airbyte-cdk 6.55.0__py3-none-any.whl → 6.55.1.post11.dev15684355943__py3-none-any.whl

airbyte_cdk/test/standard_tests/docker_base.py

@@ -0,0 +1,388 @@
+# Copyright (c) 2024 Airbyte, Inc., all rights reserved.
+"""Base class for connector test suites."""
+
+from __future__ import annotations
+
+import inspect
+import shutil
+import sys
+import tempfile
+import warnings
+from dataclasses import asdict
+from pathlib import Path
+from subprocess import CompletedProcess, SubprocessError
+from typing import Literal
+
+import orjson
+import pytest
+import yaml
+from boltons.typeutils import classproperty
+
+from airbyte_cdk.models import (
+    AirbyteCatalog,
+    ConfiguredAirbyteCatalog,
+    ConfiguredAirbyteStream,
+    DestinationSyncMode,
+    SyncMode,
+)
+from airbyte_cdk.models.airbyte_protocol_serializers import (
+    AirbyteCatalogSerializer,
+    AirbyteStreamSerializer,
+)
+from airbyte_cdk.models.connector_metadata import MetadataFile
+from airbyte_cdk.test.entrypoint_wrapper import EntrypointOutput
+from airbyte_cdk.test.models import ConnectorTestScenario
+from airbyte_cdk.test.utils.reading import catalog
+from airbyte_cdk.utils.connector_paths import (
+    ACCEPTANCE_TEST_CONFIG,
+    find_connector_root,
+)
+from airbyte_cdk.utils.docker import build_connector_image, run_docker_command
+
+
+class DockerConnectorTestSuite:
+    """Base class for connector test suites."""
+
+    @classmethod
+    def get_test_class_dir(cls) -> Path:
+        """Get the file path that contains the class."""
+        module = sys.modules[cls.__module__]
+        # Get the directory containing the test file
+        return Path(inspect.getfile(module)).parent
+
+    @classmethod
+    def get_connector_root_dir(cls) -> Path:
+        """Get the root directory of the connector."""
+        return find_connector_root([cls.get_test_class_dir(), Path.cwd()])
+
+    @classproperty
+    def acceptance_test_config_path(cls) -> Path:
+        """Get the path to the acceptance test config file."""
+        result = cls.get_connector_root_dir() / ACCEPTANCE_TEST_CONFIG
+        if result.exists():
+            return result
+
+        raise FileNotFoundError(f"Acceptance test config file not found at: {str(result)}")
+
+    @classmethod
+    def get_scenarios(
+        cls,
+    ) -> list[ConnectorTestScenario]:
+        """Get acceptance tests for a given category.
+
+        This has to be a separate function because pytest does not allow
+        parametrization of fixtures with arguments from the test class itself.
+        """
+        categories = ["connection", "spec"]
+        try:
+            acceptance_test_config_path = cls.acceptance_test_config_path
+        except FileNotFoundError as e:
+            # Destinations sometimes do not have an acceptance tests file.
+            warnings.warn(
+                f"Acceptance test config file not found: {e!s}. No scenarios will be loaded.",
+                category=UserWarning,
+                stacklevel=1,
+            )
+            return []
+
+        all_tests_config = yaml.safe_load(cls.acceptance_test_config_path.read_text())
+        if "acceptance_tests" not in all_tests_config:
+            raise ValueError(
+                f"Acceptance tests config not found in {cls.acceptance_test_config_path}."
+                f" Found only: {str(all_tests_config)}."
+            )
+
+        test_scenarios: list[ConnectorTestScenario] = []
+        for category in categories:
+            if (
+                category not in all_tests_config["acceptance_tests"]
+                or "tests" not in all_tests_config["acceptance_tests"][category]
+            ):
+                continue
+
+            for test in all_tests_config["acceptance_tests"][category]["tests"]:
+                if "config_path" not in test:
+                    # Skip tests without a config_path
+                    continue
+
+                if "iam_role" in test["config_path"]:
+                    # We skip iam_role tests for now, as they are not supported in the test suite.
+                    continue
+
+                scenario = ConnectorTestScenario.model_validate(test)
+
+                if scenario.config_path and scenario.config_path in [
+                    s.config_path for s in test_scenarios
+                ]:
+                    # Skip duplicate scenarios based on config_path
+                    continue
+
+                test_scenarios.append(scenario)
+
+        return test_scenarios
+
+    @pytest.mark.skipif(
+        shutil.which("docker") is None,
+        reason="docker CLI not found in PATH, skipping docker image tests",
+    )
+    @pytest.mark.image_tests
+    def test_docker_image_build_and_spec(
+        self,
+        connector_image_override: str | None,
+    ) -> None:
+        """Run `docker_image` acceptance tests."""
+        connector_root = self.get_connector_root_dir().absolute()
+        metadata = MetadataFile.from_file(connector_root / "metadata.yaml")
+
+        connector_image: str | None = connector_image_override
+        if not connector_image:
+            tag = "dev-latest"
+            connector_image = build_connector_image(
+                connector_name=connector_root.absolute().name,
+                connector_directory=connector_root,
+                metadata=metadata,
+                tag=tag,
+                no_verify=False,
+            )
+
+        try:
+            result: CompletedProcess[str] = run_docker_command(
+                [
+                    "docker",
+                    "run",
+                    "--rm",
+                    connector_image,
+                    "spec",
+                ],
+                check=True,  # Raise an error if the command fails
+                capture_stderr=True,
+                capture_stdout=True,
+            )
+        except SubprocessError as ex:
+            raise AssertionError(
+                f"Failed to run `spec` command in docker image {connector_image!r}. Error: {ex!s}"
+            ) from None
+
+    @pytest.mark.skipif(
+        shutil.which("docker") is None,
+        reason="docker CLI not found in PATH, skipping docker image tests",
+    )
+    @pytest.mark.image_tests
+    def test_docker_image_build_and_check(
+        self,
+        scenario: ConnectorTestScenario,
+        connector_image_override: str | None,
+    ) -> None:
+        """Run `docker_image` acceptance tests.
+
+        This test builds the connector image and runs the `check` command inside the container.
+
+        Note:
+          - It is expected for docker image caches to be reused between test runs.
+          - In the rare case that image caches need to be cleared, please clear
+            the local docker image cache using `docker image prune -a` command.
+        """
+        if scenario.expected_outcome.expect_exception():
+            pytest.skip("Skipping test_docker_image_build_and_check (expected to fail).")
+
+        tag = "dev-latest"
+        connector_root = self.get_connector_root_dir()
+        metadata = MetadataFile.from_file(connector_root / "metadata.yaml")
+        connector_image: str | None = connector_image_override
+        if not connector_image:
+            tag = "dev-latest"
+            connector_image = build_connector_image(
+                connector_name=connector_root.absolute().name,
+                connector_directory=connector_root,
+                metadata=metadata,
+                tag=tag,
+                no_verify=False,
+            )
+
+        container_config_path = "/secrets/config.json"
+        with scenario.with_temp_config_file(
+            connector_root=connector_root,
+        ) as temp_config_file:
+            _ = run_docker_command(
+                [
+                    "docker",
+                    "run",
+                    "--rm",
+                    "-v",
+                    f"{temp_config_file}:{container_config_path}",
+                    connector_image,
+                    "check",
+                    "--config",
+                    container_config_path,
+                ],
+                check=True,  # Raise an error if the command fails
+                capture_stderr=True,
+                capture_stdout=True,
+            )
+
+    @pytest.mark.skipif(
+        shutil.which("docker") is None,
+        reason="docker CLI not found in PATH, skipping docker image tests",
+    )
+    @pytest.mark.image_tests
+    def test_docker_image_build_and_read(
+        self,
+        scenario: ConnectorTestScenario,
+        connector_image_override: str | None,
+        read_from_streams: Literal["all", "none", "default"] | list[str],
+        read_scenarios: Literal["all", "none", "default"] | list[str],
+    ) -> None:
+        """Read from the connector's Docker image.
+
+        This test builds the connector image and runs the `read` command inside the container.
+
+        Note:
+          - It is expected for docker image caches to be reused between test runs.
+          - In the rare case that image caches need to be cleared, please clear
+            the local docker image cache using `docker image prune -a` command.
+          - If the --connector-image arg is provided, it will be used instead of building the image.
+        """
+        if scenario.expected_outcome.expect_exception():
+            pytest.skip("Skipping (expected to fail).")
+
+        if read_from_streams == "none":
+            pytest.skip("Skipping read test (`--read-from-streams=false`).")
+
+        if read_scenarios == "none":
+            pytest.skip("Skipping (`--read-scenarios=none`).")
+
+        default_scenario_ids = ["config", "valid_config", "default"]
+        if read_scenarios == "all":
+            pass
+        elif read_scenarios == "default":
+            if scenario.id not in default_scenario_ids:
+                pytest.skip(
+                    f"Skipping read test for scenario '{scenario.id}' "
+                    f"(not in default scenarios list '{default_scenario_ids}')."
+                )
+        elif scenario.id not in read_scenarios:
+            # pytest.skip(
+            raise ValueError(
+                f"Skipping read test for scenario '{scenario.id}' "
+                f"(not in --read-scenarios={read_scenarios})."
+            )
+
+        tag = "dev-latest"
+        connector_root = self.get_connector_root_dir()
+        connector_name = connector_root.absolute().name
+        metadata = MetadataFile.from_file(connector_root / "metadata.yaml")
+        connector_image: str | None = connector_image_override
+        if not connector_image:
+            tag = "dev-latest"
+            connector_image = build_connector_image(
+                connector_name=connector_name,
+                connector_directory=connector_root,
+                metadata=metadata,
+                tag=tag,
+                no_verify=False,
+            )
+
+        container_config_path = "/secrets/config.json"
+        container_catalog_path = "/secrets/catalog.json"
+
+        with (
+            scenario.with_temp_config_file(
+                connector_root=connector_root,
+            ) as temp_config_file,
+            tempfile.TemporaryDirectory(
+                prefix=f"{connector_name}-test",
+                ignore_cleanup_errors=True,
+            ) as temp_dir_str,
+        ):
+            temp_dir = Path(temp_dir_str)
+            discover_result = run_docker_command(
+                [
+                    "docker",
+                    "run",
+                    "--rm",
+                    "-v",
+                    f"{temp_config_file}:{container_config_path}",
+                    connector_image,
+                    "discover",
+                    "--config",
+                    container_config_path,
+                ],
+                check=True,  # Raise an error if the command fails
+                capture_stderr=True,
+                capture_stdout=True,
+            )
+            parsed_output = EntrypointOutput(messages=discover_result.stdout.splitlines())
+            try:
+                catalog_message = parsed_output.catalog  # Get catalog message
+                assert catalog_message.catalog is not None, "Catalog message missing catalog."
+                discovered_catalog: AirbyteCatalog = parsed_output.catalog.catalog
+            except Exception as ex:
+                raise AssertionError(
+                    f"Failed to load discovered catalog from {discover_result.stdout}. "
+                    f"Error: {ex!s}"
+                ) from None
+            if not discovered_catalog.streams:
+                raise ValueError(
+                    f"Discovered catalog for connector '{connector_name}' is empty. "
+                    "Please check the connector's discover implementation."
+                )
+
+            streams_list = [stream.name for stream in discovered_catalog.streams]
+            if read_from_streams == "default" and metadata.data.suggestedStreams:
+                # set `streams_list` to be the intersection of discovered and suggested streams.
+                streams_list = list(set(streams_list) & set(metadata.data.suggestedStreams.streams))
+
+            if isinstance(read_from_streams, list):
+                # If `read_from_streams` is a list, we filter the discovered streams.
+                streams_list = list(set(streams_list) & set(read_from_streams))
+
+            configured_catalog: ConfiguredAirbyteCatalog = ConfiguredAirbyteCatalog(
+                streams=[
+                    ConfiguredAirbyteStream(
+                        stream=stream,
+                        sync_mode=(
+                            stream.supported_sync_modes[0]
+                            if stream.supported_sync_modes
+                            else SyncMode.full_refresh
+                        ),
+                        destination_sync_mode=DestinationSyncMode.append,
+                    )
+                    for stream in discovered_catalog.streams
+                    if stream.name in streams_list
+                ]
+            )
+            configured_catalog_path = temp_dir / "catalog.json"
+            configured_catalog_path.write_text(
+                orjson.dumps(asdict(configured_catalog)).decode("utf-8")
+            )
+            read_result: CompletedProcess[str] = run_docker_command(
+                [
+                    "docker",
+                    "run",
+                    "--rm",
+                    "-v",
+                    f"{temp_config_file}:{container_config_path}",
+                    "-v",
+                    f"{configured_catalog_path}:{container_catalog_path}",
+                    connector_image,
+                    "read",
+                    "--config",
+                    container_config_path,
+                    "--catalog",
+                    container_catalog_path,
+                ],
+                check=False,
+                capture_stderr=True,
+                capture_stdout=True,
+            )
+            if read_result.returncode != 0:
+                raise AssertionError(
+                    f"Failed to run `read` command in docker image {connector_image!r}. "
+                    "\n-----------------"
+                    f"EXIT CODE: {read_result.returncode}\n"
+                    "STDERR:\n"
+                    f"{read_result.stderr}\n"
+                    f"STDOUT:\n"
+                    f"{read_result.stdout}\n"
+                    "\n-----------------"
+                ) from None

airbyte_cdk/test/standard_tests/pytest_hooks.py

@@ -13,12 +13,125 @@ pytest_plugins = [
 ```
 """
 
+from typing import Literal, cast
+
 import pytest
 
 
-def pytest_generate_tests(metafunc: pytest.Metafunc) -> None:
+@pytest.fixture
+def connector_image_override(request: pytest.FixtureRequest) -> str | None:
+    """Return the value of --connector-image, or None if not set."""
+    return cast(str | None, request.config.getoption("--connector-image"))
+
+
+@pytest.fixture
+def read_from_streams(
+    request: pytest.FixtureRequest,
+) -> Literal["all", "none", "default"] | list[str]:
+    """Specify if the test should read from streams.
+
+    The input can be one of the following:
+    - [Omitted] - Default to False, meaning no streams will be read.
+    - `--read-from-streams`: Read from all suggested streams.
+    - `--read-from-streams=true`: Read from all suggested streams.
+    - `--read-from-streams=suggested`: Read from all suggested streams.
+    - `--read-from-streams=default`: Read from all suggested streams.
+    - `--read-from-streams=all`: Read from all streams.
+    - `--read-from-streams=stream1,stream2`: Read from the specified streams only.
+    - `--read-from-streams=false`: Do not read from any streams.
+    - `--read-from-streams=none`: Do not read from any streams.
+    """
+    input_val: str = request.config.getoption(
+        "--read-from-streams",
+        default="default",  # type: ignore
+    )  # type: ignore
+
+    if isinstance(input_val, str):
+        if input_val.lower() == "false":
+            return "none"
+        if input_val.lower() in ["true", "suggested", "default"]:
+            # Default to 'default' (suggested) streams if the input is 'true', 'suggested', or
+            # 'default'.
+            # This is the default behavior if the option is not set.
+            return "default"
+        if input_val.lower() == "all":
+            # This will sometimes fail if the account doesn't have permissions
+            # to premium or restricted stream data.
+            return "all"
+
+        # If the input is a comma-separated list, split it into a list.
+        # This will return a one-element list if the input is a single stream name.
+        return input_val.split(",")
+
+    # Else, probably a bool; return it as is.
+    return input_val or "none"
+
+
+@pytest.fixture
+def read_scenarios(
+    request: pytest.FixtureRequest,
+) -> list[str] | Literal["all", "default"]:
+    """Return the value of `--read-scenarios`.
+
+    This argument is ignored if `--read-from-streams` is False or not set.
+
+    The input can be one of the following:
+    - [Omitted] - Default to 'config.json', meaning the default scenario will be read.
+    - `--read-scenarios=all`: Read all scenarios.
+    - `--read-scenarios=none`: Read no scenarios. (Overrides `--read-from-streams`, if set.)
+    - `--read-scenarios=scenario1,scenario2`: Read the specified scenarios only.
+
     """
-    A helper for pytest_generate_tests hook.
+    input_val = cast(
+        str,
+        request.config.getoption(
+            "--read-scenarios",
+            default="default",  # type: ignore
+        ),
+    )
+
+    if input_val.lower() == "default":
+        # Default config scenario is always 'config.json'.
+        return "default"
+
+    if input_val.lower() == "none":
+        # Default config scenario is always 'config.json'.
+        return []
+
+    return (
+        [
+            scenario_name.strip().lower().removesuffix(".json")
+            for scenario_name in input_val.split(",")
+        ]
+        if input_val
+        else []
+    )
+
+
+def pytest_addoption(parser: pytest.Parser) -> None:
+    """Add --connector-image to pytest's CLI."""
+    parser.addoption(
+        "--connector-image",
+        action="store",
+        default=None,
+        help="Use this pre-built connector Docker image instead of building one.",
+    )
+    parser.addoption(
+        "--read-from-streams",
+        action="store",
+        default=None,
+        help=read_from_streams.__doc__,
+    )
+    parser.addoption(
+        "--read-scenarios",
+        action="store",
+        default="default",
+        help=read_scenarios.__doc__,
+    )
+
+
+def pytest_generate_tests(metafunc: pytest.Metafunc) -> None:
+    """A helper for pytest_generate_tests hook.
 
     If a test method (in a class subclassed from our base class)
     declares an argument 'scenario', this function retrieves the

airbyte_cdk/test/standard_tests/source_base.py

@@ -4,6 +4,8 @@
 from dataclasses import asdict
 from typing import TYPE_CHECKING
 
+import pytest
+
 from airbyte_cdk.models import (
     AirbyteMessage,
     AirbyteStream,
@@ -48,13 +50,10 @@ class SourceTestSuiteBase(ConnectorTestSuiteBase):
             test_scenario=scenario,
             connector_root=self.get_connector_root_dir(),
         )
-        conn_status_messages: list[AirbyteMessage] = [
-            msg for msg in result._messages if msg.type == Type.CONNECTION_STATUS
-        ]  # noqa: SLF001  # Non-public API
-        num_status_messages = len(conn_status_messages)
+        num_status_messages = len(result.connection_status_messages)
         assert num_status_messages == 1, (
             f"Expected exactly one CONNECTION_STATUS message. Got {num_status_messages}: \n"
-            + "\n".join([str(m) for m in result._messages])
+            + "\n".join([str(m) for m in result.get_message_iterator()])
         )
 
     def test_discover(
@@ -62,6 +61,13 @@ class SourceTestSuiteBase(ConnectorTestSuiteBase):
         scenario: ConnectorTestScenario,
     ) -> None:
         """Standard test for `discover`."""
+        if scenario.expected_outcome.expect_exception():
+            # If the scenario expects an exception, we can't ensure it specifically would fail
+            # in discover, because some discover implementations do not need to make a connection.
+            # We skip this test in that case.
+            pytest.skip("Skipping discover test for scenario that expects an exception.")
+            return
+
         run_test_job(
             self.create_connector(scenario),
             "discover",
@@ -133,8 +139,8 @@ class SourceTestSuiteBase(ConnectorTestSuiteBase):
             catalog=configured_catalog,
         )
 
-        if not result.records:
-            raise AssertionError("Expected records but got none.")  # noqa: TRY003
+        if scenario.expected_outcome.expect_success() and not result.records:
+            raise AssertionError("Expected records but got none.")
 
     def test_fail_read_with_bad_catalog(
         self,

airbyte_cdk/test/standard_tests/util.py

@@ -10,7 +10,7 @@ from airbyte_cdk.test.standard_tests.connector_base import ConnectorTestSuiteBas
 from airbyte_cdk.test.standard_tests.declarative_sources import (
     DeclarativeSourceTestSuite,
 )
-from airbyte_cdk.test.standard_tests.destination_base import DestinationTestSuiteBase
+from airbyte_cdk.test.standard_tests.docker_base import DockerConnectorTestSuite
 from airbyte_cdk.test.standard_tests.source_base import SourceTestSuiteBase
 from airbyte_cdk.utils.connector_paths import (
     METADATA_YAML,
@@ -18,11 +18,12 @@ from airbyte_cdk.utils.connector_paths import (
 )
 
 TEST_CLASS_MAPPING: dict[
-    Literal["python", "manifest-only", "declarative"], type[ConnectorTestSuiteBase]
+    Literal["python", "manifest-only", "java"],
+    type[DockerConnectorTestSuite],
 ] = {
     "python": SourceTestSuiteBase,
     "manifest-only": DeclarativeSourceTestSuite,
-    # "declarative": DeclarativeSourceTestSuite,
+    "java": DockerConnectorTestSuite,
 }
 
 

airbyte_cdk/utils/connector_paths.py

@@ -181,11 +181,11 @@ def find_connector_root(from_paths: list[Path]) -> Path:
         # Check if the manifest file exists in the current directory
         for parent in [path, *path.parents]:
             if (parent / METADATA_YAML).exists():
-                return parent
+                return parent.absolute()
             if (parent / MANIFEST_YAML).exists():
-                return parent
+                return parent.absolute()
             if (parent / ACCEPTANCE_TEST_CONFIG).exists():
-                return parent
+                return parent.absolute()
             if parent.name == "airbyte_cdk":
                 break