refactor(tests): clean up backend test setup to make non-data-loading…

… steps atomic

ibis/backends/bigquery/tests/conftest.py

@@ -22,9 +22,6 @@
 from ibis.backends.tests.data import json_types, non_null_array_types, struct_types, win
 
 if TYPE_CHECKING:
-    import pathlib
-    from pathlib import Path
-
     import ibis.expr.types as ir
 
 DATASET_ID = "ibis_gbq_testing"
@@ -83,13 +80,13 @@ class TestConf(UnorderedComparator, BackendTest, RoundAwayFromZero):
     supports_structs = True
     supports_json = True
     check_names = False
+    deps = ("google.cloud.bigquery",)
 
     @staticmethod
     def format_table(name: str) -> str:
         return f"{DATASET_ID}.{name}"
 
-    @staticmethod
-    def _load_data(data_dir: Path, script_dir: Path, **_: Any) -> None:
+    def _load_data(self, **_: Any) -> None:
         """Load test data into a BigQuery instance."""
 
         credentials, default_project_id = google.auth.default(
@@ -198,7 +195,7 @@ def _load_data(data_dir: Path, script_dir: Path, **_: Any) -> None:
                     make_job,
                     client.load_table_from_file,
                     io.BytesIO(
-                        data_dir.joinpath("avro", "struct_table.avro").read_bytes()
+                        self.data_dir.joinpath("avro", "struct_table.avro").read_bytes()
                     ),
                     bq.TableReference(testing_dataset, "struct_table"),
                     job_config=bq.LoadJobConfig(
@@ -267,7 +264,9 @@ def _load_data(data_dir: Path, script_dir: Path, **_: Any) -> None:
                     make_job,
                     client.load_table_from_file,
                     io.BytesIO(
-                        data_dir.joinpath("parquet", f"{table}.parquet").read_bytes()
+                        self.data_dir.joinpath(
+                            "parquet", f"{table}.parquet"
+                        ).read_bytes()
                     ),
                     bq.TableReference(testing_dataset, table),
                     job_config=bq.LoadJobConfig(
@@ -286,7 +285,9 @@ def _load_data(data_dir: Path, script_dir: Path, **_: Any) -> None:
                     make_job,
                     client.load_table_from_file,
                     io.BytesIO(
-                        data_dir.joinpath("parquet", f"{table}.parquet").read_bytes()
+                        self.data_dir.joinpath(
+                            "parquet", f"{table}.parquet"
+                        ).read_bytes()
                     ),
                     bq.TableReference(testing_dataset_tokyo, table),
                     job_config=bq.LoadJobConfig(
@@ -307,7 +308,7 @@ def functional_alltypes(self) -> ir.Table:
         return t.select(~s.c("index", "Unnamed_0"))
 
     @staticmethod
-    def connect(data_directory: pathlib.Path) -> Backend:
+    def connect(*, tmpdir, worker_id, **kw) -> Backend:
         """Connect to the test project and dataset."""
         credentials, default_project_id = google.auth.default(
             scopes=EXTERNAL_DATA_SCOPES
@@ -317,7 +318,7 @@ def connect(data_directory: pathlib.Path) -> Backend:
             os.environ.get(PROJECT_ID_ENV_VAR, default_project_id) or DEFAULT_PROJECT_ID
         )
         con = ibis.bigquery.connect(
-            project_id=project_id, dataset_id=DATASET_ID, credentials=credentials
+            project_id=project_id, dataset_id=DATASET_ID, credentials=credentials, **kw
         )
         expr = ibis.literal(1)
         try:

ibis/backends/clickhouse/tests/conftest.py

@@ -2,16 +2,16 @@
 
 import contextlib
 import os
-from typing import TYPE_CHECKING, Callable
+from typing import TYPE_CHECKING, Any, Callable, Iterable
 
 import pytest
 
 import ibis
 import ibis.expr.types as ir
+from ibis import util
 from ibis.backends.tests.base import (
     RoundHalfToEven,
     ServiceBackendTest,
-    ServiceSpec,
     UnorderedComparator,
 )
 
@@ -32,28 +32,22 @@ class TestConf(UnorderedComparator, ServiceBackendTest, RoundHalfToEven):
     supported_to_timestamp_units = {'s'}
     supports_floating_modulus = False
     supports_json = False
+    data_volume = "/var/lib/clickhouse/user_files/ibis"
+    service_name = "clickhouse"
+    deps = ("clickhouse_connect",)
 
     @property
     def native_bool(self) -> bool:
         [(value,)] = self.connection.con.query("SELECT true").result_set
         return isinstance(value, bool)
 
-    @classmethod
-    def service_spec(cls, data_dir: Path) -> ServiceSpec:
-        return ServiceSpec(
-            name=cls.name(),
-            data_volume="/var/lib/clickhouse/user_files/ibis",
-            files=data_dir.joinpath("parquet").glob("*.parquet"),
-        )
+    @property
+    def test_files(self) -> Iterable[Path]:
+        return self.data_dir.joinpath("parquet").glob("*.parquet")
 
-    @staticmethod
     def _load_data(
-        data_dir: Path,
-        script_dir: Path,
-        host: str = CLICKHOUSE_HOST,
-        port: int = CLICKHOUSE_PORT,
-        user: str = CLICKHOUSE_USER,
-        password: str = CLICKHOUSE_PASS,
+        self,
+        *,
         database: str = IBIS_TEST_CLICKHOUSE_DB,
         **_,
     ) -> None:
@@ -66,35 +60,28 @@ def _load_data(
         script_dir
             Location of scripts defining schemas
         """
-        cc = pytest.importorskip("clickhouse_connect")
-
-        client = cc.get_client(
-            host=host,
-            port=port,
-            user=user,
-            password=password,
-            settings={
-                "allow_experimental_object_type": 1,
-                "output_format_json_named_tuples_as_objects": 1,
-            },
-        )
+        import clickhouse_connect as cc
+
+        con = self.connection
+        client = con.con
 
         with contextlib.suppress(cc.driver.exceptions.DatabaseError):
             client.command(f"CREATE DATABASE {database} ENGINE = Atomic")
 
-        with open(script_dir / 'schema' / 'clickhouse.sql') as schema:
-            for stmt in filter(None, map(str.strip, schema.read().split(";"))):
-                client.command(stmt)
+        util.consume(map(client.command, self.ddl_script))
+
+    def postload(self, **kw: Any):
+        # reconnect to set the database to the test database
+        self.connection = self.connect(database=IBIS_TEST_CLICKHOUSE_DB, **kw)
 
     @staticmethod
-    def connect(data_directory: Path):
-        pytest.importorskip("clickhouse_connect")
+    def connect(*, tmpdir, worker_id, **kw: Any):
         return ibis.clickhouse.connect(
             host=CLICKHOUSE_HOST,
             port=CLICKHOUSE_PORT,
             password=CLICKHOUSE_PASS,
-            database=IBIS_TEST_CLICKHOUSE_DB,
             user=CLICKHOUSE_USER,
+            **kw,
         )
 
     @staticmethod
@@ -114,27 +101,22 @@ def least(f: Callable[..., ir.Value], *args: ir.Value) -> ir.Value:
         return f(*args)
 
 
-@pytest.fixture(scope='module')
-def con(tmp_path_factory, data_directory, script_directory, worker_id):
-    return TestConf.load_data(
-        data_directory,
-        script_directory,
-        tmp_path_factory,
-        worker_id,
-    ).connect(data_directory)
+@pytest.fixture(scope='session')
+def con(tmp_path_factory, data_dir, worker_id):
+    return TestConf.load_data(data_dir, tmp_path_factory, worker_id).connection
 
 
-@pytest.fixture(scope='module')
+@pytest.fixture(scope='session')
 def db(con):
     return con.database()
 
 
-@pytest.fixture(scope='module')
-def alltypes(db):
-    return db.functional_alltypes
+@pytest.fixture(scope='session')
+def alltypes(con):
+    return con.tables.functional_alltypes
 
 
-@pytest.fixture(scope='module')
+@pytest.fixture(scope='session')
 def df(alltypes):
     return alltypes.execute()
 

ibis/backends/conftest.py

@@ -9,7 +9,7 @@
 import sys
 from functools import lru_cache
 from pathlib import Path
-from typing import Any, TextIO
+from typing import Any, Iterable
 
 import _pytest
 import numpy as np
@@ -126,19 +126,7 @@
 
 
 @pytest.fixture(scope='session')
-def script_directory() -> Path:
-    """Return the test script directory.
-
-    Returns
-    -------
-    Path
-        Test script directory
-    """
-    return Path(__file__).absolute().parents[2] / "ci"
-
-
-@pytest.fixture(scope='session')
-def data_directory() -> Path:
+def data_dir() -> Path:
     """Return the test data directory.
 
     Returns
@@ -178,7 +166,7 @@ def recreate_database(
 def init_database(
     url: sa.engine.url.URL,
     database: str,
-    schema: TextIO | None = None,
+    schema: Iterable[str] | None = None,
     recreate: bool = True,
     isolation_level: str | None = "AUTOCOMMIT",
     **kwargs: Any,
@@ -220,11 +208,7 @@ def init_database(
 
     if schema:
         with engine.begin() as conn:
-            for stmt in filter(
-                None,
-                map(str.strip, schema.read().split(';')),
-            ):
-                conn.exec_driver_sql(stmt)
+            util.consume(map(conn.exec_driver_sql, schema))
 
     return engine
 
@@ -530,11 +514,11 @@ def pytest_runtest_call(item):
 
 
 @pytest.fixture(params=_get_backends_to_test(), scope='session')
-def backend(request, data_directory, script_directory, tmp_path_factory, worker_id):
+def backend(request, data_dir, tmp_path_factory, worker_id):
     """Return an instance of BackendTest, loaded with data."""
 
     cls = _get_backend_conf(request.param)
-    return cls.load_data(data_directory, script_directory, tmp_path_factory, worker_id)
+    return cls.load_data(data_dir, tmp_path_factory, worker_id)
 
 
 @pytest.fixture(scope="session")
@@ -543,9 +527,7 @@ def con(backend):
     return backend.connection
 
 
-def _setup_backend(
-    request, data_directory, script_directory, tmp_path_factory, worker_id
-):
+def _setup_backend(request, data_dir, tmp_path_factory, worker_id):
     if (backend := request.param) == "duckdb" and platform.system() == "Windows":
         pytest.xfail(
             "windows prevents two connections to the same duckdb file "
@@ -554,20 +536,16 @@ def _setup_backend(
         return None
     else:
         cls = _get_backend_conf(backend)
-        return cls.load_data(
-            data_directory, script_directory, tmp_path_factory, worker_id
-        )
+        return cls.load_data(data_dir, tmp_path_factory, worker_id)
 
 
 @pytest.fixture(
     params=_get_backends_to_test(discard=("dask", "pandas")),
     scope='session',
 )
-def ddl_backend(request, data_directory, script_directory, tmp_path_factory, worker_id):
+def ddl_backend(request, data_dir, tmp_path_factory, worker_id):
     """Set up the backends that are SQL-based."""
-    return _setup_backend(
-        request, data_directory, script_directory, tmp_path_factory, worker_id
-    )
+    return _setup_backend(request, data_dir, tmp_path_factory, worker_id)
 
 
 @pytest.fixture(scope='session')
@@ -591,13 +569,9 @@ def ddl_con(ddl_backend):
     ),
     scope='session',
 )
-def alchemy_backend(
-    request, data_directory, script_directory, tmp_path_factory, worker_id
-):
+def alchemy_backend(request, data_dir, tmp_path_factory, worker_id):
     """Set up the SQLAlchemy-based backends."""
-    return _setup_backend(
-        request, data_directory, script_directory, tmp_path_factory, worker_id
-    )
+    return _setup_backend(request, data_dir, tmp_path_factory, worker_id)
 
 
 @pytest.fixture(scope='session')
@@ -610,10 +584,10 @@ def alchemy_con(alchemy_backend):
     params=_get_backends_to_test(keep=("dask", "pandas", "pyspark")),
     scope='session',
 )
-def udf_backend(request, data_directory, script_directory, tmp_path_factory, worker_id):
+def udf_backend(request, data_dir, tmp_path_factory, worker_id):
     """Runs the UDF-supporting backends."""
     cls = _get_backend_conf(request.param)
-    return cls.load_data(data_directory, script_directory, tmp_path_factory, worker_id)
+    return cls.load_data(data_dir, tmp_path_factory, worker_id)
 
 
 @pytest.fixture(scope='session')