Ejemplo n.º 1
0
class TestEnvironmentController():
    def setup_method(self):
        # provide mountable tmp directory for docker
        tempfile.tempdir = "/tmp" if not platform.system(
        ) == "Windows" else None
        test_datmo_dir = os.environ.get('TEST_DATMO_DIR',
                                        tempfile.gettempdir())
        self.temp_dir = tempfile.mkdtemp(dir=test_datmo_dir)
        self.project = ProjectController(self.temp_dir)
        self.environment = EnvironmentController(self.temp_dir)

    def teardown_method(self):
        pass

    def test_create(self):
        # 0) Test create when unsupported language given
        # 1) Test create when NO file exists and NO definition path exists
        # 2) Test create when NO file exists and definition path exists
        # 3) Test create when definition path exists and given
        # 4) Test create when file exists and definition path exists
        # 5) Test create when file exists but NO definition path exists
        # 6) Test create when definition path exists and given for NEW definition filepath

        self.project.init("test3", "test description")

        # 0) Test option 0
        try:
            self.environment.create({"language": "java"})
        except EnvironmentDoesNotExist:
            failed = True
        assert failed

        # 1) Test option 1
        failed = False
        try:
            self.environment.create({})
        except EnvironmentDoesNotExist:
            failed = True
        assert failed

        # Create environment definition
        definition_filepath = os.path.join(self.environment.home, "Dockerfile")
        with open(definition_filepath, "w") as f:
            f.write(to_unicode(str("FROM datmo/xgboost:cpu")))

        # 2) Test option 2
        environment_obj_1 = self.environment.create({})

        # Get file collection path
        file_collection_obj = self.environment.dal.file_collection. \
            get_by_id(environment_obj_1.file_collection_id)
        file_collection_dir = self.environment.file_driver. \
            get_collection_path(file_collection_obj.filehash)

        assert environment_obj_1
        assert environment_obj_1.id
        assert environment_obj_1.driver_type == "docker"
        assert environment_obj_1.file_collection_id
        assert environment_obj_1.definition_filename
        assert environment_obj_1.hardware_info
        assert environment_obj_1.unique_hash == file_collection_obj.filehash
        assert os.path.isfile(os.path.join(file_collection_dir, "Dockerfile"))
        assert os.path.isfile(
            os.path.join(file_collection_dir, "datmoDockerfile"))
        assert os.path.isfile(
            os.path.join(file_collection_dir, "hardware_info"))

        # 3) Test option 3
        input_dict = {
            "definition_filepath": definition_filepath,
        }

        # Create environment in the project
        environment_obj_2 = self.environment.create(input_dict)

        # Get file collection path
        file_collection_obj = self.environment.dal.file_collection. \
            get_by_id(environment_obj_2.file_collection_id)
        file_collection_dir = self.environment.file_driver. \
            get_collection_path(file_collection_obj.filehash)

        assert environment_obj_2
        assert environment_obj_2.id
        assert environment_obj_2.driver_type == "docker"
        assert environment_obj_2.file_collection_id
        assert environment_obj_2.definition_filename
        assert environment_obj_2.hardware_info
        assert environment_obj_2.unique_hash == file_collection_obj.filehash
        assert os.path.isfile(os.path.join(file_collection_dir, "Dockerfile"))
        assert os.path.isfile(
            os.path.join(file_collection_dir, "datmoDockerfile"))
        assert os.path.isfile(
            os.path.join(file_collection_dir, "hardware_info"))

        # Create script to test
        test_filepath = os.path.join(self.environment.home, "script.py")
        with open(test_filepath, "w") as f:
            f.write(to_unicode("import numpy\n"))
            f.write(to_unicode("import sklearn\n"))
            f.write(to_unicode("print('hello')\n"))

        # 4) Test option 4
        environment_obj_3 = self.environment.create({})

        # Get file collection path
        file_collection_obj = self.environment.dal.file_collection. \
            get_by_id(environment_obj_3.file_collection_id)
        file_collection_dir = self.environment.file_driver. \
            get_collection_path(file_collection_obj.filehash)

        assert environment_obj_3
        assert environment_obj_3.id
        assert environment_obj_3.driver_type == "docker"
        assert environment_obj_3.file_collection_id
        assert environment_obj_3.definition_filename
        assert environment_obj_3.hardware_info
        assert environment_obj_3.unique_hash == file_collection_obj.filehash
        assert os.path.isfile(os.path.join(file_collection_dir, "Dockerfile"))
        assert os.path.isfile(
            os.path.join(file_collection_dir, "datmoDockerfile"))
        assert os.path.isfile(
            os.path.join(file_collection_dir, "hardware_info"))

        # Remove definition filepath
        os.remove(definition_filepath)

        assert environment_obj_1.id == environment_obj_2.id
        assert environment_obj_2.id == environment_obj_3.id

        # 5) Test option 5
        environment_obj_4 = self.environment.create({})

        # Get file collection path
        file_collection_obj = self.environment.dal.file_collection. \
            get_by_id(environment_obj_4.file_collection_id)
        file_collection_dir = self.environment.file_driver. \
            get_collection_path(file_collection_obj.filehash)

        assert environment_obj_4
        assert environment_obj_4.id
        assert environment_obj_4.driver_type == "docker"
        assert environment_obj_4.file_collection_id
        assert environment_obj_4.definition_filename
        assert environment_obj_4.hardware_info
        assert environment_obj_4.unique_hash == file_collection_obj.filehash
        assert os.path.isfile(
            os.path.join(file_collection_dir, "requirements.txt"))
        assert os.path.isfile(os.path.join(file_collection_dir, "Dockerfile"))
        assert os.path.isfile(
            os.path.join(file_collection_dir, "datmoDockerfile"))
        assert os.path.isfile(
            os.path.join(file_collection_dir, "hardware_info"))

        assert environment_obj_1.id != environment_obj_4.id

        # 6) Test option 6

        # Create environment definition
        definition_filepath = os.path.join(self.environment.home, "Dockerfile")
        with open(definition_filepath, "w") as f:
            f.write(to_unicode(str("FROM cloudgear/ubuntu:14.04")))

        input_dict = {
            "definition_filepath": definition_filepath,
        }

        # Create a new environment obj
        environment_obj_5 = self.environment.create(input_dict)

        # Get file collection path
        file_collection_obj = self.environment.dal.file_collection. \
            get_by_id(environment_obj_5.file_collection_id)
        file_collection_dir = self.environment.file_driver. \
            get_collection_path(file_collection_obj.filehash)

        assert environment_obj_5
        assert environment_obj_5.id
        assert environment_obj_5.driver_type == "docker"
        assert environment_obj_5.file_collection_id
        assert environment_obj_5.definition_filename
        assert environment_obj_5.hardware_info
        assert environment_obj_5.unique_hash == file_collection_obj.filehash
        assert os.path.isfile(os.path.join(file_collection_dir, "Dockerfile"))
        assert os.path.isfile(
            os.path.join(file_collection_dir, "datmoDockerfile"))
        assert os.path.isfile(
            os.path.join(file_collection_dir, "hardware_info"))

        assert environment_obj_5.id != environment_obj_1.id
        assert environment_obj_5.id != environment_obj_4.id

    def test_build(self):
        # 1) Test build when no environment given
        # 2) Test build when definition path exists and given
        # 3) Test build when NO file exists and definition path exists
        # 4) Test build when file exists and definition path exists
        # 5) Test build when file exists but NO definition path exists
        self.project.init("test5", "test description")

        # 1) Test option 1
        failed = False
        try:
            _ = self.environment.build("does_not_exist")
        except EntityNotFound:
            failed = True
        assert failed

        # Create environment definition
        definition_filepath = os.path.join(self.environment.home, "Dockerfile")
        with open(definition_filepath, "w") as f:
            f.write(to_unicode(str("FROM datmo/xgboost:cpu")))
        input_dict = {
            "definition_filepath": definition_filepath,
        }

        # 2) Test option 2
        # Create environment in the project
        environment_obj_1 = self.environment.create(input_dict)
        result = self.environment.build(environment_obj_1.id)
        assert result

        # 3) Test option 3
        # Create environment in the project
        environment_obj_2 = self.environment.create({})
        result = self.environment.build(environment_obj_2.id)
        assert result

        # Create script to test
        test_filepath = os.path.join(self.environment.home, "script.py")
        with open(test_filepath, "w") as f:
            f.write(to_unicode("import numpy\n"))
            f.write(to_unicode("import sklearn\n"))
            f.write(to_unicode("print('hello')\n"))

        # 4) Test option 4
        environment_obj_3 = self.environment.create({})
        result = self.environment.build(environment_obj_3.id)
        assert result

        # test 2), 3), and 4) will result in the same environment
        assert environment_obj_1.id == environment_obj_2.id
        assert environment_obj_2.id == environment_obj_3.id

        # Test for building dockerfile when there exists not
        os.remove(definition_filepath)

        # 5) Test option 5
        environment_obj_4 = self.environment.create({})
        result = self.environment.build(environment_obj_4.id)
        assert result
        assert environment_obj_4.id != environment_obj_1.id

        # teardown
        self.environment.delete(environment_obj_1.id)
        self.environment.delete(environment_obj_4.id)

    def test_run(self):
        # 1) Test run simple command with simple Dockerfile
        # 2) Test run script, with autogenerated definition
        self.project.init("test5", "test description")

        # 1) Test option 1

        # Create environment definition
        definition_filepath = os.path.join(self.environment.home, "Dockerfile")
        with open(definition_filepath, "w") as f:
            f.write(to_unicode(str("FROM datmo/xgboost:cpu")))

        run_options = {
            "command": ["sh", "-c", "echo yo"],
            "ports": ["8888:8888"],
            "name": None,
            "volumes": None,
            "detach": False,
            "stdin_open": False,
            "tty": False,
            "gpu": False,
            "api": False
        }

        input_dict = {
            "definition_filepath": definition_filepath,
        }

        # Create environment in the project
        environment_obj = self.environment.create(input_dict)

        log_filepath = os.path.join(self.project.home, "task.log")

        # Build environment in the project
        _ = self.environment.build(environment_obj.id)

        # Run environment in the project
        return_code, run_id, logs = \
            self.environment.run(environment_obj.id, run_options, log_filepath)

        assert return_code == 0
        assert run_id
        assert logs

        # teardown
        self.environment.delete(environment_obj.id)

        # 2) Test option 2
        os.remove(definition_filepath)

        # Create script to test
        test_filepath = os.path.join(self.environment.home, "script.py")
        with open(test_filepath, "w") as f:
            f.write(to_unicode("import numpy\n"))
            f.write(to_unicode("import sklearn\n"))
            f.write(to_unicode("print('hello')\n"))

        # Create environment in the project
        environment_obj = self.environment.create({})
        self.environment.build(environment_obj.id)

        run_options = {
            "command": ["python", "script.py"],
            "ports": ["8888:8888"],
            "name": None,
            "volumes": {
                self.environment.home: {
                    'bind': '/home/',
                    'mode': 'rw'
                }
            },
            "detach": False,
            "stdin_open": False,
            "tty": False,
            "gpu": False,
            "api": False
        }

        # Run environment in the project
        return_code, run_id, logs = \
            self.environment.run(environment_obj.id, run_options, log_filepath)

        assert return_code == 0
        assert run_id
        assert logs

        # teardown
        self.environment.delete(environment_obj.id)

    def test_list(self):
        self.project.init("test4", "test description")

        # Create environment definition for object 1
        definition_path_1 = os.path.join(self.environment.home, "Dockerfile")
        with open(definition_path_1, "w") as f:
            f.write(to_unicode(str("FROM datmo/xgboost:cpu")))

        input_dict_1 = {
            "definition_filepath": definition_path_1,
        }

        # Create environment in the project
        environment_obj_1 = self.environment.create(input_dict_1)

        # Create environment definition for object 2
        definition_path_2 = os.path.join(self.environment.home, "Dockerfile2")
        with open(definition_path_2, "w") as f:
            f.write(to_unicode(str("FROM datmo/scikit-opencv")))

        input_dict_2 = {
            "definition_filepath": definition_path_2,
        }

        # Create second environment in the project
        environment_obj_2 = self.environment.create(input_dict_2)

        # List all environments and ensure they exist
        result = self.environment.list()

        assert len(result) == 2 and \
            environment_obj_1 in result and \
            environment_obj_2 in result

    def test_delete(self):
        self.project.init("test5", "test description")

        # Create environment definition
        definition_filepath = os.path.join(self.environment.home, "Dockerfile")
        with open(definition_filepath, "w") as f:
            f.write(to_unicode(str("FROM datmo/xgboost:cpu")))

        input_dict = {
            "definition_filepath": definition_filepath,
        }

        # Create environment in the project
        environment_obj = self.environment.create(input_dict)

        # Delete environment in the project
        result = self.environment.delete(environment_obj.id)

        # Check if environment retrieval throws error
        thrown = False
        try:
            self.environment.dal.environment.get_by_id(environment_obj.id)
        except EntityNotFound:
            thrown = True

        assert result == True and \
            thrown == True

    def test_stop(self):
        self.project.init("test5", "test description")

        # Create environment definition
        definition_filepath = os.path.join(self.environment.home, "Dockerfile")
        with open(definition_filepath, "w") as f:
            f.write(to_unicode(str("FROM datmo/xgboost:cpu")))

        run_options = {
            "command": ["sh", "-c", "echo yo"],
            "ports": ["8888:8888"],
            "name": None,
            "volumes": None,
            "detach": False,
            "stdin_open": False,
            "tty": False,
            "gpu": False,
            "api": False
        }

        # Create environment_driver definition
        env_def_path = os.path.join(self.project.home, "Dockerfile")
        with open(env_def_path, "w") as f:
            f.write(to_unicode(str("FROM datmo/xgboost:cpu")))

        input_dict = {
            "definition_filepath": definition_filepath,
        }

        # Create environment in the project
        environment_obj = self.environment.create(input_dict)

        log_filepath = os.path.join(self.project.home, "task.log")

        # Build environment in the project
        _ = self.environment.build(environment_obj.id)

        # Run environment in the project
        _, run_id, _ = \
            self.environment.run(environment_obj.id, run_options, log_filepath)

        # Stop the running environment
        return_code = self.environment.stop(run_id)
        assert return_code

        # teardown
        self.environment.delete(environment_obj.id)
Ejemplo n.º 2
0
class TestProjectController():
    def setup_method(self):
        self.temp_dir = tempfile.mkdtemp(dir=test_datmo_dir)
        Config().set_home(self.temp_dir)
        self.project_controller = ProjectController()
        self.environment_ids = []

    def teardown_method(self):
        if not check_docker_inactive(test_datmo_dir):
            self.project_controller = ProjectController()
            if self.project_controller.is_initialized:
                self.environment_controller = EnvironmentController()
                for env_id in list(set(self.environment_ids)):
                    if not self.environment_controller.delete(env_id):
                        raise Exception

    def test_init_failure_none(self):
        # Test failed case
        failed = False
        try:
            self.project_controller.init(None, None)
        except ValidationFailed:
            failed = True
        assert failed

    def test_init_failure_empty_str(self):
        # Test failed case
        failed = False
        try:
            self.project_controller.init("", "")
        except ValidationFailed:
            failed = True
        assert failed
        assert not self.project_controller.code_driver.is_initialized
        assert not self.project_controller.file_driver.is_initialized

    def test_init_failure_git_code_driver(self):
        # Create a HEAD.lock file in .git to make GitCodeDriver.init() fail
        if self.project_controller.code_driver.type == "git":
            git_dir = os.path.join(
                self.project_controller.code_driver.filepath, ".git")
            os.makedirs(git_dir)
            with open(os.path.join(git_dir, "HEAD.lock"), "a+") as f:
                f.write(to_bytes("test"))
            failed = False
            try:
                self.project_controller.init("test1", "test description")
            except Exception:
                failed = True
            assert failed
            assert not self.project_controller.code_driver.is_initialized
            assert not self.project_controller.file_driver.is_initialized

    def test_init_success(self):
        result = self.project_controller.init("test1", "test description")

        # Tested with is_initialized
        assert self.project_controller.model.name == "test1"
        assert self.project_controller.model.description == "test description"
        assert result and self.project_controller.is_initialized

        # Changeable by user, not tested in is_initialized
        assert self.project_controller.current_session.name == "default"

    # TODO: Test lower level functions (DAL, JSONStore, etc for interruptions)
    # def test_init_with_interruption(self):
    #     # Reinitializing after timed interruption during init
    #     @timeout_decorator.timeout(0.001, use_signals=False)
    #     def timed_init_with_interruption():
    #         result = self.project_controller.init("test1", "test description")
    #         return result
    #
    #     failed = False
    #     try:
    #         timed_init_with_interruption()
    #     except timeout_decorator.timeout_decorator.TimeoutError:
    #         failed = True
    #     # Tested with is_initialized
    #     assert failed
    #
    #     # Reperforming init after a wait of 2 seconds
    #     time.sleep(2)
    #     result = self.project_controller.init("test2", "test description")
    #     # Tested with is_initialized
    #     assert self.project_controller.model.name == "test2"
    #     assert self.project_controller.model.description == "test description"
    #     assert result and self.project_controller.is_initialized
    #
    #     # Changeable by user, not tested in is_initialized
    #     assert self.project_controller.current_session.name == "default"

    def test_init_reinit_failure_empty_str(self):
        _ = self.project_controller.init("test1", "test description")
        failed = True
        try:
            self.project_controller.init("", "")
        except Exception:
            failed = True
        assert failed
        assert self.project_controller.model.name == "test1"
        assert self.project_controller.model.description == "test description"
        assert self.project_controller.code_driver.is_initialized
        assert self.project_controller.file_driver.is_initialized

    def test_init_reinit_success(self):
        _ = self.project_controller.init("test1", "test description")
        # Test out functionality for re-initialize project
        result = self.project_controller.init("anything", "else")

        assert self.project_controller.model.name == "anything"
        assert self.project_controller.model.description == "else"
        assert result == True

    def test_cleanup_no_environment(self):
        self.project_controller.init("test2", "test description")
        result = self.project_controller.cleanup()

        assert not self.project_controller.code_driver.is_initialized
        assert not self.project_controller.file_driver.is_initialized
        # Ensure that containers built with this image do not exist
        # assert not self.project_controller.environment_driver.list_containers(filters={
        #     "ancestor": image_id
        # })
        assert result == True

    @pytest_docker_environment_failed_instantiation(test_datmo_dir)
    def test_cleanup_with_environment(self):
        self.project_controller.init("test2", "test description")
        result = self.project_controller.cleanup()

        assert not self.project_controller.code_driver.is_initialized
        assert not self.project_controller.file_driver.is_initialized
        assert not self.project_controller.environment_driver.list_images(
            "datmo-test2")
        # Ensure that containers built with this image do not exist
        # assert not self.project_controller.environment_driver.list_containers(filters={
        #     "ancestor": image_id
        # })
        assert result == True

    def test_status_basic(self):
        self.project_controller.init("test3", "test description")
        status_dict, current_snapshot, latest_snapshot_user_generated, latest_snapshot_auto_generated, unstaged_code, unstaged_environment, unstaged_files = \
            self.project_controller.status()

        assert status_dict
        assert isinstance(status_dict, dict)
        assert status_dict['name'] == "test3"
        assert status_dict['description'] == "test description"
        assert isinstance(status_dict['config'], dict)
        assert not current_snapshot
        assert not latest_snapshot_user_generated
        assert not latest_snapshot_auto_generated
        assert unstaged_code  # no files, but unstaged because blank commit id has not yet been created (no initial snapshot)
        assert not unstaged_environment
        assert not unstaged_files

    @pytest_docker_environment_failed_instantiation(test_datmo_dir)
    def test_status_snapshot_task(self):
        self.project_controller.init("test4", "test description")
        self.snapshot_controller = SnapshotController()
        self.task_controller = TaskController()

        # Create files to add
        self.snapshot_controller.file_driver.create("dirpath1", directory=True)
        self.snapshot_controller.file_driver.create("dirpath2", directory=True)
        self.snapshot_controller.file_driver.create("filepath1")

        # Create environment definition
        env_def_path = os.path.join(self.snapshot_controller.home,
                                    "Dockerfile")
        with open(env_def_path, "wb") as f:
            f.write(to_bytes("FROM python:3.5-alpine"))

        environment_paths = [env_def_path]

        # Create config
        config_filepath = os.path.join(self.snapshot_controller.home,
                                       "config.json")
        with open(config_filepath, "wb") as f:
            f.write(to_bytes(str("{}")))

        # Create stats
        stats_filepath = os.path.join(self.snapshot_controller.home,
                                      "stats.json")
        with open(stats_filepath, "wb") as f:
            f.write(to_bytes(str("{}")))

        input_dict = {
            "message":
                "my test snapshot",
            "paths": [
                os.path.join(self.snapshot_controller.home, "dirpath1"),
                os.path.join(self.snapshot_controller.home, "dirpath2"),
                os.path.join(self.snapshot_controller.home, "filepath1")
            ],
            "environment_paths":
                environment_paths,
            "config_filename":
                config_filepath,
            "stats_filename":
                stats_filepath,
        }

        # Create snapshot in the project, then wait, and try status
        first_snapshot = self.snapshot_controller.create(input_dict)

        status_dict, current_snapshot, latest_snapshot_user_generated, latest_snapshot_auto_generated, unstaged_code, unstaged_environment, unstaged_files = \
            self.project_controller.status()

        assert status_dict
        assert isinstance(status_dict, dict)
        assert status_dict['name'] == "test4"
        assert status_dict['description'] == "test description"
        assert isinstance(status_dict['config'], dict)
        assert not current_snapshot  # snapshot was created from other environments and files (so user is not on any current snapshot)
        assert isinstance(latest_snapshot_user_generated, Snapshot)
        assert latest_snapshot_user_generated == first_snapshot
        assert not latest_snapshot_auto_generated
        assert not unstaged_code
        assert not unstaged_environment
        assert not unstaged_files

        # Create and run a task and test if task is shown
        first_task = self.task_controller.create()

        # Create task_dict
        task_command = ["sh", "-c", "echo accuracy:0.45"]
        task_dict = {"command_list": task_command}

        updated_first_task = self.task_controller.run(
            first_task.id, task_dict=task_dict)
        before_snapshot_obj = self.task_controller.dal.snapshot.get_by_id(
            updated_first_task.before_snapshot_id)
        after_snapshot_obj = self.task_controller.dal.snapshot.get_by_id(
            updated_first_task.after_snapshot_id)
        before_environment_obj = self.task_controller.dal.environment.get_by_id(
            before_snapshot_obj.environment_id)
        after_environment_obj = self.task_controller.dal.environment.get_by_id(
            after_snapshot_obj.environment_id)
        assert before_environment_obj == after_environment_obj
        self.environment_ids.append(after_environment_obj.id)

        status_dict, current_snapshot, latest_snapshot_user_generated, latest_snapshot_auto_generated, unstaged_code, unstaged_environment, unstaged_files = \
            self.project_controller.status()

        assert status_dict
        assert isinstance(status_dict, dict)
        assert status_dict['name'] == "test4"
        assert status_dict['description'] == "test description"
        assert isinstance(status_dict['config'], dict)
        assert isinstance(current_snapshot, Snapshot)
        assert isinstance(latest_snapshot_user_generated, Snapshot)
        assert latest_snapshot_user_generated == first_snapshot
        assert isinstance(latest_snapshot_auto_generated, Snapshot)
        # current snapshot is the before snapshot for the run
        assert current_snapshot == before_snapshot_obj
        assert current_snapshot != latest_snapshot_auto_generated
        assert current_snapshot != latest_snapshot_user_generated
        # latest autogenerated snapshot is the after snapshot id
        assert latest_snapshot_auto_generated == after_snapshot_obj
        assert latest_snapshot_auto_generated != latest_snapshot_user_generated
        # user generated snapshot is not associated with any before or after snapshot
        assert latest_snapshot_user_generated != before_snapshot_obj
        assert latest_snapshot_user_generated != after_snapshot_obj
        assert not unstaged_code
        assert not unstaged_environment
        assert not unstaged_files
Ejemplo n.º 3
0
class TestSnapshotController():
    def setup_method(self):
        self.temp_dir = tempfile.mkdtemp(dir=test_datmo_dir)
        Config().set_home(self.temp_dir)
        self.environment_ids = []

    def teardown_method(self):
        if not check_docker_inactive(test_datmo_dir,
                                     Config().datmo_directory_name):
            self.__setup()
            self.environment_controller = EnvironmentController()
            for env_id in list(set(self.environment_ids)):
                if not self.environment_controller.delete(env_id):
                    raise Exception

    def __setup(self):
        Config().set_home(self.temp_dir)
        self.project_controller = ProjectController()
        self.project_controller.init("test", "test description")
        self.task_controller = TaskController()
        self.snapshot_controller = SnapshotController()

    def test_init_fail_project_not_init(self):
        Config().set_home(self.temp_dir)
        failed = False
        try:
            SnapshotController()
        except ProjectNotInitialized:
            failed = True
        assert failed

    def test_init_fail_invalid_path(self):
        test_home = "some_random_dir"
        Config().set_home(test_home)
        failed = False
        try:
            SnapshotController()
        except InvalidProjectPath:
            failed = True
        assert failed

    def test_current_snapshot(self):
        self.__setup()
        # Test failure for unstaged changes
        failed = False
        try:
            self.snapshot_controller.current_snapshot()
        except UnstagedChanges:
            failed = True
        assert failed
        # Test success after snapshot created
        snapshot_obj = self.__default_create()
        current_snapshot_obj = self.snapshot_controller.current_snapshot()
        assert current_snapshot_obj == snapshot_obj

    def test_create_fail_no_message(self):
        self.__setup()
        # Test no message
        failed = False
        try:
            self.snapshot_controller.create({})
        except RequiredArgumentMissing:
            failed = True
        assert failed

    def test_create_success_no_code(self):
        self.__setup()
        # Test default values for snapshot, fail due to code
        result = self.snapshot_controller.create(
            {"message": "my test snapshot"})
        assert result

    def test_create_success_no_code_environment(self):
        self.__setup()
        # Create environment definition
        env_def_path = os.path.join(
            self.project_controller.environment_driver.
            environment_directory_path, "Dockerfile")
        with open(env_def_path, "wb") as f:
            f.write(to_bytes("FROM python:3.5-alpine"))

        # test must pass when there is file present in root project folder
        result = self.snapshot_controller.create(
            {"message": "my test snapshot"})

        assert result

    def test_create_success_no_code_environment_files(self):
        self.__setup()
        # Create environment definition
        env_def_path = os.path.join(
            self.project_controller.environment_driver.
            environment_directory_path, "Dockerfile")
        with open(env_def_path, "wb") as f:
            f.write(to_bytes("FROM python:3.5-alpine"))

        test_file = os.path.join(
            self.project_controller.file_driver.files_directory, "test.txt")
        with open(test_file, "wb") as f:
            f.write(to_bytes(str("hello")))

        # test must pass when there is file present in root project folder
        result = self.snapshot_controller.create(
            {"message": "my test snapshot"})

        assert result

    def test_create_no_environment_detected_in_file(self):
        self.__setup()

        # Test default values for snapshot, fail due to no environment from file
        self.snapshot_controller.file_driver.create("filepath1")
        snapshot_obj_0 = self.snapshot_controller.create(
            {"message": "my test snapshot"})
        assert isinstance(snapshot_obj_0, Snapshot)
        assert snapshot_obj_0.code_id
        assert snapshot_obj_0.environment_id
        assert snapshot_obj_0.file_collection_id
        assert snapshot_obj_0.config == {}
        assert snapshot_obj_0.stats == {}

    def test_create_success_default_detected_in_file(self):
        self.__setup()
        # Test default values for snapshot when there is no environment
        test_filepath = os.path.join(self.snapshot_controller.home,
                                     "script.py")
        with open(test_filepath, "wb") as f:
            f.write(to_bytes("import os\n"))
            f.write(to_bytes("import sys\n"))
            f.write(to_bytes("print('hello')\n"))

        snapshot_obj_1 = self.snapshot_controller.create(
            {"message": "my test snapshot"})

        assert isinstance(snapshot_obj_1, Snapshot)
        assert snapshot_obj_1.code_id
        assert snapshot_obj_1.environment_id
        assert snapshot_obj_1.file_collection_id
        assert snapshot_obj_1.config == {}
        assert snapshot_obj_1.stats == {}

    def test_create_success_default_env_def(self):
        self.__setup()
        # Create environment definition
        env_def_path = os.path.join(
            self.project_controller.environment_driver.
            environment_directory_path, "Dockerfile")
        with open(env_def_path, "wb") as f:
            f.write(to_bytes("FROM python:3.5-alpine"))

        # Creating a file in project folder
        test_filepath = os.path.join(self.snapshot_controller.home,
                                     "script.py")
        with open(test_filepath, "wb") as f:
            f.write(to_bytes("import numpy\n"))
            f.write(to_bytes("import sklearn\n"))
            f.write(to_bytes("print('hello')\n"))

        # Test default values for snapshot, success
        snapshot_obj = self.snapshot_controller.create(
            {"message": "my test snapshot"})

        assert isinstance(snapshot_obj, Snapshot)
        assert snapshot_obj.code_id
        assert snapshot_obj.environment_id
        assert snapshot_obj.file_collection_id
        assert snapshot_obj.config == {}
        assert snapshot_obj.stats == {}

    def test_create_success_with_environment(self):
        self.__setup()
        # Create environment definition
        env_def_path = os.path.join(
            self.project_controller.environment_driver.
            environment_directory_path, "Dockerfile")
        with open(env_def_path, "wb") as f:
            f.write(to_bytes("FROM python:3.5-alpine"))

        # creating a file in project folder
        test_filepath = os.path.join(self.snapshot_controller.home,
                                     "script.py")
        with open(test_filepath, "wb") as f:
            f.write(to_bytes("import numpy\n"))
            f.write(to_bytes("import sklearn\n"))
            f.write(to_bytes("print('hello')\n"))

        # Test default values for snapshot, success
        snapshot_obj = self.snapshot_controller.create(
            {"message": "my test snapshot"})

        assert isinstance(snapshot_obj, Snapshot)
        assert snapshot_obj.code_id
        assert snapshot_obj.environment_id
        assert snapshot_obj.file_collection_id
        assert snapshot_obj.config == {}
        assert snapshot_obj.stats == {}

    def test_create_success_env_paths(self):
        self.__setup()
        # Create environment definition
        random_dir = os.path.join(self.snapshot_controller.home, "random_dir")
        os.makedirs(random_dir)
        env_def_path = os.path.join(random_dir, "randomDockerfile")
        with open(env_def_path, "wb") as f:
            f.write(to_bytes("FROM python:3.5-alpine"))
        environment_paths = [env_def_path + ">Dockerfile"]

        # Test default values for snapshot, success
        snapshot_obj = self.snapshot_controller.create({
            "message":
            "my test snapshot",
            "environment_paths":
            environment_paths
        })

        assert isinstance(snapshot_obj, Snapshot)
        assert snapshot_obj.code_id
        assert snapshot_obj.environment_id
        assert snapshot_obj.file_collection_id
        assert snapshot_obj.config == {}
        assert snapshot_obj.stats == {}

    def test_create_success_default_env_def_duplicate(self):
        self.__setup()
        # Test 2 snapshots with same parameters
        # Create environment definition
        env_def_path = os.path.join(
            self.project_controller.environment_driver.
            environment_directory_path, "Dockerfile")
        with open(env_def_path, "wb") as f:
            f.write(to_bytes("FROM python:3.5-alpine"))

        # Creating a file in project folder
        test_filepath = os.path.join(self.snapshot_controller.home,
                                     "script.py")
        with open(test_filepath, "wb") as f:
            f.write(to_bytes("import numpy\n"))
            f.write(to_bytes("import sklearn\n"))
            f.write(to_bytes("print('hello')\n"))

        snapshot_obj = self.snapshot_controller.create(
            {"message": "my test snapshot"})

        snapshot_obj_1 = self.snapshot_controller.create(
            {"message": "my test snapshot"})

        # Should return the same object back
        assert snapshot_obj_1.id == snapshot_obj.id
        assert snapshot_obj_1.code_id == snapshot_obj.code_id
        assert snapshot_obj_1.environment_id == \
               snapshot_obj.environment_id
        assert snapshot_obj_1.file_collection_id == \
               snapshot_obj.file_collection_id
        assert snapshot_obj_1.config == \
               snapshot_obj.config
        assert snapshot_obj_1.stats == \
               snapshot_obj.stats

    def test_create_success_given_files_env_def_config_file_stats_file(self):
        self.__setup()
        # Create environment definition
        env_def_path = os.path.join(
            self.project_controller.environment_driver.
            environment_directory_path, "Dockerfile")
        with open(env_def_path, "wb") as f:
            f.write(to_bytes("FROM python:3.5-alpine"))

        # Creating a file in project folder
        test_filepath = os.path.join(self.snapshot_controller.home,
                                     "script.py")
        with open(test_filepath, "wb") as f:
            f.write(to_bytes("import numpy\n"))
            f.write(to_bytes("import sklearn\n"))
            f.write(to_bytes("print('hello')\n"))

        snapshot_obj = self.snapshot_controller.create(
            {"message": "my test snapshot"})

        # Create files to add
        _, files_directory_name = os.path.split(
            self.project_controller.file_driver.files_directory)
        files_directory_relative_path = os.path.join(
            self.project_controller.file_driver.datmo_directory_name,
            files_directory_name)
        self.snapshot_controller.file_driver.create(os.path.join(
            files_directory_relative_path, "dirpath1"),
                                                    directory=True)
        self.snapshot_controller.file_driver.create(os.path.join(
            files_directory_relative_path, "dirpath2"),
                                                    directory=True)
        self.snapshot_controller.file_driver.create(
            os.path.join(files_directory_relative_path, "filepath1"))

        # Create config
        config_filepath = os.path.join(self.snapshot_controller.home,
                                       "config.json")
        with open(config_filepath, "wb") as f:
            f.write(to_bytes(str('{"foo":"bar"}')))

        # Create stats
        stats_filepath = os.path.join(self.snapshot_controller.home,
                                      "stats.json")
        with open(stats_filepath, "wb") as f:
            f.write(to_bytes(str('{"foo":"bar"}')))

        input_dict = {
            "message": "my test snapshot",
            "config_filepath": config_filepath,
            "stats_filepath": stats_filepath,
        }
        # Create snapshot in the project
        snapshot_obj_4 = self.snapshot_controller.create(input_dict)

        assert snapshot_obj_4 != snapshot_obj
        assert snapshot_obj_4.code_id != snapshot_obj.code_id
        assert snapshot_obj_4.environment_id == \
               snapshot_obj.environment_id
        assert snapshot_obj_4.file_collection_id != \
               snapshot_obj.file_collection_id
        assert snapshot_obj_4.config == {"foo": "bar"}
        assert snapshot_obj_4.stats == {"foo": "bar"}

    def test_create_success_given_files_env_def_different_config_stats(self):
        self.__setup()
        # Create environment definition
        env_def_path = os.path.join(
            self.project_controller.environment_driver.
            environment_directory_path, "Dockerfile")
        with open(env_def_path, "wb") as f:
            f.write(to_bytes("FROM python:3.5-alpine"))

        # Creating a file in project folder
        test_filepath = os.path.join(self.snapshot_controller.home,
                                     "script.py")
        with open(test_filepath, "wb") as f:
            f.write(to_bytes("import numpy\n"))
            f.write(to_bytes("import sklearn\n"))
            f.write(to_bytes("print('hello')\n"))

        snapshot_obj = self.snapshot_controller.create(
            {"message": "my test snapshot"})

        # Create files to add
        _, files_directory_name = os.path.split(
            self.project_controller.file_driver.files_directory)
        files_directory_relative_path = os.path.join(
            self.project_controller.file_driver.datmo_directory_name,
            files_directory_name)
        self.snapshot_controller.file_driver.create(os.path.join(
            files_directory_relative_path, "dirpath1"),
                                                    directory=True)
        self.snapshot_controller.file_driver.create(os.path.join(
            files_directory_relative_path, "dirpath2"),
                                                    directory=True)
        self.snapshot_controller.file_driver.create(
            os.path.join(files_directory_relative_path, "filepath1"))

        # Create config
        config_filepath = os.path.join(self.snapshot_controller.home,
                                       "config.json")
        with open(config_filepath, "wb") as f:
            f.write(to_bytes(str('{"foo":"bar"}')))

        # Create stats
        stats_filepath = os.path.join(self.snapshot_controller.home,
                                      "stats.json")
        with open(stats_filepath, "wb") as f:
            f.write(to_bytes(str('{"foo":"bar"}')))

        # Test different config and stats inputs
        input_dict = {
            "message": "my test snapshot",
            "config_filename": "different_name",
            "stats_filename": "different_name",
        }

        # Create snapshot in the project
        snapshot_obj_1 = self.snapshot_controller.create(input_dict)

        assert snapshot_obj_1 != snapshot_obj
        assert snapshot_obj_1.config == {}
        assert snapshot_obj_1.stats == {}

    def test_create_success_given_files_env_def_direct_config_stats(self):
        self.__setup()
        # Create environment definition
        env_def_path = os.path.join(
            self.project_controller.environment_driver.
            environment_directory_path, "Dockerfile")
        with open(env_def_path, "wb") as f:
            f.write(to_bytes("FROM python:3.5-alpine"))

        # Create files to add
        _, files_directory_name = os.path.split(
            self.project_controller.file_driver.files_directory)
        files_directory_relative_path = os.path.join(
            self.project_controller.file_driver.datmo_directory_name,
            files_directory_name)
        self.snapshot_controller.file_driver.create(os.path.join(
            files_directory_relative_path, "dirpath1"),
                                                    directory=True)
        self.snapshot_controller.file_driver.create(os.path.join(
            files_directory_relative_path, "dirpath2"),
                                                    directory=True)
        self.snapshot_controller.file_driver.create(
            os.path.join(files_directory_relative_path, "filepath1"))

        # Creating a file in project folder
        test_filepath = os.path.join(self.snapshot_controller.home,
                                     "script.py")
        with open(test_filepath, "wb") as f:
            f.write(to_bytes("import numpy\n"))
            f.write(to_bytes("import sklearn\n"))
            f.write(to_bytes("print('hello')\n"))

        # Test different config and stats inputs
        input_dict = {
            "message": "my test snapshot",
            "config": {
                "foo": "bar"
            },
            "stats": {
                "foo": "bar"
            },
        }

        # Create snapshot in the project
        snapshot_obj_6 = self.snapshot_controller.create(input_dict)

        assert snapshot_obj_6.config == {"foo": "bar"}
        assert snapshot_obj_6.stats == {"foo": "bar"}

    @pytest_docker_environment_failed_instantiation(test_datmo_dir)
    def test_create_from_task(self):
        self.__setup()
        # 0) Test if fails with TaskNotComplete error
        # 1) Test if success with empty task files, results
        # 2) Test if success with task files, results, and message
        # 3) Test if success with message, label, config and stats
        # 4) Test if success with updated stats from after_snapshot_id and task_results

        # Create task in the project
        task_obj = self.task_controller.create()

        # 0) Test option 0
        failed = False
        try:
            _ = self.snapshot_controller.create_from_task(
                message="my test snapshot", task_id=task_obj.id)
        except TaskNotComplete:
            failed = True
        assert failed

        # 1) Test option 1

        # Create task_dict
        task_command = ["sh", "-c", "echo test"]
        task_dict = {"command_list": task_command}

        # Create environment definition
        env_def_path = os.path.join(self.project_controller.home, "Dockerfile")
        with open(env_def_path, "wb") as f:
            f.write(to_bytes("FROM python:3.5-alpine"))

        updated_task_obj = self.task_controller.run(task_obj.id,
                                                    task_dict=task_dict)
        after_snapshot_obj = self.task_controller.dal.snapshot.get_by_id(
            updated_task_obj.after_snapshot_id)
        environment_obj = self.task_controller.dal.environment.get_by_id(
            after_snapshot_obj.environment_id)
        self.environment_ids.append(environment_obj.id)

        snapshot_obj = self.snapshot_controller.create_from_task(
            message="my test snapshot", task_id=updated_task_obj.id)

        assert isinstance(snapshot_obj, Snapshot)
        assert snapshot_obj.id == updated_task_obj.after_snapshot_id
        assert snapshot_obj.message == "my test snapshot"
        assert snapshot_obj.stats == updated_task_obj.results
        assert snapshot_obj.visible == True

        # Create new task and corresponding dict
        task_obj = self.task_controller.create()
        task_command = ["sh", "-c", "echo accuracy:0.45"]
        task_dict = {"command_list": task_command}

        # Create environment definition
        env_def_path = os.path.join(self.project_controller.home, "Dockerfile")
        with open(env_def_path, "wb") as f:
            f.write(to_bytes("FROM python:3.5-alpine"))

        # Test the default values
        updated_task_obj = self.task_controller.run(task_obj.id,
                                                    task_dict=task_dict)
        after_snapshot_obj = self.task_controller.dal.snapshot.get_by_id(
            updated_task_obj.after_snapshot_id)
        environment_obj = self.task_controller.dal.environment.get_by_id(
            after_snapshot_obj.environment_id)
        self.environment_ids.append(environment_obj.id)

        # 2) Test option 2
        snapshot_obj = self.snapshot_controller.create_from_task(
            message="my test snapshot", task_id=updated_task_obj.id)

        assert isinstance(snapshot_obj, Snapshot)
        assert snapshot_obj.id == updated_task_obj.after_snapshot_id
        assert snapshot_obj.message == "my test snapshot"
        assert snapshot_obj.stats == updated_task_obj.results
        assert snapshot_obj.visible == True

        # 3) Test option 3
        test_config = {"algo": "regression"}
        test_stats = {"accuracy": 0.9}
        snapshot_obj = self.snapshot_controller.create_from_task(
            message="my test snapshot",
            task_id=updated_task_obj.id,
            label="best",
            config=test_config,
            stats=test_stats)

        assert isinstance(snapshot_obj, Snapshot)
        assert snapshot_obj.id == updated_task_obj.after_snapshot_id
        assert snapshot_obj.message == "my test snapshot"
        assert snapshot_obj.label == "best"
        assert snapshot_obj.config == test_config
        assert snapshot_obj.stats == test_stats
        assert snapshot_obj.visible == True

        # 4) Test option 4
        test_config = {"algo": "regression"}
        test_stats = {"new_key": 0.9}
        task_obj_2 = self.task_controller.create()
        updated_task_obj_2 = self.task_controller.run(task_obj_2.id,
                                                      task_dict=task_dict,
                                                      snapshot_dict={
                                                          "config":
                                                          test_config,
                                                          "stats": test_stats
                                                      })
        after_snapshot_obj = self.task_controller.dal.snapshot.get_by_id(
            updated_task_obj_2.after_snapshot_id)
        environment_obj = self.task_controller.dal.environment.get_by_id(
            after_snapshot_obj.environment_id)
        self.environment_ids.append(environment_obj.id)

        snapshot_obj = self.snapshot_controller.create_from_task(
            message="my test snapshot",
            task_id=updated_task_obj_2.id,
            label="best")
        updated_stats_dict = {}
        updated_stats_dict.update(test_stats)
        updated_stats_dict.update(updated_task_obj.results)

        assert isinstance(snapshot_obj, Snapshot)
        assert snapshot_obj.id == updated_task_obj_2.after_snapshot_id
        assert snapshot_obj.message == "my test snapshot"
        assert snapshot_obj.label == "best"
        assert snapshot_obj.stats == updated_stats_dict
        assert snapshot_obj.visible == True

    def __default_create(self):
        # Create files to add
        _, files_directory_name = os.path.split(
            self.project_controller.file_driver.files_directory)
        files_directory_relative_path = os.path.join(
            self.project_controller.file_driver.datmo_directory_name,
            files_directory_name)
        self.snapshot_controller.file_driver.create(os.path.join(
            files_directory_relative_path, "dirpath1"),
                                                    directory=True)
        self.snapshot_controller.file_driver.create(os.path.join(
            files_directory_relative_path, "dirpath2"),
                                                    directory=True)
        self.snapshot_controller.file_driver.create(
            os.path.join(files_directory_relative_path, "filepath1"))
        self.snapshot_controller.file_driver.create("filepath2")
        with open(os.path.join(self.snapshot_controller.home, "filepath2"),
                  "wb") as f:
            f.write(to_bytes(str("import sys\n")))
        # Create environment_driver definition
        env_def_path = os.path.join(
            self.project_controller.environment_driver.
            environment_directory_path, "Dockerfile")
        with open(env_def_path, "wb") as f:
            f.write(to_bytes("FROM python:3.5-alpine"))

        # Create config
        config_filepath = os.path.join(self.snapshot_controller.home,
                                       "config.json")
        with open(config_filepath, "wb") as f:
            f.write(to_bytes(str("{}")))

        # Create stats
        stats_filepath = os.path.join(self.snapshot_controller.home,
                                      "stats.json")
        with open(stats_filepath, "wb") as f:
            f.write(to_bytes(str("{}")))

        input_dict = {
            "message": "my test snapshot",
            "config_filename": config_filepath,
            "stats_filename": stats_filepath,
        }

        # Create snapshot in the project
        return self.snapshot_controller.create(input_dict)

    def test_check_unstaged_changes(self):
        self.__setup()
        # Check unstaged changes
        failed = False
        try:
            self.snapshot_controller.check_unstaged_changes()
        except UnstagedChanges:
            failed = True
        assert failed
        # Check no unstaged changes
        _ = self.__default_create()
        result = self.snapshot_controller.check_unstaged_changes()
        assert result == False

    def test_checkout(self):
        self.__setup()
        # Create snapshot
        snapshot_obj_1 = self.__default_create()

        # Create duplicate snapshot in project
        self.snapshot_controller.file_driver.create("test")
        snapshot_obj_2 = self.__default_create()

        assert snapshot_obj_2 != snapshot_obj_1

        # Checkout to snapshot 1 using snapshot id
        result = self.snapshot_controller.checkout(snapshot_obj_1.id)
        # TODO: Check for which snapshot we are on

        assert result == True

    def test_list(self):
        self.__setup()
        # Create file to add to snapshot
        test_filepath_1 = os.path.join(self.snapshot_controller.home,
                                       "test.txt")
        with open(test_filepath_1, "wb") as f:
            f.write(to_bytes(str("test")))

        # Create snapshot in the project
        snapshot_obj_1 = self.__default_create()

        # Create file to add to second snapshot
        test_filepath_2 = os.path.join(self.snapshot_controller.home,
                                       "test2.txt")
        with open(test_filepath_2, "wb") as f:
            f.write(to_bytes(str("test2")))

        # Create second snapshot in the project
        snapshot_obj_2 = self.__default_create()

        # List all snapshots and ensure they exist
        result = self.snapshot_controller.list()

        assert len(result) == 2 and \
            snapshot_obj_1 in result and \
            snapshot_obj_2 in result

        # List all tasks regardless of filters in ascending
        result = self.snapshot_controller.list(sort_key='created_at',
                                               sort_order='ascending')

        assert len(result) == 2 and \
               snapshot_obj_1 in result and \
               snapshot_obj_2 in result
        assert result[0].created_at <= result[-1].created_at

        # List all tasks regardless of filters in descending
        result = self.snapshot_controller.list(sort_key='created_at',
                                               sort_order='descending')
        assert len(result) == 2 and \
               snapshot_obj_1 in result and \
               snapshot_obj_2 in result
        assert result[0].created_at >= result[-1].created_at

        # Wrong order being passed in
        failed = False
        try:
            _ = self.snapshot_controller.list(sort_key='created_at',
                                              sort_order='wrong_order')
        except InvalidArgumentType:
            failed = True
        assert failed

        # Wrong key and order being passed in
        failed = False
        try:
            _ = self.snapshot_controller.list(sort_key='wrong_key',
                                              sort_order='wrong_order')
        except InvalidArgumentType:
            failed = True
        assert failed

        # wrong key and right order being passed in
        expected_result = self.snapshot_controller.list(sort_key='created_at',
                                                        sort_order='ascending')
        result = self.snapshot_controller.list(sort_key='wrong_key',
                                               sort_order='ascending')
        expected_ids = [item.id for item in expected_result]
        ids = [item.id for item in result]
        assert set(expected_ids) == set(ids)

        # List snapshots with visible filter
        result = self.snapshot_controller.list(visible=False)
        assert len(result) == 0

        result = self.snapshot_controller.list(visible=True)
        assert len(result) == 2 and \
               snapshot_obj_1 in result and \
               snapshot_obj_2 in result

    def test_update(self):
        self.__setup()
        test_config = {"config_foo": "bar"}
        test_stats = {"stats_foo": "bar"}
        test_message = 'test_message'
        test_label = 'test_label'

        # Updating all config, stats, message and label
        # Create snapshot in the project
        snapshot_obj = self.__default_create()

        # Update snapshot in the project
        self.snapshot_controller.update(snapshot_obj.id,
                                        config=test_config,
                                        stats=test_stats,
                                        message=test_message,
                                        label=test_label)

        # Get the updated snapshot obj
        updated_snapshot_obj = self.snapshot_controller.dal.snapshot.get_by_id(
            snapshot_obj.id)
        assert updated_snapshot_obj.config == test_config
        assert updated_snapshot_obj.stats == test_stats
        assert updated_snapshot_obj.message == test_message
        assert updated_snapshot_obj.label == test_label

        # Updating config, stats
        # Create snapshot in the project
        snapshot_obj = self.__default_create()

        # Update snapshot in the project
        self.snapshot_controller.update(snapshot_obj.id,
                                        config=test_config,
                                        stats=test_stats)

        # Get the updated snapshot obj
        updated_snapshot_obj = self.snapshot_controller.dal.snapshot.get_by_id(
            snapshot_obj.id)
        assert updated_snapshot_obj.config == test_config
        assert updated_snapshot_obj.stats == test_stats

        # Updating both message and label
        # Create snapshot in the project
        snapshot_obj = self.__default_create()

        # Update snapshot in the project
        self.snapshot_controller.update(snapshot_obj.id,
                                        message=test_message,
                                        label=test_label)

        # Get the updated snapshot obj
        updated_snapshot_obj = self.snapshot_controller.dal.snapshot.get_by_id(
            snapshot_obj.id)

        assert updated_snapshot_obj.message == test_message
        assert updated_snapshot_obj.label == test_label

        # Updating only message
        # Create snapshot in the project
        snapshot_obj_1 = self.__default_create()

        # Update snapshot in the project
        self.snapshot_controller.update(snapshot_obj_1.id,
                                        message=test_message)

        # Get the updated snapshot obj
        updated_snapshot_obj_1 = self.snapshot_controller.dal.snapshot.get_by_id(
            snapshot_obj_1.id)

        assert updated_snapshot_obj_1.message == test_message

        # Updating only label
        # Create snapshot in the project
        snapshot_obj_2 = self.__default_create()

        # Update snapshot in the project
        self.snapshot_controller.update(snapshot_obj_2.id, label=test_label)

        # Get the updated snapshot obj
        updated_snapshot_obj_2 = self.snapshot_controller.dal.snapshot.get_by_id(
            snapshot_obj_2.id)

        assert updated_snapshot_obj_2.label == test_label

    def test_get(self):
        self.__setup()
        # Test failure for no snapshot
        failed = False
        try:
            self.snapshot_controller.get("random")
        except DoesNotExist:
            failed = True
        assert failed

        # Test success for snapshot
        snapshot_obj = self.__default_create()
        snapshot_obj_returned = self.snapshot_controller.get(snapshot_obj.id)
        assert snapshot_obj == snapshot_obj_returned

    def test_get_files(self):
        self.__setup()
        # Test failure case
        failed = False
        try:
            self.snapshot_controller.get_files("random")
        except DoesNotExist:
            failed = True
        assert failed

        # Test success case
        snapshot_obj = self.__default_create()
        result = self.snapshot_controller.get_files(snapshot_obj.id)
        file_collection_obj = self.task_controller.dal.file_collection.get_by_id(
            snapshot_obj.file_collection_id)

        file_names = [item.name for item in result]

        assert len(result) == 1
        for item in result:
            assert isinstance(item, TextIOWrapper)
            assert item.mode == "r"
        assert os.path.join(self.task_controller.home, ".datmo", "collections",
                            file_collection_obj.filehash,
                            "filepath1") in file_names

        result = self.snapshot_controller.get_files(snapshot_obj.id, mode="a")

        assert len(result) == 1
        for item in result:
            assert isinstance(item, TextIOWrapper)
            assert item.mode == "a"
        assert os.path.join(self.task_controller.home, ".datmo", "collections",
                            file_collection_obj.filehash,
                            "filepath1") in file_names

    def test_delete(self):
        self.__setup()
        # Create snapshot in the project
        snapshot_obj = self.__default_create()

        # Delete snapshot in the project
        result = self.snapshot_controller.delete(snapshot_obj.id)

        # Check if snapshot retrieval throws error
        thrown = False
        try:
            self.snapshot_controller.dal.snapshot.get_by_id(snapshot_obj.id)
        except EntityNotFound:
            thrown = True

        assert result == True and \
            thrown == True
Ejemplo n.º 4
0
class TestEnvironmentController():
    def setup_method(self):
        self.temp_dir = tempfile.mkdtemp(dir=test_datmo_dir)
        Config().set_home(self.temp_dir)
        self.project_controller = ProjectController()

    def teardown_method(self):
        pass

    def __setup(self):
        self.project_controller.init("test_setup", "test description")
        self.environment_controller = EnvironmentController()
        with open(os.path.join(self.temp_dir, "test.txt"), "wb") as f:
            f.write(to_bytes("hello"))
        self.random_filepath = os.path.join(
            self.environment_controller.file_driver.environment_directory,
            "test")
        with open(self.random_filepath, "wb") as f:
            f.write(to_bytes("cool"))
        self.definition_filepath = os.path.join(
            self.environment_controller.file_driver.environment_directory,
            "Dockerfile")
        with open(self.definition_filepath, "wb") as f:
            f.write(to_bytes("FROM python:3.5-alpine"))

    def test_init_fail_project_not_init(self):
        Config().set_home(self.temp_dir)
        failed = False
        try:
            EnvironmentController()
        except ProjectNotInitialized:
            failed = True
        assert failed

    def test_get_supported_environments(self):
        self.__setup()
        result = self.environment_controller.get_supported_environments()
        assert result

    def test_setup(self):
        self.project_controller.init("test_setup", "test description")
        self.environment_controller = EnvironmentController()

        # Test success setup once (no files present)
        options = {"name": "xgboost:cpu"}
        result = self.environment_controller.setup(options=options)
        output_definition_filepath = os.path.join(
            self.environment_controller.file_driver.environment_directory,
            "Dockerfile")

        assert isinstance(result, Environment)
        assert result.name == options['name']
        assert result.description == "supported base environment created by datmo"
        assert os.path.isfile(output_definition_filepath)
        assert "FROM datmo/xgboost:cpu" in open(output_definition_filepath,
                                                "r").read()

        # Test success setup again (files present, but staged)
        options = {"name": "xgboost:cpu"}
        result = self.environment_controller.setup(options=options)
        output_definition_filepath = os.path.join(
            self.environment_controller.file_driver.environment_directory,
            "Dockerfile")

        assert isinstance(result, Environment)
        assert result.name == options['name']
        assert result.description == "supported base environment created by datmo"
        assert os.path.isfile(output_definition_filepath)
        assert "FROM datmo/xgboost:cpu" in open(output_definition_filepath,
                                                "r").read()

        # Test failure in downstream function (e.g. bad inputs, no name given)
        failed = False
        try:
            self.environment_controller.setup(options={})
        except EnvironmentDoesNotExist:
            failed = True
        assert failed

        # Change environment file
        with open(output_definition_filepath, "wb") as f:
            f.write(to_bytes("new content"))

        # Test failure setup (unstaged changes)
        failed = False
        try:
            self.environment_controller.setup(options=options)
        except UnstagedChanges:
            failed = True
        assert failed

    def test_create(self):
        # 0) Test SUCCESS create when definition path exists in project environment directory (no input, no root) -- with hardware file
        # 1) Test SUCCESS create when definition path exists in project environment directory (no input, no root)
        # 5) Test SUCCESS when definition path exists in project environment directory and passed from input dict (takes input)
        # 2) Test SUCCESS create when definition path exists in root project folder (no input, no project environment dir)
        # 3) Test SUCCESS create when definition path is passed into input dict (takes input, no project environment dir)
        # 4) Test SUCCESS create when definition path is passed into input dict along with expected filename to be saved
        # 6) Test FAIL when passing same filepath with same filename into input dict

        self.__setup()

        input_dict_0 = {"name": "test", "description": "test description"}

        # 0) Test option 0 (cannot test hash because hardware is machine-dependent)
        environment_obj_0 = self.environment_controller.create(input_dict_0)
        assert environment_obj_0
        assert isinstance(environment_obj_0, Environment)
        assert environment_obj_0.id
        assert environment_obj_0.driver_type == "docker"
        assert environment_obj_0.file_collection_id
        assert environment_obj_0.definition_filename
        assert environment_obj_0.hardware_info
        assert environment_obj_0.unique_hash
        assert environment_obj_0.name == "test"
        assert environment_obj_0.description == "test description"

        # Get file collection path
        file_collection_obj = self.environment_controller.dal.file_collection. \
            get_by_id(environment_obj_0.file_collection_id)
        file_collection_dir = self.environment_controller.file_driver. \
            get_collection_path(file_collection_obj.filehash)
        assert os.path.isfile(os.path.join(file_collection_dir, "test"))
        assert os.path.isfile(os.path.join(file_collection_dir, "Dockerfile"))
        output = open(os.path.join(file_collection_dir, "Dockerfile"),
                      "r").read()
        print(repr(output))
        assert os.path.isfile(
            os.path.join(file_collection_dir, "datmoDockerfile"))
        output = open(os.path.join(file_collection_dir, "datmoDockerfile"),
                      "r").read()
        print(repr(output))
        assert os.path.isfile(
            os.path.join(file_collection_dir, "hardware_info"))
        output = open(os.path.join(file_collection_dir, "hardware_info"),
                      "r").read()
        print(repr(output))

        # 1) Test option 1
        environment_obj_0 = self.environment_controller.create(
            input_dict_0, save_hardware_file=False)
        assert environment_obj_0
        assert isinstance(environment_obj_0, Environment)
        assert environment_obj_0.id
        assert environment_obj_0.driver_type == "docker"
        assert environment_obj_0.file_collection_id
        assert environment_obj_0.definition_filename
        assert environment_obj_0.hardware_info
        # Get file collection path
        file_collection_obj = self.environment_controller.dal.file_collection. \
            get_by_id(environment_obj_0.file_collection_id)
        file_collection_dir = self.environment_controller.file_driver. \
            get_collection_path(file_collection_obj.filehash)
        assert os.path.isfile(os.path.join(file_collection_dir, "test"))
        assert os.path.isfile(os.path.join(file_collection_dir, "Dockerfile"))
        output = open(os.path.join(file_collection_dir, "Dockerfile"),
                      "r").read()
        print(repr(output))
        assert os.path.isfile(
            os.path.join(file_collection_dir, "datmoDockerfile"))
        output = open(os.path.join(file_collection_dir, "datmoDockerfile"),
                      "r").read()
        print(repr(output))
        assert environment_obj_0.unique_hash == "c309ae4f58163693a91816988d9dc88b"
        assert environment_obj_0.name == "test"
        assert environment_obj_0.description == "test description"
        # Files ["test", "Dockerfile", "datmoDockerfile"]

        # 5) Test option 5
        input_dict_1 = {
            "name": "test",
            "description": "test description",
            "paths": [self.definition_filepath],
        }

        environment_obj = self.environment_controller.create(
            input_dict_1, save_hardware_file=False)
        assert environment_obj
        assert isinstance(environment_obj, Environment)
        assert environment_obj.id
        assert environment_obj.driver_type == "docker"
        assert environment_obj.file_collection_id
        assert environment_obj.definition_filename
        assert environment_obj.hardware_info
        # Get file collection path
        file_collection_obj = self.environment_controller.dal.file_collection. \
            get_by_id(environment_obj.file_collection_id)
        file_collection_dir = self.environment_controller.file_driver. \
            get_collection_path(file_collection_obj.filehash)
        assert os.path.isfile(os.path.join(file_collection_dir, "Dockerfile"))
        output = open(os.path.join(file_collection_dir, "Dockerfile"),
                      "r").read()
        print(repr(output))
        assert os.path.isfile(
            os.path.join(file_collection_dir, "datmoDockerfile"))
        output = open(os.path.join(file_collection_dir, "datmoDockerfile"),
                      "r").read()
        print(repr(output))
        assert environment_obj.unique_hash == "6e06d7c4d77cb6ae69e7e0efa883ef4b"
        assert environment_obj.name == "test"
        assert environment_obj.description == "test description"
        # Files ["Dockerfile", "datmoDockerfile"]

        # remove the project environment directory
        shutil.rmtree(
            self.environment_controller.file_driver.environment_directory)

        # Create environment definition in root directory
        home_definition_filepath = os.path.join(
            self.environment_controller.home, "Dockerfile")
        with open(home_definition_filepath, "wb") as f:
            f.write(to_bytes("FROM python:3.5-alpine"))

        # 2) Test option 2
        environment_obj_1 = self.environment_controller.create(
            input_dict_0, save_hardware_file=False)

        assert environment_obj_1
        assert isinstance(environment_obj_1, Environment)
        assert environment_obj_1.id
        assert environment_obj_1.driver_type == "docker"
        assert environment_obj_1.file_collection_id
        assert environment_obj_1.definition_filename
        assert environment_obj_1.hardware_info
        assert environment_obj_1.unique_hash == file_collection_obj.filehash
        # Get file collection path
        file_collection_obj = self.environment_controller.dal.file_collection. \
            get_by_id(environment_obj_1.file_collection_id)
        file_collection_dir = self.environment_controller.file_driver. \
            get_collection_path(file_collection_obj.filehash)
        assert environment_obj_1.name == "test"
        assert environment_obj_1.description == "test description"
        assert os.path.isfile(os.path.join(file_collection_dir, "Dockerfile"))
        assert os.path.isfile(
            os.path.join(file_collection_dir, "datmoDockerfile"))
        assert environment_obj_1.unique_hash == "6e06d7c4d77cb6ae69e7e0efa883ef4b"

        # 3) Test option 3
        input_dict_2 = {
            "name": "test",
            "description": "test description",
            "paths": [home_definition_filepath],
        }

        # Create environment in the project
        environment_obj_2 = self.environment_controller.create(
            input_dict_2, save_hardware_file=False)

        assert environment_obj_2
        assert isinstance(environment_obj_2, Environment)
        assert environment_obj_2.id
        assert environment_obj_2.driver_type == "docker"
        assert environment_obj_2.file_collection_id
        assert environment_obj_2.definition_filename
        assert environment_obj_2.hardware_info
        assert environment_obj_2.unique_hash == file_collection_obj.filehash
        # Get file collection path
        file_collection_obj = self.environment_controller.dal.file_collection. \
            get_by_id(environment_obj_2.file_collection_id)
        file_collection_dir = self.environment_controller.file_driver. \
            get_collection_path(file_collection_obj.filehash)
        assert environment_obj_2.name == "test"
        assert environment_obj_2.description == "test description"
        assert os.path.isfile(os.path.join(file_collection_dir, "Dockerfile"))
        assert os.path.isfile(
            os.path.join(file_collection_dir, "datmoDockerfile"))
        assert environment_obj_2.unique_hash == "6e06d7c4d77cb6ae69e7e0efa883ef4b"

        # 4) Test option 4
        input_dict_3 = {
            "paths": [home_definition_filepath + ">Dockerfile"],
        }

        # Create environment in the project
        environment_obj_3 = self.environment_controller.create(
            input_dict_3, save_hardware_file=False)

        assert environment_obj_3
        assert isinstance(environment_obj_3, Environment)
        assert environment_obj_3.id
        assert environment_obj_3.driver_type == "docker"
        assert environment_obj_3.file_collection_id
        assert environment_obj_3.definition_filename
        assert environment_obj_3.hardware_info
        assert environment_obj_3.unique_hash == file_collection_obj.filehash
        # Get file collection path
        file_collection_obj = self.environment_controller.dal.file_collection. \
            get_by_id(environment_obj_3.file_collection_id)
        file_collection_dir = self.environment_controller.file_driver. \
            get_collection_path(file_collection_obj.filehash)
        assert environment_obj_3.name == "test"
        assert environment_obj_3.description == "test description"
        assert os.path.isfile(os.path.join(file_collection_dir, "Dockerfile"))
        assert os.path.isfile(
            os.path.join(file_collection_dir, "datmoDockerfile"))
        assert environment_obj_3.unique_hash == "6e06d7c4d77cb6ae69e7e0efa883ef4b"

        # 6) Test option 6
        definition_filepath = os.path.join(self.environment_controller.home,
                                           "Dockerfile")

        input_dict = {
            "paths": [
                definition_filepath + ">Dockerfile",
                definition_filepath + ">Dockerfile"
            ],
        }

        # Create environment in the project
        failed = False
        try:
            _ = self.environment_controller.create(input_dict,
                                                   save_hardware_file=False)
        except FileAlreadyExistsError:
            failed = True

        assert failed

    @pytest_docker_environment_failed_instantiation(test_datmo_dir)
    def test_build(self):
        # 1) Test build when no environment given
        # 2) Test build when definition path exists and given
        # 3) Test build when NO file exists and definition path exists
        # 4) Test build when file exists and definition path exists
        # 5) Test build when file exists but NO definition path exists
        self.project_controller.init("test5", "test description")
        self.environment_controller = EnvironmentController()
        # 1) Test option 1
        failed = False
        try:
            _ = self.environment_controller.build("does_not_exist")
        except EnvironmentDoesNotExist:
            failed = True
        assert failed

        # Create environment definition
        definition_filepath = os.path.join(self.environment_controller.home,
                                           "Dockerfile")
        random_text = str(uuid.uuid1())
        with open(definition_filepath, "wb") as f:
            f.write(to_bytes("FROM python:3.5-alpine" + "\n"))
            f.write(to_bytes(str("RUN echo " + random_text)))
        input_dict = {
            "paths": [definition_filepath],
        }

        # 2) Test option 2
        # Create environment in the project
        environment_obj_1 = self.environment_controller.create(input_dict)
        result = self.environment_controller.build(environment_obj_1.id)
        assert result

        # 3) Test option 3
        # Create environment in the project
        environment_obj_2 = self.environment_controller.create({})
        result = self.environment_controller.build(environment_obj_2.id)
        assert result

        # Create script to test
        test_filepath = os.path.join(self.environment_controller.home,
                                     "script.py")
        with open(test_filepath, "wb") as f:
            f.write(to_bytes("import numpy\n"))
            f.write(to_bytes("import sklearn\n"))
            f.write(to_bytes("print('hello')\n"))

        # 4) Test option 4
        environment_obj_3 = self.environment_controller.create({})
        result = self.environment_controller.build(environment_obj_3.id)
        assert result

        # test 2), 3), and 4) will result in the same environment
        assert environment_obj_1.id == environment_obj_2.id
        assert environment_obj_2.id == environment_obj_3.id

        # Test for building dockerfile when there exists not
        os.remove(definition_filepath)

        # 5) Test option 5
        # Create environment definition in project environment directory
        definition_filepath = os.path.join(
            self.environment_controller.file_driver.environment_directory,
            "Dockerfile")
        random_text = str(uuid.uuid1())
        with open(definition_filepath, "wb") as f:
            f.write(to_bytes("FROM python:3.5-alpine" + "\n"))
            f.write(to_bytes(str("RUN echo " + random_text)))
        environment_obj_4 = self.environment_controller.create({})
        result = self.environment_controller.build(environment_obj_4.id)
        assert result

        # teardown
        self.environment_controller.delete(environment_obj_1.id)

    @pytest_docker_environment_failed_instantiation(test_datmo_dir)
    def test_run(self):
        # Test run simple command with simple Dockerfile
        self.project_controller.init("test5", "test description")
        self.environment_controller = EnvironmentController()

        # 0) Test option 0
        # Create environment definition in project environment directory
        definition_filepath = os.path.join(
            self.environment_controller.file_driver.environment_directory,
            "Dockerfile")
        random_text = str(uuid.uuid1())
        with open(definition_filepath, "wb") as f:
            f.write(to_bytes("FROM python:3.5-alpine" + "\n"))
            f.write(to_bytes(str("RUN echo " + random_text)))

        random_name = ''.join([
            random.choice(string.ascii_letters + string.digits)
            for _ in range(32)
        ])
        run_options = {
            "command": ["sh", "-c", "echo yo"],
            "ports": ["8888:8888"],
            "name": random_name,
            "volumes": None,
            "detach": True,
            "stdin_open": False,
            "mem_limit": "4g",
            "tty": False,
            "api": False
        }

        # Create environment in the project
        environment_obj = self.environment_controller.create({})

        log_filepath = os.path.join(self.project_controller.home, "task.log")

        # Build environment in the project
        _ = self.environment_controller.build(environment_obj.id)

        # Run environment in the project
        return_code, run_id, logs = \
            self.environment_controller.run(environment_obj.id, run_options, log_filepath)

        assert return_code == 0
        assert run_id
        assert logs

        # teardown
        self.environment_controller.delete(environment_obj.id)
        shutil.rmtree(
            self.environment_controller.file_driver.environment_directory)

        # 1) Test option 1
        # Create environment definition
        definition_filepath = os.path.join(self.environment_controller.home,
                                           "Dockerfile")
        random_text = str(uuid.uuid1())
        with open(definition_filepath, "wb") as f:
            f.write(to_bytes("FROM python:3.5-alpine" + "\n"))
            f.write(to_bytes(str("RUN echo " + random_text)))

        random_name = ''.join([
            random.choice(string.ascii_letters + string.digits)
            for _ in range(32)
        ])
        run_options = {
            "command": ["sh", "-c", "echo yo"],
            "ports": ["8888:8888"],
            "name": random_name,
            "volumes": None,
            "mem_limit": "4g",
            "detach": True,
            "stdin_open": False,
            "tty": False,
            "api": False
        }

        input_dict = {
            "paths": [definition_filepath],
        }

        # Create environment in the project
        environment_obj = self.environment_controller.create(input_dict)

        log_filepath = os.path.join(self.project_controller.home, "task.log")

        # Build environment in the project
        _ = self.environment_controller.build(environment_obj.id)

        # Run environment in the project
        return_code, run_id, logs = \
            self.environment_controller.run(environment_obj.id, run_options, log_filepath)

        assert return_code == 0
        assert run_id
        assert logs

        # teardown
        self.environment_controller.delete(environment_obj.id)

        # 2) Test option 2
        os.remove(definition_filepath)

        # Create script to test
        test_filepath = os.path.join(self.environment_controller.home,
                                     "script.py")
        with open(test_filepath, "wb") as f:
            f.write(to_bytes("import os\n"))
            f.write(to_bytes("import sys\n"))
            f.write(to_bytes("print('hello')\n"))

        # Create environment in the project
        environment_obj = self.environment_controller.create({})
        self.environment_controller.build(environment_obj.id)

        random_name = ''.join([
            random.choice(string.ascii_letters + string.digits)
            for _ in range(32)
        ])
        run_options = {
            "command": ["sh", "-c", "echo yo"],
            "ports": ["8888:8888"],
            "name": random_name,
            "volumes": {
                self.environment_controller.home: {
                    'bind': '/home/',
                    'mode': 'rw'
                }
            },
            "mem_limit": "4g",
            "detach": False,
            "stdin_open": False,
            "tty": False,
            "api": False
        }

        # Run environment in the project
        return_code, run_id, logs = \
            self.environment_controller.run(environment_obj.id, run_options, log_filepath)

        assert return_code == 0
        assert run_id
        assert logs

        # teardown
        self.environment_controller.delete(environment_obj.id)

    @pytest_docker_environment_failed_instantiation(test_datmo_dir)
    def test_interactive_run(self):
        # 1) Test run interactive terminal in environment
        # 2) Test run jupyter notebook in environment
        # Create environment definition
        self.project_controller.init("test6", "test description")
        self.environment_controller = EnvironmentController()

        definition_filepath = os.path.join(self.environment_controller.home,
                                           "Dockerfile")
        random_text = str(uuid.uuid1())
        with open(definition_filepath, "wb") as f:
            f.write(to_bytes("FROM datmo/xgboost:cpu" + "\n"))
            f.write(to_bytes(str("RUN echo " + random_text)))

        input_dict = {
            "paths": [definition_filepath],
        }

        # Create environment in the project
        environment_obj = self.environment_controller.create(input_dict)
        # 1) Test option 1
        @timeout_decorator.timeout(10, use_signals=False)
        def timed_run(container_name, timed_run):
            run_options = {
                "command": [],
                "ports": ["8888:8888"],
                "name": container_name,
                "volumes": None,
                "mem_limit": "4g",
                "detach": True,
                "stdin_open": True,
                "tty": True,
                "api": False
            }

            log_filepath = os.path.join(self.project_controller.home,
                                        "task.log")

            # Build environment in the project
            _ = self.environment_controller.build(environment_obj.id)

            # Run environment in the project
            self.environment_controller.run(environment_obj.id, run_options,
                                            log_filepath)

            return timed_run

        container_name = str(uuid.uuid1())
        timed_run_result = False
        try:
            timed_run_result = timed_run(container_name, timed_run_result)
        except timeout_decorator.timeout_decorator.TimeoutError:
            timed_run_result = True

        assert timed_run_result

        # teardown
        self.environment_controller.delete(environment_obj.id)

        # 2) Test option 2
        environment_obj = self.environment_controller.create(input_dict)

        @timeout_decorator.timeout(10, use_signals=False)
        def timed_run(container_name, timed_run):
            run_options = {
                "command": ["jupyter", "notebook"],
                "ports": ["8888:8888"],
                "name": container_name,
                "volumes": None,
                "mem_limit": "4g",
                "detach": True,
                "stdin_open": False,
                "tty": False,
                "api": False
            }

            log_filepath = os.path.join(self.project_controller.home,
                                        "task.log")

            # Build environment in the project
            _ = self.environment_controller.build(environment_obj.id)

            # Run environment in the project
            self.environment_controller.run(environment_obj.id, run_options,
                                            log_filepath)

            return timed_run

        container_name = str(uuid.uuid1())
        timed_run_result = False
        try:
            timed_run_result = timed_run(container_name, timed_run_result)
        except timeout_decorator.timeout_decorator.TimeoutError:
            timed_run_result = True

        assert timed_run_result

        # Stop the running environment
        # self.environment_controller.stop(container_name)

        # teardown
        self.environment_controller.delete(environment_obj.id)

    def test_list(self):
        self.project_controller.init("test4", "test description")
        self.environment_controller = EnvironmentController()

        # Create environment definition for object 1
        definition_path_1 = os.path.join(self.environment_controller.home,
                                         "Dockerfile")
        with open(definition_path_1, "wb") as f:
            f.write(to_bytes("FROM python:3.5-alpine"))

        input_dict_1 = {
            "paths": [definition_path_1],
        }

        # Create environment in the project
        environment_obj_1 = self.environment_controller.create(input_dict_1)

        # Create environment definition for object 2
        definition_path_2 = os.path.join(self.environment_controller.home,
                                         "Dockerfile2")
        with open(definition_path_2, "wb") as f:
            f.write(to_bytes("FROM python:3.4-alpine"))

        input_dict_2 = {
            "paths": [definition_path_2 + ">Dockerfile"],
        }

        # Create second environment in the project
        environment_obj_2 = self.environment_controller.create(input_dict_2)

        # List all environments and ensure they exist
        result = self.environment_controller.list()

        assert len(result) == 2 and \
            environment_obj_1 in result and \
            environment_obj_2 in result

    def test_update(self):
        self.project_controller.init("test5", "test description")
        self.environment_controller = EnvironmentController()

        # Create environment definition
        definition_filepath = os.path.join(self.environment_controller.home,
                                           "Dockerfile")
        with open(definition_filepath, "wb") as f:
            f.write(to_bytes("FROM python:3.5-alpine"))

        input_dict = {
            "paths": [definition_filepath],
        }

        # Create environment in the project
        environment_obj = self.environment_controller.create(input_dict)

        # Test success update
        new_name = "test name"
        new_description = "test description"
        result = self.environment_controller.update(
            environment_obj.id, name=new_name, description=new_description)
        assert result
        assert isinstance(result, Environment)
        assert result.name == new_name
        assert result.description == new_description

        # Test failed update
        failed = False
        try:
            self.environment_controller.update("random_id",
                                               name=new_name,
                                               description=new_description)
        except EnvironmentDoesNotExist:
            failed = True
        assert failed

    @pytest_docker_environment_failed_instantiation(test_datmo_dir)
    def test_delete(self):
        self.project_controller.init("test5", "test description")
        self.environment_controller = EnvironmentController()

        # Create environment definition
        definition_filepath = os.path.join(self.environment_controller.home,
                                           "Dockerfile")
        with open(definition_filepath, "wb") as f:
            f.write(to_bytes("FROM python:3.5-alpine"))

        input_dict = {
            "paths": [definition_filepath],
        }

        # Create environment in the project
        environment_obj = self.environment_controller.create(input_dict)

        # Delete environment in the project
        result = self.environment_controller.delete(environment_obj.id)

        # Check if environment retrieval throws error
        thrown = False
        try:
            self.environment_controller.dal.environment.get_by_id(
                environment_obj.id)
        except EntityNotFound:
            thrown = True

        assert result == True and \
            thrown == True

    @pytest_docker_environment_failed_instantiation(test_datmo_dir)
    def test_stop_failure(self):
        # 1) Test failure with RequiredArgumentMissing
        # 2) Test failure with TooManyArgumentsFound
        self.project_controller.init("test5", "test description")
        self.environment_controller = EnvironmentController()
        # 1) Test option 1
        failed = False
        try:
            self.environment_controller.stop()
        except RequiredArgumentMissing:
            failed = True
        assert failed

        # 2) Test option 2
        failed = False
        try:
            self.environment_controller.stop(run_id="hello",
                                             match_string="there")
        except TooManyArgumentsFound:
            failed = True
        assert failed

    @pytest_docker_environment_failed_instantiation(test_datmo_dir)
    def test_stop_success(self):
        # TODO: test more run options
        # 1) Test run_id input to stop
        # 2) Test match_string input to stop
        # 3) Test all input to stop
        self.project_controller.init("test5", "test description")
        self.environment_controller = EnvironmentController()

        # Create environment definition
        definition_filepath = os.path.join(self.environment_controller.home,
                                           "Dockerfile")
        with open(definition_filepath, "wb") as f:
            f.write(to_bytes("FROM python:3.5-alpine"))

        run_options = {
            "command": ["sh", "-c", "echo yo"],
            "ports": ["8888:8888"],
            "name": "datmo-task-" + self.environment_controller.model.id +
            "-" + "test",
            "volumes": None,
            "mem_limit": "4g",
            "detach": False,
            "stdin_open": False,
            "tty": False,
            "api": False
        }

        # Create environment definition
        env_def_path = os.path.join(self.project_controller.home, "Dockerfile")
        random_text = str(uuid.uuid1())
        with open(env_def_path, "wb") as f:
            f.write(to_bytes("FROM python:3.5-alpine" + "\n"))
            f.write(to_bytes(str("RUN echo " + random_text)))

        input_dict = {
            "paths": [definition_filepath],
        }

        # Create environment in the project
        environment_obj = self.environment_controller.create(input_dict)

        log_filepath = os.path.join(self.project_controller.home, "task.log")

        # Build environment in the project
        _ = self.environment_controller.build(environment_obj.id)

        # 1) Test option 1

        _, run_id, _ = \
            self.environment_controller.run(environment_obj.id, run_options, log_filepath)
        return_code = self.environment_controller.stop(run_id=run_id)

        assert return_code

        # 2) Test option 2
        _, _, _ = \
            self.environment_controller.run(environment_obj.id, run_options, log_filepath)
        return_code = self.environment_controller.stop(
            match_string="datmo-task-" + self.environment_controller.model.id)

        assert return_code

        # 3) Test option 3
        _, _, _ = \
            self.environment_controller.run(environment_obj.id, run_options, log_filepath)
        run_options_2 = {
            "command": ["sh", "-c", "echo yo"],
            "ports": ["8888:8888"],
            "name": "datmo-task-" + self.environment_controller.model.id +
            "-" + "test2",
            "volumes": None,
            "mem_limit": "4g",
            "detach": False,
            "stdin_open": False,
            "tty": False,
            "api": False
        }
        _, _, _ = \
            self.environment_controller.run(environment_obj.id, run_options_2, log_filepath)
        return_code = self.environment_controller.stop(all=True)

        assert return_code

        # teardown
        self.environment_controller.delete(environment_obj.id)

    def test_exists_env(self):
        # Test failure, not initialized
        failed = False
        try:
            _ = self.environment_controller.create({})
        except:
            failed = True
        assert failed

        # Setup
        self.__setup()
        environment_obj = self.environment_controller.create({})

        # Check by environment id
        result = self.environment_controller.exists(
            environment_id=environment_obj.id)
        assert result

        # Check by unique hash
        result = self.environment_controller.exists(
            environment_unique_hash=environment_obj.unique_hash)
        assert result

        # Test with wrong environment id
        result = self.environment_controller.exists(
            environment_id='test_wrong_env_id')
        assert not result

    def test_calculate_project_environment_hash(self):
        # Setup
        self.__setup()
        # Test hashing the default (with hardware info)
        result = self.environment_controller._calculate_project_environment_hash(
        )
        assert result
        # Test hashing the default Dockerfile
        result = self.environment_controller._calculate_project_environment_hash(
            save_hardware_file=False)
        assert result == "c309ae4f58163693a91816988d9dc88b"
        # Test if hash is the same as that of create
        environment_obj = self.environment_controller.create(
            {}, save_hardware_file=False)
        result = self.environment_controller._calculate_project_environment_hash(
            save_hardware_file=False)
        assert result == "c309ae4f58163693a91816988d9dc88b"
        assert result == environment_obj.unique_hash

        # Test if the hash is the same if the same file is passed in as an input
        input_dict = {
            "paths": [self.definition_filepath, self.random_filepath]
        }
        environment_obj_1 = self.environment_controller.create(
            input_dict, save_hardware_file=False)
        result = self.environment_controller._calculate_project_environment_hash(
            save_hardware_file=False)
        assert result == "c309ae4f58163693a91816988d9dc88b"
        assert result == environment_obj_1.unique_hash

    def test_has_unstaged_changes(self):
        # Setup
        self.__setup()
        _ = self.environment_controller.create({})
        # Check for no unstaged changes
        result = self.environment_controller._has_unstaged_changes()
        assert not result

        # Make a change to the file (update python version)
        with open(
                os.path.join(
                    self.environment_controller.file_driver.
                    environment_directory, "Dockerfile"), "wb") as f:
            f.write(to_bytes("FROM python:3.6-alpine"))

        # Check again, should have unstaged changes
        result = self.environment_controller._has_unstaged_changes()
        assert result

    def test_check_unstaged_changes(self):
        # Setup
        self.__setup()
        obj = self.environment_controller.create({})

        # 1) After commiting the changes
        # Check for no unstaged changes because already committed
        result = self.environment_controller.check_unstaged_changes()
        assert not result

        # Add a new file
        with open(
                os.path.join(
                    self.environment_controller.file_driver.
                    environment_directory, "test2"), "wb") as f:
            f.write(to_bytes("cool"))

        # 2) Not commiting the changes, should error and raise UnstagedChanges
        failed = False
        try:
            self.environment_controller.check_unstaged_changes()
        except UnstagedChanges:
            failed = True
        assert failed

        # Remove new file
        os.remove(
            os.path.join(
                self.environment_controller.file_driver.environment_directory,
                "test2"))

        # 3) Files are the same as before but no new commit, should have no unstaged changes
        result = self.environment_controller.check_unstaged_changes()
        assert not result

        # 4) Remove another file, now it is different and should have unstaged changes
        os.remove(
            os.path.join(
                self.environment_controller.file_driver.environment_directory,
                "test"))
        failed = False
        try:
            self.environment_controller.check_unstaged_changes()
        except UnstagedChanges:
            failed = True
        assert failed

        # 5) Remove the rest of the files, now it is empty and should return as already staged
        os.remove(
            os.path.join(
                self.environment_controller.file_driver.environment_directory,
                "Dockerfile"))
        result = self.environment_controller.check_unstaged_changes()
        assert not result

    def test_checkout(self):
        # Setup and create all environment files
        self.__setup()

        # Create environment to checkout to with defaults
        environment_obj = self.environment_controller.create({})

        # Checkout success with there are no unstaged changes
        result = self.environment_controller.checkout(environment_obj.id)
        assert result
        current_hash = self.environment_controller._calculate_project_environment_hash(
        )
        assert environment_obj.unique_hash == current_hash
        # Check the filenames as well because the hash does not take this into account
        assert os.path.isfile(
            os.path.join(
                self.environment_controller.file_driver.environment_directory,
                "test"))
        assert os.path.isfile(
            os.path.join(
                self.environment_controller.file_driver.environment_directory,
                "Dockerfile"))
        assert not os.path.isfile(
            os.path.join(
                self.environment_controller.file_driver.environment_directory,
                "datmoDockerfile"))
        assert not os.path.isfile(
            os.path.join(
                self.environment_controller.file_driver.environment_directory,
                "hardware_info"))

        # Change file contents to make it unstaged
        with open(self.definition_filepath, "wb") as f:
            f.write(to_bytes("new content"))

        # Checkout failure with unstaged changes
        failed = False
        try:
            _ = self.environment_controller.checkout(environment_obj.id)
        except UnstagedChanges:
            failed = True
        assert failed

        # Create new environment to checkout to with defaults (no hardware)
        environment_obj_1 = self.environment_controller.create(
            {}, save_hardware_file=False)

        # Checkout success with there are no unstaged changes
        result = self.environment_controller.checkout(environment_obj.id)
        assert result
        current_hash = self.environment_controller._calculate_project_environment_hash(
        )
        assert environment_obj.unique_hash == current_hash
        assert environment_obj_1.unique_hash != current_hash
        # Check the filenames as well because the hash does not take this into account
        assert os.path.isfile(
            os.path.join(
                self.environment_controller.file_driver.environment_directory,
                "test"))
        assert os.path.isfile(
            os.path.join(
                self.environment_controller.file_driver.environment_directory,
                "Dockerfile"))
        assert not os.path.isfile(
            os.path.join(
                self.environment_controller.file_driver.environment_directory,
                "datmoDockerfile"))
        assert not os.path.isfile(
            os.path.join(
                self.environment_controller.file_driver.environment_directory,
                "hardware_info"))
Ejemplo n.º 5
0
class TestTaskController():
    def setup_method(self):
        self.temp_dir = tempfile.mkdtemp(dir=test_datmo_dir)
        self.environment_ids = []

    def teardown_method(self):
        if not check_docker_inactive(test_datmo_dir,
                                     Config().datmo_directory_name):
            self.__setup()
            self.environment_controller = EnvironmentController()
            for env_id in list(set(self.environment_ids)):
                if not self.environment_controller.delete(env_id):
                    raise Exception

    def __setup(self):
        Config().set_home(self.temp_dir)
        self.project_controller = ProjectController()
        self.project_controller.init("test", "test description")
        self.environment_controller = EnvironmentController()
        self.task_controller = TaskController()

    def test_init_fail_project_not_init(self):
        Config().set_home(self.temp_dir)
        failed = False
        try:
            TaskController()
        except ProjectNotInitialized:
            failed = True
        assert failed

    def test_init_fail_invalid_path(self):
        test_home = "some_random_dir"
        Config().set_home(test_home)
        failed = False
        try:
            TaskController()
        except InvalidProjectPath:
            failed = True
        assert failed

    def test_create(self):
        self.__setup()
        # Create task in the project
        task_obj = self.task_controller.create()

        assert isinstance(task_obj, Task)
        assert task_obj.created_at
        assert task_obj.updated_at

    @pytest_docker_environment_failed_instantiation(test_datmo_dir)
    def test_run_helper(self):
        self.__setup()
        # TODO: Try out more options (see below)
        # Create environment_driver id
        env_def_path = os.path.join(self.project_controller.home, "Dockerfile")
        with open(env_def_path, "wb") as f:
            f.write(to_bytes("FROM python:3.5-alpine"))

        paths = [env_def_path]
        environment_obj = self.environment_controller.create({"paths": paths})
        self.environment_ids.append(environment_obj.id)

        # Set log filepath
        log_filepath = os.path.join(self.task_controller.home, "test.log")

        # create volume to mount
        temp_test_dirpath = os.path.join(self.temp_dir, "temp")
        os.makedirs(temp_test_dirpath)

        # Test option set 1
        random_name = str(uuid.uuid1())
        options_dict = {
            "command": ["sh", "-c", "echo accuracy:0.45"],
            "ports": ["8888:8888"],
            "name": random_name,
            "volumes": {
                temp_test_dirpath: {
                    'bind': '/task/',
                    'mode': 'rw'
                }
            },
            "mem_limit": "4g",
            "detach": False,
            "stdin_open": False,
            "tty": False,
            "api": False,
            "interactive": False
        }

        return_code, run_id, logs = \
            self.task_controller._run_helper(environment_obj.id,
                                  options_dict, log_filepath)
        assert return_code == 0
        assert run_id and \
               self.task_controller.environment_driver.get_container(run_id)
        assert logs and \
               os.path.exists(log_filepath)
        self.task_controller.environment_driver.stop_remove_containers_by_term(
            term=random_name)

        # Test option set 2
        random_name_2 = str(uuid.uuid1())
        options_dict = {
            "command": ["sh", "-c", "echo accuracy:0.45"],
            "ports": ["8888:8888"],
            "name": random_name_2,
            "volumes": {
                temp_test_dirpath: {
                    'bind': '/task/',
                    'mode': 'rw'
                }
            },
            "mem_limit": "4g",
            "detach": True,
            "stdin_open": False,
            "tty": False,
            "api": True,
            "interactive": False
        }

        return_code, run_id, logs = \
            self.task_controller._run_helper(environment_obj.id,
                                  options_dict, log_filepath)
        assert return_code == 0
        assert run_id and \
               self.task_controller.environment_driver.get_container(run_id)
        assert logs and \
               os.path.exists(log_filepath)
        self.task_controller.environment_driver.stop_remove_containers_by_term(
            term=random_name_2)

    def test_parse_logs_for_results(self):
        self.__setup()
        test_logs = """
        this is a log
        accuracy is good
        accuracy : 0.94
        this did not work
        validation : 0.32
        model_type : logistic regression
        """
        result = self.task_controller._parse_logs_for_results(test_logs)

        assert isinstance(result, dict)
        assert result['accuracy'] == "0.94"
        assert result['validation'] == "0.32"
        assert result['model_type'] == "logistic regression"

        test_logs = """test"""
        result = self.task_controller._parse_logs_for_results(test_logs)
        assert result is None

    def test_update_environment_run_options(self):
        self.__setup()
        environment_run_option = {
            "command": ["python", "script.py"],
            "volumes": {
                os.path.join(self.temp_dir, "/temp_task"): {
                    'bind': '/task/',
                    'mode': 'rw'
                },
                self.temp_dir: {
                    'bind': '/home/',
                    'mode': 'rw'
                }
            }
        }
        # create data file
        data_dirpath = os.path.join(self.temp_dir, "data")
        data_file_dirpath = os.path.join(self.temp_dir, "data_folder")
        data_filepath = os.path.join(data_file_dirpath, "data_file.txt")
        os.mkdir(data_dirpath)
        os.mkdir(data_file_dirpath)
        with open(data_filepath, "wb") as f:
            f.write(to_bytes("data file"))

        data_file_path_map = [(data_filepath, "data_file.txt")]
        data_directory_path_map = [(data_dirpath, "data_directory")]

        environment_run_option = self.task_controller._update_environment_run_options(
            environment_run_option, data_file_path_map,
            data_directory_path_map)

        assert environment_run_option["volumes"][data_file_dirpath] == {
            'bind': '/data/',
            'mode': 'rw'
        }
        assert environment_run_option["volumes"][data_dirpath] == {
            'bind': '/data/data_directory',
            'mode': 'rw'
        }

        # Error by passing directory which does not exist
        data_dirpath = os.path.join(self.temp_dir, "data_dne")
        data_filepath = os.path.join(self.temp_dir, "data_dne",
                                     "data_file.txt")
        data_file_path_map = [(data_filepath, "data_file.txt")]
        data_directory_path_map = [(data_dirpath, "data_directory")]
        failed = False
        try:
            self.task_controller._update_environment_run_options(
                environment_run_option, data_file_path_map,
                data_directory_path_map)
        except TaskRunError:
            failed = True

        assert failed

    @pytest_docker_environment_failed_instantiation(test_datmo_dir)
    def test_run(self):
        self.__setup()
        # 0) Test failure case without command and without interactive
        # 1) Test success case with default values and env def file
        # 2) Test failure case if running same task (conflicting containers)
        # 3) Test failure case if running same task with snapshot_dict (conflicting containers)
        # 4) Test success case with snapshot_dict
        # 5) Test success case with saved file during task run
        # 6) Test success case with data file path being passed
        # 7) Test success case with data directory path being passed

        # TODO: look into log filepath randomness, sometimes logs are not written

        # Create task in the project
        task_obj = self.task_controller.create()

        # Create environment definition
        env_def_path = os.path.join(self.project_controller.home, "Dockerfile")
        with open(env_def_path, "wb") as f:
            f.write(to_bytes("FROM python:3.5-alpine"))

        # 0) Test option 0
        failed = False
        try:
            self.task_controller.run(task_obj.id)
        except RequiredArgumentMissing:
            failed = True
        assert failed

        failed = False
        try:
            self.task_controller.run(
                task_obj.id,
                task_dict={
                    "command": None,
                    "interactive": False,
                    "ports": None
                })
        except RequiredArgumentMissing:
            failed = True
        assert failed

        # Create task_dict
        task_command = ["sh", "-c", "echo accuracy:0.45"]
        task_dict = {"command_list": task_command}

        # 1) Test option 1
        updated_task_obj = self.task_controller.run(
            task_obj.id, task_dict=task_dict)
        after_snapshot_obj = self.task_controller.dal.snapshot.get_by_id(
            updated_task_obj.after_snapshot_id)
        environment_obj = self.task_controller.dal.environment.get_by_id(
            after_snapshot_obj.environment_id)
        self.environment_ids.append(environment_obj.id)

        assert isinstance(updated_task_obj, Task)
        assert task_obj.id == updated_task_obj.id

        assert updated_task_obj.before_snapshot_id
        assert updated_task_obj.ports == None
        assert updated_task_obj.interactive == False
        assert updated_task_obj.task_dirpath
        assert updated_task_obj.log_filepath
        assert updated_task_obj.start_time

        assert updated_task_obj.after_snapshot_id
        assert updated_task_obj.run_id
        assert updated_task_obj.logs
        assert "accuracy" in updated_task_obj.logs
        assert updated_task_obj.results
        assert updated_task_obj.results == {"accuracy": "0.45"}
        assert after_snapshot_obj.stats == {"accuracy": "0.45"}
        assert updated_task_obj.status == "SUCCESS"
        assert updated_task_obj.end_time
        assert updated_task_obj.duration

        self.task_controller.stop(task_obj.id)

        # 2) Test option 2
        failed = False
        try:
            self.task_controller.run(task_obj.id)
        except TaskRunError:
            failed = True
        assert failed

        # 3) Test option 3

        # Create files to add
        self.project_controller.file_driver.create("dirpath1", directory=True)
        self.project_controller.file_driver.create("dirpath2", directory=True)
        self.project_controller.file_driver.create("filepath1")

        # Snapshot dictionary
        snapshot_dict = {
            "paths": [
                os.path.join(self.project_controller.home, "dirpath1"),
                os.path.join(self.project_controller.home, "dirpath2"),
                os.path.join(self.project_controller.home, "filepath1")
            ],
        }

        # Run a basic task in the project
        failed = False
        try:
            self.task_controller.run(task_obj.id, snapshot_dict=snapshot_dict)
        except TaskRunError:
            failed = True
        assert failed

        # Test when the specific task id is already RUNNING
        # Create task in the project
        task_obj_1 = self.task_controller.create()
        self.task_controller.dal.task.update({
            "id": task_obj_1.id,
            "status": "RUNNING"
        })
        # Create environment_driver definition
        env_def_path = os.path.join(self.project_controller.home, "Dockerfile")
        with open(env_def_path, "wb") as f:
            f.write(to_bytes("FROM python:3.5-alpine"))

        failed = False
        try:
            self.task_controller.run(task_obj_1.id, task_dict=task_dict)
        except TaskRunError:
            failed = True
        assert failed

        # 4) Test option 4

        # Create a new task in the project
        task_obj_2 = self.task_controller.create()

        # Run another task in the project
        updated_task_obj_2 = self.task_controller.run(
            task_obj_2.id, task_dict=task_dict, snapshot_dict=snapshot_dict)
        after_snapshot_obj = self.task_controller.dal.snapshot.get_by_id(
            updated_task_obj_2.after_snapshot_id)
        environment_obj = self.task_controller.dal.environment.get_by_id(
            after_snapshot_obj.environment_id)
        self.environment_ids.append(environment_obj.id)

        assert isinstance(updated_task_obj_2, Task)
        assert task_obj_2.id == updated_task_obj_2.id

        assert updated_task_obj_2.before_snapshot_id
        assert updated_task_obj_2.ports == None
        assert updated_task_obj_2.interactive == False
        assert updated_task_obj_2.task_dirpath
        assert updated_task_obj_2.log_filepath
        assert updated_task_obj_2.start_time

        assert updated_task_obj_2.after_snapshot_id
        assert updated_task_obj_2.run_id
        assert updated_task_obj_2.logs
        assert "accuracy" in updated_task_obj_2.logs
        assert updated_task_obj_2.results
        assert updated_task_obj_2.results == {"accuracy": "0.45"}
        assert updated_task_obj_2.status == "SUCCESS"
        assert updated_task_obj_2.end_time
        assert updated_task_obj_2.duration

        self.task_controller.stop(task_obj_2.id)

        # 5) Test option 5

        # Create a basic script
        test_filepath = os.path.join(self.temp_dir, "script.py")
        with open(test_filepath, "wb") as f:
            f.write(to_bytes("import os\n"))
            f.write(to_bytes("import shutil\n"))
            f.write(to_bytes("print('hello')\n"))
            f.write(to_bytes("print(' accuracy: 0.56 ')\n"))
            f.write(
                to_bytes(
                    "with open(os.path.join('/task', 'new_file.txt'), 'a') as f:\n"
                ))
            f.write(to_bytes("    f.write('my test file')\n"))

        # Create task in the project
        task_obj_2 = self.task_controller.create()

        # Create task_dict
        task_command = ["python", "script.py"]
        task_dict = {"command_list": task_command}

        # Create environment definition
        env_def_path = os.path.join(self.project_controller.home, "Dockerfile")
        with open(env_def_path, "wb") as f:
            f.write(to_bytes("FROM python:3.5-alpine"))

        updated_task_obj_2 = self.task_controller.run(
            task_obj_2.id, task_dict=task_dict)
        after_snapshot_obj = self.task_controller.dal.snapshot.get_by_id(
            updated_task_obj_2.after_snapshot_id)
        environment_obj = self.task_controller.dal.environment.get_by_id(
            after_snapshot_obj.environment_id)
        self.environment_ids.append(environment_obj.id)

        assert isinstance(updated_task_obj_2, Task)
        assert updated_task_obj_2.before_snapshot_id
        assert updated_task_obj_2.ports == None
        assert updated_task_obj_2.interactive == False
        assert updated_task_obj_2.task_dirpath
        assert updated_task_obj_2.log_filepath
        assert updated_task_obj_2.start_time

        assert updated_task_obj_2.after_snapshot_id
        assert updated_task_obj_2.run_id
        assert updated_task_obj_2.logs
        assert "accuracy" in updated_task_obj_2.logs
        assert updated_task_obj_2.results
        assert updated_task_obj_2.results == {"accuracy": "0.56"}
        assert updated_task_obj_2.status == "SUCCESS"
        assert updated_task_obj_2.end_time
        assert updated_task_obj_2.duration

        self.task_controller.stop(task_obj_2.id)

        # test if after snapshot has the file written
        after_snapshot_obj = self.task_controller.dal.snapshot.get_by_id(
            updated_task_obj_2.after_snapshot_id)
        file_collection_obj = self.task_controller.dal.file_collection.get_by_id(
            after_snapshot_obj.file_collection_id)
        files_absolute_path = os.path.join(self.task_controller.home,
                                           file_collection_obj.path)

        assert os.path.isfile(os.path.join(files_absolute_path, "task.log"))
        assert os.path.isfile(
            os.path.join(files_absolute_path, "new_file.txt"))

        # 6) Test Option 6
        self.project_controller.file_driver.create("dirpath1", directory=True)
        self.project_controller.file_driver.create(
            os.path.join("dirpath1", "file.txt"))
        with open(
                os.path.join(self.project_controller.home, "dirpath1",
                             "file.txt"), "wb") as f:
            f.write(to_bytes('my initial line\n'))
        test_filename = "script.py"
        test_filepath = os.path.join(self.temp_dir, test_filename)
        with open(test_filepath, "wb") as f:
            f.write(to_bytes("import os\n"))
            f.write(to_bytes("print('hello')\n"))
            f.write(to_bytes("import shutil\n"))

            f.write(
                to_bytes(
                    "with open(os.path.join('/data', 'file.txt'), 'a') as f:\n"
                ))
            f.write(to_bytes("    f.write('my test file')\n"))

        # Create task in the project
        task_obj_3 = self.task_controller.create()

        # Create task_dict
        task_command = ["python", test_filename]
        task_dict = {
            "command_list":
                task_command,
            "data_file_path_map": [(os.path.join(self.project_controller.home,
                                                 "dirpath1", "file.txt"),
                                    'file.txt')]
        }

        # Create environment definition
        env_def_path = os.path.join(self.project_controller.home, "Dockerfile")
        with open(env_def_path, "wb") as f:
            f.write(to_bytes("FROM python:3.5-alpine"))

        updated_task_obj_3 = self.task_controller.run(
            task_obj_3.id, task_dict=task_dict)
        after_snapshot_obj = self.task_controller.dal.snapshot.get_by_id(
            updated_task_obj_3.after_snapshot_id)
        environment_obj = self.task_controller.dal.environment.get_by_id(
            after_snapshot_obj.environment_id)
        self.environment_ids.append(environment_obj.id)

        assert isinstance(updated_task_obj_3, Task)
        assert updated_task_obj_3.before_snapshot_id
        assert updated_task_obj_3.ports == None
        assert updated_task_obj_3.interactive == False
        assert updated_task_obj_3.task_dirpath
        assert updated_task_obj_3.log_filepath
        assert updated_task_obj_3.start_time

        assert updated_task_obj_3.after_snapshot_id
        assert updated_task_obj_3.run_id
        assert updated_task_obj_3.logs
        assert updated_task_obj_3.status == "SUCCESS"
        assert updated_task_obj_3.end_time
        assert updated_task_obj_3.duration

        self.task_controller.stop(task_obj_3.id)

        # test if after snapshot has the file written
        after_snapshot_obj = self.task_controller.dal.snapshot.get_by_id(
            updated_task_obj_3.after_snapshot_id)
        file_collection_obj = self.task_controller.dal.file_collection.get_by_id(
            after_snapshot_obj.file_collection_id)
        files_absolute_path = os.path.join(self.task_controller.home,
                                           file_collection_obj.path)

        assert os.path.isfile(os.path.join(files_absolute_path, "task.log"))
        assert os.path.isfile(
            os.path.join(self.project_controller.home, "dirpath1", "file.txt"))
        assert "my initial line" in open(
            os.path.join(self.project_controller.home, "dirpath1", "file.txt"),
            "r").read()
        assert "my test file" in open(
            os.path.join(self.project_controller.home, "dirpath1", "file.txt"),
            "r").read()

        # 7) Test Option 7
        self.project_controller.file_driver.create("dirpath1", directory=True)
        self.project_controller.file_driver.create(
            os.path.join("dirpath1", "file.txt"))
        with open(
                os.path.join(self.project_controller.home, "dirpath1",
                             "file.txt"), "wb") as f:
            f.write(to_bytes('my initial line\n'))
        test_filename = "script.py"
        test_filepath = os.path.join(self.temp_dir, test_filename)
        with open(test_filepath, "wb") as f:
            f.write(to_bytes("import os\n"))
            f.write(to_bytes("print('hello')\n"))
            f.write(to_bytes("import shutil\n"))

            f.write(
                to_bytes(
                    "with open(os.path.join('/data', 'dirpath1', 'file.txt'), 'a') as f:\n"
                ))
            f.write(to_bytes("    f.write('my test file')\n"))

        # Create task in the project
        task_obj_4 = self.task_controller.create()

        # Create task_dict
        task_command = ["python", test_filename]
        task_dict = {
            "command_list":
                task_command,
            "data_directory_path_map": [(os.path.join(
                self.project_controller.home, "dirpath1"), 'dirpath1')]
        }

        # Create environment definition
        env_def_path = os.path.join(self.project_controller.home, "Dockerfile")
        with open(env_def_path, "wb") as f:
            f.write(to_bytes("FROM python:3.5-alpine"))

        updated_task_obj_4 = self.task_controller.run(
            task_obj_4.id, task_dict=task_dict)
        after_snapshot_obj = self.task_controller.dal.snapshot.get_by_id(
            updated_task_obj_4.after_snapshot_id)
        environment_obj = self.task_controller.dal.environment.get_by_id(
            after_snapshot_obj.environment_id)
        self.environment_ids.append(environment_obj.id)

        assert isinstance(updated_task_obj_4, Task)
        assert updated_task_obj_4.before_snapshot_id
        assert updated_task_obj_4.ports == None
        assert updated_task_obj_4.interactive == False
        assert updated_task_obj_4.task_dirpath
        assert updated_task_obj_4.log_filepath
        assert updated_task_obj_4.start_time

        assert updated_task_obj_4.after_snapshot_id
        assert updated_task_obj_4.run_id
        assert updated_task_obj_4.logs
        assert updated_task_obj_4.status == "SUCCESS"
        assert updated_task_obj_4.end_time
        assert updated_task_obj_4.duration

        self.task_controller.stop(task_obj_4.id)

        # test if after snapshot has the file written
        after_snapshot_obj = self.task_controller.dal.snapshot.get_by_id(
            updated_task_obj_4.after_snapshot_id)
        file_collection_obj = self.task_controller.dal.file_collection.get_by_id(
            after_snapshot_obj.file_collection_id)
        files_absolute_path = os.path.join(self.task_controller.home,
                                           file_collection_obj.path)

        assert os.path.isfile(os.path.join(files_absolute_path, "task.log"))
        assert os.path.isfile(
            os.path.join(self.project_controller.home, "dirpath1", "file.txt"))
        assert "my initial line" in open(
            os.path.join(self.project_controller.home, "dirpath1", "file.txt"),
            "r").read()
        assert "my test file" in open(
            os.path.join(self.project_controller.home, "dirpath1", "file.txt"),
            "r").read()

    def test_list(self):
        self.__setup()
        # Create tasks in the project
        task_obj_1 = self.task_controller.create()
        task_obj_2 = self.task_controller.create()

        # List all tasks regardless of filters
        result = self.task_controller.list()

        assert len(result) == 2 and \
               task_obj_1 in result and \
               task_obj_2 in result

        # List all tasks regardless of filters in ascending
        result = self.task_controller.list(
            sort_key='created_at', sort_order='ascending')

        assert len(result) == 2 and \
               task_obj_1 in result and \
               task_obj_2 in result
        assert result[0].created_at <= result[-1].created_at

        # List all tasks regardless of filters in descending
        result = self.task_controller.list(
            sort_key='created_at', sort_order='descending')
        assert len(result) == 2 and \
               task_obj_1 in result and \
               task_obj_2 in result
        assert result[0].created_at >= result[-1].created_at

        # Wrong order being passed in
        failed = False
        try:
            _ = self.task_controller.list(
                sort_key='created_at', sort_order='wrong_order')
        except InvalidArgumentType:
            failed = True
        assert failed

        # Wrong key and order being passed in
        failed = False
        try:
            _ = self.task_controller.list(
                sort_key='wrong_key', sort_order='wrong_order')
        except InvalidArgumentType:
            failed = True
        assert failed

        # wrong key and right order being passed in
        expected_result = self.task_controller.list(
            sort_key='created_at', sort_order='ascending')
        result = self.task_controller.list(
            sort_key='wrong_key', sort_order='ascending')
        expected_ids = [item.id for item in expected_result]
        ids = [item.id for item in result]
        assert set(expected_ids) == set(ids)

        # List all tasks
        result = self.task_controller.list()

        assert len(result) == 2 and \
               task_obj_1 in result and \
               task_obj_2 in result

    def test_get(self):
        self.__setup()
        # Test failure for no task
        failed = False
        try:
            self.task_controller.get("random")
        except DoesNotExist:
            failed = True
        assert failed

        # Test success for task
        task_obj = self.task_controller.create()
        task_obj_returned = self.task_controller.get(task_obj.id)
        assert task_obj == task_obj_returned

    @pytest_docker_environment_failed_instantiation(test_datmo_dir)
    def test_get_files(self):
        self.__setup()
        # Test failure case
        failed = False
        try:
            self.task_controller.get_files("random")
        except DoesNotExist:
            failed = True
        assert failed

        # Create task in the project
        task_obj = self.task_controller.create()

        # Create environment definition
        env_def_path = os.path.join(self.project_controller.home, "Dockerfile")
        with open(env_def_path, "wb") as f:
            f.write(to_bytes("FROM python:3.5-alpine"))

        # Create file to add
        self.project_controller.file_driver.create("dirpath1", directory=True)
        self.project_controller.file_driver.create(
            os.path.join("dirpath1", "filepath1"))

        # Snapshot dictionary
        snapshot_dict = {
            "paths": [
                os.path.join(self.project_controller.home, "dirpath1",
                             "filepath1")
            ],
        }

        # Create task_dict
        task_command = ["sh", "-c", "echo accuracy:0.45"]
        task_dict = {"command_list": task_command}

        # Test the default values
        updated_task_obj = self.task_controller.run(
            task_obj.id, task_dict=task_dict, snapshot_dict=snapshot_dict)
        after_snapshot_obj = self.task_controller.dal.snapshot.get_by_id(
            updated_task_obj.after_snapshot_id)
        environment_obj = self.task_controller.dal.environment.get_by_id(
            after_snapshot_obj.environment_id)
        self.environment_ids.append(environment_obj.id)

        # TODO: Test case for during run and before_snapshot run
        # Get files for the task after run is complete (default)
        result = self.task_controller.get_files(updated_task_obj.id)

        after_snapshot_obj = self.task_controller.dal.snapshot.get_by_id(
            updated_task_obj.after_snapshot_id)
        file_collection_obj = self.task_controller.dal.file_collection.get_by_id(
            after_snapshot_obj.file_collection_id)

        file_names = [item.name for item in result]

        assert len(result) == 2
        for item in result:
            assert isinstance(item, TextIOWrapper)
            assert item.mode == "r"
        assert os.path.join(self.task_controller.home, ".datmo", "collections",
                            file_collection_obj.filehash,
                            "task.log") in file_names
        assert os.path.join(self.task_controller.home, ".datmo", "collections",
                            file_collection_obj.filehash,
                            "filepath1") in file_names

        # Get files for the task after run is complete for different mode
        result = self.task_controller.get_files(updated_task_obj.id, mode="a")

        assert len(result) == 2
        for item in result:
            assert isinstance(item, TextIOWrapper)
            assert item.mode == "a"
        assert os.path.join(self.task_controller.home, ".datmo", "collections",
                            file_collection_obj.filehash,
                            "task.log") in file_names
        assert os.path.join(self.task_controller.home, ".datmo", "collections",
                            file_collection_obj.filehash,
                            "filepath1") in file_names

        self.task_controller.stop(task_obj.id)

    def test_update(self):
        self.__setup()
        # Create task in the project
        task_obj = self.task_controller.create()
        assert isinstance(task_obj, Task)

        # Test 1: When no meta data is passed
        updated_task_obj = self.task_controller.update(task_obj.id)
        assert updated_task_obj.workspace is None

        # Test 2: When meta data for workspace is passed
        test_workspace = "notebook"
        test_command = "python script.py"
        updated_task_obj = self.task_controller.update(
            task_obj.id, workspace=test_workspace, command=test_command)
        assert updated_task_obj.workspace == test_workspace
        assert updated_task_obj.command == test_command
        assert updated_task_obj.command_list == ["python", "script.py"]

        # Test 3: When meta data for workspace is passed
        test_interactive = True
        updated_task_obj = self.task_controller.update(
            task_obj.id, interactive=test_interactive)
        assert updated_task_obj.interactive == test_interactive

        # Test 4: When meta data for workspace is passed
        test_command_list = ["python", "script.py"]
        updated_task_obj = self.task_controller.update(
            task_obj.id, command_list=test_command_list)
        assert updated_task_obj.command_list == ["python", "script.py"]

    @pytest_docker_environment_failed_instantiation(test_datmo_dir)
    def test_delete(self):
        self.__setup()
        # Create tasks in the project
        task_obj = self.task_controller.create()

        # Delete task from the project
        result = self.task_controller.delete(task_obj.id)

        # Check if task retrieval throws error
        thrown = False
        try:
            self.task_controller.dal.snapshot.get_by_id(task_obj.id)
        except EntityNotFound:
            thrown = True

        assert result == True and \
               thrown == True

    @pytest_docker_environment_failed_instantiation(test_datmo_dir)
    def test_stop_failure(self):
        self.__setup()
        # 1) Test required arguments not provided
        # 2) Test too many arguments found
        # 3) Test incorrect task id given

        # 1) Test option 1
        failed = False
        try:
            self.task_controller.stop()
        except RequiredArgumentMissing:
            failed = True
        assert failed

        # 2) Test option 2
        failed = False
        try:
            self.task_controller.stop(task_id="test_task_id", all=True)
        except TooManyArgumentsFound:
            failed = True
        assert failed

        # 3) Test option 3
        thrown = False
        try:
            self.task_controller.stop(task_id="incorrect_task_id")
        except DoesNotExist:
            thrown = True
        assert thrown

    @pytest_docker_environment_failed_instantiation(test_datmo_dir)
    def test_stop_success(self):
        self.__setup()
        # 1) Test stop with task_id
        # 2) Test stop with all given

        # Create task in the project
        task_obj = self.task_controller.create()

        # Create environment driver definition
        env_def_path = os.path.join(self.project_controller.home, "Dockerfile")
        with open(env_def_path, "wb") as f:
            f.write(to_bytes("FROM python:3.5-alpine"))

        # Create task_dict
        task_command = ["sh", "-c", "echo accuracy:0.45"]
        task_dict = {"command_list": task_command}

        # 1) Test option 1
        updated_task_obj = self.task_controller.run(
            task_obj.id, task_dict=task_dict)
        after_snapshot_obj = self.task_controller.dal.snapshot.get_by_id(
            updated_task_obj.after_snapshot_id)
        environment_obj = self.task_controller.dal.environment.get_by_id(
            after_snapshot_obj.environment_id)
        self.environment_ids.append(environment_obj.id)
        task_id = updated_task_obj.id
        result = self.task_controller.stop(task_id=task_id)
        after_task_obj = self.task_controller.dal.task.get_by_id(task_id)

        assert result
        assert after_task_obj.status == "STOPPED"

        # 2) Test option 2
        task_obj_2 = self.task_controller.create()
        updated_task_obj = self.task_controller.run(
            task_obj_2.id, task_dict=task_dict)
        after_snapshot_obj = self.task_controller.dal.snapshot.get_by_id(
            updated_task_obj.after_snapshot_id)
        environment_obj = self.task_controller.dal.environment.get_by_id(
            after_snapshot_obj.environment_id)
        self.environment_ids.append(environment_obj.id)
        result = self.task_controller.stop(all=True)
        all_task_objs = self.task_controller.dal.task.query({})

        assert result
        for task_obj in all_task_objs:
            assert task_obj.status == "STOPPED"
Ejemplo n.º 6
0
class TestEnvironmentController():
    def setup_method(self):
        # provide mountable tmp directory for docker
        tempfile.tempdir = "/tmp" if not platform.system(
        ) == "Windows" else None
        test_datmo_dir = os.environ.get('TEST_DATMO_DIR',
                                        tempfile.gettempdir())
        self.temp_dir = tempfile.mkdtemp(dir=test_datmo_dir)
        self.project = ProjectController(self.temp_dir)
        self.environment = EnvironmentController(self.temp_dir)

    def teardown_method(self):
        pass

    def test_create(self):
        # 0) Test create when unsupported language given
        # 1) Test create when NO file exists and NO definition path exists
        # 2) Test create when NO file exists and definition path exists
        # 3) Test create when definition path exists and given
        # 4) Test create when file exists and definition path exists
        # 5) Test create when file exists but NO definition path exists
        # 6) Test create when definition path exists and given for NEW definition filepath

        self.project.init("test3", "test description")

        # 0) Test option 0
        try:
            self.environment.create({"language": "java"})
        except EnvironmentDoesNotExist:
            failed = True
        assert failed

        # 1) Test option 1
        # Creates environment with python3 based docker image
        environment_obj_0 = self.environment.create({})
        assert environment_obj_0
        assert environment_obj_0.id
        assert environment_obj_0.driver_type == "docker"
        assert environment_obj_0.file_collection_id
        assert environment_obj_0.definition_filename
        assert environment_obj_0.hardware_info

        # Create environment definition
        definition_filepath = os.path.join(self.environment.home, "Dockerfile")
        with open(definition_filepath, "w") as f:
            f.write(to_unicode(str("FROM datmo/xgboost:cpu")))

        # 2) Test option 2
        environment_obj_1 = self.environment.create({})

        # Get file collection path
        file_collection_obj = self.environment.dal.file_collection. \
            get_by_id(environment_obj_1.file_collection_id)
        file_collection_dir = self.environment.file_driver. \
            get_collection_path(file_collection_obj.filehash)

        assert environment_obj_1
        assert environment_obj_1.id
        assert environment_obj_1.driver_type == "docker"
        assert environment_obj_1.file_collection_id
        assert environment_obj_1.definition_filename
        assert environment_obj_1.hardware_info
        assert environment_obj_1.unique_hash == file_collection_obj.filehash
        assert os.path.isfile(os.path.join(file_collection_dir, "Dockerfile"))
        assert os.path.isfile(
            os.path.join(file_collection_dir, "datmoDockerfile"))
        assert os.path.isfile(
            os.path.join(file_collection_dir, "hardware_info"))

        # 3) Test option 3
        input_dict = {
            "definition_filepath": definition_filepath,
        }

        # Create environment in the project
        environment_obj_2 = self.environment.create(input_dict)

        # Get file collection path
        file_collection_obj = self.environment.dal.file_collection. \
            get_by_id(environment_obj_2.file_collection_id)
        file_collection_dir = self.environment.file_driver. \
            get_collection_path(file_collection_obj.filehash)

        assert environment_obj_2
        assert environment_obj_2.id
        assert environment_obj_2.driver_type == "docker"
        assert environment_obj_2.file_collection_id
        assert environment_obj_2.definition_filename
        assert environment_obj_2.hardware_info
        assert environment_obj_2.unique_hash == file_collection_obj.filehash
        assert os.path.isfile(os.path.join(file_collection_dir, "Dockerfile"))
        assert os.path.isfile(
            os.path.join(file_collection_dir, "datmoDockerfile"))
        assert os.path.isfile(
            os.path.join(file_collection_dir, "hardware_info"))

        # Create script to test
        test_filepath = os.path.join(self.environment.home, "script.py")
        with open(test_filepath, "w") as f:
            f.write(to_unicode("import numpy\n"))
            f.write(to_unicode("import sklearn\n"))
            f.write(to_unicode("print('hello')\n"))

        # 4) Test option 4
        environment_obj_3 = self.environment.create({})

        # Get file collection path
        file_collection_obj = self.environment.dal.file_collection. \
            get_by_id(environment_obj_3.file_collection_id)
        file_collection_dir = self.environment.file_driver. \
            get_collection_path(file_collection_obj.filehash)

        assert environment_obj_3
        assert environment_obj_3.id
        assert environment_obj_3.driver_type == "docker"
        assert environment_obj_3.file_collection_id
        assert environment_obj_3.definition_filename
        assert environment_obj_3.hardware_info
        assert environment_obj_3.unique_hash == file_collection_obj.filehash
        assert os.path.isfile(os.path.join(file_collection_dir, "Dockerfile"))
        assert os.path.isfile(
            os.path.join(file_collection_dir, "datmoDockerfile"))
        assert os.path.isfile(
            os.path.join(file_collection_dir, "hardware_info"))

        # Remove definition filepath
        os.remove(definition_filepath)

        assert environment_obj_1.id == environment_obj_2.id
        assert environment_obj_2.id == environment_obj_3.id

        # 5) Test option 5
        environment_obj_4 = self.environment.create({})

        # Get file collection path
        file_collection_obj = self.environment.dal.file_collection. \
            get_by_id(environment_obj_4.file_collection_id)
        file_collection_dir = self.environment.file_driver. \
            get_collection_path(file_collection_obj.filehash)

        assert environment_obj_4
        assert environment_obj_4.id
        assert environment_obj_4.driver_type == "docker"
        assert environment_obj_4.file_collection_id
        assert environment_obj_4.definition_filename
        assert environment_obj_4.hardware_info
        assert environment_obj_4.unique_hash == file_collection_obj.filehash
        assert os.path.isfile(
            os.path.join(file_collection_dir, "datmorequirements.txt"))
        assert os.path.isfile(os.path.join(file_collection_dir, "Dockerfile"))
        assert os.path.isfile(
            os.path.join(file_collection_dir, "datmoDockerfile"))
        assert os.path.isfile(
            os.path.join(file_collection_dir, "hardware_info"))

        assert environment_obj_1.id != environment_obj_4.id

        # 6) Test option 6

        # Create environment definition
        definition_filepath = os.path.join(self.environment.home, "Dockerfile")
        with open(definition_filepath, "w") as f:
            f.write(to_unicode(str("FROM cloudgear/ubuntu:14.04")))

        input_dict = {
            "definition_filepath": definition_filepath,
        }

        # Create a new environment obj
        environment_obj_5 = self.environment.create(input_dict)

        # Get file collection path
        file_collection_obj = self.environment.dal.file_collection. \
            get_by_id(environment_obj_5.file_collection_id)
        file_collection_dir = self.environment.file_driver. \
            get_collection_path(file_collection_obj.filehash)

        assert environment_obj_5
        assert environment_obj_5.id
        assert environment_obj_5.driver_type == "docker"
        assert environment_obj_5.file_collection_id
        assert environment_obj_5.definition_filename
        assert environment_obj_5.hardware_info
        assert environment_obj_5.unique_hash == file_collection_obj.filehash
        assert os.path.isfile(os.path.join(file_collection_dir, "Dockerfile"))
        assert os.path.isfile(
            os.path.join(file_collection_dir, "datmoDockerfile"))
        assert os.path.isfile(
            os.path.join(file_collection_dir, "hardware_info"))

        assert environment_obj_5.id != environment_obj_1.id
        assert environment_obj_5.id != environment_obj_4.id

    def test_build(self):
        # 1) Test build when no environment given
        # 2) Test build when definition path exists and given
        # 3) Test build when NO file exists and definition path exists
        # 4) Test build when file exists and definition path exists
        # 5) Test build when file exists but NO definition path exists
        self.project.init("test5", "test description")

        # 1) Test option 1
        failed = False
        try:
            _ = self.environment.build("does_not_exist")
        except EntityNotFound:
            failed = True
        assert failed

        # Create environment definition
        definition_filepath = os.path.join(self.environment.home, "Dockerfile")
        with open(definition_filepath, "w") as f:
            f.write(to_unicode(str("FROM datmo/xgboost:cpu")))
        input_dict = {
            "definition_filepath": definition_filepath,
        }

        # 2) Test option 2
        # Create environment in the project
        environment_obj_1 = self.environment.create(input_dict)
        result = self.environment.build(environment_obj_1.id)
        assert result

        # 3) Test option 3
        # Create environment in the project
        environment_obj_2 = self.environment.create({})
        result = self.environment.build(environment_obj_2.id)
        assert result

        # Create script to test
        test_filepath = os.path.join(self.environment.home, "script.py")
        with open(test_filepath, "w") as f:
            f.write(to_unicode("import numpy\n"))
            f.write(to_unicode("import sklearn\n"))
            f.write(to_unicode("print('hello')\n"))

        # 4) Test option 4
        environment_obj_3 = self.environment.create({})
        result = self.environment.build(environment_obj_3.id)
        assert result

        # test 2), 3), and 4) will result in the same environment
        assert environment_obj_1.id == environment_obj_2.id
        assert environment_obj_2.id == environment_obj_3.id

        # Test for building dockerfile when there exists not
        os.remove(definition_filepath)

        # 5) Test option 5
        environment_obj_4 = self.environment.create({})
        result = self.environment.build(environment_obj_4.id)
        # 2) Test run script, with autogenerated definition
        assert result
        assert environment_obj_4.id != environment_obj_1.id

        # teardown
        self.environment.delete(environment_obj_1.id)
        self.environment.delete(environment_obj_4.id)

    def test_run(self):
        # 1) Test run simple command with simple Dockerfile
        self.project.init("test5", "test description")

        # 1) Test option 1

        # Create environment definition
        definition_filepath = os.path.join(self.environment.home, "Dockerfile")
        with open(definition_filepath, "w") as f:
            f.write(to_unicode(str("FROM datmo/xgboost:cpu")))

        random_name = ''.join([
            random.choice(string.ascii_letters + string.digits)
            for _ in range(32)
        ])
        run_options = {
            "command": ["sh", "-c", "echo yo"],
            "ports": ["8888:8888"],
            "name": random_name,
            "volumes": None,
            "detach": True,
            "stdin_open": False,
            "tty": False,
            "api": False
        }

        input_dict = {
            "definition_filepath": definition_filepath,
        }

        # Create environment in the project
        environment_obj = self.environment.create(input_dict)

        log_filepath = os.path.join(self.project.home, "task.log")

        # Build environment in the project
        _ = self.environment.build(environment_obj.id)

        # Run environment in the project
        return_code, run_id, logs = \
            self.environment.run(environment_obj.id, run_options, log_filepath)

        assert return_code == 0
        assert run_id
        assert logs

        # teardown
        self.environment.delete(environment_obj.id)

        # 2) Test option 2
        os.remove(definition_filepath)

        # Create script to test
        test_filepath = os.path.join(self.environment.home, "script.py")
        with open(test_filepath, "w") as f:
            f.write(to_unicode("import numpy\n"))
            f.write(to_unicode("import sklearn\n"))
            f.write(to_unicode("print('hello')\n"))

        # Create environment in the project
        environment_obj = self.environment.create({})
        self.environment.build(environment_obj.id)

        random_name = ''.join([
            random.choice(string.ascii_letters + string.digits)
            for _ in range(32)
        ])
        run_options = {
            "command": ["python", "script.py"],
            "ports": ["8888:8888"],
            "name": random_name,
            "volumes": {
                self.environment.home: {
                    'bind': '/home/',
                    'mode': 'rw'
                }
            },
            "detach": False,
            "stdin_open": False,
            "tty": False,
            "api": False
        }

        # Run environment in the project
        return_code, run_id, logs = \
            self.environment.run(environment_obj.id, run_options, log_filepath)

        assert return_code == 0
        assert run_id
        assert logs

        # teardown
        self.environment.delete(environment_obj.id)

    def test_interactive_run(self):
        # 1) Test run interactive terminal in environment
        # 2) Test run jupyter notebook in environment
        # Create environment definition
        self.project.init("test6", "test description")

        definition_filepath = os.path.join(self.environment.home, "Dockerfile")
        with open(definition_filepath, "w") as f:
            f.write(to_unicode(str("FROM datmo/xgboost:cpu")))

        input_dict = {
            "definition_filepath": definition_filepath,
        }

        # Create environment in the project
        environment_obj = self.environment.create(input_dict)
        # 1) Test option 1
        @timeout_decorator.timeout(10, use_signals=False)
        def timed_run(container_name, timed_run):
            run_options = {
                "command": [],
                "ports": ["8888:8888"],
                "name": container_name,
                "volumes": None,
                "detach": True,
                "stdin_open": True,
                "tty": True,
                "api": False
            }

            log_filepath = os.path.join(self.project.home, "task.log")

            # Build environment in the project
            _ = self.environment.build(environment_obj.id)

            # Run environment in the project
            self.environment.run(environment_obj.id, run_options, log_filepath)

            return timed_run

        container_name = str(uuid.uuid1())
        timed_run_result = False
        try:
            timed_run_result = timed_run(container_name, timed_run_result)
        except timeout_decorator.timeout_decorator.TimeoutError:
            timed_run_result = True

        assert timed_run_result

        # teardown
        self.environment.delete(environment_obj.id)

        # 2) Test option 2
        environment_obj = self.environment.create(input_dict)

        @timeout_decorator.timeout(10, use_signals=False)
        def timed_run(container_name, timed_run):
            run_options = {
                "command": ["jupyter", "notebook"],
                "ports": ["8888:8888"],
                "name": container_name,
                "volumes": None,
                "detach": True,
                "stdin_open": False,
                "tty": False,
                "api": False
            }

            log_filepath = os.path.join(self.project.home, "task.log")

            # Build environment in the project
            _ = self.environment.build(environment_obj.id)

            # Run environment in the project
            self.environment.run(environment_obj.id, run_options, log_filepath)

            return timed_run

        container_name = str(uuid.uuid1())
        timed_run_result = False
        try:
            timed_run_result = timed_run(container_name, timed_run_result)
        except timeout_decorator.timeout_decorator.TimeoutError:
            timed_run_result = True

        assert timed_run_result

        # Stop the running environment
        # self.environment.stop(container_name)

        # teardown
        self.environment.delete(environment_obj.id)

    def test_list(self):
        self.project.init("test4", "test description")

        # Create environment definition for object 1
        definition_path_1 = os.path.join(self.environment.home, "Dockerfile")
        with open(definition_path_1, "w") as f:
            f.write(to_unicode(str("FROM datmo/xgboost:cpu")))

        input_dict_1 = {
            "definition_filepath": definition_path_1,
        }

        # Create environment in the project
        environment_obj_1 = self.environment.create(input_dict_1)

        # Create environment definition for object 2
        definition_path_2 = os.path.join(self.environment.home, "Dockerfile2")
        with open(definition_path_2, "w") as f:
            f.write(to_unicode(str("FROM datmo/scikit-opencv")))

        input_dict_2 = {
            "definition_filepath": definition_path_2,
        }

        # Create second environment in the project
        environment_obj_2 = self.environment.create(input_dict_2)

        # List all environments and ensure they exist
        result = self.environment.list()

        assert len(result) == 2 and \
            environment_obj_1 in result and \
            environment_obj_2 in result

    def test_delete(self):
        self.project.init("test5", "test description")

        # Create environment definition
        definition_filepath = os.path.join(self.environment.home, "Dockerfile")
        with open(definition_filepath, "w") as f:
            f.write(to_unicode(str("FROM datmo/xgboost:cpu")))

        input_dict = {
            "definition_filepath": definition_filepath,
        }

        # Create environment in the project
        environment_obj = self.environment.create(input_dict)

        # Delete environment in the project
        result = self.environment.delete(environment_obj.id)

        # Check if environment retrieval throws error
        thrown = False
        try:
            self.environment.dal.environment.get_by_id(environment_obj.id)
        except EntityNotFound:
            thrown = True

        assert result == True and \
            thrown == True

    def test_stop_failure(self):
        # 1) Test failure with RequiredArgumentMissing
        # 2) Test failure with TooManyArgumentsFound

        # 1) Test option 1
        failed = False
        try:
            self.environment.stop()
        except RequiredArgumentMissing:
            failed = True
        assert failed

        # 2) Test option 2
        failed = False
        try:
            self.environment.stop(run_id="hello", match_string="there")
        except TooManyArgumentsFound:
            failed = True
        assert failed

    def test_stop_success(self):
        # TODO: test more run options
        # 1) Test run_id input to stop
        # 2) Test match_string input to stop
        # 3) Test all input to stop
        self.project.init("test5", "test description")

        # Create environment definition
        definition_filepath = os.path.join(self.environment.home, "Dockerfile")
        with open(definition_filepath, "w") as f:
            f.write(to_unicode(str("FROM datmo/xgboost:cpu")))

        run_options = {
            "command": ["sh", "-c", "echo yo"],
            "ports": ["8888:8888"],
            "name": "datmo-task-" + self.environment.model.id + "-" + "test",
            "volumes": None,
            "detach": False,
            "stdin_open": False,
            "tty": False,
            "api": False
        }

        # Create environment definition
        env_def_path = os.path.join(self.project.home, "Dockerfile")
        with open(env_def_path, "w") as f:
            f.write(to_unicode(str("FROM datmo/xgboost:cpu")))

        input_dict = {
            "definition_filepath": definition_filepath,
        }

        # Create environment in the project
        environment_obj = self.environment.create(input_dict)

        log_filepath = os.path.join(self.project.home, "task.log")

        # Build environment in the project
        _ = self.environment.build(environment_obj.id)

        # 1) Test option 1

        _, run_id, _ = \
            self.environment.run(environment_obj.id, run_options, log_filepath)
        return_code = self.environment.stop(run_id=run_id)

        assert return_code

        # 2) Test option 2
        _, _, _ = \
            self.environment.run(environment_obj.id, run_options, log_filepath)
        return_code = self.environment.stop(match_string="datmo-task-" +
                                            self.environment.model.id)

        assert return_code

        # 3) Test option 3
        _, _, _ = \
            self.environment.run(environment_obj.id, run_options, log_filepath)
        run_options_2 = {
            "command": ["sh", "-c", "echo yo"],
            "ports": ["8888:8888"],
            "name": "datmo-task-" + self.environment.model.id + "-" + "test2",
            "volumes": None,
            "detach": False,
            "stdin_open": False,
            "tty": False,
            "api": False
        }
        _, _, _ = \
            self.environment.run(environment_obj.id, run_options_2, log_filepath)
        return_code = self.environment.stop(all=True)

        assert return_code

        # teardown
        self.environment.delete(environment_obj.id)
Ejemplo n.º 7
0
class EnvironmentCommand(ProjectCommand):
    def __init__(self, cli_helper):
        super(EnvironmentCommand, self).__init__(cli_helper)

    def environment(self):
        self.parse(["environment", "--help"])
        return True

    @Helper.notify_no_project_found
    def setup(self, **kwargs):
        self.environment_controller = EnvironmentController()
        name = kwargs.get("name", None)
        available_environments = self.environment_controller.get_supported_environments(
        )
        if not name:
            name = self.cli_helper.prompt_available_environments(
                available_environments)
        try:
            options = {"name": name}
            environment_obj = self.environment_controller.setup(
                options=options)
            self.cli_helper.echo(
                __("info", "cli.environment.setup.success",
                   (environment_obj.name, environment_obj.id)))
            return environment_obj
        except EnvironmentDoesNotExist:
            self.cli_helper.echo(
                __("error", "cli.environment.setup.argument", name))

    @Helper.notify_no_project_found
    def create(self, **kwargs):
        self.environment_controller = EnvironmentController()
        self.cli_helper.echo(__("info", "cli.environment.create"))
        created_environment_obj = self.environment_controller.create(kwargs)
        environments = self.environment_controller.list()
        for environment_obj in environments:
            if created_environment_obj == environment_obj:
                self.cli_helper.echo(
                    __("info", "cli.environment.create.alreadyexist",
                       created_environment_obj.id))
                return created_environment_obj
        self.cli_helper.echo(
            __("info", "cli.environment.create.success",
               created_environment_obj.id))
        return created_environment_obj

    @Helper.notify_no_project_found
    def update(self, **kwargs):
        self.environment_controller = EnvironmentController()
        environment_id = kwargs.get('id')
        name = kwargs.get('name', None)
        description = kwargs.get("description", None)
        result = self.environment_controller.update(environment_id,
                                                    name=name,
                                                    description=description)
        return result

    @Helper.notify_environment_active(EnvironmentController)
    @Helper.notify_no_project_found
    def delete(self, **kwargs):
        self.environment_controller = EnvironmentController()
        environment_id = kwargs.get('id')
        if self.environment_controller.delete(environment_id):
            self.cli_helper.echo(
                __("info", "cli.environment.delete.success", environment_id))
            return True

    @Helper.notify_no_project_found
    def ls(self, **kwargs):
        self.environment_controller = EnvironmentController()
        print_format = kwargs.get('format', "table")
        download = kwargs.get('download', None)
        download_path = kwargs.get('download_path', None)
        environment_objs = self.environment_controller.list()
        header_list = ["id", "created at", "name", "description"]
        item_dict_list = []
        for environment_obj in environment_objs:
            environment_obj_name = printable_object(environment_obj.name)
            environment_obj_description = printable_object(
                environment_obj.description)
            item_dict_list.append({
                "id":
                environment_obj.id,
                "created at":
                prettify_datetime(environment_obj.created_at),
                "name":
                environment_obj_name,
                "description":
                environment_obj_description
            })
        if download:
            if not download_path:
                # download to current working directory with timestamp
                current_time = datetime.utcnow()
                epoch_time = datetime.utcfromtimestamp(0)
                current_time_unix_time_ms = (
                    current_time - epoch_time).total_seconds() * 1000.0
                download_path = os.path.join(
                    self.environment_controller.home,
                    "environment_ls_" + str(current_time_unix_time_ms))
            self.cli_helper.print_items(header_list,
                                        item_dict_list,
                                        print_format=print_format,
                                        output_path=download_path)
            return environment_objs
        self.cli_helper.print_items(header_list,
                                    item_dict_list,
                                    print_format=print_format)
        return environment_objs