airbyte-cdk 0.72.1__py3-none-any.whl → 6.13.1.dev4107__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 +421 -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 +65 -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 +25 -3
- 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 +159 -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 +1695 -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 +228 -72
- 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 +58 -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 +52 -15
- airbyte_cdk/sources/file_based/file_based_source.py +163 -33
- airbyte_cdk/sources/file_based/file_based_stream_reader.py +83 -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 +147 -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 +175 -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-6.13.1.dev4107.dist-info/METADATA +109 -0
- airbyte_cdk-6.13.1.dev4107.dist-info/RECORD +349 -0
- {airbyte_cdk-0.72.1.dist-info → airbyte_cdk-6.13.1.dev4107.dist-info}/WHEEL +1 -2
- airbyte_cdk-6.13.1.dev4107.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-6.13.1.dev4107.dist-info}/LICENSE.txt +0 -0
@@ -1,320 +0,0 @@
|
|
1
|
-
#
|
2
|
-
# Copyright (c) 2023 Airbyte, Inc., all rights reserved.
|
3
|
-
#
|
4
|
-
|
5
|
-
from pathlib import PosixPath
|
6
|
-
|
7
|
-
import pytest
|
8
|
-
from _pytest.capture import CaptureFixture
|
9
|
-
from airbyte_cdk.sources.abstract_source import AbstractSource
|
10
|
-
from freezegun import freeze_time
|
11
|
-
from unit_tests.sources.file_based.scenarios.avro_scenarios import (
|
12
|
-
avro_all_types_scenario,
|
13
|
-
avro_file_with_double_as_number_scenario,
|
14
|
-
multiple_avro_combine_schema_scenario,
|
15
|
-
multiple_streams_avro_scenario,
|
16
|
-
single_avro_scenario,
|
17
|
-
)
|
18
|
-
from unit_tests.sources.file_based.scenarios.check_scenarios import (
|
19
|
-
error_empty_stream_scenario,
|
20
|
-
error_listing_files_scenario,
|
21
|
-
error_multi_stream_scenario,
|
22
|
-
error_reading_file_scenario,
|
23
|
-
error_record_validation_user_provided_schema_scenario,
|
24
|
-
success_csv_scenario,
|
25
|
-
success_extensionless_scenario,
|
26
|
-
success_multi_stream_scenario,
|
27
|
-
success_user_provided_schema_scenario,
|
28
|
-
)
|
29
|
-
from unit_tests.sources.file_based.scenarios.concurrent_incremental_scenarios import (
|
30
|
-
multi_csv_different_timestamps_scenario_concurrent,
|
31
|
-
multi_csv_include_missing_files_within_history_range_concurrent_cursor_is_newer,
|
32
|
-
multi_csv_include_missing_files_within_history_range_concurrent_cursor_is_older,
|
33
|
-
multi_csv_per_timestamp_scenario_concurrent,
|
34
|
-
multi_csv_remove_old_files_if_history_is_full_scenario_concurrent_cursor_is_newer,
|
35
|
-
multi_csv_remove_old_files_if_history_is_full_scenario_concurrent_cursor_is_older,
|
36
|
-
multi_csv_same_timestamp_more_files_than_history_size_scenario_concurrent_cursor_is_newer,
|
37
|
-
multi_csv_same_timestamp_more_files_than_history_size_scenario_concurrent_cursor_is_older,
|
38
|
-
multi_csv_same_timestamp_scenario_concurrent,
|
39
|
-
multi_csv_skip_file_if_already_in_history_concurrent,
|
40
|
-
multi_csv_sync_files_within_history_time_window_if_history_is_incomplete_different_timestamps_scenario_concurrent_cursor_is_newer,
|
41
|
-
multi_csv_sync_files_within_history_time_window_if_history_is_incomplete_different_timestamps_scenario_concurrent_cursor_is_older,
|
42
|
-
multi_csv_sync_files_within_time_window_if_history_is_incomplete__different_timestamps_scenario_concurrent_cursor_is_newer,
|
43
|
-
multi_csv_sync_files_within_time_window_if_history_is_incomplete__different_timestamps_scenario_concurrent_cursor_is_older,
|
44
|
-
multi_csv_sync_recent_files_if_history_is_incomplete_scenario_concurrent_cursor_is_newer,
|
45
|
-
multi_csv_sync_recent_files_if_history_is_incomplete_scenario_concurrent_cursor_is_older,
|
46
|
-
single_csv_file_is_skipped_if_same_modified_at_as_in_history_concurrent,
|
47
|
-
single_csv_file_is_synced_if_modified_at_is_more_recent_than_in_history_concurrent,
|
48
|
-
single_csv_input_state_is_earlier_scenario_concurrent,
|
49
|
-
single_csv_input_state_is_later_scenario_concurrent,
|
50
|
-
single_csv_no_input_state_scenario_concurrent,
|
51
|
-
)
|
52
|
-
from unit_tests.sources.file_based.scenarios.csv_scenarios import (
|
53
|
-
csv_analytics_scenario,
|
54
|
-
csv_autogenerate_column_names_scenario,
|
55
|
-
csv_custom_bool_values_scenario,
|
56
|
-
csv_custom_delimiter_in_double_quotes_scenario,
|
57
|
-
csv_custom_delimiter_with_escape_char_scenario,
|
58
|
-
csv_custom_format_scenario,
|
59
|
-
csv_custom_null_values_scenario,
|
60
|
-
csv_double_quote_is_set_scenario,
|
61
|
-
csv_escape_char_is_set_scenario,
|
62
|
-
csv_multi_stream_scenario,
|
63
|
-
csv_newline_in_values_not_quoted_scenario,
|
64
|
-
csv_newline_in_values_quoted_value_scenario,
|
65
|
-
csv_no_records_scenario,
|
66
|
-
csv_single_stream_scenario,
|
67
|
-
csv_skip_after_header_scenario,
|
68
|
-
csv_skip_before_and_after_header_scenario,
|
69
|
-
csv_skip_before_header_scenario,
|
70
|
-
csv_string_are_not_null_if_strings_can_be_null_is_false_scenario,
|
71
|
-
csv_string_can_be_null_with_input_schemas_scenario,
|
72
|
-
csv_string_not_null_if_no_null_values_scenario,
|
73
|
-
csv_strings_can_be_null_not_quoted_scenario,
|
74
|
-
earlier_csv_scenario,
|
75
|
-
empty_schema_inference_scenario,
|
76
|
-
invalid_csv_multi_scenario,
|
77
|
-
invalid_csv_scenario,
|
78
|
-
multi_csv_scenario,
|
79
|
-
multi_csv_stream_n_file_exceeds_limit_for_inference,
|
80
|
-
multi_stream_custom_format,
|
81
|
-
schemaless_csv_multi_stream_scenario,
|
82
|
-
schemaless_csv_scenario,
|
83
|
-
schemaless_with_user_input_schema_fails_connection_check_multi_stream_scenario,
|
84
|
-
schemaless_with_user_input_schema_fails_connection_check_scenario,
|
85
|
-
single_csv_scenario,
|
86
|
-
)
|
87
|
-
from unit_tests.sources.file_based.scenarios.incremental_scenarios import (
|
88
|
-
multi_csv_different_timestamps_scenario,
|
89
|
-
multi_csv_include_missing_files_within_history_range,
|
90
|
-
multi_csv_per_timestamp_scenario,
|
91
|
-
multi_csv_remove_old_files_if_history_is_full_scenario,
|
92
|
-
multi_csv_same_timestamp_more_files_than_history_size_scenario,
|
93
|
-
multi_csv_same_timestamp_scenario,
|
94
|
-
multi_csv_skip_file_if_already_in_history,
|
95
|
-
multi_csv_sync_files_within_history_time_window_if_history_is_incomplete_different_timestamps_scenario,
|
96
|
-
multi_csv_sync_files_within_time_window_if_history_is_incomplete__different_timestamps_scenario,
|
97
|
-
multi_csv_sync_recent_files_if_history_is_incomplete_scenario,
|
98
|
-
single_csv_file_is_skipped_if_same_modified_at_as_in_history,
|
99
|
-
single_csv_file_is_synced_if_modified_at_is_more_recent_than_in_history,
|
100
|
-
single_csv_input_state_is_earlier_scenario,
|
101
|
-
single_csv_input_state_is_later_scenario,
|
102
|
-
single_csv_no_input_state_scenario,
|
103
|
-
)
|
104
|
-
from unit_tests.sources.file_based.scenarios.jsonl_scenarios import (
|
105
|
-
invalid_jsonl_scenario,
|
106
|
-
jsonl_multi_stream_scenario,
|
107
|
-
jsonl_user_input_schema_scenario,
|
108
|
-
multi_jsonl_stream_n_bytes_exceeds_limit_for_inference,
|
109
|
-
multi_jsonl_stream_n_file_exceeds_limit_for_inference,
|
110
|
-
multi_jsonl_with_different_keys_scenario,
|
111
|
-
schemaless_jsonl_multi_stream_scenario,
|
112
|
-
schemaless_jsonl_scenario,
|
113
|
-
single_jsonl_scenario,
|
114
|
-
)
|
115
|
-
from unit_tests.sources.file_based.scenarios.parquet_scenarios import (
|
116
|
-
multi_parquet_scenario,
|
117
|
-
parquet_file_with_decimal_as_float_scenario,
|
118
|
-
parquet_file_with_decimal_as_string_scenario,
|
119
|
-
parquet_file_with_decimal_no_config_scenario,
|
120
|
-
parquet_various_types_scenario,
|
121
|
-
parquet_with_invalid_config_scenario,
|
122
|
-
single_parquet_scenario,
|
123
|
-
single_partitioned_parquet_scenario,
|
124
|
-
)
|
125
|
-
from unit_tests.sources.file_based.scenarios.scenario_builder import TestScenario
|
126
|
-
from unit_tests.sources.file_based.scenarios.unstructured_scenarios import (
|
127
|
-
corrupted_file_scenario,
|
128
|
-
no_file_extension_unstructured_scenario,
|
129
|
-
simple_markdown_scenario,
|
130
|
-
simple_txt_scenario,
|
131
|
-
simple_unstructured_scenario,
|
132
|
-
unstructured_invalid_file_type_discover_scenario_no_skip,
|
133
|
-
unstructured_invalid_file_type_discover_scenario_skip,
|
134
|
-
unstructured_invalid_file_type_read_scenario,
|
135
|
-
)
|
136
|
-
from unit_tests.sources.file_based.scenarios.user_input_schema_scenarios import (
|
137
|
-
multi_stream_user_input_schema_scenario_emit_nonconforming_records,
|
138
|
-
multi_stream_user_input_schema_scenario_schema_is_invalid,
|
139
|
-
multi_stream_user_input_schema_scenario_skip_nonconforming_records,
|
140
|
-
single_stream_user_input_schema_scenario_emit_nonconforming_records,
|
141
|
-
single_stream_user_input_schema_scenario_schema_is_invalid,
|
142
|
-
single_stream_user_input_schema_scenario_skip_nonconforming_records,
|
143
|
-
valid_multi_stream_user_input_schema_scenario,
|
144
|
-
valid_single_stream_user_input_schema_scenario,
|
145
|
-
)
|
146
|
-
from unit_tests.sources.file_based.scenarios.validation_policy_scenarios import (
|
147
|
-
emit_record_scenario_multi_stream,
|
148
|
-
emit_record_scenario_single_stream,
|
149
|
-
skip_record_scenario_multi_stream,
|
150
|
-
skip_record_scenario_single_stream,
|
151
|
-
wait_for_rediscovery_scenario_multi_stream,
|
152
|
-
wait_for_rediscovery_scenario_single_stream,
|
153
|
-
)
|
154
|
-
from unit_tests.sources.file_based.test_scenarios import verify_check, verify_discover, verify_read, verify_spec
|
155
|
-
|
156
|
-
discover_failure_scenarios = [
|
157
|
-
earlier_csv_scenario,
|
158
|
-
empty_schema_inference_scenario,
|
159
|
-
]
|
160
|
-
|
161
|
-
discover_success_scenarios = [
|
162
|
-
csv_no_records_scenario,
|
163
|
-
csv_multi_stream_scenario,
|
164
|
-
csv_single_stream_scenario,
|
165
|
-
invalid_csv_scenario,
|
166
|
-
invalid_csv_multi_scenario,
|
167
|
-
single_csv_scenario,
|
168
|
-
multi_csv_scenario,
|
169
|
-
multi_csv_stream_n_file_exceeds_limit_for_inference,
|
170
|
-
single_csv_input_state_is_earlier_scenario,
|
171
|
-
single_csv_no_input_state_scenario,
|
172
|
-
single_csv_input_state_is_later_scenario,
|
173
|
-
multi_csv_same_timestamp_scenario,
|
174
|
-
multi_csv_different_timestamps_scenario,
|
175
|
-
multi_csv_per_timestamp_scenario,
|
176
|
-
multi_csv_skip_file_if_already_in_history,
|
177
|
-
multi_csv_include_missing_files_within_history_range,
|
178
|
-
multi_csv_remove_old_files_if_history_is_full_scenario,
|
179
|
-
multi_csv_same_timestamp_more_files_than_history_size_scenario,
|
180
|
-
multi_csv_sync_recent_files_if_history_is_incomplete_scenario,
|
181
|
-
multi_csv_sync_files_within_time_window_if_history_is_incomplete__different_timestamps_scenario,
|
182
|
-
multi_csv_sync_files_within_history_time_window_if_history_is_incomplete_different_timestamps_scenario,
|
183
|
-
single_csv_file_is_skipped_if_same_modified_at_as_in_history,
|
184
|
-
single_csv_file_is_synced_if_modified_at_is_more_recent_than_in_history,
|
185
|
-
csv_custom_format_scenario,
|
186
|
-
multi_stream_custom_format,
|
187
|
-
single_parquet_scenario,
|
188
|
-
multi_parquet_scenario,
|
189
|
-
parquet_various_types_scenario,
|
190
|
-
parquet_file_with_decimal_no_config_scenario,
|
191
|
-
parquet_file_with_decimal_as_string_scenario,
|
192
|
-
parquet_file_with_decimal_as_float_scenario,
|
193
|
-
schemaless_csv_scenario,
|
194
|
-
schemaless_csv_multi_stream_scenario,
|
195
|
-
schemaless_with_user_input_schema_fails_connection_check_multi_stream_scenario,
|
196
|
-
schemaless_with_user_input_schema_fails_connection_check_scenario,
|
197
|
-
single_stream_user_input_schema_scenario_schema_is_invalid,
|
198
|
-
single_stream_user_input_schema_scenario_emit_nonconforming_records,
|
199
|
-
single_stream_user_input_schema_scenario_skip_nonconforming_records,
|
200
|
-
multi_stream_user_input_schema_scenario_emit_nonconforming_records,
|
201
|
-
multi_stream_user_input_schema_scenario_skip_nonconforming_records,
|
202
|
-
multi_stream_user_input_schema_scenario_schema_is_invalid,
|
203
|
-
valid_multi_stream_user_input_schema_scenario,
|
204
|
-
valid_single_stream_user_input_schema_scenario,
|
205
|
-
single_jsonl_scenario,
|
206
|
-
multi_jsonl_with_different_keys_scenario,
|
207
|
-
multi_jsonl_stream_n_file_exceeds_limit_for_inference,
|
208
|
-
multi_jsonl_stream_n_bytes_exceeds_limit_for_inference,
|
209
|
-
invalid_jsonl_scenario,
|
210
|
-
jsonl_multi_stream_scenario,
|
211
|
-
jsonl_user_input_schema_scenario,
|
212
|
-
schemaless_jsonl_scenario,
|
213
|
-
schemaless_jsonl_multi_stream_scenario,
|
214
|
-
csv_string_can_be_null_with_input_schemas_scenario,
|
215
|
-
csv_string_are_not_null_if_strings_can_be_null_is_false_scenario,
|
216
|
-
csv_string_not_null_if_no_null_values_scenario,
|
217
|
-
csv_strings_can_be_null_not_quoted_scenario,
|
218
|
-
csv_newline_in_values_quoted_value_scenario,
|
219
|
-
csv_escape_char_is_set_scenario,
|
220
|
-
csv_double_quote_is_set_scenario,
|
221
|
-
csv_custom_delimiter_with_escape_char_scenario,
|
222
|
-
csv_custom_delimiter_in_double_quotes_scenario,
|
223
|
-
csv_skip_before_header_scenario,
|
224
|
-
csv_skip_after_header_scenario,
|
225
|
-
csv_skip_before_and_after_header_scenario,
|
226
|
-
csv_custom_bool_values_scenario,
|
227
|
-
csv_custom_null_values_scenario,
|
228
|
-
single_avro_scenario,
|
229
|
-
avro_all_types_scenario,
|
230
|
-
multiple_avro_combine_schema_scenario,
|
231
|
-
multiple_streams_avro_scenario,
|
232
|
-
avro_file_with_double_as_number_scenario,
|
233
|
-
csv_newline_in_values_not_quoted_scenario,
|
234
|
-
csv_autogenerate_column_names_scenario,
|
235
|
-
parquet_with_invalid_config_scenario,
|
236
|
-
single_partitioned_parquet_scenario,
|
237
|
-
simple_markdown_scenario,
|
238
|
-
simple_txt_scenario,
|
239
|
-
simple_unstructured_scenario,
|
240
|
-
corrupted_file_scenario,
|
241
|
-
no_file_extension_unstructured_scenario,
|
242
|
-
unstructured_invalid_file_type_discover_scenario_no_skip,
|
243
|
-
unstructured_invalid_file_type_discover_scenario_skip,
|
244
|
-
unstructured_invalid_file_type_read_scenario,
|
245
|
-
multi_csv_different_timestamps_scenario_concurrent,
|
246
|
-
multi_csv_include_missing_files_within_history_range_concurrent_cursor_is_newer,
|
247
|
-
multi_csv_include_missing_files_within_history_range_concurrent_cursor_is_older,
|
248
|
-
multi_csv_per_timestamp_scenario_concurrent,
|
249
|
-
multi_csv_remove_old_files_if_history_is_full_scenario_concurrent_cursor_is_newer,
|
250
|
-
multi_csv_remove_old_files_if_history_is_full_scenario_concurrent_cursor_is_older,
|
251
|
-
multi_csv_same_timestamp_more_files_than_history_size_scenario_concurrent_cursor_is_newer,
|
252
|
-
multi_csv_same_timestamp_more_files_than_history_size_scenario_concurrent_cursor_is_older,
|
253
|
-
multi_csv_same_timestamp_scenario_concurrent,
|
254
|
-
multi_csv_skip_file_if_already_in_history_concurrent,
|
255
|
-
multi_csv_sync_files_within_history_time_window_if_history_is_incomplete_different_timestamps_scenario_concurrent_cursor_is_newer,
|
256
|
-
multi_csv_sync_files_within_history_time_window_if_history_is_incomplete_different_timestamps_scenario_concurrent_cursor_is_older,
|
257
|
-
multi_csv_sync_files_within_time_window_if_history_is_incomplete__different_timestamps_scenario_concurrent_cursor_is_newer,
|
258
|
-
multi_csv_sync_files_within_time_window_if_history_is_incomplete__different_timestamps_scenario_concurrent_cursor_is_older,
|
259
|
-
multi_csv_sync_recent_files_if_history_is_incomplete_scenario_concurrent_cursor_is_newer,
|
260
|
-
multi_csv_sync_recent_files_if_history_is_incomplete_scenario_concurrent_cursor_is_older,
|
261
|
-
single_csv_file_is_skipped_if_same_modified_at_as_in_history_concurrent,
|
262
|
-
single_csv_file_is_synced_if_modified_at_is_more_recent_than_in_history_concurrent,
|
263
|
-
single_csv_input_state_is_earlier_scenario_concurrent,
|
264
|
-
single_csv_input_state_is_later_scenario_concurrent,
|
265
|
-
single_csv_no_input_state_scenario_concurrent,
|
266
|
-
]
|
267
|
-
|
268
|
-
discover_scenarios = discover_failure_scenarios + discover_success_scenarios
|
269
|
-
|
270
|
-
read_scenarios = discover_success_scenarios + [
|
271
|
-
emit_record_scenario_multi_stream,
|
272
|
-
emit_record_scenario_single_stream,
|
273
|
-
skip_record_scenario_multi_stream,
|
274
|
-
skip_record_scenario_single_stream,
|
275
|
-
csv_analytics_scenario,
|
276
|
-
wait_for_rediscovery_scenario_multi_stream,
|
277
|
-
wait_for_rediscovery_scenario_single_stream,
|
278
|
-
]
|
279
|
-
|
280
|
-
spec_scenarios = [
|
281
|
-
single_csv_scenario,
|
282
|
-
]
|
283
|
-
|
284
|
-
check_scenarios = [
|
285
|
-
error_empty_stream_scenario,
|
286
|
-
error_listing_files_scenario,
|
287
|
-
error_reading_file_scenario,
|
288
|
-
error_record_validation_user_provided_schema_scenario,
|
289
|
-
error_multi_stream_scenario,
|
290
|
-
success_csv_scenario,
|
291
|
-
success_extensionless_scenario,
|
292
|
-
success_multi_stream_scenario,
|
293
|
-
success_user_provided_schema_scenario,
|
294
|
-
schemaless_with_user_input_schema_fails_connection_check_multi_stream_scenario,
|
295
|
-
schemaless_with_user_input_schema_fails_connection_check_scenario,
|
296
|
-
valid_single_stream_user_input_schema_scenario,
|
297
|
-
single_avro_scenario,
|
298
|
-
earlier_csv_scenario,
|
299
|
-
]
|
300
|
-
|
301
|
-
|
302
|
-
@pytest.mark.parametrize("scenario", discover_scenarios, ids=[s.name for s in discover_scenarios])
|
303
|
-
def test_file_based_discover(capsys: CaptureFixture[str], tmp_path: PosixPath, scenario: TestScenario[AbstractSource]) -> None:
|
304
|
-
verify_discover(capsys, tmp_path, scenario)
|
305
|
-
|
306
|
-
|
307
|
-
@pytest.mark.parametrize("scenario", read_scenarios, ids=[s.name for s in read_scenarios])
|
308
|
-
@freeze_time("2023-06-09T00:00:00Z")
|
309
|
-
def test_file_based_read(scenario: TestScenario[AbstractSource]) -> None:
|
310
|
-
verify_read(scenario)
|
311
|
-
|
312
|
-
|
313
|
-
@pytest.mark.parametrize("scenario", spec_scenarios, ids=[c.name for c in spec_scenarios])
|
314
|
-
def test_file_based_spec(capsys: CaptureFixture[str], scenario: TestScenario[AbstractSource]) -> None:
|
315
|
-
verify_spec(capsys, scenario)
|
316
|
-
|
317
|
-
|
318
|
-
@pytest.mark.parametrize("scenario", check_scenarios, ids=[c.name for c in check_scenarios])
|
319
|
-
def test_file_based_check(capsys: CaptureFixture[str], tmp_path: PosixPath, scenario: TestScenario[AbstractSource]) -> None:
|
320
|
-
verify_check(capsys, tmp_path, scenario)
|
@@ -1,272 +0,0 @@
|
|
1
|
-
#
|
2
|
-
# Copyright (c) 2023 Airbyte, Inc., all rights reserved.
|
3
|
-
#
|
4
|
-
|
5
|
-
from io import IOBase
|
6
|
-
from typing import Any, Iterable, List, Mapping, Optional, Set
|
7
|
-
|
8
|
-
import pytest
|
9
|
-
from airbyte_cdk.sources.file_based.config.abstract_file_based_spec import AbstractFileBasedSpec
|
10
|
-
from airbyte_cdk.sources.file_based.file_based_stream_reader import AbstractFileBasedStreamReader
|
11
|
-
from airbyte_cdk.sources.file_based.remote_file import RemoteFile
|
12
|
-
from pydantic import AnyUrl
|
13
|
-
from unit_tests.sources.file_based.helpers import make_remote_files
|
14
|
-
|
15
|
-
reader = AbstractFileBasedStreamReader
|
16
|
-
|
17
|
-
"""
|
18
|
-
The rules are:
|
19
|
-
|
20
|
-
- All files at top-level: /*
|
21
|
-
- All files at top-level of mydir: mydir/*
|
22
|
-
- All files anywhere under mydir: mydir/**/*
|
23
|
-
- All files in any directory: **/*
|
24
|
-
- All files in any directory that end in .csv: **/*.csv
|
25
|
-
- All files in any directory that have a .csv extension: **/*.csv*
|
26
|
-
"""
|
27
|
-
|
28
|
-
FILEPATHS = [
|
29
|
-
"a",
|
30
|
-
"a.csv",
|
31
|
-
"a.csv.gz",
|
32
|
-
"a.jsonl",
|
33
|
-
"a/b",
|
34
|
-
"a/b.csv",
|
35
|
-
"a/b.csv.gz",
|
36
|
-
"a/b.jsonl",
|
37
|
-
"a/c",
|
38
|
-
"a/c.csv",
|
39
|
-
"a/c.csv.gz",
|
40
|
-
"a/c.jsonl",
|
41
|
-
"a/b/c",
|
42
|
-
"a/b/c.csv",
|
43
|
-
"a/b/c.csv.gz",
|
44
|
-
"a/b/c.jsonl",
|
45
|
-
"a/c/c",
|
46
|
-
"a/c/c.csv",
|
47
|
-
"a/c/c.csv.gz",
|
48
|
-
"a/c/c.jsonl",
|
49
|
-
"a/b/c/d",
|
50
|
-
"a/b/c/d.csv",
|
51
|
-
"a/b/c/d.csv.gz",
|
52
|
-
"a/b/c/d.jsonl",
|
53
|
-
]
|
54
|
-
FILES = make_remote_files(FILEPATHS)
|
55
|
-
|
56
|
-
DEFAULT_CONFIG = {
|
57
|
-
"streams": [],
|
58
|
-
}
|
59
|
-
|
60
|
-
|
61
|
-
class TestStreamReader(AbstractFileBasedStreamReader):
|
62
|
-
@property
|
63
|
-
def config(self) -> Optional[AbstractFileBasedSpec]:
|
64
|
-
return self._config
|
65
|
-
|
66
|
-
@config.setter
|
67
|
-
def config(self, value: AbstractFileBasedSpec) -> None:
|
68
|
-
self._config = value
|
69
|
-
|
70
|
-
def get_matching_files(self, globs: List[str]) -> Iterable[RemoteFile]:
|
71
|
-
pass
|
72
|
-
|
73
|
-
def open_file(self, file: RemoteFile) -> IOBase:
|
74
|
-
pass
|
75
|
-
|
76
|
-
|
77
|
-
class TestSpec(AbstractFileBasedSpec):
|
78
|
-
@classmethod
|
79
|
-
def documentation_url(cls) -> AnyUrl:
|
80
|
-
return AnyUrl(scheme="https", url="https://docs.airbyte.com/integrations/sources/test") # type: ignore
|
81
|
-
|
82
|
-
|
83
|
-
@pytest.mark.parametrize(
|
84
|
-
"globs,config,expected_matches,expected_path_prefixes",
|
85
|
-
[
|
86
|
-
pytest.param([], DEFAULT_CONFIG, set(), set(), id="no-globs"),
|
87
|
-
pytest.param([""], DEFAULT_CONFIG, set(), set(), id="empty-string"),
|
88
|
-
pytest.param(["**"], DEFAULT_CONFIG, set(FILEPATHS), set(), id="**"),
|
89
|
-
pytest.param(
|
90
|
-
["**/*.csv"], DEFAULT_CONFIG, {"a.csv", "a/b.csv", "a/c.csv", "a/b/c.csv", "a/c/c.csv", "a/b/c/d.csv"}, set(), id="**/*.csv"
|
91
|
-
),
|
92
|
-
pytest.param(
|
93
|
-
["**/*.csv*"],
|
94
|
-
DEFAULT_CONFIG,
|
95
|
-
{
|
96
|
-
"a.csv",
|
97
|
-
"a.csv.gz",
|
98
|
-
"a/b.csv",
|
99
|
-
"a/b.csv.gz",
|
100
|
-
"a/c.csv",
|
101
|
-
"a/c.csv.gz",
|
102
|
-
"a/b/c.csv",
|
103
|
-
"a/b/c.csv.gz",
|
104
|
-
"a/c/c.csv",
|
105
|
-
"a/c/c.csv.gz",
|
106
|
-
"a/b/c/d.csv",
|
107
|
-
"a/b/c/d.csv.gz",
|
108
|
-
},
|
109
|
-
set(),
|
110
|
-
id="**/*.csv*",
|
111
|
-
),
|
112
|
-
pytest.param(["*"], DEFAULT_CONFIG, {"a", "a.csv", "a.csv.gz", "a.jsonl"}, set(), id="*"),
|
113
|
-
pytest.param(["*.csv"], DEFAULT_CONFIG, {"a.csv"}, set(), id="*.csv"),
|
114
|
-
pytest.param(["*.csv*"], DEFAULT_CONFIG, {"a.csv", "a.csv.gz"}, set(), id="*.csv*"),
|
115
|
-
pytest.param(
|
116
|
-
["*/*"],
|
117
|
-
DEFAULT_CONFIG,
|
118
|
-
{"a/b", "a/b.csv", "a/b.csv.gz", "a/b.jsonl", "a/c", "a/c.csv", "a/c.csv.gz", "a/c.jsonl"},
|
119
|
-
set(),
|
120
|
-
id="*/*",
|
121
|
-
),
|
122
|
-
pytest.param(["*/*.csv"], DEFAULT_CONFIG, {"a/b.csv", "a/c.csv"}, set(), id="*/*.csv"),
|
123
|
-
pytest.param(["*/*.csv*"], DEFAULT_CONFIG, {"a/b.csv", "a/b.csv.gz", "a/c.csv", "a/c.csv.gz"}, set(), id="*/*.csv*"),
|
124
|
-
pytest.param(
|
125
|
-
["*/**"],
|
126
|
-
DEFAULT_CONFIG,
|
127
|
-
{
|
128
|
-
"a/b",
|
129
|
-
"a/b.csv",
|
130
|
-
"a/b.csv.gz",
|
131
|
-
"a/b.jsonl",
|
132
|
-
"a/c",
|
133
|
-
"a/c.csv",
|
134
|
-
"a/c.csv.gz",
|
135
|
-
"a/c.jsonl",
|
136
|
-
"a/b/c",
|
137
|
-
"a/b/c.csv",
|
138
|
-
"a/b/c.csv.gz",
|
139
|
-
"a/b/c.jsonl",
|
140
|
-
"a/c/c",
|
141
|
-
"a/c/c.csv",
|
142
|
-
"a/c/c.csv.gz",
|
143
|
-
"a/c/c.jsonl",
|
144
|
-
"a/b/c/d",
|
145
|
-
"a/b/c/d.csv",
|
146
|
-
"a/b/c/d.csv.gz",
|
147
|
-
"a/b/c/d.jsonl",
|
148
|
-
},
|
149
|
-
set(),
|
150
|
-
id="*/**",
|
151
|
-
),
|
152
|
-
pytest.param(
|
153
|
-
["a/*"],
|
154
|
-
DEFAULT_CONFIG,
|
155
|
-
{"a/b", "a/b.csv", "a/b.csv.gz", "a/b.jsonl", "a/c", "a/c.csv", "a/c.csv.gz", "a/c.jsonl"},
|
156
|
-
{"a/"},
|
157
|
-
id="a/*",
|
158
|
-
),
|
159
|
-
pytest.param(["a/*.csv"], DEFAULT_CONFIG, {"a/b.csv", "a/c.csv"}, {"a/"}, id="a/*.csv"),
|
160
|
-
pytest.param(["a/*.csv*"], DEFAULT_CONFIG, {"a/b.csv", "a/b.csv.gz", "a/c.csv", "a/c.csv.gz"}, {"a/"}, id="a/*.csv*"),
|
161
|
-
pytest.param(["a/b/*"], DEFAULT_CONFIG, {"a/b/c", "a/b/c.csv", "a/b/c.csv.gz", "a/b/c.jsonl"}, {"a/b/"}, id="a/b/*"),
|
162
|
-
pytest.param(["a/b/*.csv"], DEFAULT_CONFIG, {"a/b/c.csv"}, {"a/b/"}, id="a/b/*.csv"),
|
163
|
-
pytest.param(["a/b/*.csv*"], DEFAULT_CONFIG, {"a/b/c.csv", "a/b/c.csv.gz"}, {"a/b/"}, id="a/b/*.csv*"),
|
164
|
-
pytest.param(
|
165
|
-
["a/*/*"],
|
166
|
-
DEFAULT_CONFIG,
|
167
|
-
{"a/b/c", "a/b/c.csv", "a/b/c.csv.gz", "a/b/c.jsonl", "a/c/c", "a/c/c.csv", "a/c/c.csv.gz", "a/c/c.jsonl"},
|
168
|
-
{"a/"},
|
169
|
-
id="a/*/*",
|
170
|
-
),
|
171
|
-
pytest.param(["a/*/*.csv"], DEFAULT_CONFIG, {"a/b/c.csv", "a/c/c.csv"}, {"a/"}, id="a/*/*.csv"),
|
172
|
-
pytest.param(["a/*/*.csv*"], DEFAULT_CONFIG, {"a/b/c.csv", "a/b/c.csv.gz", "a/c/c.csv", "a/c/c.csv.gz"}, {"a/"}, id="a/*/*.csv*"),
|
173
|
-
pytest.param(
|
174
|
-
["a/**/*"],
|
175
|
-
DEFAULT_CONFIG,
|
176
|
-
{
|
177
|
-
"a/b",
|
178
|
-
"a/b.csv",
|
179
|
-
"a/b.csv.gz",
|
180
|
-
"a/b.jsonl",
|
181
|
-
"a/c",
|
182
|
-
"a/c.csv",
|
183
|
-
"a/c.csv.gz",
|
184
|
-
"a/c.jsonl",
|
185
|
-
"a/b/c",
|
186
|
-
"a/b/c.csv",
|
187
|
-
"a/b/c.csv.gz",
|
188
|
-
"a/b/c.jsonl",
|
189
|
-
"a/c/c",
|
190
|
-
"a/c/c.csv",
|
191
|
-
"a/c/c.csv.gz",
|
192
|
-
"a/c/c.jsonl",
|
193
|
-
"a/b/c/d",
|
194
|
-
"a/b/c/d.csv",
|
195
|
-
"a/b/c/d.csv.gz",
|
196
|
-
"a/b/c/d.jsonl",
|
197
|
-
},
|
198
|
-
{"a/"},
|
199
|
-
id="a/**/*",
|
200
|
-
),
|
201
|
-
pytest.param(
|
202
|
-
["a/**/*.csv"], DEFAULT_CONFIG, {"a/b.csv", "a/c.csv", "a/b/c.csv", "a/c/c.csv", "a/b/c/d.csv"}, {"a/"}, id="a/**/*.csv"
|
203
|
-
),
|
204
|
-
pytest.param(
|
205
|
-
["a/**/*.csv*"],
|
206
|
-
DEFAULT_CONFIG,
|
207
|
-
{
|
208
|
-
"a/b.csv",
|
209
|
-
"a/b.csv.gz",
|
210
|
-
"a/c.csv",
|
211
|
-
"a/c.csv.gz",
|
212
|
-
"a/b/c.csv",
|
213
|
-
"a/b/c.csv.gz",
|
214
|
-
"a/c/c.csv",
|
215
|
-
"a/c/c.csv.gz",
|
216
|
-
"a/b/c/d.csv",
|
217
|
-
"a/b/c/d.csv.gz",
|
218
|
-
},
|
219
|
-
{"a/"},
|
220
|
-
id="a/**/*.csv*",
|
221
|
-
),
|
222
|
-
pytest.param(
|
223
|
-
["**/*.csv", "**/*.gz"],
|
224
|
-
DEFAULT_CONFIG,
|
225
|
-
{
|
226
|
-
"a.csv",
|
227
|
-
"a.csv.gz",
|
228
|
-
"a/b.csv",
|
229
|
-
"a/b.csv.gz",
|
230
|
-
"a/c.csv",
|
231
|
-
"a/c.csv.gz",
|
232
|
-
"a/b/c.csv",
|
233
|
-
"a/b/c.csv.gz",
|
234
|
-
"a/c/c.csv",
|
235
|
-
"a/c/c.csv.gz",
|
236
|
-
"a/b/c/d.csv",
|
237
|
-
"a/b/c/d.csv.gz",
|
238
|
-
},
|
239
|
-
set(),
|
240
|
-
id="**/*.csv,**/*.gz",
|
241
|
-
),
|
242
|
-
pytest.param(["*.csv", "*.gz"], DEFAULT_CONFIG, {"a.csv", "a.csv.gz"}, set(), id="*.csv,*.gz"),
|
243
|
-
pytest.param(
|
244
|
-
["a/*.csv", "a/*/*.csv"], DEFAULT_CONFIG, {"a/b.csv", "a/c.csv", "a/b/c.csv", "a/c/c.csv"}, {"a/"}, id="a/*.csv,a/*/*.csv"
|
245
|
-
),
|
246
|
-
pytest.param(["a/*.csv", "a/b/*.csv"], DEFAULT_CONFIG, {"a/b.csv", "a/c.csv", "a/b/c.csv"}, {"a/", "a/b/"}, id="a/*.csv,a/b/*.csv"),
|
247
|
-
pytest.param(
|
248
|
-
["**/*.csv"],
|
249
|
-
{"start_date": "2023-06-01T03:54:07.000Z", "streams": []},
|
250
|
-
{"a.csv", "a/b.csv", "a/c.csv", "a/b/c.csv", "a/c/c.csv", "a/b/c/d.csv"},
|
251
|
-
set(),
|
252
|
-
id="all_csvs_modified_after_start_date",
|
253
|
-
),
|
254
|
-
pytest.param(
|
255
|
-
["**/*.csv"], {"start_date": "2023-06-10T03:54:07.000Z", "streams": []}, set(), set(), id="all_csvs_modified_before_start_date"
|
256
|
-
),
|
257
|
-
pytest.param(
|
258
|
-
["**/*.csv"],
|
259
|
-
{"start_date": "2023-06-05T03:54:07.000Z", "streams": []},
|
260
|
-
{"a.csv", "a/b.csv", "a/c.csv", "a/b/c.csv", "a/c/c.csv", "a/b/c/d.csv"},
|
261
|
-
set(),
|
262
|
-
id="all_csvs_modified_exactly_on_start_date",
|
263
|
-
),
|
264
|
-
],
|
265
|
-
)
|
266
|
-
def test_globs_and_prefixes_from_globs(
|
267
|
-
globs: List[str], config: Mapping[str, Any], expected_matches: Set[str], expected_path_prefixes: Set[str]
|
268
|
-
) -> None:
|
269
|
-
reader = TestStreamReader()
|
270
|
-
reader.config = TestSpec(**config)
|
271
|
-
assert set([f.uri for f in reader.filter_files_by_globs_and_start_date(FILES, globs)]) == expected_matches
|
272
|
-
assert set(reader.get_prefixes_from_globs(globs)) == expected_path_prefixes
|