Merge pull request #166 from keboola/vb-CT-1526-add-snowflake-vector-datatype

CT-1526 add snowflake vector datatype

Author: vojtabiberle

src/Definition/Exasol.php

@@ -328,7 +328,7 @@ private function validateLength(string $type, $length = null): void
                 }
 
                 if (preg_match('/(?<val>[1-9]+\d*)\s*(?<unit>BYTE|BIT)/i', (string) $length, $matched)) {
-                    $val = $matched['val'];
+                    $val = (int) $matched['val'];
                     $unit = strtoupper($matched['unit']);
 
                     $limits = [

src/Definition/Snowflake.php

@@ -60,6 +60,7 @@ class Snowflake extends Common
     public const TYPE_ARRAY = 'ARRAY';
     public const TYPE_GEOGRAPHY = 'GEOGRAPHY';
     public const TYPE_GEOMETRY = 'GEOMETRY';
+    public const TYPE_VECTOR = 'VECTOR';
     public const TYPES = [
         self::TYPE_NUMBER,
         self::TYPE_DEC,
@@ -103,6 +104,7 @@ class Snowflake extends Common
         self::TYPE_ARRAY,
         self::TYPE_GEOGRAPHY,
         self::TYPE_GEOMETRY,
+        self::TYPE_VECTOR,
     ];
     public const MAX_VARCHAR_LENGTH = 16777216;
     public const MAX_VARBINARY_LENGTH = 8388608;
@@ -350,6 +352,24 @@ private function validateLength(string $type, $length = null): void
                     break;
                 }
                 break;
+            case self::TYPE_VECTOR:
+                $valid = false;
+                if ($length === null || !is_string($length)) {
+                    break;
+                }
+                /** matches:
+                 * TYPE - INT|FLOAT - case insensitive
+                 * ,
+                 * any white space zero or infinite times
+                 * any digit with 0 to 4 places
+                 */
+                if (preg_match('/^(?<TYPE>INT|FLOAT),[^\S\r\n]*(?<DIM>[\d]{1,4})$/i', $length, $matches)) {
+                    $dimension = (int) $matches['DIM'];
+                    if ($dimension > 0 && $dimension <= 4096) {
+                        $valid = true;
+                    }
+                }
+                break;
             default:
                 if (!is_null($length) && $length !== '') {
                     $valid = false;

tests/SnowflakeDatatypeTest.php

@@ -268,7 +268,7 @@ public function testInvalidBinaryLengths($length): void
     public function testBasetypes(): void
     {
         foreach (Snowflake::TYPES as $type) {
-            $basetype = (new Snowflake($type))->getBasetype();
+            $basetype = (new Snowflake($type, $this->getTypeDefaultOptions($type)))->getBasetype();
             switch ($type) {
                 case 'INT':
                 case 'INTEGER':
@@ -484,7 +484,8 @@ public function testArrayFromLength(string $type, ?string $length, array $expect
      */
     public function testBackendBasetypeFromAlias(string $type, string $expectedType): void
     {
-        $definition = new Snowflake($type);
+
+        $definition = new Snowflake($type, $this->getTypeDefaultOptions($type));
         $this->assertSame($expectedType, $definition->getBackendBasetype());
     }
 
@@ -569,4 +570,24 @@ public function provideTestGetTypeFromAlias(): Generator
             ];
         }
     }
+
+    /**
+     * @return array{
+     *      length?:string|null|array,
+     *      nullable?:bool,
+     *      default?:string|null
+     *  }
+     */
+    private function getTypeDefaultOptions(string $type): array
+    {
+        $options = [];
+        if ($type === Snowflake::TYPE_VECTOR) {
+            // VECTOR don't have any meaningfully default option
+            $options = [
+                'length' => 'INT,3',
+            ];
+        }
+
+        return $options;
+    }
 }