'notStartsWith';
     public const TYPE_ENDS_WITH = 'endsWith';
     public const TYPE_NOT_ENDS_WITH = 'notEndsWith';
+    public const TYPE_REGEX = 'regex';
 
     // Spatial methods
     public const TYPE_CROSSES = 'crosses';
@@ -109,6 +110,7 @@ class Query
         self::TYPE_CURSOR_BEFORE,
         self::TYPE_AND,
         self::TYPE_OR,
+        self::TYPE_REGEX
     ];
 
     public const VECTOR_TYPES = [
@@ -1178,4 +1180,16 @@ public static function vectorEuclidean(string $attribute, array $vector): self
     {
         return new self(self::TYPE_VECTOR_EUCLIDEAN, $attribute, [$vector]);
     }
+
+    /**
+     * Helper method to create Query with regex method
+     *
+     * @param string $attribute
+     * @param string $pattern
+     * @return Query
+     */
+    public static function regex(string $attribute, string $pattern): self
+    {
+        return new self(self::TYPE_REGEX, $attribute, [$pattern]);
+    }
 }