feat(duckdb): add support for read_mysql (#8656)

ibis/backends/duckdb/__init__.py

@@ -5,6 +5,7 @@
 import ast
 import contextlib
 import os
+import urllib
 import warnings
 from operator import itemgetter
 from pathlib import Path
@@ -948,7 +949,6 @@ def list_tables(
             )
             .sql(self.name, pretty=True)
         )
-
         out = self.con.execute(sql).fetch_arrow_table()
 
         return self._filter_with_like(out[col].to_pylist(), like)
@@ -988,6 +988,47 @@ def read_postgres(
 
         return self.table(table_name)
 
+    def read_mysql(
+        self,
+        *,
+        uri: str,
+        catalog: str,
+        table_name: str | None = None,
+    ) -> ir.Table:
+        """Register a table from a MySQL instance into a DuckDB table.
+
+        Parameters
+        ----------
+        uri
+            A mysql URI of the form `mysql://user:password@host:port/database`
+        catalog
+            User-defined alias given to the MySQL database that is being attached
+            to DuckDB
+        table_name
+            The table to read
+
+        Returns
+        -------
+        ir.Table
+            The just-registered table.
+        """
+
+        parsed = urllib.parse.urlparse(uri)
+
+        if table_name is None:
+            raise ValueError("`table_name` is required when registering a mysql table")
+
+        self._load_extensions(["mysql"])
+
+        database = parsed.path.strip("/")
+
+        query_con = f"""ATTACH 'host={parsed.hostname} user={parsed.username} password={parsed.password} port={parsed.port} database={database}' AS {catalog} (TYPE mysql)"""
+
+        with self._safe_raw_sql(query_con):
+            pass
+
+        return self.table(table_name, schema=database, database=catalog)
+
     def read_sqlite(self, path: str | Path, table_name: str | None = None) -> ir.Table:
         """Register a table from a SQLite database into a DuckDB table.
 

ibis/backends/duckdb/tests/test_register.py

@@ -178,13 +178,48 @@ def pgurl():  # pragma: no cover
     os.environ.get("DUCKDB_POSTGRES") is None, reason="avoiding CI shenanigans"
 )
 def test_read_postgres(con, pgurl):  # pragma: no cover
+    # we don't run this test in CI, only locally, to avoid bringing a postgres
+    # container up just for this test.  To run locally set env variable to True
+    # and once a postgres container is up run the test.
     table = con.read_postgres(
         f"postgres://{pgurl.username}:{pgurl.password}@{pgurl.host}:{pgurl.port}",
         table_name="duckdb_test",
     )
     assert table.count().execute()
 
 
+@pytest.fixture(scope="session")
+def mysqlurl():  # pragma: no cover
+    mysqlcon = ibis.mysql.connect(user="ibis", password="ibis", database="ibis_testing")
+
+    df = pd.DataFrame({"x": [1.0, 2.0, 3.0, 1.0], "y": ["a", "b", "c", "a"]})
+    s = ibis.schema(dict(x="float64", y="str"))
+
+    mysqlcon.create_table("duckdb_test", df, schema=s, overwrite=True)
+    yield mysqlcon.con
+    mysqlcon.drop_table("duckdb_test", force=True)
+
+
+@pytest.mark.skipif(
+    os.environ.get("DUCKDB_MYSQL") is None, reason="avoiding CI shenanigans"
+)
+def test_read_mysql(con, mysqlurl):  # pragma: no cover
+    # we don't run this test in CI, only locally, to avoid bringing a mysql
+    # container up just for this test.  To run locally set env variable to True
+    # and once a mysql container is up run the test.
+
+    # TODO(ncclementi) replace for mysqlurl.host when this is fix
+    # https://github.com/duckdb/duckdb_mysql/issues/44
+    hostname = "127.0.0.1"
+
+    table = con.read_mysql(
+        uri=f"mysql://{mysqlurl.user.decode()}:{mysqlurl.password.decode()}@{hostname}:{mysqlurl.port}/ibis_testing",
+        catalog="mysqldb",
+        table_name="duckdb_test",
+    )
+    assert table.count().execute()
+
+
 @pytest.mark.xfail(
     LINUX and SANDBOXED,
     reason="nix on linux cannot download duckdb extensions or data due to sandboxing",