DISCORD-15363566 fix(sources): resolve column type, format and size the way Appwrite does

src/Migration/Resources/Database/Column.php

@@ -2,6 +2,7 @@
 
 namespace Utopia\Migration\Resources\Database;
 
+use Utopia\Database\Database as UtopiaDatabase;
 use Utopia\Migration\Resource;
 use Utopia\Migration\Transfer;
 
@@ -31,6 +32,82 @@ abstract class Column extends Resource
     public const TYPE_OBJECT = 'object';
     public const TYPE_VECTOR = 'vector';
 
+    /**
+     * Types whose size is fixed by the type itself. Appwrite leaves the size
+     * off the API response for these, so it has to be derived from the type.
+     *
+     * Mirrors Appwrite\Utopia\Database\Attribute::SIZES.
+     *
+     * @var array<string, int>
+     */
+    public const SIZES = [
+        self::TYPE_TEXT => 65535,
+        self::TYPE_MEDIUMTEXT => 16777215,
+        self::TYPE_LONGTEXT => 2147483647,
+    ];
+
+    /**
+     * String formats, mapped to the size Appwrite creates them with. Each
+     * format is also accepted as a shorthand type on an inline column
+     * definition, where it means a string of that format.
+     *
+     * Mirrors Appwrite\Utopia\Database\Attribute::FORMAT_SIZES.
+     *
+     * @var array<string, int>
+     */
+    public const FORMAT_SIZES = [
+        self::TYPE_EMAIL => 254,
+        self::TYPE_ENUM => UtopiaDatabase::LENGTH_KEY,
+        self::TYPE_IP => 39,
+        self::TYPE_URL => 2000,
+    ];
+
+    /**
+     * Size to fall back on for a varchar that arrives without one. Appwrite
+     * requires an explicit size for varchar, so this only guards a source
+     * that omits it.
+     */
+    public const DEFAULT_VARCHAR_SIZE = 255;
+
+    /**
+     * Resolve a raw column definition into the type, format and size Appwrite
+     * stores for it. A format shorthand (`email`, `url`, `ip`, `enum`) becomes
+     * a string of that format, and an omitted size is filled in from the type
+     * or the format.
+     *
+     * Mirrors Appwrite\Utopia\Database\Attribute::resolve() so a column read
+     * from a source ends up with the size the destination would have stored.
+     *
+     * @param array<string, mixed> $column
+     * @return array{type: string, format: string, size: int}
+     */
+    public static function resolve(array $column): array
+    {
+        $type = \is_string($column['type'] ?? null) ? $column['type'] : '';
+        $format = \is_string($column['format'] ?? null) ? $column['format'] : '';
+
+        if (isset(self::FORMAT_SIZES[$type])) {
+            $format = $type;
+            $type = self::TYPE_STRING;
+        }
+
+        $size = $column['size'] ?? null;
+        $size = \is_numeric($size) ? (int) $size : 0;
+
+        if (isset(self::SIZES[$type])) {
+            // Fixed width types ignore any size the source reported.
+            $size = self::SIZES[$type];
+        } elseif ($size < 1) {
+            $size = self::FORMAT_SIZES[$format] ?? $size;
+        }
+
+        return [
+            'type' => $type,
+            'format' => $format,
+            'size' => $size,
+        ];
+    }
+
     /**
      * @param string $key
      * @param Table $table

src/Migration/Resources/Database/Columns/Email.php

@@ -13,7 +13,7 @@ public function __construct(
         bool    $required = false,
         ?string $default = null,
         bool    $array = false,
-        int     $size = 254,
+        int     $size = Column::FORMAT_SIZES[Column::TYPE_EMAIL],
         string  $createdAt = '',
         string  $updatedAt = ''
     ) {

src/Migration/Resources/Database/Columns/Enum.php

@@ -17,7 +17,7 @@ public function __construct(
         bool    $required = false,
         ?string $default = null,
         bool    $array = false,
-        int     $size = 256,
+        int     $size = Column::FORMAT_SIZES[Column::TYPE_ENUM],
         string  $createdAt = '',
         string  $updatedAt = ''
     ) {

src/Migration/Resources/Database/Columns/IP.php

@@ -13,7 +13,7 @@ public function __construct(
         bool    $required = false,
         ?string $default = null,
         bool    $array = false,
-        int     $size = 39,
+        int     $size = Column::FORMAT_SIZES[Column::TYPE_IP],
         string  $createdAt = '',
         string  $updatedAt = ''
     ) {

src/Migration/Resources/Database/Columns/LongText.php

@@ -14,7 +14,7 @@ public function __construct(
         bool    $required = false,
         ?string $default = null,
         bool    $array = false,
-        int     $size = 2147483647,
+        int     $size = Column::SIZES[Column::TYPE_LONGTEXT],
         string  $format = '',
         string  $createdAt = '',
         string  $updatedAt = ''

src/Migration/Resources/Database/Columns/MediumText.php

@@ -14,7 +14,7 @@ public function __construct(
         bool    $required = false,
         ?string $default = null,
         bool    $array = false,
-        int     $size = 16777215,
+        int     $size = Column::SIZES[Column::TYPE_MEDIUMTEXT],
         string  $format = '',
         string  $createdAt = '',
         string  $updatedAt = ''

src/Migration/Resources/Database/Columns/RegularText.php

@@ -14,7 +14,7 @@ public function __construct(
         bool    $required = false,
         ?string $default = null,
         bool    $array = false,
-        int     $size = 65535,
+        int     $size = Column::SIZES[Column::TYPE_TEXT],
         string  $format = '',
         string  $createdAt = '',
         string  $updatedAt = ''

src/Migration/Resources/Database/Columns/URL.php

@@ -13,7 +13,7 @@ public function __construct(
         bool    $required = false,
         ?string $default = null,
         bool    $array = false,
-        int     $size = 2000,
+        int     $size = Column::FORMAT_SIZES[Column::TYPE_URL],
         string  $createdAt = '',
         string  $updatedAt = ''
     ) {

src/Migration/Resources/Database/Columns/Varchar.php

@@ -14,7 +14,7 @@ public function __construct(
         bool    $required = false,
         ?string $default = null,
         bool    $array = false,
-        int     $size = 255,
+        int     $size = Column::DEFAULT_VARCHAR_SIZE,
         string  $format = '',
         string  $createdAt = '',
         string  $updatedAt = ''

src/Migration/Sources/Appwrite.php

@@ -3269,15 +3269,20 @@ public static function getRecord(string $databaseType, array $record): Resource
 
     public static function getColumn(Table $table, mixed $column): Column
     {
-        return match ($column['type']) {
-            Column::TYPE_STRING => match ($column['format'] ?? '') {
+        // Appwrite accepts a format (`email`, `url`, `ip`, `enum`) as a type on
+        // an inline column definition, and reports no size for the types whose
+        // size the type itself implies. Resolve both the way the server does.
+        ['type' => $type, 'format' => $format, 'size' => $size] = Column::resolve($column);
+
+        return match ($type) {
+            Column::TYPE_STRING => match ($format) {
                 Column::TYPE_EMAIL => new Email(
                     $column['key'],
                     $table,
                     required: $column['required'],
                     default: $column['default'],
                     array: $column['array'],
-                    size: $column['size'] ?? 254,
+                    size: $size,
                     createdAt: $column['$createdAt'] ?? '',
                     updatedAt: $column['$updatedAt'] ?? '',
                 ),
@@ -3288,7 +3293,7 @@ public static function getColumn(Table $table, mixed $column): Column
                     required: $column['required'],
                     default: $column['default'],
                     array: $column['array'],
-                    size: $column['size'] ?? UtopiaDatabase::LENGTH_KEY,
+                    size: $size,
                     createdAt: $column['$createdAt'] ?? '',
                     updatedAt: $column['$updatedAt'] ?? '',
                 ),
@@ -3298,7 +3303,7 @@ public static function getColumn(Table $table, mixed $column): Column
                     required: $column['required'],
                     default: $column['default'],
                     array: $column['array'],
-                    size: $column['size'] ?? 2000,
+                    size: $size,
                     createdAt: $column['$createdAt'] ?? '',
                     updatedAt: $column['$updatedAt'] ?? '',
                 ),
@@ -3308,7 +3313,7 @@ public static function getColumn(Table $table, mixed $column): Column
                     required: $column['required'],
                     default: $column['default'],
                     array: $column['array'],
-                    size: $column['size'] ?? 39,
+                    size: $size,
                     createdAt: $column['$createdAt'] ?? '',
                     updatedAt: $column['$updatedAt'] ?? '',
                 ),
@@ -3318,7 +3323,7 @@ public static function getColumn(Table $table, mixed $column): Column
                     required: $column['required'],
                     default: $column['default'],
                     array: $column['array'],
-                    size: $column['size'] ?? 0,
+                    size: $size,
                     createdAt: $column['$createdAt'] ?? '',
                     updatedAt: $column['$updatedAt'] ?? '',
                 ),
@@ -3446,7 +3451,7 @@ public static function getColumn(Table $table, mixed $column): Column
                 required: $column['required'],
                 default: $column['default'],
                 array: $column['array'],
-                size: $column['size'] ?? 255,
+                size: $size ?: Column::DEFAULT_VARCHAR_SIZE,
                 createdAt: $column['$createdAt'] ?? '',
                 updatedAt: $column['$updatedAt'] ?? '',
             ),
@@ -3457,7 +3462,7 @@ public static function getColumn(Table $table, mixed $column): Column
                 required: $column['required'],
                 default: $column['default'],
                 array: $column['array'],
-                size: $column['size'] ?? 65535,
+                size: $size,
                 createdAt: $column['$createdAt'] ?? '',
                 updatedAt: $column['$updatedAt'] ?? '',
             ),
@@ -3468,7 +3473,7 @@ public static function getColumn(Table $table, mixed $column): Column
                 required: $column['required'],
                 default: $column['default'],
                 array: $column['array'],
-                size: $column['size'] ?? 16777215,
+                size: $size,
                 createdAt: $column['$createdAt'] ?? '',
                 updatedAt: $column['$updatedAt'] ?? '',
             ),
@@ -3479,13 +3484,13 @@ public static function getColumn(Table $table, mixed $column): Column
                 required: $column['required'],
                 default: $column['default'],
                 array: $column['array'],
-                size: $column['size'] ?? 2147483647,
+                size: $size,
                 createdAt: $column['$createdAt'] ?? '',
                 updatedAt: $column['$updatedAt'] ?? '',
             ),
 
 
-            default => throw new \InvalidArgumentException("Unsupported column type: {$column['type']}"),
+            default => throw new \InvalidArgumentException("Unsupported column type: {$type}"),
         };
     }
 

tests/Migration/Unit/Resources/ColumnTest.php

@@ -0,0 +1,109 @@
+<?php
+
+namespace Utopia\Tests\Unit\Resources;
+
+use PHPUnit\Framework\TestCase;
+use Utopia\Database\Database as UtopiaDatabase;
+use Utopia\Migration\Resources\Database\Column;
+
+/**
+ * Lock-in for Column::resolve(), which mirrors the type/format/size mapping
+ * Appwrite applies in Appwrite\Utopia\Database\Attribute. If the two drift, a
+ * migrated column lands on the destination with a different size than the one
+ * the source project created it with.
+ */
+class ColumnTest extends TestCase
+{
+    public function testFixedWidthTypesDeriveTheirSize(): void
+    {
+        $this->assertSame(
+            ['type' => Column::TYPE_TEXT, 'format' => '', 'size' => 65535],
+            Column::resolve(['key' => 'body', 'type' => Column::TYPE_TEXT]),
+        );
+
+        $this->assertSame(
+            ['type' => Column::TYPE_MEDIUMTEXT, 'format' => '', 'size' => 16777215],
+            Column::resolve(['key' => 'summary', 'type' => Column::TYPE_MEDIUMTEXT]),
+        );
+
+        $this->assertSame(
+            ['type' => Column::TYPE_LONGTEXT, 'format' => '', 'size' => 2147483647],
+            Column::resolve(['key' => 'archive', 'type' => Column::TYPE_LONGTEXT]),
+        );
+    }
+
+    public function testFixedWidthTypesIgnoreAReportedSize(): void
+    {
+        $this->assertSame(
+            ['type' => Column::TYPE_TEXT, 'format' => '', 'size' => 65535],
+            Column::resolve(['key' => 'body', 'type' => Column::TYPE_TEXT, 'size' => 128]),
+        );
+    }
+
+    public function testFormatShorthandsBecomeAString(): void
+    {
+        $this->assertSame(
+            ['type' => Column::TYPE_STRING, 'format' => Column::TYPE_EMAIL, 'size' => 254],
+            Column::resolve(['key' => 'email', 'type' => Column::TYPE_EMAIL]),
+        );
+
+        $this->assertSame(
+            ['type' => Column::TYPE_STRING, 'format' => Column::TYPE_URL, 'size' => 2000],
+            Column::resolve(['key' => 'website', 'type' => Column::TYPE_URL]),
+        );
+
+        $this->assertSame(
+            ['type' => Column::TYPE_STRING, 'format' => Column::TYPE_IP, 'size' => 39],
+            Column::resolve(['key' => 'address', 'type' => Column::TYPE_IP]),
+        );
+
+        $this->assertSame(
+            ['type' => Column::TYPE_STRING, 'format' => Column::TYPE_ENUM, 'size' => UtopiaDatabase::LENGTH_KEY],
+            Column::resolve(['key' => 'status', 'type' => Column::TYPE_ENUM]),
+        );
+    }
+
+    public function testFormattedStringWithoutSizeFallsBackToTheFormatSize(): void
+    {
+        $this->assertSame(
+            ['type' => Column::TYPE_STRING, 'format' => Column::TYPE_EMAIL, 'size' => 254],
+            Column::resolve([
+                'key' => 'email',
+                'type' => Column::TYPE_STRING,
+                'format' => Column::TYPE_EMAIL,
+            ]),
+        );
+    }
+
+    public function testExplicitSizeWins(): void
+    {
+        $this->assertSame(
+            ['type' => Column::TYPE_STRING, 'format' => Column::TYPE_EMAIL, 'size' => 512],
+            Column::resolve(['key' => 'email', 'type' => Column::TYPE_EMAIL, 'size' => 512]),
+        );
+
+        $this->assertSame(
+            ['type' => Column::TYPE_VARCHAR, 'format' => '', 'size' => 64],
+            Column::resolve(['key' => 'slug', 'type' => Column::TYPE_VARCHAR, 'size' => 64]),
+        );
+
+        // A size that survived a round trip through a string stays a size.
+        $this->assertSame(
+            ['type' => Column::TYPE_VARCHAR, 'format' => '', 'size' => 64],
+            Column::resolve(['key' => 'slug', 'type' => Column::TYPE_VARCHAR, 'size' => '64']),
+        );
+    }
+
+    public function testUnsizedAndUnknownDefinitionsResolveToZero(): void
+    {
+        $this->assertSame(
+            ['type' => Column::TYPE_VARCHAR, 'format' => '', 'size' => 0],
+            Column::resolve(['key' => 'slug', 'type' => Column::TYPE_VARCHAR]),
+        );
+
+        $this->assertSame(
+            ['type' => '', 'format' => '', 'size' => 0],
+            Column::resolve(['key' => 'unknown']),
+        );
+    }
+}