Merge pull request #119 from Maxteabag/feature/motherduck-support

feat: add MotherDuck cloud DuckDB support

Author: Maxteabag

sqlit/domains/connections/domain/config.py

@@ -20,6 +20,7 @@ class DatabaseType(str, Enum):
     FLIGHT = "flight"
     HANA = "hana"
     MARIADB = "mariadb"
+    MOTHERDUCK = "motherduck"
     MSSQL = "mssql"
     MYSQL = "mysql"
     ORACLE = "oracle"
@@ -52,6 +53,7 @@ class DatabaseType(str, Enum):
     DatabaseType.TRINO,
     DatabaseType.PRESTO,
     DatabaseType.DUCKDB,
+    DatabaseType.MOTHERDUCK,
     DatabaseType.REDSHIFT,
     DatabaseType.CLICKHOUSE,
     DatabaseType.COCKROACHDB,

sqlit/domains/connections/providers/motherduck/__init__.py

@@ -0,0 +1 @@
+"""MotherDuck provider package."""

sqlit/domains/connections/providers/motherduck/adapter.py

@@ -0,0 +1,113 @@
+"""MotherDuck adapter for cloud DuckDB."""
+
+from __future__ import annotations
+
+from typing import TYPE_CHECKING, Any
+
+from sqlit.domains.connections.providers.adapters.base import TableInfo
+from sqlit.domains.connections.providers.duckdb.adapter import DuckDBAdapter
+
+if TYPE_CHECKING:
+    from sqlit.domains.connections.domain.config import ConnectionConfig
+
+
+class MotherDuckAdapter(DuckDBAdapter):
+    """Adapter for MotherDuck cloud DuckDB service."""
+
+    @property
+    def name(self) -> str:
+        return "MotherDuck"
+
+    @property
+    def supports_process_worker(self) -> bool:
+        """MotherDuck handles concurrency server-side."""
+        return True
+
+    @property
+    def supports_multiple_databases(self) -> bool:
+        """MotherDuck supports multiple databases."""
+        return True
+
+    def connect(self, config: ConnectionConfig) -> Any:
+        """Connect to MotherDuck cloud database."""
+        duckdb = self._import_driver_module(
+            "duckdb",
+            driver_name=self.name,
+            extra_name=self.install_extra,
+            package_name=self.install_package,
+        )
+
+        # Get default database from options
+        database = config.get_option("default_database", "")
+
+        # Get token from tcp_endpoint.password (stored in keyring)
+        token = ""
+        if config.tcp_endpoint:
+            token = config.tcp_endpoint.password or ""
+
+        if not database:
+            raise ValueError("MotherDuck connections require a database name.")
+        if not token:
+            raise ValueError("MotherDuck connections require an access token.")
+
+        conn_str = f"md:{database}?motherduck_token={token}"
+
+        duckdb_any: Any = duckdb
+        return duckdb_any.connect(conn_str)
+
+    def get_databases(self, conn: Any) -> list[str]:
+        """List all MotherDuck databases."""
+        result = conn.execute("SELECT database_name FROM duckdb_databases() WHERE NOT internal")
+        return [row[0] for row in result.fetchall()]
+
+    def get_tables(self, conn: Any, database: str | None = None) -> list[TableInfo]:
+        """Get tables from a specific MotherDuck database."""
+        if database:
+            result = conn.execute(
+                "SELECT table_schema, table_name FROM information_schema.tables "
+                "WHERE table_catalog = ? "
+                "AND table_type = 'BASE TABLE' "
+                "AND table_schema NOT IN ('pg_catalog', 'information_schema') "
+                "ORDER BY table_schema, table_name",
+                (database,),
+            )
+        else:
+            result = conn.execute(
+                "SELECT table_schema, table_name FROM information_schema.tables "
+                "WHERE table_type = 'BASE TABLE' "
+                "AND table_schema NOT IN ('pg_catalog', 'information_schema') "
+                "ORDER BY table_schema, table_name"
+            )
+        return [(row[0], row[1]) for row in result.fetchall()]
+
+    def get_views(self, conn: Any, database: str | None = None) -> list[TableInfo]:
+        """Get views from a specific MotherDuck database."""
+        if database:
+            result = conn.execute(
+                "SELECT table_schema, table_name FROM information_schema.tables "
+                "WHERE table_catalog = ? "
+                "AND table_type = 'VIEW' "
+                "AND table_schema NOT IN ('pg_catalog', 'information_schema') "
+                "ORDER BY table_schema, table_name",
+                (database,),
+            )
+        else:
+            result = conn.execute(
+                "SELECT table_schema, table_name FROM information_schema.tables "
+                "WHERE table_type = 'VIEW' "
+                "AND table_schema NOT IN ('pg_catalog', 'information_schema') "
+                "ORDER BY table_schema, table_name"
+            )
+        return [(row[0], row[1]) for row in result.fetchall()]
+
+    def build_select_query(
+        self, table: str, limit: int, database: str | None = None, schema: str | None = None
+    ) -> str:
+        """Build SELECT LIMIT query for MotherDuck.
+
+        MotherDuck requires three-part names: database.schema.table
+        """
+        schema = schema or "main"
+        if database:
+            return f'SELECT * FROM "{database}"."{schema}"."{table}" LIMIT {limit}'
+        return f'SELECT * FROM "{schema}"."{table}" LIMIT {limit}'

sqlit/domains/connections/providers/motherduck/provider.py

@@ -0,0 +1,38 @@
+"""Provider registration for MotherDuck."""
+
+from sqlit.domains.connections.providers.adapter_provider import build_adapter_provider
+from sqlit.domains.connections.providers.catalog import register_provider
+from sqlit.domains.connections.providers.model import DatabaseProvider, ProviderSpec
+from sqlit.domains.connections.providers.motherduck.schema import SCHEMA
+
+
+def _display_info(config) -> str:
+    """Display info for MotherDuck connections."""
+    database = config.get_option("database", "") or config.database or ""
+    if database:
+        return f"md:{database}"
+    return "MotherDuck"
+
+
+def _provider_factory(spec: ProviderSpec) -> DatabaseProvider:
+    from sqlit.domains.connections.providers.motherduck.adapter import MotherDuckAdapter
+
+    return build_adapter_provider(spec, SCHEMA, MotherDuckAdapter())
+
+
+SPEC = ProviderSpec(
+    db_type="motherduck",
+    display_name="MotherDuck",
+    schema_path=("sqlit.domains.connections.providers.motherduck.schema", "SCHEMA"),
+    supports_ssh=False,
+    is_file_based=False,
+    has_advanced_auth=False,
+    default_port="",
+    requires_auth=True,
+    badge_label="MotherDuck",
+    url_schemes=("motherduck", "md"),
+    provider_factory=_provider_factory,
+    display_info=_display_info,
+)
+
+register_provider(SPEC)

sqlit/domains/connections/providers/motherduck/schema.py

@@ -0,0 +1,29 @@
+"""Connection schema for MotherDuck."""
+
+from sqlit.domains.connections.providers.schema_helpers import (
+    ConnectionSchema,
+    FieldType,
+    SchemaField,
+)
+
+SCHEMA = ConnectionSchema(
+    db_type="motherduck",
+    display_name="MotherDuck",
+    fields=(
+        SchemaField(
+            name="default_database",
+            label="Default Database",
+            placeholder="my_database",
+            required=True,
+        ),
+        SchemaField(
+            name="password",
+            label="Access Token",
+            field_type=FieldType.PASSWORD,
+            required=True,
+        ),
+    ),
+    supports_ssh=False,
+    is_file_based=False,
+    requires_auth=True,
+)

tests/unit/test_motherduck_adapter.py

@@ -0,0 +1,89 @@
+"""Unit tests for MotherDuck adapter."""
+
+from __future__ import annotations
+
+
+def test_motherduck_provider_registered():
+    """Test that MotherDuck provider is properly registered."""
+    from sqlit.domains.connections.providers.catalog import get_supported_db_types
+
+    db_types = get_supported_db_types()
+    assert "motherduck" in db_types
+
+
+def test_motherduck_provider_metadata():
+    """Test MotherDuck provider metadata."""
+    from sqlit.domains.connections.providers.catalog import get_provider
+
+    provider = get_provider("motherduck")
+    assert provider.metadata.display_name == "MotherDuck"
+    assert provider.metadata.is_file_based is False
+    assert provider.metadata.supports_ssh is False
+    assert provider.metadata.requires_auth is True
+    assert "md" in provider.metadata.url_schemes
+    assert "motherduck" in provider.metadata.url_schemes
+
+
+def test_motherduck_database_type_enum():
+    """Test MotherDuck is in DatabaseType enum."""
+    from sqlit.domains.connections.domain.config import DatabaseType
+
+    assert DatabaseType.MOTHERDUCK.value == "motherduck"
+
+
+def test_motherduck_schema_uses_password_field():
+    """Test MotherDuck schema uses standard password field for token."""
+    from sqlit.domains.connections.providers.motherduck.schema import SCHEMA
+
+    field_names = [f.name for f in SCHEMA.fields]
+    assert "default_database" in field_names
+    assert "password" in field_names  # Uses standard password field for token
+
+    # Password field should be labeled as "Access Token"
+    password_field = next(f for f in SCHEMA.fields if f.name == "password")
+    assert password_field.label == "Access Token"
+
+    # Database field should be labeled as "Default Database"
+    db_field = next(f for f in SCHEMA.fields if f.name == "default_database")
+    assert db_field.label == "Default Database"
+
+
+def test_motherduck_supports_multiple_databases():
+    """Test MotherDuck reports support for multiple databases."""
+    from sqlit.domains.connections.providers.motherduck.adapter import MotherDuckAdapter
+
+    adapter = MotherDuckAdapter()
+    assert adapter.supports_multiple_databases is True
+
+
+def test_motherduck_build_select_query_with_database():
+    """Test MotherDuck uses three-part names (database.schema.table)."""
+    from sqlit.domains.connections.providers.motherduck.adapter import MotherDuckAdapter
+
+    adapter = MotherDuckAdapter()
+
+    # With database - should use three-part name
+    query = adapter.build_select_query("hacker_news", 100, database="sample_data", schema="hn")
+    assert query == 'SELECT * FROM "sample_data"."hn"."hacker_news" LIMIT 100'
+
+
+def test_motherduck_build_select_query_without_database():
+    """Test MotherDuck falls back to two-part names without database."""
+    from sqlit.domains.connections.providers.motherduck.adapter import MotherDuckAdapter
+
+    adapter = MotherDuckAdapter()
+
+    # Without database - should use two-part name
+    query = adapter.build_select_query("my_table", 50, schema="main")
+    assert query == 'SELECT * FROM "main"."my_table" LIMIT 50'
+
+
+def test_motherduck_build_select_query_default_schema():
+    """Test MotherDuck defaults to 'main' schema."""
+    from sqlit.domains.connections.providers.motherduck.adapter import MotherDuckAdapter
+
+    adapter = MotherDuckAdapter()
+
+    # No schema specified - should default to main
+    query = adapter.build_select_query("my_table", 25, database="my_db")
+    assert query == 'SELECT * FROM "my_db"."main"."my_table" LIMIT 25'