airbyte-cdk 0.72.1__py3-none-any.whl → 6.17.1.dev0__py3-none-any.whl
Sign up to get free protection for your applications and to get access to all the features.
- airbyte_cdk/__init__.py +355 -6
- airbyte_cdk/cli/__init__.py +1 -0
- airbyte_cdk/cli/source_declarative_manifest/__init__.py +5 -0
- airbyte_cdk/cli/source_declarative_manifest/_run.py +230 -0
- airbyte_cdk/cli/source_declarative_manifest/spec.json +17 -0
- airbyte_cdk/config_observation.py +29 -10
- airbyte_cdk/connector.py +24 -24
- airbyte_cdk/connector_builder/README.md +53 -0
- airbyte_cdk/connector_builder/connector_builder_handler.py +37 -11
- airbyte_cdk/connector_builder/main.py +45 -13
- airbyte_cdk/connector_builder/message_grouper.py +189 -50
- airbyte_cdk/connector_builder/models.py +3 -2
- airbyte_cdk/destinations/__init__.py +4 -3
- airbyte_cdk/destinations/destination.py +54 -20
- airbyte_cdk/destinations/vector_db_based/README.md +37 -0
- airbyte_cdk/destinations/vector_db_based/config.py +40 -17
- airbyte_cdk/destinations/vector_db_based/document_processor.py +56 -17
- airbyte_cdk/destinations/vector_db_based/embedder.py +57 -15
- airbyte_cdk/destinations/vector_db_based/test_utils.py +14 -4
- airbyte_cdk/destinations/vector_db_based/utils.py +8 -2
- airbyte_cdk/destinations/vector_db_based/writer.py +24 -5
- airbyte_cdk/entrypoint.py +153 -44
- airbyte_cdk/exception_handler.py +21 -3
- airbyte_cdk/logger.py +30 -44
- airbyte_cdk/models/__init__.py +13 -2
- airbyte_cdk/models/airbyte_protocol.py +86 -1
- airbyte_cdk/models/airbyte_protocol_serializers.py +44 -0
- airbyte_cdk/models/file_transfer_record_message.py +13 -0
- airbyte_cdk/models/well_known_types.py +1 -1
- airbyte_cdk/sources/__init__.py +5 -1
- airbyte_cdk/sources/abstract_source.py +125 -79
- airbyte_cdk/sources/concurrent_source/__init__.py +7 -2
- airbyte_cdk/sources/concurrent_source/concurrent_read_processor.py +102 -36
- airbyte_cdk/sources/concurrent_source/concurrent_source.py +29 -36
- airbyte_cdk/sources/concurrent_source/concurrent_source_adapter.py +94 -10
- airbyte_cdk/sources/concurrent_source/stream_thread_exception.py +25 -0
- airbyte_cdk/sources/concurrent_source/thread_pool_manager.py +20 -14
- airbyte_cdk/sources/config.py +3 -2
- airbyte_cdk/sources/connector_state_manager.py +49 -83
- airbyte_cdk/sources/declarative/async_job/job.py +52 -0
- airbyte_cdk/sources/declarative/async_job/job_orchestrator.py +497 -0
- airbyte_cdk/sources/declarative/async_job/job_tracker.py +75 -0
- airbyte_cdk/sources/declarative/async_job/repository.py +35 -0
- airbyte_cdk/sources/declarative/async_job/status.py +24 -0
- airbyte_cdk/sources/declarative/async_job/timer.py +39 -0
- airbyte_cdk/sources/declarative/auth/__init__.py +2 -3
- airbyte_cdk/sources/declarative/auth/declarative_authenticator.py +3 -1
- airbyte_cdk/sources/declarative/auth/jwt.py +191 -0
- airbyte_cdk/sources/declarative/auth/oauth.py +60 -20
- airbyte_cdk/sources/declarative/auth/selective_authenticator.py +10 -2
- airbyte_cdk/sources/declarative/auth/token.py +28 -10
- airbyte_cdk/sources/declarative/auth/token_provider.py +9 -8
- airbyte_cdk/sources/declarative/checks/check_stream.py +16 -8
- airbyte_cdk/sources/declarative/checks/connection_checker.py +4 -2
- airbyte_cdk/sources/declarative/concurrency_level/__init__.py +7 -0
- airbyte_cdk/sources/declarative/concurrency_level/concurrency_level.py +50 -0
- airbyte_cdk/sources/declarative/concurrent_declarative_source.py +490 -0
- airbyte_cdk/sources/declarative/datetime/datetime_parser.py +4 -0
- airbyte_cdk/sources/declarative/datetime/min_max_datetime.py +26 -6
- airbyte_cdk/sources/declarative/declarative_component_schema.yaml +1185 -85
- airbyte_cdk/sources/declarative/declarative_source.py +5 -2
- airbyte_cdk/sources/declarative/declarative_stream.py +95 -9
- airbyte_cdk/sources/declarative/decoders/__init__.py +23 -2
- airbyte_cdk/sources/declarative/decoders/composite_raw_decoder.py +97 -0
- airbyte_cdk/sources/declarative/decoders/decoder.py +11 -4
- airbyte_cdk/sources/declarative/decoders/json_decoder.py +92 -5
- airbyte_cdk/sources/declarative/decoders/noop_decoder.py +21 -0
- airbyte_cdk/sources/declarative/decoders/pagination_decoder_decorator.py +39 -0
- airbyte_cdk/sources/declarative/decoders/xml_decoder.py +98 -0
- airbyte_cdk/sources/declarative/extractors/__init__.py +12 -1
- airbyte_cdk/sources/declarative/extractors/dpath_extractor.py +29 -24
- airbyte_cdk/sources/declarative/extractors/http_selector.py +4 -5
- airbyte_cdk/sources/declarative/extractors/record_extractor.py +2 -3
- airbyte_cdk/sources/declarative/extractors/record_filter.py +63 -8
- airbyte_cdk/sources/declarative/extractors/record_selector.py +85 -26
- airbyte_cdk/sources/declarative/extractors/response_to_file_extractor.py +177 -0
- airbyte_cdk/sources/declarative/extractors/type_transformer.py +55 -0
- airbyte_cdk/sources/declarative/incremental/__init__.py +31 -3
- airbyte_cdk/sources/declarative/incremental/concurrent_partition_cursor.py +346 -0
- airbyte_cdk/sources/declarative/incremental/datetime_based_cursor.py +156 -48
- airbyte_cdk/sources/declarative/incremental/declarative_cursor.py +13 -0
- airbyte_cdk/sources/declarative/incremental/global_substream_cursor.py +350 -0
- airbyte_cdk/sources/declarative/incremental/per_partition_cursor.py +173 -74
- airbyte_cdk/sources/declarative/incremental/per_partition_with_global.py +200 -0
- airbyte_cdk/sources/declarative/incremental/resumable_full_refresh_cursor.py +122 -0
- airbyte_cdk/sources/declarative/interpolation/filters.py +27 -1
- airbyte_cdk/sources/declarative/interpolation/interpolated_boolean.py +23 -5
- airbyte_cdk/sources/declarative/interpolation/interpolated_mapping.py +12 -8
- airbyte_cdk/sources/declarative/interpolation/interpolated_nested_mapping.py +13 -6
- airbyte_cdk/sources/declarative/interpolation/interpolated_string.py +21 -6
- airbyte_cdk/sources/declarative/interpolation/interpolation.py +9 -3
- airbyte_cdk/sources/declarative/interpolation/jinja.py +72 -37
- airbyte_cdk/sources/declarative/interpolation/macros.py +72 -17
- airbyte_cdk/sources/declarative/manifest_declarative_source.py +193 -52
- airbyte_cdk/sources/declarative/migrations/legacy_to_per_partition_state_migration.py +98 -0
- airbyte_cdk/sources/declarative/migrations/state_migration.py +24 -0
- airbyte_cdk/sources/declarative/models/__init__.py +1 -1
- airbyte_cdk/sources/declarative/models/declarative_component_schema.py +1319 -603
- airbyte_cdk/sources/declarative/parsers/custom_exceptions.py +2 -2
- airbyte_cdk/sources/declarative/parsers/manifest_component_transformer.py +26 -4
- airbyte_cdk/sources/declarative/parsers/manifest_reference_resolver.py +26 -15
- airbyte_cdk/sources/declarative/parsers/model_to_component_factory.py +1759 -225
- airbyte_cdk/sources/declarative/partition_routers/__init__.py +24 -4
- airbyte_cdk/sources/declarative/partition_routers/async_job_partition_router.py +65 -0
- airbyte_cdk/sources/declarative/partition_routers/cartesian_product_stream_slicer.py +176 -0
- airbyte_cdk/sources/declarative/partition_routers/list_partition_router.py +39 -9
- airbyte_cdk/sources/declarative/partition_routers/partition_router.py +62 -0
- airbyte_cdk/sources/declarative/partition_routers/single_partition_router.py +15 -3
- airbyte_cdk/sources/declarative/partition_routers/substream_partition_router.py +222 -39
- airbyte_cdk/sources/declarative/requesters/error_handlers/__init__.py +19 -5
- airbyte_cdk/sources/declarative/requesters/error_handlers/backoff_strategies/__init__.py +3 -1
- airbyte_cdk/sources/declarative/requesters/error_handlers/backoff_strategies/constant_backoff_strategy.py +19 -7
- airbyte_cdk/sources/declarative/requesters/error_handlers/backoff_strategies/exponential_backoff_strategy.py +19 -7
- airbyte_cdk/sources/declarative/requesters/error_handlers/backoff_strategies/header_helper.py +4 -2
- airbyte_cdk/sources/declarative/requesters/error_handlers/backoff_strategies/wait_time_from_header_backoff_strategy.py +41 -9
- airbyte_cdk/sources/declarative/requesters/error_handlers/backoff_strategies/wait_until_time_from_header_backoff_strategy.py +29 -14
- airbyte_cdk/sources/declarative/requesters/error_handlers/backoff_strategy.py +5 -13
- airbyte_cdk/sources/declarative/requesters/error_handlers/composite_error_handler.py +32 -16
- airbyte_cdk/sources/declarative/requesters/error_handlers/default_error_handler.py +46 -56
- airbyte_cdk/sources/declarative/requesters/error_handlers/default_http_response_filter.py +40 -0
- airbyte_cdk/sources/declarative/requesters/error_handlers/error_handler.py +6 -32
- airbyte_cdk/sources/declarative/requesters/error_handlers/http_response_filter.py +119 -41
- airbyte_cdk/sources/declarative/requesters/http_job_repository.py +228 -0
- airbyte_cdk/sources/declarative/requesters/http_requester.py +98 -344
- airbyte_cdk/sources/declarative/requesters/paginators/__init__.py +14 -3
- airbyte_cdk/sources/declarative/requesters/paginators/default_paginator.py +105 -46
- airbyte_cdk/sources/declarative/requesters/paginators/no_pagination.py +14 -8
- airbyte_cdk/sources/declarative/requesters/paginators/paginator.py +19 -8
- airbyte_cdk/sources/declarative/requesters/paginators/strategies/__init__.py +9 -3
- airbyte_cdk/sources/declarative/requesters/paginators/strategies/cursor_pagination_strategy.py +53 -21
- airbyte_cdk/sources/declarative/requesters/paginators/strategies/offset_increment.py +42 -19
- airbyte_cdk/sources/declarative/requesters/paginators/strategies/page_increment.py +25 -12
- airbyte_cdk/sources/declarative/requesters/paginators/strategies/pagination_strategy.py +13 -10
- airbyte_cdk/sources/declarative/requesters/paginators/strategies/stop_condition.py +26 -13
- airbyte_cdk/sources/declarative/requesters/request_options/__init__.py +15 -2
- airbyte_cdk/sources/declarative/requesters/request_options/datetime_based_request_options_provider.py +91 -0
- airbyte_cdk/sources/declarative/requesters/request_options/default_request_options_provider.py +60 -0
- airbyte_cdk/sources/declarative/requesters/request_options/interpolated_nested_request_input_provider.py +31 -14
- airbyte_cdk/sources/declarative/requesters/request_options/interpolated_request_input_provider.py +27 -15
- airbyte_cdk/sources/declarative/requesters/request_options/interpolated_request_options_provider.py +63 -10
- airbyte_cdk/sources/declarative/requesters/request_options/request_options_provider.py +1 -1
- airbyte_cdk/sources/declarative/requesters/requester.py +9 -17
- airbyte_cdk/sources/declarative/resolvers/__init__.py +41 -0
- airbyte_cdk/sources/declarative/resolvers/components_resolver.py +55 -0
- airbyte_cdk/sources/declarative/resolvers/config_components_resolver.py +136 -0
- airbyte_cdk/sources/declarative/resolvers/http_components_resolver.py +112 -0
- airbyte_cdk/sources/declarative/retrievers/__init__.py +6 -2
- airbyte_cdk/sources/declarative/retrievers/async_retriever.py +100 -0
- airbyte_cdk/sources/declarative/retrievers/retriever.py +1 -3
- airbyte_cdk/sources/declarative/retrievers/simple_retriever.py +229 -73
- airbyte_cdk/sources/declarative/schema/__init__.py +14 -1
- airbyte_cdk/sources/declarative/schema/default_schema_loader.py +5 -3
- airbyte_cdk/sources/declarative/schema/dynamic_schema_loader.py +236 -0
- airbyte_cdk/sources/declarative/schema/json_file_schema_loader.py +8 -8
- airbyte_cdk/sources/declarative/spec/spec.py +12 -5
- airbyte_cdk/sources/declarative/stream_slicers/__init__.py +1 -2
- airbyte_cdk/sources/declarative/stream_slicers/declarative_partition_generator.py +88 -0
- airbyte_cdk/sources/declarative/stream_slicers/stream_slicer.py +9 -14
- airbyte_cdk/sources/declarative/transformations/add_fields.py +19 -11
- airbyte_cdk/sources/declarative/transformations/flatten_fields.py +52 -0
- airbyte_cdk/sources/declarative/transformations/keys_replace_transformation.py +61 -0
- airbyte_cdk/sources/declarative/transformations/keys_to_lower_transformation.py +22 -0
- airbyte_cdk/sources/declarative/transformations/keys_to_snake_transformation.py +68 -0
- airbyte_cdk/sources/declarative/transformations/remove_fields.py +13 -10
- airbyte_cdk/sources/declarative/transformations/transformation.py +5 -5
- airbyte_cdk/sources/declarative/types.py +19 -110
- airbyte_cdk/sources/declarative/yaml_declarative_source.py +31 -10
- airbyte_cdk/sources/embedded/base_integration.py +16 -5
- airbyte_cdk/sources/embedded/catalog.py +16 -4
- airbyte_cdk/sources/embedded/runner.py +19 -3
- airbyte_cdk/sources/embedded/tools.py +5 -2
- airbyte_cdk/sources/file_based/README.md +152 -0
- airbyte_cdk/sources/file_based/__init__.py +24 -0
- airbyte_cdk/sources/file_based/availability_strategy/__init__.py +9 -2
- airbyte_cdk/sources/file_based/availability_strategy/abstract_file_based_availability_strategy.py +22 -6
- airbyte_cdk/sources/file_based/availability_strategy/default_file_based_availability_strategy.py +46 -10
- airbyte_cdk/sources/file_based/config/abstract_file_based_spec.py +47 -10
- airbyte_cdk/sources/file_based/config/avro_format.py +2 -1
- airbyte_cdk/sources/file_based/config/csv_format.py +29 -10
- airbyte_cdk/sources/file_based/config/excel_format.py +18 -0
- airbyte_cdk/sources/file_based/config/file_based_stream_config.py +16 -4
- airbyte_cdk/sources/file_based/config/jsonl_format.py +2 -1
- airbyte_cdk/sources/file_based/config/parquet_format.py +2 -1
- airbyte_cdk/sources/file_based/config/unstructured_format.py +13 -5
- airbyte_cdk/sources/file_based/discovery_policy/__init__.py +6 -2
- airbyte_cdk/sources/file_based/discovery_policy/abstract_discovery_policy.py +2 -4
- airbyte_cdk/sources/file_based/discovery_policy/default_discovery_policy.py +7 -2
- airbyte_cdk/sources/file_based/exceptions.py +18 -15
- airbyte_cdk/sources/file_based/file_based_source.py +140 -33
- airbyte_cdk/sources/file_based/file_based_stream_reader.py +69 -5
- airbyte_cdk/sources/file_based/file_types/__init__.py +14 -1
- airbyte_cdk/sources/file_based/file_types/avro_parser.py +75 -24
- airbyte_cdk/sources/file_based/file_types/csv_parser.py +116 -34
- airbyte_cdk/sources/file_based/file_types/excel_parser.py +196 -0
- airbyte_cdk/sources/file_based/file_types/file_transfer.py +37 -0
- airbyte_cdk/sources/file_based/file_types/file_type_parser.py +4 -1
- airbyte_cdk/sources/file_based/file_types/jsonl_parser.py +24 -8
- airbyte_cdk/sources/file_based/file_types/parquet_parser.py +60 -18
- airbyte_cdk/sources/file_based/file_types/unstructured_parser.py +141 -41
- airbyte_cdk/sources/file_based/remote_file.py +1 -1
- airbyte_cdk/sources/file_based/schema_helpers.py +38 -10
- airbyte_cdk/sources/file_based/schema_validation_policies/__init__.py +3 -1
- airbyte_cdk/sources/file_based/schema_validation_policies/abstract_schema_validation_policy.py +3 -1
- airbyte_cdk/sources/file_based/schema_validation_policies/default_schema_validation_policies.py +16 -5
- airbyte_cdk/sources/file_based/stream/abstract_file_based_stream.py +50 -13
- airbyte_cdk/sources/file_based/stream/concurrent/adapters.py +67 -27
- airbyte_cdk/sources/file_based/stream/concurrent/cursor/__init__.py +5 -1
- airbyte_cdk/sources/file_based/stream/concurrent/cursor/abstract_concurrent_file_based_cursor.py +14 -23
- airbyte_cdk/sources/file_based/stream/concurrent/cursor/file_based_concurrent_cursor.py +54 -18
- airbyte_cdk/sources/file_based/stream/concurrent/cursor/file_based_final_state_cursor.py +21 -9
- airbyte_cdk/sources/file_based/stream/cursor/abstract_file_based_cursor.py +3 -1
- airbyte_cdk/sources/file_based/stream/cursor/default_file_based_cursor.py +27 -10
- airbyte_cdk/sources/file_based/stream/default_file_based_stream.py +147 -45
- airbyte_cdk/sources/http_logger.py +8 -3
- airbyte_cdk/sources/message/__init__.py +7 -1
- airbyte_cdk/sources/message/repository.py +18 -4
- airbyte_cdk/sources/source.py +42 -38
- airbyte_cdk/sources/streams/__init__.py +2 -2
- airbyte_cdk/sources/streams/availability_strategy.py +54 -3
- airbyte_cdk/sources/streams/call_rate.py +64 -21
- airbyte_cdk/sources/streams/checkpoint/__init__.py +26 -0
- airbyte_cdk/sources/streams/checkpoint/checkpoint_reader.py +335 -0
- airbyte_cdk/sources/{declarative/incremental → streams/checkpoint}/cursor.py +17 -14
- airbyte_cdk/sources/streams/checkpoint/per_partition_key_serializer.py +22 -0
- airbyte_cdk/sources/streams/checkpoint/resumable_full_refresh_cursor.py +51 -0
- airbyte_cdk/sources/streams/checkpoint/substream_resumable_full_refresh_cursor.py +110 -0
- airbyte_cdk/sources/streams/concurrent/README.md +7 -0
- airbyte_cdk/sources/streams/concurrent/abstract_stream.py +7 -2
- airbyte_cdk/sources/streams/concurrent/adapters.py +84 -75
- airbyte_cdk/sources/streams/concurrent/availability_strategy.py +30 -2
- airbyte_cdk/sources/streams/concurrent/cursor.py +298 -42
- airbyte_cdk/sources/streams/concurrent/default_stream.py +12 -3
- airbyte_cdk/sources/streams/concurrent/exceptions.py +3 -0
- airbyte_cdk/sources/streams/concurrent/helpers.py +14 -3
- airbyte_cdk/sources/streams/concurrent/partition_enqueuer.py +12 -3
- airbyte_cdk/sources/streams/concurrent/partition_reader.py +10 -3
- airbyte_cdk/sources/streams/concurrent/partitions/partition.py +1 -16
- airbyte_cdk/sources/streams/concurrent/partitions/stream_slicer.py +21 -0
- airbyte_cdk/sources/streams/concurrent/partitions/types.py +15 -5
- airbyte_cdk/sources/streams/concurrent/state_converters/abstract_stream_state_converter.py +109 -17
- airbyte_cdk/sources/streams/concurrent/state_converters/datetime_stream_state_converter.py +90 -72
- airbyte_cdk/sources/streams/core.py +412 -87
- airbyte_cdk/sources/streams/http/__init__.py +2 -1
- airbyte_cdk/sources/streams/http/availability_strategy.py +12 -101
- airbyte_cdk/sources/streams/http/error_handlers/__init__.py +22 -0
- airbyte_cdk/sources/streams/http/error_handlers/backoff_strategy.py +28 -0
- airbyte_cdk/sources/streams/http/error_handlers/default_backoff_strategy.py +17 -0
- airbyte_cdk/sources/streams/http/error_handlers/default_error_mapping.py +86 -0
- airbyte_cdk/sources/streams/http/error_handlers/error_handler.py +42 -0
- airbyte_cdk/sources/streams/http/error_handlers/error_message_parser.py +19 -0
- airbyte_cdk/sources/streams/http/error_handlers/http_status_error_handler.py +110 -0
- airbyte_cdk/sources/streams/http/error_handlers/json_error_message_parser.py +52 -0
- airbyte_cdk/sources/streams/http/error_handlers/response_models.py +65 -0
- airbyte_cdk/sources/streams/http/exceptions.py +27 -7
- airbyte_cdk/sources/streams/http/http.py +369 -246
- airbyte_cdk/sources/streams/http/http_client.py +531 -0
- airbyte_cdk/sources/streams/http/rate_limiting.py +76 -12
- airbyte_cdk/sources/streams/http/requests_native_auth/abstract_oauth.py +28 -9
- airbyte_cdk/sources/streams/http/requests_native_auth/abstract_token.py +2 -1
- airbyte_cdk/sources/streams/http/requests_native_auth/oauth.py +90 -35
- airbyte_cdk/sources/streams/http/requests_native_auth/token.py +13 -3
- airbyte_cdk/sources/types.py +154 -0
- airbyte_cdk/sources/utils/record_helper.py +36 -21
- airbyte_cdk/sources/utils/schema_helpers.py +13 -6
- airbyte_cdk/sources/utils/slice_logger.py +4 -1
- airbyte_cdk/sources/utils/transform.py +54 -20
- airbyte_cdk/sql/_util/hashing.py +34 -0
- airbyte_cdk/sql/_util/name_normalizers.py +92 -0
- airbyte_cdk/sql/constants.py +32 -0
- airbyte_cdk/sql/exceptions.py +235 -0
- airbyte_cdk/sql/secrets.py +123 -0
- airbyte_cdk/sql/shared/__init__.py +15 -0
- airbyte_cdk/sql/shared/catalog_providers.py +145 -0
- airbyte_cdk/sql/shared/sql_processor.py +786 -0
- airbyte_cdk/sql/types.py +160 -0
- airbyte_cdk/test/catalog_builder.py +70 -18
- airbyte_cdk/test/entrypoint_wrapper.py +117 -42
- airbyte_cdk/test/mock_http/__init__.py +1 -1
- airbyte_cdk/test/mock_http/matcher.py +6 -0
- airbyte_cdk/test/mock_http/mocker.py +57 -10
- airbyte_cdk/test/mock_http/request.py +19 -3
- airbyte_cdk/test/mock_http/response.py +3 -1
- airbyte_cdk/test/mock_http/response_builder.py +32 -16
- airbyte_cdk/test/state_builder.py +18 -10
- airbyte_cdk/test/utils/__init__.py +1 -0
- airbyte_cdk/test/utils/data.py +24 -0
- airbyte_cdk/test/utils/http_mocking.py +16 -0
- airbyte_cdk/test/utils/manifest_only_fixtures.py +60 -0
- airbyte_cdk/test/utils/reading.py +26 -0
- airbyte_cdk/utils/__init__.py +2 -1
- airbyte_cdk/utils/airbyte_secrets_utils.py +5 -3
- airbyte_cdk/utils/analytics_message.py +10 -2
- airbyte_cdk/utils/datetime_format_inferrer.py +4 -1
- airbyte_cdk/utils/event_timing.py +10 -10
- airbyte_cdk/utils/mapping_helpers.py +3 -1
- airbyte_cdk/utils/message_utils.py +20 -11
- airbyte_cdk/utils/print_buffer.py +75 -0
- airbyte_cdk/utils/schema_inferrer.py +198 -28
- airbyte_cdk/utils/slice_hasher.py +30 -0
- airbyte_cdk/utils/spec_schema_transformations.py +6 -3
- airbyte_cdk/utils/stream_status_utils.py +8 -1
- airbyte_cdk/utils/traced_exception.py +61 -21
- airbyte_cdk-6.17.1.dev0.dist-info/METADATA +109 -0
- airbyte_cdk-6.17.1.dev0.dist-info/RECORD +350 -0
- {airbyte_cdk-0.72.1.dist-info → airbyte_cdk-6.17.1.dev0.dist-info}/WHEEL +1 -2
- airbyte_cdk-6.17.1.dev0.dist-info/entry_points.txt +3 -0
- airbyte_cdk/sources/declarative/create_partial.py +0 -92
- airbyte_cdk/sources/declarative/parsers/class_types_registry.py +0 -102
- airbyte_cdk/sources/declarative/parsers/default_implementation_registry.py +0 -64
- airbyte_cdk/sources/declarative/requesters/error_handlers/response_action.py +0 -16
- airbyte_cdk/sources/declarative/requesters/error_handlers/response_status.py +0 -68
- airbyte_cdk/sources/declarative/stream_slicers/cartesian_product_stream_slicer.py +0 -114
- airbyte_cdk/sources/deprecated/base_source.py +0 -94
- airbyte_cdk/sources/deprecated/client.py +0 -99
- airbyte_cdk/sources/singer/__init__.py +0 -8
- airbyte_cdk/sources/singer/singer_helpers.py +0 -304
- airbyte_cdk/sources/singer/source.py +0 -186
- airbyte_cdk/sources/streams/concurrent/partitions/record.py +0 -23
- airbyte_cdk/sources/streams/http/auth/__init__.py +0 -17
- airbyte_cdk/sources/streams/http/auth/core.py +0 -29
- airbyte_cdk/sources/streams/http/auth/oauth.py +0 -113
- airbyte_cdk/sources/streams/http/auth/token.py +0 -47
- airbyte_cdk/sources/streams/utils/stream_helper.py +0 -40
- airbyte_cdk/sources/utils/catalog_helpers.py +0 -22
- airbyte_cdk/sources/utils/schema_models.py +0 -84
- airbyte_cdk-0.72.1.dist-info/METADATA +0 -243
- airbyte_cdk-0.72.1.dist-info/RECORD +0 -466
- airbyte_cdk-0.72.1.dist-info/top_level.txt +0 -3
- source_declarative_manifest/main.py +0 -29
- unit_tests/connector_builder/__init__.py +0 -3
- unit_tests/connector_builder/test_connector_builder_handler.py +0 -871
- unit_tests/connector_builder/test_message_grouper.py +0 -713
- unit_tests/connector_builder/utils.py +0 -27
- unit_tests/destinations/test_destination.py +0 -243
- unit_tests/singer/test_singer_helpers.py +0 -56
- unit_tests/singer/test_singer_source.py +0 -112
- unit_tests/sources/__init__.py +0 -0
- unit_tests/sources/concurrent_source/__init__.py +0 -3
- unit_tests/sources/concurrent_source/test_concurrent_source_adapter.py +0 -106
- unit_tests/sources/declarative/__init__.py +0 -3
- unit_tests/sources/declarative/auth/__init__.py +0 -3
- unit_tests/sources/declarative/auth/test_oauth.py +0 -331
- unit_tests/sources/declarative/auth/test_selective_authenticator.py +0 -39
- unit_tests/sources/declarative/auth/test_session_token_auth.py +0 -182
- unit_tests/sources/declarative/auth/test_token_auth.py +0 -200
- unit_tests/sources/declarative/auth/test_token_provider.py +0 -73
- unit_tests/sources/declarative/checks/__init__.py +0 -3
- unit_tests/sources/declarative/checks/test_check_stream.py +0 -146
- unit_tests/sources/declarative/decoders/__init__.py +0 -0
- unit_tests/sources/declarative/decoders/test_json_decoder.py +0 -16
- unit_tests/sources/declarative/external_component.py +0 -13
- unit_tests/sources/declarative/extractors/__init__.py +0 -3
- unit_tests/sources/declarative/extractors/test_dpath_extractor.py +0 -55
- unit_tests/sources/declarative/extractors/test_record_filter.py +0 -55
- unit_tests/sources/declarative/extractors/test_record_selector.py +0 -179
- unit_tests/sources/declarative/incremental/__init__.py +0 -0
- unit_tests/sources/declarative/incremental/test_datetime_based_cursor.py +0 -860
- unit_tests/sources/declarative/incremental/test_per_partition_cursor.py +0 -406
- unit_tests/sources/declarative/incremental/test_per_partition_cursor_integration.py +0 -332
- unit_tests/sources/declarative/interpolation/__init__.py +0 -3
- unit_tests/sources/declarative/interpolation/test_filters.py +0 -80
- unit_tests/sources/declarative/interpolation/test_interpolated_boolean.py +0 -40
- unit_tests/sources/declarative/interpolation/test_interpolated_mapping.py +0 -35
- unit_tests/sources/declarative/interpolation/test_interpolated_nested_mapping.py +0 -45
- unit_tests/sources/declarative/interpolation/test_interpolated_string.py +0 -25
- unit_tests/sources/declarative/interpolation/test_jinja.py +0 -240
- unit_tests/sources/declarative/interpolation/test_macros.py +0 -73
- unit_tests/sources/declarative/parsers/__init__.py +0 -3
- unit_tests/sources/declarative/parsers/test_manifest_component_transformer.py +0 -406
- unit_tests/sources/declarative/parsers/test_manifest_reference_resolver.py +0 -139
- unit_tests/sources/declarative/parsers/test_model_to_component_factory.py +0 -1847
- unit_tests/sources/declarative/parsers/testing_components.py +0 -36
- unit_tests/sources/declarative/partition_routers/__init__.py +0 -3
- unit_tests/sources/declarative/partition_routers/test_list_partition_router.py +0 -155
- unit_tests/sources/declarative/partition_routers/test_single_partition_router.py +0 -14
- unit_tests/sources/declarative/partition_routers/test_substream_partition_router.py +0 -404
- unit_tests/sources/declarative/requesters/__init__.py +0 -3
- unit_tests/sources/declarative/requesters/error_handlers/__init__.py +0 -3
- unit_tests/sources/declarative/requesters/error_handlers/backoff_strategies/__init__.py +0 -3
- unit_tests/sources/declarative/requesters/error_handlers/backoff_strategies/test_constant_backoff.py +0 -34
- unit_tests/sources/declarative/requesters/error_handlers/backoff_strategies/test_exponential_backoff.py +0 -36
- unit_tests/sources/declarative/requesters/error_handlers/backoff_strategies/test_header_helper.py +0 -38
- unit_tests/sources/declarative/requesters/error_handlers/backoff_strategies/test_wait_time_from_header.py +0 -35
- unit_tests/sources/declarative/requesters/error_handlers/backoff_strategies/test_wait_until_time_from_header.py +0 -64
- unit_tests/sources/declarative/requesters/error_handlers/test_composite_error_handler.py +0 -213
- unit_tests/sources/declarative/requesters/error_handlers/test_default_error_handler.py +0 -178
- unit_tests/sources/declarative/requesters/error_handlers/test_http_response_filter.py +0 -121
- unit_tests/sources/declarative/requesters/error_handlers/test_response_status.py +0 -44
- unit_tests/sources/declarative/requesters/paginators/__init__.py +0 -3
- unit_tests/sources/declarative/requesters/paginators/test_cursor_pagination_strategy.py +0 -64
- unit_tests/sources/declarative/requesters/paginators/test_default_paginator.py +0 -313
- unit_tests/sources/declarative/requesters/paginators/test_no_paginator.py +0 -12
- unit_tests/sources/declarative/requesters/paginators/test_offset_increment.py +0 -58
- unit_tests/sources/declarative/requesters/paginators/test_page_increment.py +0 -70
- unit_tests/sources/declarative/requesters/paginators/test_request_option.py +0 -43
- unit_tests/sources/declarative/requesters/paginators/test_stop_condition.py +0 -105
- unit_tests/sources/declarative/requesters/request_options/__init__.py +0 -3
- unit_tests/sources/declarative/requesters/request_options/test_interpolated_request_options_provider.py +0 -101
- unit_tests/sources/declarative/requesters/test_http_requester.py +0 -974
- unit_tests/sources/declarative/requesters/test_interpolated_request_input_provider.py +0 -32
- unit_tests/sources/declarative/retrievers/__init__.py +0 -3
- unit_tests/sources/declarative/retrievers/test_simple_retriever.py +0 -542
- unit_tests/sources/declarative/schema/__init__.py +0 -6
- unit_tests/sources/declarative/schema/source_test/SourceTest.py +0 -8
- unit_tests/sources/declarative/schema/source_test/__init__.py +0 -3
- unit_tests/sources/declarative/schema/test_default_schema_loader.py +0 -32
- unit_tests/sources/declarative/schema/test_inline_schema_loader.py +0 -19
- unit_tests/sources/declarative/schema/test_json_file_schema_loader.py +0 -26
- unit_tests/sources/declarative/states/__init__.py +0 -3
- unit_tests/sources/declarative/stream_slicers/__init__.py +0 -3
- unit_tests/sources/declarative/stream_slicers/test_cartesian_product_stream_slicer.py +0 -225
- unit_tests/sources/declarative/test_create_partial.py +0 -83
- unit_tests/sources/declarative/test_declarative_stream.py +0 -103
- unit_tests/sources/declarative/test_manifest_declarative_source.py +0 -1260
- unit_tests/sources/declarative/test_types.py +0 -39
- unit_tests/sources/declarative/test_yaml_declarative_source.py +0 -148
- unit_tests/sources/file_based/__init__.py +0 -0
- unit_tests/sources/file_based/availability_strategy/__init__.py +0 -0
- unit_tests/sources/file_based/availability_strategy/test_default_file_based_availability_strategy.py +0 -100
- unit_tests/sources/file_based/config/__init__.py +0 -0
- unit_tests/sources/file_based/config/test_abstract_file_based_spec.py +0 -28
- unit_tests/sources/file_based/config/test_csv_format.py +0 -34
- unit_tests/sources/file_based/config/test_file_based_stream_config.py +0 -84
- unit_tests/sources/file_based/discovery_policy/__init__.py +0 -0
- unit_tests/sources/file_based/discovery_policy/test_default_discovery_policy.py +0 -31
- unit_tests/sources/file_based/file_types/__init__.py +0 -0
- unit_tests/sources/file_based/file_types/test_avro_parser.py +0 -243
- unit_tests/sources/file_based/file_types/test_csv_parser.py +0 -546
- unit_tests/sources/file_based/file_types/test_jsonl_parser.py +0 -158
- unit_tests/sources/file_based/file_types/test_parquet_parser.py +0 -274
- unit_tests/sources/file_based/file_types/test_unstructured_parser.py +0 -593
- unit_tests/sources/file_based/helpers.py +0 -70
- unit_tests/sources/file_based/in_memory_files_source.py +0 -211
- unit_tests/sources/file_based/scenarios/__init__.py +0 -0
- unit_tests/sources/file_based/scenarios/avro_scenarios.py +0 -744
- unit_tests/sources/file_based/scenarios/check_scenarios.py +0 -220
- unit_tests/sources/file_based/scenarios/concurrent_incremental_scenarios.py +0 -2844
- unit_tests/sources/file_based/scenarios/csv_scenarios.py +0 -3105
- unit_tests/sources/file_based/scenarios/file_based_source_builder.py +0 -91
- unit_tests/sources/file_based/scenarios/incremental_scenarios.py +0 -1926
- unit_tests/sources/file_based/scenarios/jsonl_scenarios.py +0 -930
- unit_tests/sources/file_based/scenarios/parquet_scenarios.py +0 -754
- unit_tests/sources/file_based/scenarios/scenario_builder.py +0 -234
- unit_tests/sources/file_based/scenarios/unstructured_scenarios.py +0 -608
- unit_tests/sources/file_based/scenarios/user_input_schema_scenarios.py +0 -746
- unit_tests/sources/file_based/scenarios/validation_policy_scenarios.py +0 -726
- unit_tests/sources/file_based/stream/__init__.py +0 -0
- unit_tests/sources/file_based/stream/concurrent/__init__.py +0 -0
- unit_tests/sources/file_based/stream/concurrent/test_adapters.py +0 -362
- unit_tests/sources/file_based/stream/concurrent/test_file_based_concurrent_cursor.py +0 -458
- unit_tests/sources/file_based/stream/test_default_file_based_cursor.py +0 -310
- unit_tests/sources/file_based/stream/test_default_file_based_stream.py +0 -244
- unit_tests/sources/file_based/test_file_based_scenarios.py +0 -320
- unit_tests/sources/file_based/test_file_based_stream_reader.py +0 -272
- unit_tests/sources/file_based/test_scenarios.py +0 -253
- unit_tests/sources/file_based/test_schema_helpers.py +0 -346
- unit_tests/sources/fixtures/__init__.py +0 -3
- unit_tests/sources/fixtures/source_test_fixture.py +0 -153
- unit_tests/sources/message/__init__.py +0 -0
- unit_tests/sources/message/test_repository.py +0 -153
- unit_tests/sources/streams/__init__.py +0 -0
- unit_tests/sources/streams/concurrent/__init__.py +0 -3
- unit_tests/sources/streams/concurrent/scenarios/__init__.py +0 -3
- unit_tests/sources/streams/concurrent/scenarios/incremental_scenarios.py +0 -250
- unit_tests/sources/streams/concurrent/scenarios/stream_facade_builder.py +0 -140
- unit_tests/sources/streams/concurrent/scenarios/stream_facade_scenarios.py +0 -452
- unit_tests/sources/streams/concurrent/scenarios/test_concurrent_scenarios.py +0 -76
- unit_tests/sources/streams/concurrent/scenarios/thread_based_concurrent_stream_scenarios.py +0 -418
- unit_tests/sources/streams/concurrent/scenarios/thread_based_concurrent_stream_source_builder.py +0 -142
- unit_tests/sources/streams/concurrent/scenarios/utils.py +0 -55
- unit_tests/sources/streams/concurrent/test_adapters.py +0 -380
- unit_tests/sources/streams/concurrent/test_concurrent_read_processor.py +0 -684
- unit_tests/sources/streams/concurrent/test_cursor.py +0 -139
- unit_tests/sources/streams/concurrent/test_datetime_state_converter.py +0 -369
- unit_tests/sources/streams/concurrent/test_default_stream.py +0 -197
- unit_tests/sources/streams/concurrent/test_partition_enqueuer.py +0 -90
- unit_tests/sources/streams/concurrent/test_partition_reader.py +0 -67
- unit_tests/sources/streams/concurrent/test_thread_pool_manager.py +0 -106
- unit_tests/sources/streams/http/__init__.py +0 -0
- unit_tests/sources/streams/http/auth/__init__.py +0 -0
- unit_tests/sources/streams/http/auth/test_auth.py +0 -173
- unit_tests/sources/streams/http/requests_native_auth/__init__.py +0 -0
- unit_tests/sources/streams/http/requests_native_auth/test_requests_native_auth.py +0 -423
- unit_tests/sources/streams/http/test_availability_strategy.py +0 -180
- unit_tests/sources/streams/http/test_http.py +0 -635
- unit_tests/sources/streams/test_availability_strategy.py +0 -70
- unit_tests/sources/streams/test_call_rate.py +0 -300
- unit_tests/sources/streams/test_stream_read.py +0 -405
- unit_tests/sources/streams/test_streams_core.py +0 -184
- unit_tests/sources/test_abstract_source.py +0 -1442
- unit_tests/sources/test_concurrent_source.py +0 -112
- unit_tests/sources/test_config.py +0 -92
- unit_tests/sources/test_connector_state_manager.py +0 -482
- unit_tests/sources/test_http_logger.py +0 -252
- unit_tests/sources/test_integration_source.py +0 -86
- unit_tests/sources/test_source.py +0 -684
- unit_tests/sources/test_source_read.py +0 -460
- unit_tests/test/__init__.py +0 -0
- unit_tests/test/mock_http/__init__.py +0 -0
- unit_tests/test/mock_http/test_matcher.py +0 -53
- unit_tests/test/mock_http/test_mocker.py +0 -214
- unit_tests/test/mock_http/test_request.py +0 -117
- unit_tests/test/mock_http/test_response_builder.py +0 -177
- unit_tests/test/test_entrypoint_wrapper.py +0 -240
- unit_tests/utils/__init__.py +0 -0
- unit_tests/utils/test_datetime_format_inferrer.py +0 -60
- unit_tests/utils/test_mapping_helpers.py +0 -54
- unit_tests/utils/test_message_utils.py +0 -91
- unit_tests/utils/test_rate_limiting.py +0 -26
- unit_tests/utils/test_schema_inferrer.py +0 -202
- unit_tests/utils/test_secret_utils.py +0 -135
- unit_tests/utils/test_stream_status_utils.py +0 -61
- unit_tests/utils/test_traced_exception.py +0 -107
- /airbyte_cdk/sources/{deprecated → declarative/async_job}/__init__.py +0 -0
- {source_declarative_manifest → airbyte_cdk/sources/declarative/migrations}/__init__.py +0 -0
- {unit_tests/destinations → airbyte_cdk/sql}/__init__.py +0 -0
- {unit_tests/singer → airbyte_cdk/sql/_util}/__init__.py +0 -0
- {airbyte_cdk-0.72.1.dist-info → airbyte_cdk-6.17.1.dev0.dist-info}/LICENSE.txt +0 -0
@@ -1,860 +0,0 @@
|
|
1
|
-
#
|
2
|
-
# Copyright (c) 2023 Airbyte, Inc., all rights reserved.
|
3
|
-
#
|
4
|
-
|
5
|
-
import datetime
|
6
|
-
import unittest
|
7
|
-
|
8
|
-
import pytest
|
9
|
-
from airbyte_cdk.sources.declarative.datetime.min_max_datetime import MinMaxDatetime
|
10
|
-
from airbyte_cdk.sources.declarative.incremental import DatetimeBasedCursor
|
11
|
-
from airbyte_cdk.sources.declarative.interpolation.interpolated_string import InterpolatedString
|
12
|
-
from airbyte_cdk.sources.declarative.requesters.request_option import RequestOption, RequestOptionType
|
13
|
-
from airbyte_cdk.sources.declarative.types import Record, StreamSlice
|
14
|
-
|
15
|
-
datetime_format = "%Y-%m-%dT%H:%M:%S.%f%z"
|
16
|
-
cursor_granularity = "PT0.000001S"
|
17
|
-
FAKE_NOW = datetime.datetime(2022, 1, 1, tzinfo=datetime.timezone.utc)
|
18
|
-
|
19
|
-
config = {"start_date": "2021-01-01T00:00:00.000000+0000", "start_date_ymd": "2021-01-01"}
|
20
|
-
end_date_now = InterpolatedString(string="{{ today_utc() }}", parameters={})
|
21
|
-
cursor_field = "created"
|
22
|
-
timezone = datetime.timezone.utc
|
23
|
-
NO_STATE = {}
|
24
|
-
ANY_SLICE = {}
|
25
|
-
|
26
|
-
|
27
|
-
class MockedNowDatetime(datetime.datetime):
|
28
|
-
@classmethod
|
29
|
-
def now(cls, tz=None):
|
30
|
-
return FAKE_NOW
|
31
|
-
|
32
|
-
|
33
|
-
@pytest.fixture()
|
34
|
-
def mock_datetime_now(monkeypatch):
|
35
|
-
monkeypatch.setattr(datetime, "datetime", MockedNowDatetime)
|
36
|
-
|
37
|
-
|
38
|
-
@pytest.mark.parametrize(
|
39
|
-
"test_name, stream_state, start, end, step, cursor_field, lookback_window, datetime_format, cursor_granularity, expected_slices",
|
40
|
-
[
|
41
|
-
(
|
42
|
-
"test_1_day",
|
43
|
-
NO_STATE,
|
44
|
-
MinMaxDatetime(datetime="{{ config['start_date'] }}", parameters={}),
|
45
|
-
MinMaxDatetime(datetime="2021-01-10T00:00:00.000000+0000", parameters={}),
|
46
|
-
"P1D",
|
47
|
-
cursor_field,
|
48
|
-
None,
|
49
|
-
datetime_format,
|
50
|
-
cursor_granularity,
|
51
|
-
[
|
52
|
-
{"start_time": "2021-01-01T00:00:00.000000+0000", "end_time": "2021-01-01T23:59:59.999999+0000"},
|
53
|
-
{"start_time": "2021-01-02T00:00:00.000000+0000", "end_time": "2021-01-02T23:59:59.999999+0000"},
|
54
|
-
{"start_time": "2021-01-03T00:00:00.000000+0000", "end_time": "2021-01-03T23:59:59.999999+0000"},
|
55
|
-
{"start_time": "2021-01-04T00:00:00.000000+0000", "end_time": "2021-01-04T23:59:59.999999+0000"},
|
56
|
-
{"start_time": "2021-01-05T00:00:00.000000+0000", "end_time": "2021-01-05T23:59:59.999999+0000"},
|
57
|
-
{"start_time": "2021-01-06T00:00:00.000000+0000", "end_time": "2021-01-06T23:59:59.999999+0000"},
|
58
|
-
{"start_time": "2021-01-07T00:00:00.000000+0000", "end_time": "2021-01-07T23:59:59.999999+0000"},
|
59
|
-
{"start_time": "2021-01-08T00:00:00.000000+0000", "end_time": "2021-01-08T23:59:59.999999+0000"},
|
60
|
-
{"start_time": "2021-01-09T00:00:00.000000+0000", "end_time": "2021-01-09T23:59:59.999999+0000"},
|
61
|
-
{"start_time": "2021-01-10T00:00:00.000000+0000", "end_time": "2021-01-10T00:00:00.000000+0000"},
|
62
|
-
],
|
63
|
-
),
|
64
|
-
(
|
65
|
-
"test_2_day",
|
66
|
-
NO_STATE,
|
67
|
-
MinMaxDatetime(datetime="{{ config['start_date'] }}", parameters={}),
|
68
|
-
MinMaxDatetime(datetime="2021-01-10T00:00:00.000000+0000", parameters={}),
|
69
|
-
"P2D",
|
70
|
-
cursor_field,
|
71
|
-
None,
|
72
|
-
datetime_format,
|
73
|
-
cursor_granularity,
|
74
|
-
[
|
75
|
-
{"start_time": "2021-01-01T00:00:00.000000+0000", "end_time": "2021-01-02T23:59:59.999999+0000"},
|
76
|
-
{"start_time": "2021-01-03T00:00:00.000000+0000", "end_time": "2021-01-04T23:59:59.999999+0000"},
|
77
|
-
{"start_time": "2021-01-05T00:00:00.000000+0000", "end_time": "2021-01-06T23:59:59.999999+0000"},
|
78
|
-
{"start_time": "2021-01-07T00:00:00.000000+0000", "end_time": "2021-01-08T23:59:59.999999+0000"},
|
79
|
-
{"start_time": "2021-01-09T00:00:00.000000+0000", "end_time": "2021-01-10T00:00:00.000000+0000"},
|
80
|
-
],
|
81
|
-
),
|
82
|
-
(
|
83
|
-
"test_1_week",
|
84
|
-
NO_STATE,
|
85
|
-
MinMaxDatetime(datetime="{{ config['start_date'] }}", parameters={}),
|
86
|
-
MinMaxDatetime(datetime="2021-02-10T00:00:00.000000+0000", parameters={}),
|
87
|
-
"P1W",
|
88
|
-
cursor_field,
|
89
|
-
None,
|
90
|
-
datetime_format,
|
91
|
-
cursor_granularity,
|
92
|
-
[
|
93
|
-
{"start_time": "2021-01-01T00:00:00.000000+0000", "end_time": "2021-01-07T23:59:59.999999+0000"},
|
94
|
-
{"start_time": "2021-01-08T00:00:00.000000+0000", "end_time": "2021-01-14T23:59:59.999999+0000"},
|
95
|
-
{"start_time": "2021-01-15T00:00:00.000000+0000", "end_time": "2021-01-21T23:59:59.999999+0000"},
|
96
|
-
{"start_time": "2021-01-22T00:00:00.000000+0000", "end_time": "2021-01-28T23:59:59.999999+0000"},
|
97
|
-
{"start_time": "2021-01-29T00:00:00.000000+0000", "end_time": "2021-02-04T23:59:59.999999+0000"},
|
98
|
-
{"start_time": "2021-02-05T00:00:00.000000+0000", "end_time": "2021-02-10T00:00:00.000000+0000"},
|
99
|
-
],
|
100
|
-
),
|
101
|
-
(
|
102
|
-
"test_1_month",
|
103
|
-
NO_STATE,
|
104
|
-
MinMaxDatetime(datetime="{{ config['start_date'] }}", parameters={}),
|
105
|
-
MinMaxDatetime(datetime="2021-06-10T00:00:00.000000+0000", parameters={}),
|
106
|
-
"P1M",
|
107
|
-
cursor_field,
|
108
|
-
None,
|
109
|
-
datetime_format,
|
110
|
-
cursor_granularity,
|
111
|
-
[
|
112
|
-
{"start_time": "2021-01-01T00:00:00.000000+0000", "end_time": "2021-01-31T23:59:59.999999+0000"},
|
113
|
-
{"start_time": "2021-02-01T00:00:00.000000+0000", "end_time": "2021-02-28T23:59:59.999999+0000"},
|
114
|
-
{"start_time": "2021-03-01T00:00:00.000000+0000", "end_time": "2021-03-31T23:59:59.999999+0000"},
|
115
|
-
{"start_time": "2021-04-01T00:00:00.000000+0000", "end_time": "2021-04-30T23:59:59.999999+0000"},
|
116
|
-
{"start_time": "2021-05-01T00:00:00.000000+0000", "end_time": "2021-05-31T23:59:59.999999+0000"},
|
117
|
-
{"start_time": "2021-06-01T00:00:00.000000+0000", "end_time": "2021-06-10T00:00:00.000000+0000"},
|
118
|
-
],
|
119
|
-
),
|
120
|
-
(
|
121
|
-
"test_1_year",
|
122
|
-
NO_STATE,
|
123
|
-
MinMaxDatetime(datetime="{{ config['start_date'] }}", parameters={}),
|
124
|
-
MinMaxDatetime(datetime="2022-06-10T00:00:00.000000+0000", parameters={}),
|
125
|
-
"P1Y",
|
126
|
-
cursor_field,
|
127
|
-
None,
|
128
|
-
datetime_format,
|
129
|
-
cursor_granularity,
|
130
|
-
[
|
131
|
-
{"start_time": "2021-01-01T00:00:00.000000+0000", "end_time": "2021-12-31T23:59:59.999999+0000"},
|
132
|
-
{"start_time": "2022-01-01T00:00:00.000000+0000", "end_time": "2022-01-01T00:00:00.000000+0000"},
|
133
|
-
],
|
134
|
-
),
|
135
|
-
(
|
136
|
-
"test_from_stream_state",
|
137
|
-
{cursor_field: "2021-01-05T00:00:00.000000+0000"},
|
138
|
-
MinMaxDatetime(datetime="2020-01-05T00:00:00.000000+0000", parameters={}),
|
139
|
-
MinMaxDatetime(datetime="2021-01-10T00:00:00.000000+0000", parameters={}),
|
140
|
-
"P1D",
|
141
|
-
cursor_field,
|
142
|
-
None,
|
143
|
-
datetime_format,
|
144
|
-
cursor_granularity,
|
145
|
-
[
|
146
|
-
{"start_time": "2021-01-05T00:00:00.000000+0000", "end_time": "2021-01-05T23:59:59.999999+0000"},
|
147
|
-
{"start_time": "2021-01-06T00:00:00.000000+0000", "end_time": "2021-01-06T23:59:59.999999+0000"},
|
148
|
-
{"start_time": "2021-01-07T00:00:00.000000+0000", "end_time": "2021-01-07T23:59:59.999999+0000"},
|
149
|
-
{"start_time": "2021-01-08T00:00:00.000000+0000", "end_time": "2021-01-08T23:59:59.999999+0000"},
|
150
|
-
{"start_time": "2021-01-09T00:00:00.000000+0000", "end_time": "2021-01-09T23:59:59.999999+0000"},
|
151
|
-
{"start_time": "2021-01-10T00:00:00.000000+0000", "end_time": "2021-01-10T00:00:00.000000+0000"},
|
152
|
-
],
|
153
|
-
),
|
154
|
-
(
|
155
|
-
"test_12_day",
|
156
|
-
NO_STATE,
|
157
|
-
MinMaxDatetime(datetime="{{ config['start_date'] }}", parameters={}),
|
158
|
-
MinMaxDatetime(datetime="2021-01-10T00:00:00.000000+0000", parameters={}),
|
159
|
-
"P12D",
|
160
|
-
cursor_field,
|
161
|
-
None,
|
162
|
-
datetime_format,
|
163
|
-
cursor_granularity,
|
164
|
-
[
|
165
|
-
{"start_time": "2021-01-01T00:00:00.000000+0000", "end_time": "2021-01-10T00:00:00.000000+0000"},
|
166
|
-
],
|
167
|
-
),
|
168
|
-
(
|
169
|
-
"test_end_time_greater_than_now",
|
170
|
-
NO_STATE,
|
171
|
-
MinMaxDatetime(datetime="2021-12-28T00:00:00.000000+0000", parameters={}),
|
172
|
-
MinMaxDatetime(datetime=f"{(FAKE_NOW + datetime.timedelta(days=1)).strftime(datetime_format)}", parameters={}),
|
173
|
-
"P1D",
|
174
|
-
cursor_field,
|
175
|
-
None,
|
176
|
-
datetime_format,
|
177
|
-
cursor_granularity,
|
178
|
-
[
|
179
|
-
{"start_time": "2021-12-28T00:00:00.000000+0000", "end_time": "2021-12-28T23:59:59.999999+0000"},
|
180
|
-
{"start_time": "2021-12-29T00:00:00.000000+0000", "end_time": "2021-12-29T23:59:59.999999+0000"},
|
181
|
-
{"start_time": "2021-12-30T00:00:00.000000+0000", "end_time": "2021-12-30T23:59:59.999999+0000"},
|
182
|
-
{"start_time": "2021-12-31T00:00:00.000000+0000", "end_time": "2021-12-31T23:59:59.999999+0000"},
|
183
|
-
{"start_time": "2022-01-01T00:00:00.000000+0000", "end_time": "2022-01-01T00:00:00.000000+0000"},
|
184
|
-
],
|
185
|
-
),
|
186
|
-
(
|
187
|
-
"test_start_date_greater_than_end_time",
|
188
|
-
NO_STATE,
|
189
|
-
MinMaxDatetime(datetime="2021-01-10T00:00:00.000000+0000", parameters={}),
|
190
|
-
MinMaxDatetime(datetime="2021-01-05T00:00:00.000000+0000", parameters={}),
|
191
|
-
"P1D",
|
192
|
-
cursor_field,
|
193
|
-
None,
|
194
|
-
datetime_format,
|
195
|
-
cursor_granularity,
|
196
|
-
[
|
197
|
-
{"start_time": "2021-01-05T00:00:00.000000+0000", "end_time": "2021-01-05T00:00:00.000000+0000"},
|
198
|
-
],
|
199
|
-
),
|
200
|
-
(
|
201
|
-
"test_cursor_date_greater_than_start_date",
|
202
|
-
{cursor_field: "2021-01-05T00:00:00.000000+0000"},
|
203
|
-
MinMaxDatetime(datetime="2021-01-01T00:00:00.000000+0000", parameters={}),
|
204
|
-
MinMaxDatetime(datetime="2021-01-10T00:00:00.000000+0000", parameters={}),
|
205
|
-
"P1D",
|
206
|
-
cursor_field,
|
207
|
-
None,
|
208
|
-
datetime_format,
|
209
|
-
cursor_granularity,
|
210
|
-
[
|
211
|
-
{"start_time": "2021-01-05T00:00:00.000000+0000", "end_time": "2021-01-05T23:59:59.999999+0000"},
|
212
|
-
{"start_time": "2021-01-06T00:00:00.000000+0000", "end_time": "2021-01-06T23:59:59.999999+0000"},
|
213
|
-
{"start_time": "2021-01-07T00:00:00.000000+0000", "end_time": "2021-01-07T23:59:59.999999+0000"},
|
214
|
-
{"start_time": "2021-01-08T00:00:00.000000+0000", "end_time": "2021-01-08T23:59:59.999999+0000"},
|
215
|
-
{"start_time": "2021-01-09T00:00:00.000000+0000", "end_time": "2021-01-09T23:59:59.999999+0000"},
|
216
|
-
{"start_time": "2021-01-10T00:00:00.000000+0000", "end_time": "2021-01-10T00:00:00.000000+0000"},
|
217
|
-
],
|
218
|
-
),
|
219
|
-
(
|
220
|
-
"test_cursor_date_greater_than_start_date_multiday_step",
|
221
|
-
{cursor_field: "2021-01-05T00:00:00.000000+0000"},
|
222
|
-
MinMaxDatetime(datetime="2021-01-03T00:00:00.000000+0000", parameters={}),
|
223
|
-
MinMaxDatetime(datetime="2021-01-10T00:00:00.000000+0000", parameters={}),
|
224
|
-
"P2D",
|
225
|
-
cursor_field,
|
226
|
-
None,
|
227
|
-
datetime_format,
|
228
|
-
cursor_granularity,
|
229
|
-
[
|
230
|
-
{"start_time": "2021-01-05T00:00:00.000000+0000", "end_time": "2021-01-06T23:59:59.999999+0000"},
|
231
|
-
{"start_time": "2021-01-07T00:00:00.000000+0000", "end_time": "2021-01-08T23:59:59.999999+0000"},
|
232
|
-
{"start_time": "2021-01-09T00:00:00.000000+0000", "end_time": "2021-01-10T00:00:00.000000+0000"},
|
233
|
-
],
|
234
|
-
),
|
235
|
-
(
|
236
|
-
"test_with_lookback_window_from_start_date",
|
237
|
-
NO_STATE,
|
238
|
-
MinMaxDatetime(datetime="2021-01-05", datetime_format="%Y-%m-%d", parameters={}),
|
239
|
-
MinMaxDatetime(datetime="2021-01-05", datetime_format="%Y-%m-%d", parameters={}),
|
240
|
-
"P1D",
|
241
|
-
cursor_field,
|
242
|
-
"P3D",
|
243
|
-
datetime_format,
|
244
|
-
cursor_granularity,
|
245
|
-
[
|
246
|
-
{"start_time": "2021-01-02T00:00:00.000000+0000", "end_time": "2021-01-02T23:59:59.999999+0000"},
|
247
|
-
{"start_time": "2021-01-03T00:00:00.000000+0000", "end_time": "2021-01-03T23:59:59.999999+0000"},
|
248
|
-
{"start_time": "2021-01-04T00:00:00.000000+0000", "end_time": "2021-01-04T23:59:59.999999+0000"},
|
249
|
-
{"start_time": "2021-01-05T00:00:00.000000+0000", "end_time": "2021-01-05T00:00:00.000000+0000"},
|
250
|
-
],
|
251
|
-
),
|
252
|
-
(
|
253
|
-
"test_with_lookback_window_from_cursor",
|
254
|
-
{cursor_field: "2021-01-05T00:00:00.000000+0000"},
|
255
|
-
MinMaxDatetime(datetime="2021-01-01T00:00:00.000000+0000", parameters={}),
|
256
|
-
MinMaxDatetime(datetime="2021-01-06T00:00:00.000000+0000", parameters={}),
|
257
|
-
"P1D",
|
258
|
-
cursor_field,
|
259
|
-
"P3D",
|
260
|
-
datetime_format,
|
261
|
-
cursor_granularity,
|
262
|
-
[
|
263
|
-
{"start_time": "2021-01-02T00:00:00.000000+0000", "end_time": "2021-01-02T23:59:59.999999+0000"},
|
264
|
-
{"start_time": "2021-01-03T00:00:00.000000+0000", "end_time": "2021-01-03T23:59:59.999999+0000"},
|
265
|
-
{"start_time": "2021-01-04T00:00:00.000000+0000", "end_time": "2021-01-04T23:59:59.999999+0000"},
|
266
|
-
{"start_time": "2021-01-05T00:00:00.000000+0000", "end_time": "2021-01-05T23:59:59.999999+0000"},
|
267
|
-
{"start_time": "2021-01-06T00:00:00.000000+0000", "end_time": "2021-01-06T00:00:00.000000+0000"},
|
268
|
-
],
|
269
|
-
),
|
270
|
-
(
|
271
|
-
"test_with_lookback_window_defaults_to_0d",
|
272
|
-
{},
|
273
|
-
MinMaxDatetime(datetime="2021-01-01", datetime_format="%Y-%m-%d", parameters={}),
|
274
|
-
MinMaxDatetime(datetime="2021-01-05", datetime_format="%Y-%m-%d", parameters={}),
|
275
|
-
"P1D",
|
276
|
-
cursor_field,
|
277
|
-
"{{ config['does_not_exist'] }}",
|
278
|
-
datetime_format,
|
279
|
-
cursor_granularity,
|
280
|
-
[
|
281
|
-
{"start_time": "2021-01-01T00:00:00.000000+0000", "end_time": "2021-01-01T23:59:59.999999+0000"},
|
282
|
-
{"start_time": "2021-01-02T00:00:00.000000+0000", "end_time": "2021-01-02T23:59:59.999999+0000"},
|
283
|
-
{"start_time": "2021-01-03T00:00:00.000000+0000", "end_time": "2021-01-03T23:59:59.999999+0000"},
|
284
|
-
{"start_time": "2021-01-04T00:00:00.000000+0000", "end_time": "2021-01-04T23:59:59.999999+0000"},
|
285
|
-
{"start_time": "2021-01-05T00:00:00.000000+0000", "end_time": "2021-01-05T00:00:00.000000+0000"},
|
286
|
-
],
|
287
|
-
),
|
288
|
-
(
|
289
|
-
"test_start_is_after_stream_state",
|
290
|
-
{cursor_field: "2021-01-05T00:00:00.000000+0000"},
|
291
|
-
MinMaxDatetime(datetime="2021-01-01T00:00:00.000000+0000", parameters={}),
|
292
|
-
MinMaxDatetime(datetime="2021-01-10T00:00:00.000000+0000", parameters={}),
|
293
|
-
"P1D",
|
294
|
-
cursor_field,
|
295
|
-
None,
|
296
|
-
datetime_format,
|
297
|
-
cursor_granularity,
|
298
|
-
[
|
299
|
-
{"start_time": "2021-01-05T00:00:00.000000+0000", "end_time": "2021-01-05T23:59:59.999999+0000"},
|
300
|
-
{"start_time": "2021-01-06T00:00:00.000000+0000", "end_time": "2021-01-06T23:59:59.999999+0000"},
|
301
|
-
{"start_time": "2021-01-07T00:00:00.000000+0000", "end_time": "2021-01-07T23:59:59.999999+0000"},
|
302
|
-
{"start_time": "2021-01-08T00:00:00.000000+0000", "end_time": "2021-01-08T23:59:59.999999+0000"},
|
303
|
-
{"start_time": "2021-01-09T00:00:00.000000+0000", "end_time": "2021-01-09T23:59:59.999999+0000"},
|
304
|
-
{"start_time": "2021-01-10T00:00:00.000000+0000", "end_time": "2021-01-10T00:00:00.000000+0000"},
|
305
|
-
],
|
306
|
-
),
|
307
|
-
],
|
308
|
-
)
|
309
|
-
def test_stream_slices(
|
310
|
-
mock_datetime_now,
|
311
|
-
test_name,
|
312
|
-
stream_state,
|
313
|
-
start,
|
314
|
-
end,
|
315
|
-
step,
|
316
|
-
cursor_field,
|
317
|
-
lookback_window,
|
318
|
-
datetime_format,
|
319
|
-
cursor_granularity,
|
320
|
-
expected_slices,
|
321
|
-
):
|
322
|
-
lookback_window = InterpolatedString(string=lookback_window, parameters={}) if lookback_window else None
|
323
|
-
cursor = DatetimeBasedCursor(
|
324
|
-
start_datetime=start,
|
325
|
-
end_datetime=end,
|
326
|
-
step=step,
|
327
|
-
cursor_field=cursor_field,
|
328
|
-
datetime_format=datetime_format,
|
329
|
-
cursor_granularity=cursor_granularity,
|
330
|
-
lookback_window=lookback_window,
|
331
|
-
config=config,
|
332
|
-
parameters={},
|
333
|
-
)
|
334
|
-
cursor.set_initial_state(stream_state)
|
335
|
-
stream_slices = cursor.stream_slices()
|
336
|
-
|
337
|
-
assert stream_slices == expected_slices
|
338
|
-
|
339
|
-
|
340
|
-
@pytest.mark.parametrize(
|
341
|
-
"test_name, previous_cursor, stream_slice, observed_records, expected_state",
|
342
|
-
[
|
343
|
-
(
|
344
|
-
"test_close_slice_previous_cursor_is_highest",
|
345
|
-
"2023-01-01",
|
346
|
-
StreamSlice(partition={}, cursor_slice={"start_time": "2021-01-01", "end_time": "2022-01-01"}),
|
347
|
-
[{cursor_field: "2021-01-01"}],
|
348
|
-
{cursor_field: "2023-01-01"},
|
349
|
-
),
|
350
|
-
(
|
351
|
-
"test_close_slice_stream_slice_partition_end_is_highest",
|
352
|
-
"2020-01-01",
|
353
|
-
StreamSlice(partition={}, cursor_slice={"start_time": "2021-01-01", "end_time": "2023-01-01"}),
|
354
|
-
[{cursor_field: "2021-01-01"}],
|
355
|
-
{cursor_field: "2021-01-01"},
|
356
|
-
),
|
357
|
-
(
|
358
|
-
"test_close_slice_latest_record_cursor_value_is_higher_than_slice_end",
|
359
|
-
"2021-01-01",
|
360
|
-
StreamSlice(partition={}, cursor_slice={"start_time": "2021-01-01", "end_time": "2022-01-01"}),
|
361
|
-
[{cursor_field: "2023-01-01"}],
|
362
|
-
{cursor_field: "2021-01-01"},
|
363
|
-
),
|
364
|
-
(
|
365
|
-
"test_close_slice_with_no_records_observed",
|
366
|
-
"2021-01-01",
|
367
|
-
StreamSlice(partition={}, cursor_slice={"start_time": "2021-01-01", "end_time": "2022-01-01"}),
|
368
|
-
[],
|
369
|
-
{cursor_field: "2021-01-01"},
|
370
|
-
),
|
371
|
-
(
|
372
|
-
"test_close_slice_with_no_records_observed_and_no_previous_state",
|
373
|
-
None,
|
374
|
-
StreamSlice(partition={}, cursor_slice={"start_time": "2021-01-01", "end_time": "2022-01-01"}),
|
375
|
-
[],
|
376
|
-
{},
|
377
|
-
),
|
378
|
-
(
|
379
|
-
"test_close_slice_without_previous_cursor",
|
380
|
-
None,
|
381
|
-
StreamSlice(partition={}, cursor_slice={"start_time": "2021-01-01", "end_time": "2023-01-01"}),
|
382
|
-
[{cursor_field: "2022-01-01"}],
|
383
|
-
{cursor_field: "2022-01-01"},
|
384
|
-
),
|
385
|
-
(
|
386
|
-
"test_close_slice_with_out_of_order_records",
|
387
|
-
"2021-01-01",
|
388
|
-
StreamSlice(partition={}, cursor_slice={"start_time": "2021-01-01", "end_time": "2022-01-01"}),
|
389
|
-
[{cursor_field: "2021-04-01"}, {cursor_field: "2021-02-01"}, {cursor_field: "2021-03-01"}],
|
390
|
-
{cursor_field: "2021-04-01"},
|
391
|
-
),
|
392
|
-
(
|
393
|
-
"test_close_slice_with_some_records_out_of_slice_boundaries",
|
394
|
-
"2021-01-01",
|
395
|
-
StreamSlice(partition={}, cursor_slice={"start_time": "2021-01-01", "end_time": "2022-01-01"}),
|
396
|
-
[{cursor_field: "2021-02-01"}, {cursor_field: "2021-03-01"}, {cursor_field: "2023-01-01"}],
|
397
|
-
{cursor_field: "2021-03-01"},
|
398
|
-
),
|
399
|
-
(
|
400
|
-
"test_close_slice_with_all_records_out_of_slice_boundaries",
|
401
|
-
"2021-01-01",
|
402
|
-
StreamSlice(partition={}, cursor_slice={"start_time": "2021-01-01", "end_time": "2022-01-01"}),
|
403
|
-
[{cursor_field: "2023-01-01"}],
|
404
|
-
{cursor_field: "2021-01-01"},
|
405
|
-
),
|
406
|
-
(
|
407
|
-
"test_close_slice_with_all_records_out_of_slice_and_no_previous_cursor",
|
408
|
-
None,
|
409
|
-
StreamSlice(partition={}, cursor_slice={"start_time": "2021-01-01", "end_time": "2022-01-01"}),
|
410
|
-
[{cursor_field: "2023-01-01"}],
|
411
|
-
{},
|
412
|
-
),
|
413
|
-
],
|
414
|
-
)
|
415
|
-
def test_close_slice(test_name, previous_cursor, stream_slice, observed_records, expected_state):
|
416
|
-
cursor = DatetimeBasedCursor(
|
417
|
-
start_datetime=MinMaxDatetime(datetime="2021-01-01T00:00:00.000000+0000", parameters={}),
|
418
|
-
cursor_field=InterpolatedString(string=cursor_field, parameters={}),
|
419
|
-
datetime_format="%Y-%m-%d",
|
420
|
-
config=config,
|
421
|
-
parameters={},
|
422
|
-
partition_field_start="start_time",
|
423
|
-
partition_field_end="end_time",
|
424
|
-
)
|
425
|
-
cursor.set_initial_state({cursor_field: previous_cursor})
|
426
|
-
for record_data in observed_records:
|
427
|
-
record = Record(record_data, stream_slice)
|
428
|
-
cursor.observe(stream_slice, record)
|
429
|
-
last_record = observed_records[-1] if observed_records else None
|
430
|
-
cursor.close_slice(stream_slice, Record(last_record, stream_slice) if last_record else None)
|
431
|
-
updated_state = cursor.get_stream_state()
|
432
|
-
assert updated_state == expected_state
|
433
|
-
|
434
|
-
|
435
|
-
def test_close_slice_fails_if_slice_has_a_partition():
|
436
|
-
cursor = DatetimeBasedCursor(
|
437
|
-
start_datetime=MinMaxDatetime(datetime="2021-01-01T00:00:00.000000+0000", parameters={}),
|
438
|
-
cursor_field=InterpolatedString(string=cursor_field, parameters={}),
|
439
|
-
datetime_format="%Y-%m-%d",
|
440
|
-
config=config,
|
441
|
-
parameters={},
|
442
|
-
)
|
443
|
-
stream_slice = StreamSlice(partition={"key": "value"}, cursor_slice={"end_time": "2022-01-01"})
|
444
|
-
with pytest.raises(ValueError):
|
445
|
-
cursor.close_slice(stream_slice, Record({"id": 1}, stream_slice))
|
446
|
-
|
447
|
-
|
448
|
-
def test_compares_cursor_values_by_chronological_order():
|
449
|
-
cursor = DatetimeBasedCursor(
|
450
|
-
start_datetime=MinMaxDatetime(datetime="2021-01-01T00:00:00.000000+0000", parameters={}),
|
451
|
-
cursor_field=cursor_field,
|
452
|
-
datetime_format="%d-%m-%Y",
|
453
|
-
config=config,
|
454
|
-
parameters={},
|
455
|
-
)
|
456
|
-
|
457
|
-
_slice = StreamSlice(partition={}, cursor_slice={"start_time": "01-01-2023", "end_time": "01-04-2023"})
|
458
|
-
first_record = Record({cursor_field: "21-02-2023"}, _slice)
|
459
|
-
cursor.observe(_slice, first_record)
|
460
|
-
second_record = Record({cursor_field: "01-03-2023"}, _slice)
|
461
|
-
cursor.observe(_slice, second_record)
|
462
|
-
cursor.close_slice(_slice, second_record)
|
463
|
-
|
464
|
-
assert cursor.get_stream_state()[cursor_field] == "01-03-2023"
|
465
|
-
|
466
|
-
|
467
|
-
def test_given_different_format_and_slice_is_highest_when_close_slice_then_state_uses_record_format():
|
468
|
-
cursor = DatetimeBasedCursor(
|
469
|
-
start_datetime=MinMaxDatetime(datetime="2021-01-01T00:00:00.000000+0000", parameters={}),
|
470
|
-
cursor_field=cursor_field,
|
471
|
-
datetime_format="%Y-%m-%dT%H:%M:%S.%fZ",
|
472
|
-
cursor_datetime_formats=["%Y-%m-%d"],
|
473
|
-
config=config,
|
474
|
-
parameters={},
|
475
|
-
)
|
476
|
-
|
477
|
-
_slice = StreamSlice(partition={}, cursor_slice={"start_time": "2023-01-01T17:30:19.000Z", "end_time": "2023-01-04T17:30:19.000Z"})
|
478
|
-
record_cursor_value = "2023-01-03"
|
479
|
-
record = Record({cursor_field: record_cursor_value}, _slice)
|
480
|
-
cursor.observe(_slice, record)
|
481
|
-
cursor.close_slice(_slice, record)
|
482
|
-
|
483
|
-
assert cursor.get_stream_state()[cursor_field] == "2023-01-03"
|
484
|
-
|
485
|
-
|
486
|
-
@pytest.mark.parametrize(
|
487
|
-
"test_name, inject_into, field_name, expected_req_params, expected_headers, expected_body_json, expected_body_data",
|
488
|
-
[
|
489
|
-
("test_start_time_inject_into_none", None, None, {}, {}, {}, {}),
|
490
|
-
(
|
491
|
-
"test_start_time_passed_by_req_param",
|
492
|
-
RequestOptionType.request_parameter,
|
493
|
-
"start_time",
|
494
|
-
{"start_time": "2021-01-01T00:00:00.000000+0000", "endtime": "2021-01-04T00:00:00.000000+0000"},
|
495
|
-
{},
|
496
|
-
{},
|
497
|
-
{},
|
498
|
-
),
|
499
|
-
(
|
500
|
-
"test_start_time_inject_into_header",
|
501
|
-
RequestOptionType.header,
|
502
|
-
"start_time",
|
503
|
-
{},
|
504
|
-
{"start_time": "2021-01-01T00:00:00.000000+0000", "endtime": "2021-01-04T00:00:00.000000+0000"},
|
505
|
-
{},
|
506
|
-
{},
|
507
|
-
),
|
508
|
-
(
|
509
|
-
"test_start_time_inject_intoy_body_json",
|
510
|
-
RequestOptionType.body_json,
|
511
|
-
"start_time",
|
512
|
-
{},
|
513
|
-
{},
|
514
|
-
{"start_time": "2021-01-01T00:00:00.000000+0000", "endtime": "2021-01-04T00:00:00.000000+0000"},
|
515
|
-
{},
|
516
|
-
),
|
517
|
-
(
|
518
|
-
"test_start_time_inject_into_body_data",
|
519
|
-
RequestOptionType.body_data,
|
520
|
-
"start_time",
|
521
|
-
{},
|
522
|
-
{},
|
523
|
-
{},
|
524
|
-
{"start_time": "2021-01-01T00:00:00.000000+0000", "endtime": "2021-01-04T00:00:00.000000+0000"},
|
525
|
-
),
|
526
|
-
],
|
527
|
-
)
|
528
|
-
def test_request_option(test_name, inject_into, field_name, expected_req_params, expected_headers, expected_body_json, expected_body_data):
|
529
|
-
start_request_option = RequestOption(inject_into=inject_into, parameters={}, field_name=field_name) if inject_into else None
|
530
|
-
end_request_option = RequestOption(inject_into=inject_into, parameters={}, field_name="endtime") if inject_into else None
|
531
|
-
slicer = DatetimeBasedCursor(
|
532
|
-
start_datetime=MinMaxDatetime(datetime="2021-01-01T00:00:00.000000+0000", parameters={}),
|
533
|
-
end_datetime=MinMaxDatetime(datetime="2021-01-10T00:00:00.000000+0000", parameters={}),
|
534
|
-
step="P1D",
|
535
|
-
cursor_field=InterpolatedString(string=cursor_field, parameters={}),
|
536
|
-
datetime_format=datetime_format,
|
537
|
-
cursor_granularity=cursor_granularity,
|
538
|
-
lookback_window=InterpolatedString(string="P0D", parameters={}),
|
539
|
-
start_time_option=start_request_option,
|
540
|
-
end_time_option=end_request_option,
|
541
|
-
config=config,
|
542
|
-
parameters={},
|
543
|
-
)
|
544
|
-
stream_slice = {"start_time": "2021-01-01T00:00:00.000000+0000", "end_time": "2021-01-04T00:00:00.000000+0000"}
|
545
|
-
assert expected_req_params == slicer.get_request_params(stream_slice=stream_slice)
|
546
|
-
assert expected_headers == slicer.get_request_headers(stream_slice=stream_slice)
|
547
|
-
assert expected_body_json == slicer.get_request_body_json(stream_slice=stream_slice)
|
548
|
-
assert expected_body_data == slicer.get_request_body_data(stream_slice=stream_slice)
|
549
|
-
|
550
|
-
|
551
|
-
@pytest.mark.parametrize(
|
552
|
-
"stream_slice", [
|
553
|
-
pytest.param(None, id="test_none_stream_slice"),
|
554
|
-
pytest.param({}, id="test_none_stream_slice"),
|
555
|
-
]
|
556
|
-
)
|
557
|
-
def test_request_option_with_empty_stream_slice(stream_slice):
|
558
|
-
start_request_option = RequestOption(inject_into=RequestOptionType.request_parameter, parameters={}, field_name="starttime")
|
559
|
-
end_request_option = RequestOption(inject_into=RequestOptionType.request_parameter, parameters={}, field_name="endtime")
|
560
|
-
slicer = DatetimeBasedCursor(
|
561
|
-
start_datetime=MinMaxDatetime(datetime="2021-01-01T00:00:00.000000+0000", parameters={}),
|
562
|
-
end_datetime=MinMaxDatetime(datetime="2021-01-10T00:00:00.000000+0000", parameters={}),
|
563
|
-
step="P1D",
|
564
|
-
cursor_field=InterpolatedString(string=cursor_field, parameters={}),
|
565
|
-
datetime_format=datetime_format,
|
566
|
-
cursor_granularity=cursor_granularity,
|
567
|
-
lookback_window=InterpolatedString(string="P0D", parameters={}),
|
568
|
-
start_time_option=start_request_option,
|
569
|
-
end_time_option=end_request_option,
|
570
|
-
config=config,
|
571
|
-
parameters={},
|
572
|
-
)
|
573
|
-
assert {} == slicer.get_request_params(stream_slice=stream_slice)
|
574
|
-
|
575
|
-
|
576
|
-
@pytest.mark.parametrize(
|
577
|
-
"test_name, input_date, date_format, date_format_granularity, expected_output_date",
|
578
|
-
[
|
579
|
-
(
|
580
|
-
"test_parse_date_iso",
|
581
|
-
"2021-01-01T00:00:00.000000+0000",
|
582
|
-
"%Y-%m-%dT%H:%M:%S.%f%z",
|
583
|
-
"PT0.000001S",
|
584
|
-
datetime.datetime(2021, 1, 1, 0, 0, tzinfo=datetime.timezone.utc),
|
585
|
-
),
|
586
|
-
(
|
587
|
-
"test_parse_timestamp",
|
588
|
-
"1609459200",
|
589
|
-
"%s",
|
590
|
-
"PT1S",
|
591
|
-
datetime.datetime(2021, 1, 1, 0, 0, tzinfo=datetime.timezone.utc),
|
592
|
-
),
|
593
|
-
("test_parse_date_number", "20210101", "%Y%m%d", "P1D", datetime.datetime(2021, 1, 1, 0, 0, tzinfo=datetime.timezone.utc)),
|
594
|
-
],
|
595
|
-
)
|
596
|
-
def test_parse_date_legacy_merge_datetime_format_in_cursor_datetime_format(
|
597
|
-
test_name, input_date, date_format, date_format_granularity, expected_output_date
|
598
|
-
):
|
599
|
-
slicer = DatetimeBasedCursor(
|
600
|
-
start_datetime=MinMaxDatetime("2021-01-01T00:00:00.000000+0000", parameters={}),
|
601
|
-
end_datetime=MinMaxDatetime("2021-01-10T00:00:00.000000+0000", parameters={}),
|
602
|
-
step="P1D",
|
603
|
-
cursor_field=InterpolatedString(cursor_field, parameters={}),
|
604
|
-
datetime_format=date_format,
|
605
|
-
cursor_granularity=date_format_granularity,
|
606
|
-
lookback_window=InterpolatedString("P0D", parameters={}),
|
607
|
-
config=config,
|
608
|
-
parameters={},
|
609
|
-
)
|
610
|
-
output_date = slicer.parse_date(input_date)
|
611
|
-
assert expected_output_date == output_date
|
612
|
-
|
613
|
-
|
614
|
-
@pytest.mark.parametrize(
|
615
|
-
"test_name, input_date, date_formats, expected_output_date",
|
616
|
-
[
|
617
|
-
(
|
618
|
-
"test_match_first_format",
|
619
|
-
"2021-01-01T00:00:00.000000+0000",
|
620
|
-
["%Y-%m-%dT%H:%M:%S.%f%z", "%s"],
|
621
|
-
datetime.datetime(2021, 1, 1, 0, 0, tzinfo=datetime.timezone.utc),
|
622
|
-
),
|
623
|
-
(
|
624
|
-
"test_match_second_format",
|
625
|
-
"1609459200",
|
626
|
-
["%Y-%m-%dT%H:%M:%S.%f%z", "%s"],
|
627
|
-
datetime.datetime(2021, 1, 1, 0, 0, tzinfo=datetime.timezone.utc),
|
628
|
-
),
|
629
|
-
],
|
630
|
-
)
|
631
|
-
def test_parse_date(test_name, input_date, date_formats, expected_output_date):
|
632
|
-
slicer = DatetimeBasedCursor(
|
633
|
-
start_datetime=MinMaxDatetime("2021-01-01T00:00:00.000000+0000", parameters={}),
|
634
|
-
cursor_field=InterpolatedString(cursor_field, parameters={}),
|
635
|
-
datetime_format="%Y-%m-%d",
|
636
|
-
cursor_datetime_formats=date_formats,
|
637
|
-
config=config,
|
638
|
-
parameters={},
|
639
|
-
)
|
640
|
-
assert slicer.parse_date(input_date) == expected_output_date
|
641
|
-
|
642
|
-
|
643
|
-
def test_given_unknown_format_when_parse_date_then_raise_error():
|
644
|
-
slicer = DatetimeBasedCursor(
|
645
|
-
start_datetime=MinMaxDatetime("2021-01-01T00:00:00.000000+0000", parameters={}),
|
646
|
-
cursor_field=InterpolatedString(cursor_field, parameters={}),
|
647
|
-
datetime_format="%Y-%m-%d",
|
648
|
-
cursor_datetime_formats=["%Y-%m-%d", "%s"],
|
649
|
-
config=config,
|
650
|
-
parameters={},
|
651
|
-
)
|
652
|
-
with pytest.raises(ValueError):
|
653
|
-
slicer.parse_date("2021-01-01T00:00:00.000000+0000")
|
654
|
-
|
655
|
-
|
656
|
-
@pytest.mark.parametrize(
|
657
|
-
"test_name, input_dt, datetimeformat, datetimeformat_granularity, expected_output",
|
658
|
-
[
|
659
|
-
("test_format_timestamp", datetime.datetime(2021, 1, 1, 0, 0, tzinfo=datetime.timezone.utc), "%s", "PT1S", "1609459200"),
|
660
|
-
("test_format_string", datetime.datetime(2021, 1, 1, 0, 0, tzinfo=datetime.timezone.utc), "%Y-%m-%d", "P1D", "2021-01-01"),
|
661
|
-
("test_format_to_number", datetime.datetime(2021, 1, 1, 0, 0, tzinfo=datetime.timezone.utc), "%Y%m%d", "P1D", "20210101"),
|
662
|
-
],
|
663
|
-
)
|
664
|
-
def test_format_datetime(test_name, input_dt, datetimeformat, datetimeformat_granularity, expected_output):
|
665
|
-
slicer = DatetimeBasedCursor(
|
666
|
-
start_datetime=MinMaxDatetime("2021-01-01T00:00:00.000000+0000", parameters={}),
|
667
|
-
end_datetime=MinMaxDatetime("2021-01-10T00:00:00.000000+0000", parameters={}),
|
668
|
-
step="P1D",
|
669
|
-
cursor_field=InterpolatedString(cursor_field, parameters={}),
|
670
|
-
datetime_format=datetimeformat,
|
671
|
-
cursor_granularity=datetimeformat_granularity,
|
672
|
-
lookback_window=InterpolatedString("P0D", parameters={}),
|
673
|
-
config=config,
|
674
|
-
parameters={},
|
675
|
-
)
|
676
|
-
|
677
|
-
output_date = slicer._format_datetime(input_dt)
|
678
|
-
assert expected_output == output_date
|
679
|
-
|
680
|
-
|
681
|
-
def test_step_but_no_cursor_granularity():
|
682
|
-
with pytest.raises(ValueError):
|
683
|
-
DatetimeBasedCursor(
|
684
|
-
start_datetime=MinMaxDatetime("2021-01-01T00:00:00.000000+0000", parameters={}),
|
685
|
-
end_datetime=MinMaxDatetime("2021-01-10T00:00:00.000000+0000", parameters={}),
|
686
|
-
step="P1D",
|
687
|
-
cursor_field=InterpolatedString(cursor_field, parameters={}),
|
688
|
-
datetime_format="%Y-%m-%d",
|
689
|
-
config=config,
|
690
|
-
parameters={},
|
691
|
-
)
|
692
|
-
|
693
|
-
|
694
|
-
def test_cursor_granularity_but_no_step():
|
695
|
-
with pytest.raises(ValueError):
|
696
|
-
DatetimeBasedCursor(
|
697
|
-
start_datetime=MinMaxDatetime("2021-01-01T00:00:00.000000+0000", parameters={}),
|
698
|
-
end_datetime=MinMaxDatetime("2021-01-10T00:00:00.000000+0000", parameters={}),
|
699
|
-
cursor_granularity="P1D",
|
700
|
-
cursor_field=InterpolatedString(cursor_field, parameters={}),
|
701
|
-
datetime_format="%Y-%m-%d",
|
702
|
-
config=config,
|
703
|
-
parameters={},
|
704
|
-
)
|
705
|
-
|
706
|
-
|
707
|
-
def test_given_multiple_cursor_datetime_format_then_slice_using_first_format():
|
708
|
-
cursor = DatetimeBasedCursor(
|
709
|
-
start_datetime=MinMaxDatetime("2021-01-01", parameters={}),
|
710
|
-
end_datetime=MinMaxDatetime("2023-01-10", parameters={}),
|
711
|
-
cursor_field=InterpolatedString(cursor_field, parameters={}),
|
712
|
-
datetime_format="%Y-%m-%d",
|
713
|
-
cursor_datetime_formats=["%Y-%m-%dT%H:%M:%S", "%Y-%m-%d"],
|
714
|
-
config=config,
|
715
|
-
parameters={},
|
716
|
-
)
|
717
|
-
stream_slices = cursor.stream_slices()
|
718
|
-
assert stream_slices == [{"start_time": "2021-01-01", "end_time": "2023-01-10"}]
|
719
|
-
|
720
|
-
|
721
|
-
def test_no_cursor_granularity_and_no_step_then_only_return_one_slice():
|
722
|
-
cursor = DatetimeBasedCursor(
|
723
|
-
start_datetime=MinMaxDatetime("2021-01-01", parameters={}),
|
724
|
-
end_datetime=MinMaxDatetime("2023-01-01", parameters={}),
|
725
|
-
cursor_field=InterpolatedString(cursor_field, parameters={}),
|
726
|
-
datetime_format="%Y-%m-%d",
|
727
|
-
config=config,
|
728
|
-
parameters={},
|
729
|
-
)
|
730
|
-
stream_slices = cursor.stream_slices()
|
731
|
-
assert stream_slices == [{"start_time": "2021-01-01", "end_time": "2023-01-01"}]
|
732
|
-
|
733
|
-
|
734
|
-
def test_no_end_datetime(mock_datetime_now):
|
735
|
-
cursor = DatetimeBasedCursor(
|
736
|
-
start_datetime=MinMaxDatetime("2021-01-01", parameters={}),
|
737
|
-
cursor_field=InterpolatedString(cursor_field, parameters={}),
|
738
|
-
datetime_format="%Y-%m-%d",
|
739
|
-
config=config,
|
740
|
-
parameters={},
|
741
|
-
)
|
742
|
-
stream_slices = cursor.stream_slices()
|
743
|
-
assert stream_slices == [{"start_time": "2021-01-01", "end_time": FAKE_NOW.strftime("%Y-%m-%d")}]
|
744
|
-
|
745
|
-
|
746
|
-
def test_given_no_state_and_start_before_cursor_value_when_should_be_synced_then_return_true():
|
747
|
-
cursor = DatetimeBasedCursor(
|
748
|
-
start_datetime=MinMaxDatetime("2021-01-01", parameters={}),
|
749
|
-
cursor_field=InterpolatedString(cursor_field, parameters={}),
|
750
|
-
datetime_format="%Y-%m-%d",
|
751
|
-
config=config,
|
752
|
-
parameters={},
|
753
|
-
)
|
754
|
-
assert cursor.should_be_synced(Record({cursor_field: "2022-01-01"}, ANY_SLICE))
|
755
|
-
|
756
|
-
|
757
|
-
def test_given_no_state_and_start_after_cursor_value_when_should_be_synced_then_return_false():
|
758
|
-
cursor = DatetimeBasedCursor(
|
759
|
-
start_datetime=MinMaxDatetime("2022-01-01", parameters={}),
|
760
|
-
cursor_field=InterpolatedString(cursor_field, parameters={}),
|
761
|
-
datetime_format="%Y-%m-%d",
|
762
|
-
config=config,
|
763
|
-
parameters={},
|
764
|
-
)
|
765
|
-
assert not cursor.should_be_synced(Record({cursor_field: "2021-01-01"}, ANY_SLICE))
|
766
|
-
|
767
|
-
|
768
|
-
def test_given_state_earliest_to_start_datetime_when_should_be_synced_then_use_state_as_earliest_boundary():
|
769
|
-
cursor = DatetimeBasedCursor(
|
770
|
-
start_datetime=MinMaxDatetime("2021-01-01", parameters={}),
|
771
|
-
cursor_field=InterpolatedString(cursor_field, parameters={}),
|
772
|
-
datetime_format="%Y-%m-%d",
|
773
|
-
config=config,
|
774
|
-
parameters={},
|
775
|
-
)
|
776
|
-
cursor.set_initial_state({cursor_field: "2023-01-01"})
|
777
|
-
assert not cursor.should_be_synced(Record({cursor_field: "2022-01-01"}, ANY_SLICE))
|
778
|
-
|
779
|
-
|
780
|
-
def test_given_start_datetime_earliest_to_state_when_should_be_synced_then_use_start_datetime_as_earliest_boundary():
|
781
|
-
cursor = DatetimeBasedCursor(
|
782
|
-
start_datetime=MinMaxDatetime("2023-01-01", parameters={}),
|
783
|
-
cursor_field=InterpolatedString(cursor_field, parameters={}),
|
784
|
-
datetime_format="%Y-%m-%d",
|
785
|
-
config=config,
|
786
|
-
parameters={},
|
787
|
-
)
|
788
|
-
cursor.set_initial_state({cursor_field: "2021-01-01"})
|
789
|
-
assert not cursor.should_be_synced(Record({cursor_field: "2022-01-01"}, ANY_SLICE))
|
790
|
-
|
791
|
-
|
792
|
-
def test_given_end_datetime_before_cursor_value_when_should_be_synced_then_return_false():
|
793
|
-
cursor = DatetimeBasedCursor(
|
794
|
-
start_datetime=MinMaxDatetime("2023-01-01", parameters={}),
|
795
|
-
end_datetime=MinMaxDatetime("2025-01-01", parameters={}),
|
796
|
-
cursor_field=InterpolatedString(cursor_field, parameters={}),
|
797
|
-
datetime_format="%Y-%m-%d",
|
798
|
-
config=config,
|
799
|
-
parameters={},
|
800
|
-
)
|
801
|
-
assert not cursor.should_be_synced(Record({cursor_field: "2030-01-01"}, ANY_SLICE))
|
802
|
-
|
803
|
-
|
804
|
-
def test_given_record_without_cursor_value_when_should_be_synced_then_return_true():
|
805
|
-
cursor = DatetimeBasedCursor(
|
806
|
-
start_datetime=MinMaxDatetime("3000-01-01", parameters={}),
|
807
|
-
cursor_field=InterpolatedString(cursor_field, parameters={}),
|
808
|
-
datetime_format="%Y-%m-%d",
|
809
|
-
config=config,
|
810
|
-
parameters={},
|
811
|
-
)
|
812
|
-
assert cursor.should_be_synced(Record({"record without cursor value": "any"}, ANY_SLICE))
|
813
|
-
|
814
|
-
|
815
|
-
def test_given_first_greater_than_second_then_return_true():
|
816
|
-
cursor = DatetimeBasedCursor(
|
817
|
-
start_datetime=MinMaxDatetime("3000-01-01", parameters={}),
|
818
|
-
cursor_field="cursor_field",
|
819
|
-
datetime_format="%Y-%m-%d",
|
820
|
-
config=config,
|
821
|
-
parameters={},
|
822
|
-
)
|
823
|
-
assert cursor.is_greater_than_or_equal(Record({"cursor_field": "2023-01-01"}, {}), Record({"cursor_field": "2021-01-01"}, {}))
|
824
|
-
|
825
|
-
|
826
|
-
def test_given_first_lesser_than_second_then_return_false():
|
827
|
-
cursor = DatetimeBasedCursor(
|
828
|
-
start_datetime=MinMaxDatetime("3000-01-01", parameters={}),
|
829
|
-
cursor_field="cursor_field",
|
830
|
-
datetime_format="%Y-%m-%d",
|
831
|
-
config=config,
|
832
|
-
parameters={},
|
833
|
-
)
|
834
|
-
assert not cursor.is_greater_than_or_equal(Record({"cursor_field": "2021-01-01"}, {}), Record({"cursor_field": "2023-01-01"}, {}))
|
835
|
-
|
836
|
-
|
837
|
-
def test_given_no_cursor_value_for_second_than_second_then_return_true():
|
838
|
-
cursor = DatetimeBasedCursor(
|
839
|
-
start_datetime=MinMaxDatetime("3000-01-01", parameters={}),
|
840
|
-
cursor_field="cursor_field",
|
841
|
-
datetime_format="%Y-%m-%d",
|
842
|
-
config=config,
|
843
|
-
parameters={},
|
844
|
-
)
|
845
|
-
assert cursor.is_greater_than_or_equal(Record({"cursor_field": "2021-01-01"}, {}), Record({}, {}))
|
846
|
-
|
847
|
-
|
848
|
-
def test_given_no_cursor_value_for_first_than_second_then_return_false():
|
849
|
-
cursor = DatetimeBasedCursor(
|
850
|
-
start_datetime=MinMaxDatetime("3000-01-01", parameters={}),
|
851
|
-
cursor_field="cursor_field",
|
852
|
-
datetime_format="%Y-%m-%d",
|
853
|
-
config=config,
|
854
|
-
parameters={},
|
855
|
-
)
|
856
|
-
assert not cursor.is_greater_than_or_equal(Record({}, {}), Record({"cursor_field": "2021-01-01"}, {}))
|
857
|
-
|
858
|
-
|
859
|
-
if __name__ == "__main__":
|
860
|
-
unittest.main()
|