airbyte-cdk 0.72.1__py3-none-any.whl → 6.13.1.dev4106__py3-none-any.whl

Sign up to get free protection for your applications and to get access to all the features.
Files changed (517) hide show
  1. airbyte_cdk/__init__.py +355 -6
  2. airbyte_cdk/cli/__init__.py +1 -0
  3. airbyte_cdk/cli/source_declarative_manifest/__init__.py +5 -0
  4. airbyte_cdk/cli/source_declarative_manifest/_run.py +230 -0
  5. airbyte_cdk/cli/source_declarative_manifest/spec.json +17 -0
  6. airbyte_cdk/config_observation.py +29 -10
  7. airbyte_cdk/connector.py +24 -24
  8. airbyte_cdk/connector_builder/README.md +53 -0
  9. airbyte_cdk/connector_builder/connector_builder_handler.py +37 -11
  10. airbyte_cdk/connector_builder/main.py +45 -13
  11. airbyte_cdk/connector_builder/message_grouper.py +189 -50
  12. airbyte_cdk/connector_builder/models.py +3 -2
  13. airbyte_cdk/destinations/__init__.py +4 -3
  14. airbyte_cdk/destinations/destination.py +54 -20
  15. airbyte_cdk/destinations/vector_db_based/README.md +37 -0
  16. airbyte_cdk/destinations/vector_db_based/config.py +40 -17
  17. airbyte_cdk/destinations/vector_db_based/document_processor.py +56 -17
  18. airbyte_cdk/destinations/vector_db_based/embedder.py +57 -15
  19. airbyte_cdk/destinations/vector_db_based/test_utils.py +14 -4
  20. airbyte_cdk/destinations/vector_db_based/utils.py +8 -2
  21. airbyte_cdk/destinations/vector_db_based/writer.py +24 -5
  22. airbyte_cdk/entrypoint.py +153 -44
  23. airbyte_cdk/exception_handler.py +21 -3
  24. airbyte_cdk/logger.py +30 -44
  25. airbyte_cdk/models/__init__.py +13 -2
  26. airbyte_cdk/models/airbyte_protocol.py +86 -1
  27. airbyte_cdk/models/airbyte_protocol_serializers.py +44 -0
  28. airbyte_cdk/models/file_transfer_record_message.py +13 -0
  29. airbyte_cdk/models/well_known_types.py +1 -1
  30. airbyte_cdk/sources/__init__.py +5 -1
  31. airbyte_cdk/sources/abstract_source.py +125 -79
  32. airbyte_cdk/sources/concurrent_source/__init__.py +7 -2
  33. airbyte_cdk/sources/concurrent_source/concurrent_read_processor.py +102 -36
  34. airbyte_cdk/sources/concurrent_source/concurrent_source.py +29 -36
  35. airbyte_cdk/sources/concurrent_source/concurrent_source_adapter.py +94 -10
  36. airbyte_cdk/sources/concurrent_source/stream_thread_exception.py +25 -0
  37. airbyte_cdk/sources/concurrent_source/thread_pool_manager.py +20 -14
  38. airbyte_cdk/sources/config.py +3 -2
  39. airbyte_cdk/sources/connector_state_manager.py +49 -83
  40. airbyte_cdk/sources/declarative/async_job/job.py +52 -0
  41. airbyte_cdk/sources/declarative/async_job/job_orchestrator.py +497 -0
  42. airbyte_cdk/sources/declarative/async_job/job_tracker.py +75 -0
  43. airbyte_cdk/sources/declarative/async_job/repository.py +35 -0
  44. airbyte_cdk/sources/declarative/async_job/status.py +24 -0
  45. airbyte_cdk/sources/declarative/async_job/timer.py +39 -0
  46. airbyte_cdk/sources/declarative/auth/__init__.py +2 -3
  47. airbyte_cdk/sources/declarative/auth/declarative_authenticator.py +3 -1
  48. airbyte_cdk/sources/declarative/auth/jwt.py +191 -0
  49. airbyte_cdk/sources/declarative/auth/oauth.py +60 -20
  50. airbyte_cdk/sources/declarative/auth/selective_authenticator.py +10 -2
  51. airbyte_cdk/sources/declarative/auth/token.py +28 -10
  52. airbyte_cdk/sources/declarative/auth/token_provider.py +9 -8
  53. airbyte_cdk/sources/declarative/checks/check_stream.py +16 -8
  54. airbyte_cdk/sources/declarative/checks/connection_checker.py +4 -2
  55. airbyte_cdk/sources/declarative/concurrency_level/__init__.py +7 -0
  56. airbyte_cdk/sources/declarative/concurrency_level/concurrency_level.py +50 -0
  57. airbyte_cdk/sources/declarative/concurrent_declarative_source.py +421 -0
  58. airbyte_cdk/sources/declarative/datetime/datetime_parser.py +4 -0
  59. airbyte_cdk/sources/declarative/datetime/min_max_datetime.py +26 -6
  60. airbyte_cdk/sources/declarative/declarative_component_schema.yaml +1185 -85
  61. airbyte_cdk/sources/declarative/declarative_source.py +5 -2
  62. airbyte_cdk/sources/declarative/declarative_stream.py +95 -9
  63. airbyte_cdk/sources/declarative/decoders/__init__.py +23 -2
  64. airbyte_cdk/sources/declarative/decoders/composite_raw_decoder.py +97 -0
  65. airbyte_cdk/sources/declarative/decoders/decoder.py +11 -4
  66. airbyte_cdk/sources/declarative/decoders/json_decoder.py +92 -5
  67. airbyte_cdk/sources/declarative/decoders/noop_decoder.py +21 -0
  68. airbyte_cdk/sources/declarative/decoders/pagination_decoder_decorator.py +39 -0
  69. airbyte_cdk/sources/declarative/decoders/xml_decoder.py +98 -0
  70. airbyte_cdk/sources/declarative/extractors/__init__.py +12 -1
  71. airbyte_cdk/sources/declarative/extractors/dpath_extractor.py +29 -24
  72. airbyte_cdk/sources/declarative/extractors/http_selector.py +4 -5
  73. airbyte_cdk/sources/declarative/extractors/record_extractor.py +2 -3
  74. airbyte_cdk/sources/declarative/extractors/record_filter.py +65 -8
  75. airbyte_cdk/sources/declarative/extractors/record_selector.py +85 -26
  76. airbyte_cdk/sources/declarative/extractors/response_to_file_extractor.py +177 -0
  77. airbyte_cdk/sources/declarative/extractors/type_transformer.py +55 -0
  78. airbyte_cdk/sources/declarative/incremental/__init__.py +25 -3
  79. airbyte_cdk/sources/declarative/incremental/datetime_based_cursor.py +156 -48
  80. airbyte_cdk/sources/declarative/incremental/declarative_cursor.py +13 -0
  81. airbyte_cdk/sources/declarative/incremental/global_substream_cursor.py +350 -0
  82. airbyte_cdk/sources/declarative/incremental/per_partition_cursor.py +159 -74
  83. airbyte_cdk/sources/declarative/incremental/per_partition_with_global.py +200 -0
  84. airbyte_cdk/sources/declarative/incremental/resumable_full_refresh_cursor.py +122 -0
  85. airbyte_cdk/sources/declarative/interpolation/filters.py +27 -1
  86. airbyte_cdk/sources/declarative/interpolation/interpolated_boolean.py +23 -5
  87. airbyte_cdk/sources/declarative/interpolation/interpolated_mapping.py +12 -8
  88. airbyte_cdk/sources/declarative/interpolation/interpolated_nested_mapping.py +13 -6
  89. airbyte_cdk/sources/declarative/interpolation/interpolated_string.py +21 -6
  90. airbyte_cdk/sources/declarative/interpolation/interpolation.py +9 -3
  91. airbyte_cdk/sources/declarative/interpolation/jinja.py +72 -37
  92. airbyte_cdk/sources/declarative/interpolation/macros.py +72 -17
  93. airbyte_cdk/sources/declarative/manifest_declarative_source.py +193 -52
  94. airbyte_cdk/sources/declarative/migrations/legacy_to_per_partition_state_migration.py +98 -0
  95. airbyte_cdk/sources/declarative/migrations/state_migration.py +24 -0
  96. airbyte_cdk/sources/declarative/models/__init__.py +1 -1
  97. airbyte_cdk/sources/declarative/models/declarative_component_schema.py +1319 -603
  98. airbyte_cdk/sources/declarative/parsers/custom_exceptions.py +2 -2
  99. airbyte_cdk/sources/declarative/parsers/manifest_component_transformer.py +26 -4
  100. airbyte_cdk/sources/declarative/parsers/manifest_reference_resolver.py +26 -15
  101. airbyte_cdk/sources/declarative/parsers/model_to_component_factory.py +1695 -225
  102. airbyte_cdk/sources/declarative/partition_routers/__init__.py +24 -4
  103. airbyte_cdk/sources/declarative/partition_routers/async_job_partition_router.py +65 -0
  104. airbyte_cdk/sources/declarative/partition_routers/cartesian_product_stream_slicer.py +176 -0
  105. airbyte_cdk/sources/declarative/partition_routers/list_partition_router.py +39 -9
  106. airbyte_cdk/sources/declarative/partition_routers/partition_router.py +62 -0
  107. airbyte_cdk/sources/declarative/partition_routers/single_partition_router.py +15 -3
  108. airbyte_cdk/sources/declarative/partition_routers/substream_partition_router.py +222 -39
  109. airbyte_cdk/sources/declarative/requesters/error_handlers/__init__.py +19 -5
  110. airbyte_cdk/sources/declarative/requesters/error_handlers/backoff_strategies/__init__.py +3 -1
  111. airbyte_cdk/sources/declarative/requesters/error_handlers/backoff_strategies/constant_backoff_strategy.py +19 -7
  112. airbyte_cdk/sources/declarative/requesters/error_handlers/backoff_strategies/exponential_backoff_strategy.py +19 -7
  113. airbyte_cdk/sources/declarative/requesters/error_handlers/backoff_strategies/header_helper.py +4 -2
  114. airbyte_cdk/sources/declarative/requesters/error_handlers/backoff_strategies/wait_time_from_header_backoff_strategy.py +41 -9
  115. airbyte_cdk/sources/declarative/requesters/error_handlers/backoff_strategies/wait_until_time_from_header_backoff_strategy.py +29 -14
  116. airbyte_cdk/sources/declarative/requesters/error_handlers/backoff_strategy.py +5 -13
  117. airbyte_cdk/sources/declarative/requesters/error_handlers/composite_error_handler.py +32 -16
  118. airbyte_cdk/sources/declarative/requesters/error_handlers/default_error_handler.py +46 -56
  119. airbyte_cdk/sources/declarative/requesters/error_handlers/default_http_response_filter.py +40 -0
  120. airbyte_cdk/sources/declarative/requesters/error_handlers/error_handler.py +6 -32
  121. airbyte_cdk/sources/declarative/requesters/error_handlers/http_response_filter.py +119 -41
  122. airbyte_cdk/sources/declarative/requesters/http_job_repository.py +228 -0
  123. airbyte_cdk/sources/declarative/requesters/http_requester.py +98 -344
  124. airbyte_cdk/sources/declarative/requesters/paginators/__init__.py +14 -3
  125. airbyte_cdk/sources/declarative/requesters/paginators/default_paginator.py +105 -46
  126. airbyte_cdk/sources/declarative/requesters/paginators/no_pagination.py +14 -8
  127. airbyte_cdk/sources/declarative/requesters/paginators/paginator.py +19 -8
  128. airbyte_cdk/sources/declarative/requesters/paginators/strategies/__init__.py +9 -3
  129. airbyte_cdk/sources/declarative/requesters/paginators/strategies/cursor_pagination_strategy.py +53 -21
  130. airbyte_cdk/sources/declarative/requesters/paginators/strategies/offset_increment.py +42 -19
  131. airbyte_cdk/sources/declarative/requesters/paginators/strategies/page_increment.py +25 -12
  132. airbyte_cdk/sources/declarative/requesters/paginators/strategies/pagination_strategy.py +13 -10
  133. airbyte_cdk/sources/declarative/requesters/paginators/strategies/stop_condition.py +26 -13
  134. airbyte_cdk/sources/declarative/requesters/request_options/__init__.py +15 -2
  135. airbyte_cdk/sources/declarative/requesters/request_options/datetime_based_request_options_provider.py +91 -0
  136. airbyte_cdk/sources/declarative/requesters/request_options/default_request_options_provider.py +60 -0
  137. airbyte_cdk/sources/declarative/requesters/request_options/interpolated_nested_request_input_provider.py +31 -14
  138. airbyte_cdk/sources/declarative/requesters/request_options/interpolated_request_input_provider.py +27 -15
  139. airbyte_cdk/sources/declarative/requesters/request_options/interpolated_request_options_provider.py +63 -10
  140. airbyte_cdk/sources/declarative/requesters/request_options/request_options_provider.py +1 -1
  141. airbyte_cdk/sources/declarative/requesters/requester.py +9 -17
  142. airbyte_cdk/sources/declarative/resolvers/__init__.py +41 -0
  143. airbyte_cdk/sources/declarative/resolvers/components_resolver.py +55 -0
  144. airbyte_cdk/sources/declarative/resolvers/config_components_resolver.py +136 -0
  145. airbyte_cdk/sources/declarative/resolvers/http_components_resolver.py +112 -0
  146. airbyte_cdk/sources/declarative/retrievers/__init__.py +6 -2
  147. airbyte_cdk/sources/declarative/retrievers/async_retriever.py +100 -0
  148. airbyte_cdk/sources/declarative/retrievers/retriever.py +1 -3
  149. airbyte_cdk/sources/declarative/retrievers/simple_retriever.py +228 -72
  150. airbyte_cdk/sources/declarative/schema/__init__.py +14 -1
  151. airbyte_cdk/sources/declarative/schema/default_schema_loader.py +5 -3
  152. airbyte_cdk/sources/declarative/schema/dynamic_schema_loader.py +236 -0
  153. airbyte_cdk/sources/declarative/schema/json_file_schema_loader.py +8 -8
  154. airbyte_cdk/sources/declarative/spec/spec.py +12 -5
  155. airbyte_cdk/sources/declarative/stream_slicers/__init__.py +1 -2
  156. airbyte_cdk/sources/declarative/stream_slicers/declarative_partition_generator.py +88 -0
  157. airbyte_cdk/sources/declarative/stream_slicers/stream_slicer.py +9 -14
  158. airbyte_cdk/sources/declarative/transformations/add_fields.py +19 -11
  159. airbyte_cdk/sources/declarative/transformations/flatten_fields.py +52 -0
  160. airbyte_cdk/sources/declarative/transformations/keys_replace_transformation.py +61 -0
  161. airbyte_cdk/sources/declarative/transformations/keys_to_lower_transformation.py +22 -0
  162. airbyte_cdk/sources/declarative/transformations/keys_to_snake_transformation.py +68 -0
  163. airbyte_cdk/sources/declarative/transformations/remove_fields.py +13 -10
  164. airbyte_cdk/sources/declarative/transformations/transformation.py +5 -5
  165. airbyte_cdk/sources/declarative/types.py +19 -110
  166. airbyte_cdk/sources/declarative/yaml_declarative_source.py +31 -10
  167. airbyte_cdk/sources/embedded/base_integration.py +16 -5
  168. airbyte_cdk/sources/embedded/catalog.py +16 -4
  169. airbyte_cdk/sources/embedded/runner.py +19 -3
  170. airbyte_cdk/sources/embedded/tools.py +5 -2
  171. airbyte_cdk/sources/file_based/README.md +152 -0
  172. airbyte_cdk/sources/file_based/__init__.py +24 -0
  173. airbyte_cdk/sources/file_based/availability_strategy/__init__.py +9 -2
  174. airbyte_cdk/sources/file_based/availability_strategy/abstract_file_based_availability_strategy.py +22 -6
  175. airbyte_cdk/sources/file_based/availability_strategy/default_file_based_availability_strategy.py +46 -10
  176. airbyte_cdk/sources/file_based/config/abstract_file_based_spec.py +58 -10
  177. airbyte_cdk/sources/file_based/config/avro_format.py +2 -1
  178. airbyte_cdk/sources/file_based/config/csv_format.py +29 -10
  179. airbyte_cdk/sources/file_based/config/excel_format.py +18 -0
  180. airbyte_cdk/sources/file_based/config/file_based_stream_config.py +16 -4
  181. airbyte_cdk/sources/file_based/config/jsonl_format.py +2 -1
  182. airbyte_cdk/sources/file_based/config/parquet_format.py +2 -1
  183. airbyte_cdk/sources/file_based/config/unstructured_format.py +13 -5
  184. airbyte_cdk/sources/file_based/discovery_policy/__init__.py +6 -2
  185. airbyte_cdk/sources/file_based/discovery_policy/abstract_discovery_policy.py +2 -4
  186. airbyte_cdk/sources/file_based/discovery_policy/default_discovery_policy.py +7 -2
  187. airbyte_cdk/sources/file_based/exceptions.py +52 -15
  188. airbyte_cdk/sources/file_based/file_based_source.py +163 -33
  189. airbyte_cdk/sources/file_based/file_based_stream_reader.py +83 -5
  190. airbyte_cdk/sources/file_based/file_types/__init__.py +14 -1
  191. airbyte_cdk/sources/file_based/file_types/avro_parser.py +75 -24
  192. airbyte_cdk/sources/file_based/file_types/csv_parser.py +116 -34
  193. airbyte_cdk/sources/file_based/file_types/excel_parser.py +196 -0
  194. airbyte_cdk/sources/file_based/file_types/file_transfer.py +37 -0
  195. airbyte_cdk/sources/file_based/file_types/file_type_parser.py +4 -1
  196. airbyte_cdk/sources/file_based/file_types/jsonl_parser.py +24 -8
  197. airbyte_cdk/sources/file_based/file_types/parquet_parser.py +60 -18
  198. airbyte_cdk/sources/file_based/file_types/unstructured_parser.py +145 -41
  199. airbyte_cdk/sources/file_based/remote_file.py +1 -1
  200. airbyte_cdk/sources/file_based/schema_helpers.py +38 -10
  201. airbyte_cdk/sources/file_based/schema_validation_policies/__init__.py +3 -1
  202. airbyte_cdk/sources/file_based/schema_validation_policies/abstract_schema_validation_policy.py +3 -1
  203. airbyte_cdk/sources/file_based/schema_validation_policies/default_schema_validation_policies.py +16 -5
  204. airbyte_cdk/sources/file_based/stream/abstract_file_based_stream.py +50 -13
  205. airbyte_cdk/sources/file_based/stream/concurrent/adapters.py +67 -27
  206. airbyte_cdk/sources/file_based/stream/concurrent/cursor/__init__.py +5 -1
  207. airbyte_cdk/sources/file_based/stream/concurrent/cursor/abstract_concurrent_file_based_cursor.py +14 -23
  208. airbyte_cdk/sources/file_based/stream/concurrent/cursor/file_based_concurrent_cursor.py +54 -18
  209. airbyte_cdk/sources/file_based/stream/concurrent/cursor/file_based_final_state_cursor.py +21 -9
  210. airbyte_cdk/sources/file_based/stream/cursor/abstract_file_based_cursor.py +3 -1
  211. airbyte_cdk/sources/file_based/stream/cursor/default_file_based_cursor.py +27 -10
  212. airbyte_cdk/sources/file_based/stream/default_file_based_stream.py +175 -45
  213. airbyte_cdk/sources/http_logger.py +8 -3
  214. airbyte_cdk/sources/message/__init__.py +7 -1
  215. airbyte_cdk/sources/message/repository.py +18 -4
  216. airbyte_cdk/sources/source.py +42 -38
  217. airbyte_cdk/sources/streams/__init__.py +2 -2
  218. airbyte_cdk/sources/streams/availability_strategy.py +54 -3
  219. airbyte_cdk/sources/streams/call_rate.py +64 -21
  220. airbyte_cdk/sources/streams/checkpoint/__init__.py +26 -0
  221. airbyte_cdk/sources/streams/checkpoint/checkpoint_reader.py +335 -0
  222. airbyte_cdk/sources/{declarative/incremental → streams/checkpoint}/cursor.py +17 -14
  223. airbyte_cdk/sources/streams/checkpoint/per_partition_key_serializer.py +22 -0
  224. airbyte_cdk/sources/streams/checkpoint/resumable_full_refresh_cursor.py +51 -0
  225. airbyte_cdk/sources/streams/checkpoint/substream_resumable_full_refresh_cursor.py +110 -0
  226. airbyte_cdk/sources/streams/concurrent/README.md +7 -0
  227. airbyte_cdk/sources/streams/concurrent/abstract_stream.py +7 -2
  228. airbyte_cdk/sources/streams/concurrent/adapters.py +84 -75
  229. airbyte_cdk/sources/streams/concurrent/availability_strategy.py +30 -2
  230. airbyte_cdk/sources/streams/concurrent/cursor.py +298 -42
  231. airbyte_cdk/sources/streams/concurrent/default_stream.py +12 -3
  232. airbyte_cdk/sources/streams/concurrent/exceptions.py +3 -0
  233. airbyte_cdk/sources/streams/concurrent/helpers.py +14 -3
  234. airbyte_cdk/sources/streams/concurrent/partition_enqueuer.py +12 -3
  235. airbyte_cdk/sources/streams/concurrent/partition_reader.py +10 -3
  236. airbyte_cdk/sources/streams/concurrent/partitions/partition.py +1 -16
  237. airbyte_cdk/sources/streams/concurrent/partitions/stream_slicer.py +21 -0
  238. airbyte_cdk/sources/streams/concurrent/partitions/types.py +15 -5
  239. airbyte_cdk/sources/streams/concurrent/state_converters/abstract_stream_state_converter.py +109 -17
  240. airbyte_cdk/sources/streams/concurrent/state_converters/datetime_stream_state_converter.py +90 -72
  241. airbyte_cdk/sources/streams/core.py +412 -87
  242. airbyte_cdk/sources/streams/http/__init__.py +2 -1
  243. airbyte_cdk/sources/streams/http/availability_strategy.py +12 -101
  244. airbyte_cdk/sources/streams/http/error_handlers/__init__.py +22 -0
  245. airbyte_cdk/sources/streams/http/error_handlers/backoff_strategy.py +28 -0
  246. airbyte_cdk/sources/streams/http/error_handlers/default_backoff_strategy.py +17 -0
  247. airbyte_cdk/sources/streams/http/error_handlers/default_error_mapping.py +86 -0
  248. airbyte_cdk/sources/streams/http/error_handlers/error_handler.py +42 -0
  249. airbyte_cdk/sources/streams/http/error_handlers/error_message_parser.py +19 -0
  250. airbyte_cdk/sources/streams/http/error_handlers/http_status_error_handler.py +110 -0
  251. airbyte_cdk/sources/streams/http/error_handlers/json_error_message_parser.py +52 -0
  252. airbyte_cdk/sources/streams/http/error_handlers/response_models.py +65 -0
  253. airbyte_cdk/sources/streams/http/exceptions.py +27 -7
  254. airbyte_cdk/sources/streams/http/http.py +369 -246
  255. airbyte_cdk/sources/streams/http/http_client.py +531 -0
  256. airbyte_cdk/sources/streams/http/rate_limiting.py +76 -12
  257. airbyte_cdk/sources/streams/http/requests_native_auth/abstract_oauth.py +28 -9
  258. airbyte_cdk/sources/streams/http/requests_native_auth/abstract_token.py +2 -1
  259. airbyte_cdk/sources/streams/http/requests_native_auth/oauth.py +90 -35
  260. airbyte_cdk/sources/streams/http/requests_native_auth/token.py +13 -3
  261. airbyte_cdk/sources/types.py +154 -0
  262. airbyte_cdk/sources/utils/record_helper.py +36 -21
  263. airbyte_cdk/sources/utils/schema_helpers.py +13 -6
  264. airbyte_cdk/sources/utils/slice_logger.py +4 -1
  265. airbyte_cdk/sources/utils/transform.py +54 -20
  266. airbyte_cdk/sql/_util/hashing.py +34 -0
  267. airbyte_cdk/sql/_util/name_normalizers.py +92 -0
  268. airbyte_cdk/sql/constants.py +32 -0
  269. airbyte_cdk/sql/exceptions.py +235 -0
  270. airbyte_cdk/sql/secrets.py +123 -0
  271. airbyte_cdk/sql/shared/__init__.py +15 -0
  272. airbyte_cdk/sql/shared/catalog_providers.py +145 -0
  273. airbyte_cdk/sql/shared/sql_processor.py +786 -0
  274. airbyte_cdk/sql/types.py +160 -0
  275. airbyte_cdk/test/catalog_builder.py +70 -18
  276. airbyte_cdk/test/entrypoint_wrapper.py +117 -42
  277. airbyte_cdk/test/mock_http/__init__.py +1 -1
  278. airbyte_cdk/test/mock_http/matcher.py +6 -0
  279. airbyte_cdk/test/mock_http/mocker.py +57 -10
  280. airbyte_cdk/test/mock_http/request.py +19 -3
  281. airbyte_cdk/test/mock_http/response.py +3 -1
  282. airbyte_cdk/test/mock_http/response_builder.py +32 -16
  283. airbyte_cdk/test/state_builder.py +18 -10
  284. airbyte_cdk/test/utils/__init__.py +1 -0
  285. airbyte_cdk/test/utils/data.py +24 -0
  286. airbyte_cdk/test/utils/http_mocking.py +16 -0
  287. airbyte_cdk/test/utils/manifest_only_fixtures.py +60 -0
  288. airbyte_cdk/test/utils/reading.py +26 -0
  289. airbyte_cdk/utils/__init__.py +2 -1
  290. airbyte_cdk/utils/airbyte_secrets_utils.py +5 -3
  291. airbyte_cdk/utils/analytics_message.py +10 -2
  292. airbyte_cdk/utils/datetime_format_inferrer.py +4 -1
  293. airbyte_cdk/utils/event_timing.py +10 -10
  294. airbyte_cdk/utils/mapping_helpers.py +3 -1
  295. airbyte_cdk/utils/message_utils.py +20 -11
  296. airbyte_cdk/utils/print_buffer.py +75 -0
  297. airbyte_cdk/utils/schema_inferrer.py +198 -28
  298. airbyte_cdk/utils/slice_hasher.py +30 -0
  299. airbyte_cdk/utils/spec_schema_transformations.py +6 -3
  300. airbyte_cdk/utils/stream_status_utils.py +8 -1
  301. airbyte_cdk/utils/traced_exception.py +61 -21
  302. airbyte_cdk-6.13.1.dev4106.dist-info/METADATA +109 -0
  303. airbyte_cdk-6.13.1.dev4106.dist-info/RECORD +349 -0
  304. {airbyte_cdk-0.72.1.dist-info → airbyte_cdk-6.13.1.dev4106.dist-info}/WHEEL +1 -2
  305. airbyte_cdk-6.13.1.dev4106.dist-info/entry_points.txt +3 -0
  306. airbyte_cdk/sources/declarative/create_partial.py +0 -92
  307. airbyte_cdk/sources/declarative/parsers/class_types_registry.py +0 -102
  308. airbyte_cdk/sources/declarative/parsers/default_implementation_registry.py +0 -64
  309. airbyte_cdk/sources/declarative/requesters/error_handlers/response_action.py +0 -16
  310. airbyte_cdk/sources/declarative/requesters/error_handlers/response_status.py +0 -68
  311. airbyte_cdk/sources/declarative/stream_slicers/cartesian_product_stream_slicer.py +0 -114
  312. airbyte_cdk/sources/deprecated/base_source.py +0 -94
  313. airbyte_cdk/sources/deprecated/client.py +0 -99
  314. airbyte_cdk/sources/singer/__init__.py +0 -8
  315. airbyte_cdk/sources/singer/singer_helpers.py +0 -304
  316. airbyte_cdk/sources/singer/source.py +0 -186
  317. airbyte_cdk/sources/streams/concurrent/partitions/record.py +0 -23
  318. airbyte_cdk/sources/streams/http/auth/__init__.py +0 -17
  319. airbyte_cdk/sources/streams/http/auth/core.py +0 -29
  320. airbyte_cdk/sources/streams/http/auth/oauth.py +0 -113
  321. airbyte_cdk/sources/streams/http/auth/token.py +0 -47
  322. airbyte_cdk/sources/streams/utils/stream_helper.py +0 -40
  323. airbyte_cdk/sources/utils/catalog_helpers.py +0 -22
  324. airbyte_cdk/sources/utils/schema_models.py +0 -84
  325. airbyte_cdk-0.72.1.dist-info/METADATA +0 -243
  326. airbyte_cdk-0.72.1.dist-info/RECORD +0 -466
  327. airbyte_cdk-0.72.1.dist-info/top_level.txt +0 -3
  328. source_declarative_manifest/main.py +0 -29
  329. unit_tests/connector_builder/__init__.py +0 -3
  330. unit_tests/connector_builder/test_connector_builder_handler.py +0 -871
  331. unit_tests/connector_builder/test_message_grouper.py +0 -713
  332. unit_tests/connector_builder/utils.py +0 -27
  333. unit_tests/destinations/test_destination.py +0 -243
  334. unit_tests/singer/test_singer_helpers.py +0 -56
  335. unit_tests/singer/test_singer_source.py +0 -112
  336. unit_tests/sources/__init__.py +0 -0
  337. unit_tests/sources/concurrent_source/__init__.py +0 -3
  338. unit_tests/sources/concurrent_source/test_concurrent_source_adapter.py +0 -106
  339. unit_tests/sources/declarative/__init__.py +0 -3
  340. unit_tests/sources/declarative/auth/__init__.py +0 -3
  341. unit_tests/sources/declarative/auth/test_oauth.py +0 -331
  342. unit_tests/sources/declarative/auth/test_selective_authenticator.py +0 -39
  343. unit_tests/sources/declarative/auth/test_session_token_auth.py +0 -182
  344. unit_tests/sources/declarative/auth/test_token_auth.py +0 -200
  345. unit_tests/sources/declarative/auth/test_token_provider.py +0 -73
  346. unit_tests/sources/declarative/checks/__init__.py +0 -3
  347. unit_tests/sources/declarative/checks/test_check_stream.py +0 -146
  348. unit_tests/sources/declarative/decoders/__init__.py +0 -0
  349. unit_tests/sources/declarative/decoders/test_json_decoder.py +0 -16
  350. unit_tests/sources/declarative/external_component.py +0 -13
  351. unit_tests/sources/declarative/extractors/__init__.py +0 -3
  352. unit_tests/sources/declarative/extractors/test_dpath_extractor.py +0 -55
  353. unit_tests/sources/declarative/extractors/test_record_filter.py +0 -55
  354. unit_tests/sources/declarative/extractors/test_record_selector.py +0 -179
  355. unit_tests/sources/declarative/incremental/__init__.py +0 -0
  356. unit_tests/sources/declarative/incremental/test_datetime_based_cursor.py +0 -860
  357. unit_tests/sources/declarative/incremental/test_per_partition_cursor.py +0 -406
  358. unit_tests/sources/declarative/incremental/test_per_partition_cursor_integration.py +0 -332
  359. unit_tests/sources/declarative/interpolation/__init__.py +0 -3
  360. unit_tests/sources/declarative/interpolation/test_filters.py +0 -80
  361. unit_tests/sources/declarative/interpolation/test_interpolated_boolean.py +0 -40
  362. unit_tests/sources/declarative/interpolation/test_interpolated_mapping.py +0 -35
  363. unit_tests/sources/declarative/interpolation/test_interpolated_nested_mapping.py +0 -45
  364. unit_tests/sources/declarative/interpolation/test_interpolated_string.py +0 -25
  365. unit_tests/sources/declarative/interpolation/test_jinja.py +0 -240
  366. unit_tests/sources/declarative/interpolation/test_macros.py +0 -73
  367. unit_tests/sources/declarative/parsers/__init__.py +0 -3
  368. unit_tests/sources/declarative/parsers/test_manifest_component_transformer.py +0 -406
  369. unit_tests/sources/declarative/parsers/test_manifest_reference_resolver.py +0 -139
  370. unit_tests/sources/declarative/parsers/test_model_to_component_factory.py +0 -1847
  371. unit_tests/sources/declarative/parsers/testing_components.py +0 -36
  372. unit_tests/sources/declarative/partition_routers/__init__.py +0 -3
  373. unit_tests/sources/declarative/partition_routers/test_list_partition_router.py +0 -155
  374. unit_tests/sources/declarative/partition_routers/test_single_partition_router.py +0 -14
  375. unit_tests/sources/declarative/partition_routers/test_substream_partition_router.py +0 -404
  376. unit_tests/sources/declarative/requesters/__init__.py +0 -3
  377. unit_tests/sources/declarative/requesters/error_handlers/__init__.py +0 -3
  378. unit_tests/sources/declarative/requesters/error_handlers/backoff_strategies/__init__.py +0 -3
  379. unit_tests/sources/declarative/requesters/error_handlers/backoff_strategies/test_constant_backoff.py +0 -34
  380. unit_tests/sources/declarative/requesters/error_handlers/backoff_strategies/test_exponential_backoff.py +0 -36
  381. unit_tests/sources/declarative/requesters/error_handlers/backoff_strategies/test_header_helper.py +0 -38
  382. unit_tests/sources/declarative/requesters/error_handlers/backoff_strategies/test_wait_time_from_header.py +0 -35
  383. unit_tests/sources/declarative/requesters/error_handlers/backoff_strategies/test_wait_until_time_from_header.py +0 -64
  384. unit_tests/sources/declarative/requesters/error_handlers/test_composite_error_handler.py +0 -213
  385. unit_tests/sources/declarative/requesters/error_handlers/test_default_error_handler.py +0 -178
  386. unit_tests/sources/declarative/requesters/error_handlers/test_http_response_filter.py +0 -121
  387. unit_tests/sources/declarative/requesters/error_handlers/test_response_status.py +0 -44
  388. unit_tests/sources/declarative/requesters/paginators/__init__.py +0 -3
  389. unit_tests/sources/declarative/requesters/paginators/test_cursor_pagination_strategy.py +0 -64
  390. unit_tests/sources/declarative/requesters/paginators/test_default_paginator.py +0 -313
  391. unit_tests/sources/declarative/requesters/paginators/test_no_paginator.py +0 -12
  392. unit_tests/sources/declarative/requesters/paginators/test_offset_increment.py +0 -58
  393. unit_tests/sources/declarative/requesters/paginators/test_page_increment.py +0 -70
  394. unit_tests/sources/declarative/requesters/paginators/test_request_option.py +0 -43
  395. unit_tests/sources/declarative/requesters/paginators/test_stop_condition.py +0 -105
  396. unit_tests/sources/declarative/requesters/request_options/__init__.py +0 -3
  397. unit_tests/sources/declarative/requesters/request_options/test_interpolated_request_options_provider.py +0 -101
  398. unit_tests/sources/declarative/requesters/test_http_requester.py +0 -974
  399. unit_tests/sources/declarative/requesters/test_interpolated_request_input_provider.py +0 -32
  400. unit_tests/sources/declarative/retrievers/__init__.py +0 -3
  401. unit_tests/sources/declarative/retrievers/test_simple_retriever.py +0 -542
  402. unit_tests/sources/declarative/schema/__init__.py +0 -6
  403. unit_tests/sources/declarative/schema/source_test/SourceTest.py +0 -8
  404. unit_tests/sources/declarative/schema/source_test/__init__.py +0 -3
  405. unit_tests/sources/declarative/schema/test_default_schema_loader.py +0 -32
  406. unit_tests/sources/declarative/schema/test_inline_schema_loader.py +0 -19
  407. unit_tests/sources/declarative/schema/test_json_file_schema_loader.py +0 -26
  408. unit_tests/sources/declarative/states/__init__.py +0 -3
  409. unit_tests/sources/declarative/stream_slicers/__init__.py +0 -3
  410. unit_tests/sources/declarative/stream_slicers/test_cartesian_product_stream_slicer.py +0 -225
  411. unit_tests/sources/declarative/test_create_partial.py +0 -83
  412. unit_tests/sources/declarative/test_declarative_stream.py +0 -103
  413. unit_tests/sources/declarative/test_manifest_declarative_source.py +0 -1260
  414. unit_tests/sources/declarative/test_types.py +0 -39
  415. unit_tests/sources/declarative/test_yaml_declarative_source.py +0 -148
  416. unit_tests/sources/file_based/__init__.py +0 -0
  417. unit_tests/sources/file_based/availability_strategy/__init__.py +0 -0
  418. unit_tests/sources/file_based/availability_strategy/test_default_file_based_availability_strategy.py +0 -100
  419. unit_tests/sources/file_based/config/__init__.py +0 -0
  420. unit_tests/sources/file_based/config/test_abstract_file_based_spec.py +0 -28
  421. unit_tests/sources/file_based/config/test_csv_format.py +0 -34
  422. unit_tests/sources/file_based/config/test_file_based_stream_config.py +0 -84
  423. unit_tests/sources/file_based/discovery_policy/__init__.py +0 -0
  424. unit_tests/sources/file_based/discovery_policy/test_default_discovery_policy.py +0 -31
  425. unit_tests/sources/file_based/file_types/__init__.py +0 -0
  426. unit_tests/sources/file_based/file_types/test_avro_parser.py +0 -243
  427. unit_tests/sources/file_based/file_types/test_csv_parser.py +0 -546
  428. unit_tests/sources/file_based/file_types/test_jsonl_parser.py +0 -158
  429. unit_tests/sources/file_based/file_types/test_parquet_parser.py +0 -274
  430. unit_tests/sources/file_based/file_types/test_unstructured_parser.py +0 -593
  431. unit_tests/sources/file_based/helpers.py +0 -70
  432. unit_tests/sources/file_based/in_memory_files_source.py +0 -211
  433. unit_tests/sources/file_based/scenarios/__init__.py +0 -0
  434. unit_tests/sources/file_based/scenarios/avro_scenarios.py +0 -744
  435. unit_tests/sources/file_based/scenarios/check_scenarios.py +0 -220
  436. unit_tests/sources/file_based/scenarios/concurrent_incremental_scenarios.py +0 -2844
  437. unit_tests/sources/file_based/scenarios/csv_scenarios.py +0 -3105
  438. unit_tests/sources/file_based/scenarios/file_based_source_builder.py +0 -91
  439. unit_tests/sources/file_based/scenarios/incremental_scenarios.py +0 -1926
  440. unit_tests/sources/file_based/scenarios/jsonl_scenarios.py +0 -930
  441. unit_tests/sources/file_based/scenarios/parquet_scenarios.py +0 -754
  442. unit_tests/sources/file_based/scenarios/scenario_builder.py +0 -234
  443. unit_tests/sources/file_based/scenarios/unstructured_scenarios.py +0 -608
  444. unit_tests/sources/file_based/scenarios/user_input_schema_scenarios.py +0 -746
  445. unit_tests/sources/file_based/scenarios/validation_policy_scenarios.py +0 -726
  446. unit_tests/sources/file_based/stream/__init__.py +0 -0
  447. unit_tests/sources/file_based/stream/concurrent/__init__.py +0 -0
  448. unit_tests/sources/file_based/stream/concurrent/test_adapters.py +0 -362
  449. unit_tests/sources/file_based/stream/concurrent/test_file_based_concurrent_cursor.py +0 -458
  450. unit_tests/sources/file_based/stream/test_default_file_based_cursor.py +0 -310
  451. unit_tests/sources/file_based/stream/test_default_file_based_stream.py +0 -244
  452. unit_tests/sources/file_based/test_file_based_scenarios.py +0 -320
  453. unit_tests/sources/file_based/test_file_based_stream_reader.py +0 -272
  454. unit_tests/sources/file_based/test_scenarios.py +0 -253
  455. unit_tests/sources/file_based/test_schema_helpers.py +0 -346
  456. unit_tests/sources/fixtures/__init__.py +0 -3
  457. unit_tests/sources/fixtures/source_test_fixture.py +0 -153
  458. unit_tests/sources/message/__init__.py +0 -0
  459. unit_tests/sources/message/test_repository.py +0 -153
  460. unit_tests/sources/streams/__init__.py +0 -0
  461. unit_tests/sources/streams/concurrent/__init__.py +0 -3
  462. unit_tests/sources/streams/concurrent/scenarios/__init__.py +0 -3
  463. unit_tests/sources/streams/concurrent/scenarios/incremental_scenarios.py +0 -250
  464. unit_tests/sources/streams/concurrent/scenarios/stream_facade_builder.py +0 -140
  465. unit_tests/sources/streams/concurrent/scenarios/stream_facade_scenarios.py +0 -452
  466. unit_tests/sources/streams/concurrent/scenarios/test_concurrent_scenarios.py +0 -76
  467. unit_tests/sources/streams/concurrent/scenarios/thread_based_concurrent_stream_scenarios.py +0 -418
  468. unit_tests/sources/streams/concurrent/scenarios/thread_based_concurrent_stream_source_builder.py +0 -142
  469. unit_tests/sources/streams/concurrent/scenarios/utils.py +0 -55
  470. unit_tests/sources/streams/concurrent/test_adapters.py +0 -380
  471. unit_tests/sources/streams/concurrent/test_concurrent_read_processor.py +0 -684
  472. unit_tests/sources/streams/concurrent/test_cursor.py +0 -139
  473. unit_tests/sources/streams/concurrent/test_datetime_state_converter.py +0 -369
  474. unit_tests/sources/streams/concurrent/test_default_stream.py +0 -197
  475. unit_tests/sources/streams/concurrent/test_partition_enqueuer.py +0 -90
  476. unit_tests/sources/streams/concurrent/test_partition_reader.py +0 -67
  477. unit_tests/sources/streams/concurrent/test_thread_pool_manager.py +0 -106
  478. unit_tests/sources/streams/http/__init__.py +0 -0
  479. unit_tests/sources/streams/http/auth/__init__.py +0 -0
  480. unit_tests/sources/streams/http/auth/test_auth.py +0 -173
  481. unit_tests/sources/streams/http/requests_native_auth/__init__.py +0 -0
  482. unit_tests/sources/streams/http/requests_native_auth/test_requests_native_auth.py +0 -423
  483. unit_tests/sources/streams/http/test_availability_strategy.py +0 -180
  484. unit_tests/sources/streams/http/test_http.py +0 -635
  485. unit_tests/sources/streams/test_availability_strategy.py +0 -70
  486. unit_tests/sources/streams/test_call_rate.py +0 -300
  487. unit_tests/sources/streams/test_stream_read.py +0 -405
  488. unit_tests/sources/streams/test_streams_core.py +0 -184
  489. unit_tests/sources/test_abstract_source.py +0 -1442
  490. unit_tests/sources/test_concurrent_source.py +0 -112
  491. unit_tests/sources/test_config.py +0 -92
  492. unit_tests/sources/test_connector_state_manager.py +0 -482
  493. unit_tests/sources/test_http_logger.py +0 -252
  494. unit_tests/sources/test_integration_source.py +0 -86
  495. unit_tests/sources/test_source.py +0 -684
  496. unit_tests/sources/test_source_read.py +0 -460
  497. unit_tests/test/__init__.py +0 -0
  498. unit_tests/test/mock_http/__init__.py +0 -0
  499. unit_tests/test/mock_http/test_matcher.py +0 -53
  500. unit_tests/test/mock_http/test_mocker.py +0 -214
  501. unit_tests/test/mock_http/test_request.py +0 -117
  502. unit_tests/test/mock_http/test_response_builder.py +0 -177
  503. unit_tests/test/test_entrypoint_wrapper.py +0 -240
  504. unit_tests/utils/__init__.py +0 -0
  505. unit_tests/utils/test_datetime_format_inferrer.py +0 -60
  506. unit_tests/utils/test_mapping_helpers.py +0 -54
  507. unit_tests/utils/test_message_utils.py +0 -91
  508. unit_tests/utils/test_rate_limiting.py +0 -26
  509. unit_tests/utils/test_schema_inferrer.py +0 -202
  510. unit_tests/utils/test_secret_utils.py +0 -135
  511. unit_tests/utils/test_stream_status_utils.py +0 -61
  512. unit_tests/utils/test_traced_exception.py +0 -107
  513. /airbyte_cdk/sources/{deprecated → declarative/async_job}/__init__.py +0 -0
  514. {source_declarative_manifest → airbyte_cdk/sources/declarative/migrations}/__init__.py +0 -0
  515. {unit_tests/destinations → airbyte_cdk/sql}/__init__.py +0 -0
  516. {unit_tests/singer → airbyte_cdk/sql/_util}/__init__.py +0 -0
  517. {airbyte_cdk-0.72.1.dist-info → airbyte_cdk-6.13.1.dev4106.dist-info}/LICENSE.txt +0 -0
@@ -1,684 +0,0 @@
1
- #
2
- # Copyright (c) 2023 Airbyte, Inc., all rights reserved.
3
- #
4
- import logging
5
- import unittest
6
- from unittest.mock import Mock, call
7
-
8
- import freezegun
9
- from airbyte_cdk.models import (
10
- AirbyteLogMessage,
11
- AirbyteMessage,
12
- AirbyteRecordMessage,
13
- AirbyteStream,
14
- AirbyteStreamStatus,
15
- AirbyteStreamStatusTraceMessage,
16
- AirbyteTraceMessage,
17
- )
18
- from airbyte_cdk.models import Level as LogLevel
19
- from airbyte_cdk.models import StreamDescriptor, SyncMode, TraceType
20
- from airbyte_cdk.models import Type as MessageType
21
- from airbyte_cdk.sources.concurrent_source.concurrent_read_processor import ConcurrentReadProcessor
22
- from airbyte_cdk.sources.concurrent_source.partition_generation_completed_sentinel import PartitionGenerationCompletedSentinel
23
- from airbyte_cdk.sources.concurrent_source.thread_pool_manager import ThreadPoolManager
24
- from airbyte_cdk.sources.message import LogMessage, MessageRepository
25
- from airbyte_cdk.sources.streams.concurrent.abstract_stream import AbstractStream
26
- from airbyte_cdk.sources.streams.concurrent.partition_enqueuer import PartitionEnqueuer
27
- from airbyte_cdk.sources.streams.concurrent.partition_reader import PartitionReader
28
- from airbyte_cdk.sources.streams.concurrent.partitions.partition import Partition
29
- from airbyte_cdk.sources.streams.concurrent.partitions.record import Record
30
- from airbyte_cdk.sources.streams.concurrent.partitions.types import PartitionCompleteSentinel
31
- from airbyte_cdk.sources.utils.slice_logger import SliceLogger
32
-
33
- _STREAM_NAME = "stream"
34
- _ANOTHER_STREAM_NAME = "stream2"
35
- _ANY_AIRBYTE_MESSAGE = Mock(spec=AirbyteMessage)
36
-
37
-
38
- class TestConcurrentReadProcessor(unittest.TestCase):
39
- def setUp(self):
40
- self._partition_enqueuer = Mock(spec=PartitionEnqueuer)
41
- self._thread_pool_manager = Mock(spec=ThreadPoolManager)
42
-
43
- self._an_open_partition = Mock(spec=Partition)
44
- self._log_message = Mock(spec=LogMessage)
45
- self._an_open_partition.to_slice.return_value = self._log_message
46
- self._an_open_partition.stream_name.return_value = _STREAM_NAME
47
-
48
- self._a_closed_partition = Mock(spec=Partition)
49
- self._a_closed_partition.stream_name.return_value = _ANOTHER_STREAM_NAME
50
-
51
- self._logger = Mock(spec=logging.Logger)
52
- self._slice_logger = Mock(spec=SliceLogger)
53
- self._slice_logger.create_slice_log_message.return_value = self._log_message
54
- self._message_repository = Mock(spec=MessageRepository)
55
- self._message_repository.consume_queue.return_value = []
56
- self._partition_reader = Mock(spec=PartitionReader)
57
-
58
- self._stream = Mock(spec=AbstractStream)
59
- self._stream.name = _STREAM_NAME
60
- self._stream.as_airbyte_stream.return_value = AirbyteStream(
61
- name=_STREAM_NAME,
62
- json_schema={},
63
- supported_sync_modes=[SyncMode.full_refresh],
64
- )
65
- self._another_stream = Mock(spec=AbstractStream)
66
- self._another_stream.name = _ANOTHER_STREAM_NAME
67
- self._another_stream.as_airbyte_stream.return_value = AirbyteStream(
68
- name=_ANOTHER_STREAM_NAME,
69
- json_schema={},
70
- supported_sync_modes=[SyncMode.full_refresh],
71
- )
72
-
73
- self._record_data = {"id": 1, "value": "A"}
74
- self._record = Mock(spec=Record)
75
- self._record.stream_name = _STREAM_NAME
76
- self._record.data = self._record_data
77
-
78
- def test_stream_is_not_done_initially(self):
79
- stream_instances_to_read_from = [self._stream]
80
- handler = ConcurrentReadProcessor(
81
- stream_instances_to_read_from,
82
- self._partition_enqueuer,
83
- self._thread_pool_manager,
84
- self._logger,
85
- self._slice_logger,
86
- self._message_repository,
87
- self._partition_reader,
88
- )
89
- assert not handler._is_stream_done(self._stream.name)
90
-
91
- def test_handle_partition_done_no_other_streams_to_generate_partitions_for(self):
92
- stream_instances_to_read_from = [self._stream]
93
-
94
- handler = ConcurrentReadProcessor(
95
- stream_instances_to_read_from,
96
- self._partition_enqueuer,
97
- self._thread_pool_manager,
98
- self._logger,
99
- self._slice_logger,
100
- self._message_repository,
101
- self._partition_reader,
102
- )
103
- handler.start_next_partition_generator()
104
- handler.on_partition(self._an_open_partition)
105
-
106
- sentinel = PartitionGenerationCompletedSentinel(self._stream)
107
- messages = list(handler.on_partition_generation_completed(sentinel))
108
-
109
- expected_messages = []
110
- assert expected_messages == messages
111
-
112
- @freezegun.freeze_time("2020-01-01T00:00:00")
113
- def test_handle_last_stream_partition_done(self):
114
- in_order_validation_mock = Mock()
115
- in_order_validation_mock.attach_mock(self._another_stream, "_another_stream")
116
- in_order_validation_mock.attach_mock(self._message_repository, "_message_repository")
117
- self._message_repository.consume_queue.return_value = iter([_ANY_AIRBYTE_MESSAGE])
118
- stream_instances_to_read_from = [self._another_stream]
119
-
120
- handler = ConcurrentReadProcessor(
121
- stream_instances_to_read_from,
122
- self._partition_enqueuer,
123
- self._thread_pool_manager,
124
- self._logger,
125
- self._slice_logger,
126
- self._message_repository,
127
- self._partition_reader,
128
- )
129
- handler.start_next_partition_generator()
130
-
131
- sentinel = PartitionGenerationCompletedSentinel(self._another_stream)
132
- messages = list(handler.on_partition_generation_completed(sentinel))
133
-
134
- expected_messages = [
135
- _ANY_AIRBYTE_MESSAGE,
136
- AirbyteMessage(
137
- type=MessageType.TRACE,
138
- trace=AirbyteTraceMessage(
139
- type=TraceType.STREAM_STATUS,
140
- emitted_at=1577836800000.0,
141
- stream_status=AirbyteStreamStatusTraceMessage(
142
- stream_descriptor=StreamDescriptor(name=_ANOTHER_STREAM_NAME),
143
- status=AirbyteStreamStatus(AirbyteStreamStatus.COMPLETE),
144
- ),
145
- ),
146
- ),
147
- ]
148
- assert expected_messages == messages
149
- assert in_order_validation_mock.mock_calls.index(
150
- call._another_stream.cursor.ensure_at_least_one_state_emitted
151
- ) < in_order_validation_mock.mock_calls.index(call._message_repository.consume_queue)
152
-
153
- def test_handle_partition(self):
154
- stream_instances_to_read_from = [self._another_stream]
155
-
156
- handler = ConcurrentReadProcessor(
157
- stream_instances_to_read_from,
158
- self._partition_enqueuer,
159
- self._thread_pool_manager,
160
- self._logger,
161
- self._slice_logger,
162
- self._message_repository,
163
- self._partition_reader,
164
- )
165
-
166
- handler.on_partition(self._a_closed_partition)
167
-
168
- self._thread_pool_manager.submit.assert_called_with(self._partition_reader.process_partition, self._a_closed_partition)
169
- assert self._a_closed_partition in handler._streams_to_running_partitions[_ANOTHER_STREAM_NAME]
170
-
171
- def test_handle_partition_emits_log_message_if_it_should_be_logged(self):
172
- stream_instances_to_read_from = [self._stream]
173
- self._slice_logger = Mock(spec=SliceLogger)
174
- self._slice_logger.should_log_slice_message.return_value = True
175
- self._slice_logger.create_slice_log_message.return_value = self._log_message
176
-
177
- handler = ConcurrentReadProcessor(
178
- stream_instances_to_read_from,
179
- self._partition_enqueuer,
180
- self._thread_pool_manager,
181
- self._logger,
182
- self._slice_logger,
183
- self._message_repository,
184
- self._partition_reader,
185
- )
186
-
187
- handler.on_partition(self._an_open_partition)
188
-
189
- self._thread_pool_manager.submit.assert_called_with(self._partition_reader.process_partition, self._an_open_partition)
190
- self._message_repository.emit_message.assert_called_with(self._log_message)
191
-
192
- assert self._an_open_partition in handler._streams_to_running_partitions[_STREAM_NAME]
193
-
194
- @freezegun.freeze_time("2020-01-01T00:00:00")
195
- def test_handle_on_partition_complete_sentinel_with_messages_from_repository(self):
196
- stream_instances_to_read_from = [self._stream]
197
- partition = Mock(spec=Partition)
198
- log_message = Mock(spec=LogMessage)
199
- partition.to_slice.return_value = log_message
200
- partition.stream_name.return_value = _STREAM_NAME
201
-
202
- handler = ConcurrentReadProcessor(
203
- stream_instances_to_read_from,
204
- self._partition_enqueuer,
205
- self._thread_pool_manager,
206
- self._logger,
207
- self._slice_logger,
208
- self._message_repository,
209
- self._partition_reader,
210
- )
211
- handler.start_next_partition_generator()
212
- handler.on_partition(partition)
213
-
214
- sentinel = PartitionCompleteSentinel(partition)
215
-
216
- self._message_repository.consume_queue.return_value = [
217
- AirbyteMessage(type=MessageType.LOG, log=AirbyteLogMessage(level=LogLevel.INFO, message="message emitted from the repository"))
218
- ]
219
-
220
- messages = list(handler.on_partition_complete_sentinel(sentinel))
221
-
222
- expected_messages = [
223
- AirbyteMessage(type=MessageType.LOG, log=AirbyteLogMessage(level=LogLevel.INFO, message="message emitted from the repository"))
224
- ]
225
- assert expected_messages == messages
226
-
227
- partition.close.assert_called_once()
228
-
229
- @freezegun.freeze_time("2020-01-01T00:00:00")
230
- def test_handle_on_partition_complete_sentinel_yields_status_message_if_the_stream_is_done(self):
231
- self._streams_currently_generating_partitions = [self._another_stream]
232
- stream_instances_to_read_from = [self._another_stream]
233
- log_message = Mock(spec=LogMessage)
234
- self._a_closed_partition.to_slice.return_value = log_message
235
- self._message_repository.consume_queue.return_value = []
236
-
237
- handler = ConcurrentReadProcessor(
238
- stream_instances_to_read_from,
239
- self._partition_enqueuer,
240
- self._thread_pool_manager,
241
- self._logger,
242
- self._slice_logger,
243
- self._message_repository,
244
- self._partition_reader,
245
- )
246
- handler.start_next_partition_generator()
247
- handler.on_partition(self._a_closed_partition)
248
- list(handler.on_partition_generation_completed(PartitionGenerationCompletedSentinel(self._another_stream)))
249
-
250
- sentinel = PartitionCompleteSentinel(self._a_closed_partition)
251
-
252
- messages = list(handler.on_partition_complete_sentinel(sentinel))
253
-
254
- expected_messages = [
255
- AirbyteMessage(
256
- type=MessageType.TRACE,
257
- trace=AirbyteTraceMessage(
258
- type=TraceType.STREAM_STATUS,
259
- stream_status=AirbyteStreamStatusTraceMessage(
260
- stream_descriptor=StreamDescriptor(
261
- name=_ANOTHER_STREAM_NAME,
262
- ),
263
- status=AirbyteStreamStatus.COMPLETE,
264
- ),
265
- emitted_at=1577836800000.0,
266
- ),
267
- )
268
- ]
269
- assert expected_messages == messages
270
- self._a_closed_partition.close.assert_called_once()
271
-
272
- @freezegun.freeze_time("2020-01-01T00:00:00")
273
- def test_handle_on_partition_complete_sentinel_yields_no_status_message_if_the_stream_is_not_done(self):
274
- stream_instances_to_read_from = [self._stream]
275
- partition = Mock(spec=Partition)
276
- log_message = Mock(spec=LogMessage)
277
- partition.to_slice.return_value = log_message
278
- partition.stream_name.return_value = _STREAM_NAME
279
-
280
- handler = ConcurrentReadProcessor(
281
- stream_instances_to_read_from,
282
- self._partition_enqueuer,
283
- self._thread_pool_manager,
284
- self._logger,
285
- self._slice_logger,
286
- self._message_repository,
287
- self._partition_reader,
288
- )
289
- handler.start_next_partition_generator()
290
-
291
- sentinel = PartitionCompleteSentinel(partition)
292
-
293
- messages = list(handler.on_partition_complete_sentinel(sentinel))
294
-
295
- expected_messages = []
296
- assert expected_messages == messages
297
- partition.close.assert_called_once()
298
-
299
- @freezegun.freeze_time("2020-01-01T00:00:00")
300
- def test_on_record_no_status_message_no_repository_messge(self):
301
- stream_instances_to_read_from = [self._stream]
302
- partition = Mock(spec=Partition)
303
- log_message = Mock(spec=LogMessage)
304
- partition.to_slice.return_value = log_message
305
- partition.stream_name.return_value = _STREAM_NAME
306
- self._message_repository.consume_queue.return_value = []
307
-
308
- handler = ConcurrentReadProcessor(
309
- stream_instances_to_read_from,
310
- self._partition_enqueuer,
311
- self._thread_pool_manager,
312
- self._logger,
313
- self._slice_logger,
314
- self._message_repository,
315
- self._partition_reader,
316
- )
317
-
318
- # Simulate a first record
319
- list(handler.on_record(self._record))
320
-
321
- messages = list(handler.on_record(self._record))
322
-
323
- expected_messages = [
324
- AirbyteMessage(
325
- type=MessageType.RECORD,
326
- record=AirbyteRecordMessage(
327
- stream=_STREAM_NAME,
328
- data=self._record_data,
329
- emitted_at=1577836800000,
330
- ),
331
- )
332
- ]
333
- assert expected_messages == messages
334
-
335
- @freezegun.freeze_time("2020-01-01T00:00:00")
336
- def test_on_record_with_repository_messge(self):
337
- stream_instances_to_read_from = [self._stream]
338
- partition = Mock(spec=Partition)
339
- log_message = Mock(spec=LogMessage)
340
- partition.to_slice.return_value = log_message
341
- partition.stream_name.return_value = _STREAM_NAME
342
- slice_logger = Mock(spec=SliceLogger)
343
- slice_logger.should_log_slice_message.return_value = True
344
- slice_logger.create_slice_log_message.return_value = log_message
345
- self._message_repository.consume_queue.return_value = [
346
- AirbyteMessage(type=MessageType.LOG, log=AirbyteLogMessage(level=LogLevel.INFO, message="message emitted from the repository"))
347
- ]
348
-
349
- handler = ConcurrentReadProcessor(
350
- stream_instances_to_read_from,
351
- self._partition_enqueuer,
352
- self._thread_pool_manager,
353
- self._logger,
354
- self._slice_logger,
355
- self._message_repository,
356
- self._partition_reader,
357
- )
358
-
359
- stream = Mock(spec=AbstractStream)
360
- stream.name = _STREAM_NAME
361
- stream.as_airbyte_stream.return_value = AirbyteStream(
362
- name=_STREAM_NAME,
363
- json_schema={},
364
- supported_sync_modes=[SyncMode.full_refresh],
365
- )
366
-
367
- # Simulate a first record
368
- list(handler.on_record(self._record))
369
-
370
- messages = list(handler.on_record(self._record))
371
-
372
- expected_messages = [
373
- AirbyteMessage(
374
- type=MessageType.RECORD,
375
- record=AirbyteRecordMessage(
376
- stream=_STREAM_NAME,
377
- data=self._record_data,
378
- emitted_at=1577836800000,
379
- ),
380
- ),
381
- AirbyteMessage(type=MessageType.LOG, log=AirbyteLogMessage(level=LogLevel.INFO, message="message emitted from the repository")),
382
- ]
383
- assert expected_messages == messages
384
- assert handler._record_counter[_STREAM_NAME] == 2
385
-
386
- @freezegun.freeze_time("2020-01-01T00:00:00")
387
- def test_on_record_emits_status_message_on_first_record_no_repository_message(self):
388
- self._streams_currently_generating_partitions = [_STREAM_NAME]
389
- stream_instances_to_read_from = [self._stream]
390
- partition = Mock(spec=Partition)
391
- partition.stream_name.return_value = _STREAM_NAME
392
-
393
- handler = ConcurrentReadProcessor(
394
- stream_instances_to_read_from,
395
- self._partition_enqueuer,
396
- self._thread_pool_manager,
397
- self._logger,
398
- self._slice_logger,
399
- self._message_repository,
400
- self._partition_reader,
401
- )
402
-
403
- messages = list(handler.on_record(self._record))
404
-
405
- expected_messages = [
406
- AirbyteMessage(
407
- type=MessageType.TRACE,
408
- trace=AirbyteTraceMessage(
409
- type=TraceType.STREAM_STATUS,
410
- emitted_at=1577836800000.0,
411
- stream_status=AirbyteStreamStatusTraceMessage(
412
- stream_descriptor=StreamDescriptor(name=_STREAM_NAME), status=AirbyteStreamStatus(AirbyteStreamStatus.RUNNING)
413
- ),
414
- ),
415
- ),
416
- AirbyteMessage(
417
- type=MessageType.RECORD,
418
- record=AirbyteRecordMessage(
419
- stream=_STREAM_NAME,
420
- data=self._record_data,
421
- emitted_at=1577836800000,
422
- ),
423
- ),
424
- ]
425
- assert expected_messages == messages
426
-
427
- @freezegun.freeze_time("2020-01-01T00:00:00")
428
- def test_on_record_emits_status_message_on_first_record_with_repository_message(self):
429
- stream_instances_to_read_from = [self._stream]
430
- partition = Mock(spec=Partition)
431
- log_message = Mock(spec=LogMessage)
432
- partition.to_slice.return_value = log_message
433
- partition.stream_name.return_value = _STREAM_NAME
434
- self._message_repository.consume_queue.return_value = [
435
- AirbyteMessage(type=MessageType.LOG, log=AirbyteLogMessage(level=LogLevel.INFO, message="message emitted from the repository"))
436
- ]
437
-
438
- handler = ConcurrentReadProcessor(
439
- stream_instances_to_read_from,
440
- self._partition_enqueuer,
441
- self._thread_pool_manager,
442
- self._logger,
443
- self._slice_logger,
444
- self._message_repository,
445
- self._partition_reader,
446
- )
447
-
448
- stream = Mock(spec=AbstractStream)
449
- stream.name = _STREAM_NAME
450
- stream.as_airbyte_stream.return_value = AirbyteStream(
451
- name=_STREAM_NAME,
452
- json_schema={},
453
- supported_sync_modes=[SyncMode.full_refresh],
454
- )
455
-
456
- messages = list(handler.on_record(self._record))
457
-
458
- expected_messages = [
459
- AirbyteMessage(
460
- type=MessageType.TRACE,
461
- trace=AirbyteTraceMessage(
462
- type=TraceType.STREAM_STATUS,
463
- emitted_at=1577836800000.0,
464
- stream_status=AirbyteStreamStatusTraceMessage(
465
- stream_descriptor=StreamDescriptor(name=_STREAM_NAME), status=AirbyteStreamStatus(AirbyteStreamStatus.RUNNING)
466
- ),
467
- ),
468
- ),
469
- AirbyteMessage(
470
- type=MessageType.RECORD,
471
- record=AirbyteRecordMessage(
472
- stream=_STREAM_NAME,
473
- data=self._record_data,
474
- emitted_at=1577836800000,
475
- ),
476
- ),
477
- AirbyteMessage(type=MessageType.LOG, log=AirbyteLogMessage(level=LogLevel.INFO, message="message emitted from the repository")),
478
- ]
479
- assert expected_messages == messages
480
-
481
- @freezegun.freeze_time("2020-01-01T00:00:00")
482
- def test_on_exception_stops_streams_and_raises_an_exception(self):
483
- stream_instances_to_read_from = [self._stream, self._another_stream]
484
-
485
- handler = ConcurrentReadProcessor(
486
- stream_instances_to_read_from,
487
- self._partition_enqueuer,
488
- self._thread_pool_manager,
489
- self._logger,
490
- self._slice_logger,
491
- self._message_repository,
492
- self._partition_reader,
493
- )
494
-
495
- handler.start_next_partition_generator()
496
-
497
- another_stream = Mock(spec=AbstractStream)
498
- another_stream.name = _STREAM_NAME
499
- another_stream.as_airbyte_stream.return_value = AirbyteStream(
500
- name=_ANOTHER_STREAM_NAME,
501
- json_schema={},
502
- supported_sync_modes=[SyncMode.full_refresh],
503
- )
504
-
505
- exception = RuntimeError("Something went wrong")
506
-
507
- messages = []
508
-
509
- with self.assertRaises(RuntimeError):
510
- for m in handler.on_exception(exception):
511
- messages.append(m)
512
-
513
- expected_message = [
514
- AirbyteMessage(
515
- type=MessageType.TRACE,
516
- trace=AirbyteTraceMessage(
517
- type=TraceType.STREAM_STATUS,
518
- emitted_at=1577836800000.0,
519
- stream_status=AirbyteStreamStatusTraceMessage(
520
- stream_descriptor=StreamDescriptor(name=_STREAM_NAME), status=AirbyteStreamStatus(AirbyteStreamStatus.INCOMPLETE)
521
- ),
522
- ),
523
- ),
524
- AirbyteMessage(
525
- type=MessageType.TRACE,
526
- trace=AirbyteTraceMessage(
527
- type=TraceType.STREAM_STATUS,
528
- emitted_at=1577836800000.0,
529
- stream_status=AirbyteStreamStatusTraceMessage(
530
- stream_descriptor=StreamDescriptor(name=_ANOTHER_STREAM_NAME),
531
- status=AirbyteStreamStatus(AirbyteStreamStatus.INCOMPLETE),
532
- ),
533
- ),
534
- ),
535
- ]
536
-
537
- assert messages == expected_message
538
- self._thread_pool_manager.shutdown.assert_called_once()
539
-
540
- @freezegun.freeze_time("2020-01-01T00:00:00")
541
- def test_on_exception_does_not_stop_streams_that_are_already_done(self):
542
- stream_instances_to_read_from = [self._stream, self._another_stream]
543
-
544
- handler = ConcurrentReadProcessor(
545
- stream_instances_to_read_from,
546
- self._partition_enqueuer,
547
- self._thread_pool_manager,
548
- self._logger,
549
- self._slice_logger,
550
- self._message_repository,
551
- self._partition_reader,
552
- )
553
-
554
- handler.start_next_partition_generator()
555
- handler.on_partition(self._an_open_partition)
556
- list(handler.on_partition_generation_completed(PartitionGenerationCompletedSentinel(self._stream)))
557
- list(handler.on_partition_generation_completed(PartitionGenerationCompletedSentinel(self._another_stream)))
558
-
559
- another_stream = Mock(spec=AbstractStream)
560
- another_stream.name = _STREAM_NAME
561
- another_stream.as_airbyte_stream.return_value = AirbyteStream(
562
- name=_ANOTHER_STREAM_NAME,
563
- json_schema={},
564
- supported_sync_modes=[SyncMode.full_refresh],
565
- )
566
-
567
- exception = RuntimeError("Something went wrong")
568
-
569
- messages = []
570
-
571
- with self.assertRaises(RuntimeError):
572
- for m in handler.on_exception(exception):
573
- messages.append(m)
574
-
575
- expected_message = [
576
- AirbyteMessage(
577
- type=MessageType.TRACE,
578
- trace=AirbyteTraceMessage(
579
- type=TraceType.STREAM_STATUS,
580
- emitted_at=1577836800000.0,
581
- stream_status=AirbyteStreamStatusTraceMessage(
582
- stream_descriptor=StreamDescriptor(name=_STREAM_NAME), status=AirbyteStreamStatus(AirbyteStreamStatus.INCOMPLETE)
583
- ),
584
- ),
585
- )
586
- ]
587
-
588
- assert messages == expected_message
589
- self._thread_pool_manager.shutdown.assert_called_once()
590
-
591
- def test_is_done_is_false_if_there_are_any_instances_to_read_from(self):
592
- stream_instances_to_read_from = [self._stream]
593
-
594
- handler = ConcurrentReadProcessor(
595
- stream_instances_to_read_from,
596
- self._partition_enqueuer,
597
- self._thread_pool_manager,
598
- self._logger,
599
- self._slice_logger,
600
- self._message_repository,
601
- self._partition_reader,
602
- )
603
-
604
- assert not handler.is_done()
605
-
606
- def test_is_done_is_false_if_there_are_streams_still_generating_partitions(self):
607
- stream_instances_to_read_from = [self._stream]
608
-
609
- handler = ConcurrentReadProcessor(
610
- stream_instances_to_read_from,
611
- self._partition_enqueuer,
612
- self._thread_pool_manager,
613
- self._logger,
614
- self._slice_logger,
615
- self._message_repository,
616
- self._partition_reader,
617
- )
618
-
619
- handler.start_next_partition_generator()
620
-
621
- assert not handler.is_done()
622
-
623
- def test_is_done_is_false_if_all_partitions_are_not_closed(self):
624
- stream_instances_to_read_from = [self._stream]
625
-
626
- handler = ConcurrentReadProcessor(
627
- stream_instances_to_read_from,
628
- self._partition_enqueuer,
629
- self._thread_pool_manager,
630
- self._logger,
631
- self._slice_logger,
632
- self._message_repository,
633
- self._partition_reader,
634
- )
635
-
636
- handler.start_next_partition_generator()
637
- handler.on_partition(self._an_open_partition)
638
- handler.on_partition_generation_completed(PartitionGenerationCompletedSentinel(self._stream))
639
-
640
- assert not handler.is_done()
641
-
642
- def test_is_done_is_true_if_all_partitions_are_closed_and_no_streams_are_generating_partitions_and_none_are_still_to_run(self):
643
- stream_instances_to_read_from = []
644
-
645
- handler = ConcurrentReadProcessor(
646
- stream_instances_to_read_from,
647
- self._partition_enqueuer,
648
- self._thread_pool_manager,
649
- self._logger,
650
- self._slice_logger,
651
- self._message_repository,
652
- self._partition_reader,
653
- )
654
-
655
- assert handler.is_done()
656
-
657
- @freezegun.freeze_time("2020-01-01T00:00:00")
658
- def test_start_next_partition_generator(self):
659
- stream_instances_to_read_from = [self._stream]
660
- handler = ConcurrentReadProcessor(
661
- stream_instances_to_read_from,
662
- self._partition_enqueuer,
663
- self._thread_pool_manager,
664
- self._logger,
665
- self._slice_logger,
666
- self._message_repository,
667
- self._partition_reader,
668
- )
669
-
670
- status_message = handler.start_next_partition_generator()
671
-
672
- assert status_message == AirbyteMessage(
673
- type=MessageType.TRACE,
674
- trace=AirbyteTraceMessage(
675
- type=TraceType.STREAM_STATUS,
676
- emitted_at=1577836800000.0,
677
- stream_status=AirbyteStreamStatusTraceMessage(
678
- stream_descriptor=StreamDescriptor(name=_STREAM_NAME), status=AirbyteStreamStatus(AirbyteStreamStatus.STARTED)
679
- ),
680
- ),
681
- )
682
-
683
- assert _STREAM_NAME in handler._streams_currently_generating_partitions
684
- self._thread_pool_manager.submit.assert_called_with(self._partition_enqueuer.generate_partitions, self._stream)