airbyte-cdk 0.72.0__py3-none-any.whl → 6.17.1.dev0__py3-none-any.whl

Sign up to get free protection for your applications and to get access to all the features.
Files changed (518) 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 +490 -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 +1213 -88
  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 +63 -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 +31 -3
  79. airbyte_cdk/sources/declarative/incremental/concurrent_partition_cursor.py +346 -0
  80. airbyte_cdk/sources/declarative/incremental/datetime_based_cursor.py +156 -48
  81. airbyte_cdk/sources/declarative/incremental/declarative_cursor.py +13 -0
  82. airbyte_cdk/sources/declarative/incremental/global_substream_cursor.py +350 -0
  83. airbyte_cdk/sources/declarative/incremental/per_partition_cursor.py +173 -74
  84. airbyte_cdk/sources/declarative/incremental/per_partition_with_global.py +200 -0
  85. airbyte_cdk/sources/declarative/incremental/resumable_full_refresh_cursor.py +122 -0
  86. airbyte_cdk/sources/declarative/interpolation/filters.py +27 -1
  87. airbyte_cdk/sources/declarative/interpolation/interpolated_boolean.py +23 -5
  88. airbyte_cdk/sources/declarative/interpolation/interpolated_mapping.py +12 -8
  89. airbyte_cdk/sources/declarative/interpolation/interpolated_nested_mapping.py +13 -6
  90. airbyte_cdk/sources/declarative/interpolation/interpolated_string.py +21 -6
  91. airbyte_cdk/sources/declarative/interpolation/interpolation.py +9 -3
  92. airbyte_cdk/sources/declarative/interpolation/jinja.py +72 -37
  93. airbyte_cdk/sources/declarative/interpolation/macros.py +72 -17
  94. airbyte_cdk/sources/declarative/manifest_declarative_source.py +193 -52
  95. airbyte_cdk/sources/declarative/migrations/legacy_to_per_partition_state_migration.py +98 -0
  96. airbyte_cdk/sources/declarative/migrations/state_migration.py +24 -0
  97. airbyte_cdk/sources/declarative/models/__init__.py +1 -1
  98. airbyte_cdk/sources/declarative/models/declarative_component_schema.py +1329 -595
  99. airbyte_cdk/sources/declarative/parsers/custom_exceptions.py +2 -2
  100. airbyte_cdk/sources/declarative/parsers/manifest_component_transformer.py +26 -4
  101. airbyte_cdk/sources/declarative/parsers/manifest_reference_resolver.py +26 -15
  102. airbyte_cdk/sources/declarative/parsers/model_to_component_factory.py +1763 -226
  103. airbyte_cdk/sources/declarative/partition_routers/__init__.py +24 -4
  104. airbyte_cdk/sources/declarative/partition_routers/async_job_partition_router.py +65 -0
  105. airbyte_cdk/sources/declarative/partition_routers/cartesian_product_stream_slicer.py +176 -0
  106. airbyte_cdk/sources/declarative/partition_routers/list_partition_router.py +39 -9
  107. airbyte_cdk/sources/declarative/partition_routers/partition_router.py +62 -0
  108. airbyte_cdk/sources/declarative/partition_routers/single_partition_router.py +15 -3
  109. airbyte_cdk/sources/declarative/partition_routers/substream_partition_router.py +222 -39
  110. airbyte_cdk/sources/declarative/requesters/error_handlers/__init__.py +19 -5
  111. airbyte_cdk/sources/declarative/requesters/error_handlers/backoff_strategies/__init__.py +3 -1
  112. airbyte_cdk/sources/declarative/requesters/error_handlers/backoff_strategies/constant_backoff_strategy.py +19 -7
  113. airbyte_cdk/sources/declarative/requesters/error_handlers/backoff_strategies/exponential_backoff_strategy.py +19 -7
  114. airbyte_cdk/sources/declarative/requesters/error_handlers/backoff_strategies/header_helper.py +4 -2
  115. airbyte_cdk/sources/declarative/requesters/error_handlers/backoff_strategies/wait_time_from_header_backoff_strategy.py +41 -9
  116. airbyte_cdk/sources/declarative/requesters/error_handlers/backoff_strategies/wait_until_time_from_header_backoff_strategy.py +29 -14
  117. airbyte_cdk/sources/declarative/requesters/error_handlers/backoff_strategy.py +5 -13
  118. airbyte_cdk/sources/declarative/requesters/error_handlers/composite_error_handler.py +32 -16
  119. airbyte_cdk/sources/declarative/requesters/error_handlers/default_error_handler.py +46 -56
  120. airbyte_cdk/sources/declarative/requesters/error_handlers/default_http_response_filter.py +40 -0
  121. airbyte_cdk/sources/declarative/requesters/error_handlers/error_handler.py +6 -32
  122. airbyte_cdk/sources/declarative/requesters/error_handlers/http_response_filter.py +119 -41
  123. airbyte_cdk/sources/declarative/requesters/http_job_repository.py +228 -0
  124. airbyte_cdk/sources/declarative/requesters/http_requester.py +98 -344
  125. airbyte_cdk/sources/declarative/requesters/paginators/__init__.py +14 -3
  126. airbyte_cdk/sources/declarative/requesters/paginators/default_paginator.py +105 -46
  127. airbyte_cdk/sources/declarative/requesters/paginators/no_pagination.py +14 -8
  128. airbyte_cdk/sources/declarative/requesters/paginators/paginator.py +19 -8
  129. airbyte_cdk/sources/declarative/requesters/paginators/strategies/__init__.py +9 -3
  130. airbyte_cdk/sources/declarative/requesters/paginators/strategies/cursor_pagination_strategy.py +53 -21
  131. airbyte_cdk/sources/declarative/requesters/paginators/strategies/offset_increment.py +42 -19
  132. airbyte_cdk/sources/declarative/requesters/paginators/strategies/page_increment.py +25 -12
  133. airbyte_cdk/sources/declarative/requesters/paginators/strategies/pagination_strategy.py +13 -10
  134. airbyte_cdk/sources/declarative/requesters/paginators/strategies/stop_condition.py +26 -13
  135. airbyte_cdk/sources/declarative/requesters/request_options/__init__.py +15 -2
  136. airbyte_cdk/sources/declarative/requesters/request_options/datetime_based_request_options_provider.py +91 -0
  137. airbyte_cdk/sources/declarative/requesters/request_options/default_request_options_provider.py +60 -0
  138. airbyte_cdk/sources/declarative/requesters/request_options/interpolated_nested_request_input_provider.py +31 -14
  139. airbyte_cdk/sources/declarative/requesters/request_options/interpolated_request_input_provider.py +27 -15
  140. airbyte_cdk/sources/declarative/requesters/request_options/interpolated_request_options_provider.py +63 -10
  141. airbyte_cdk/sources/declarative/requesters/request_options/request_options_provider.py +1 -1
  142. airbyte_cdk/sources/declarative/requesters/requester.py +9 -17
  143. airbyte_cdk/sources/declarative/resolvers/__init__.py +41 -0
  144. airbyte_cdk/sources/declarative/resolvers/components_resolver.py +55 -0
  145. airbyte_cdk/sources/declarative/resolvers/config_components_resolver.py +136 -0
  146. airbyte_cdk/sources/declarative/resolvers/http_components_resolver.py +112 -0
  147. airbyte_cdk/sources/declarative/retrievers/__init__.py +6 -2
  148. airbyte_cdk/sources/declarative/retrievers/async_retriever.py +100 -0
  149. airbyte_cdk/sources/declarative/retrievers/retriever.py +1 -3
  150. airbyte_cdk/sources/declarative/retrievers/simple_retriever.py +229 -73
  151. airbyte_cdk/sources/declarative/schema/__init__.py +14 -1
  152. airbyte_cdk/sources/declarative/schema/default_schema_loader.py +5 -3
  153. airbyte_cdk/sources/declarative/schema/dynamic_schema_loader.py +236 -0
  154. airbyte_cdk/sources/declarative/schema/json_file_schema_loader.py +8 -8
  155. airbyte_cdk/sources/declarative/spec/spec.py +12 -5
  156. airbyte_cdk/sources/declarative/stream_slicers/__init__.py +1 -2
  157. airbyte_cdk/sources/declarative/stream_slicers/declarative_partition_generator.py +88 -0
  158. airbyte_cdk/sources/declarative/stream_slicers/stream_slicer.py +9 -14
  159. airbyte_cdk/sources/declarative/transformations/add_fields.py +19 -11
  160. airbyte_cdk/sources/declarative/transformations/flatten_fields.py +52 -0
  161. airbyte_cdk/sources/declarative/transformations/keys_replace_transformation.py +61 -0
  162. airbyte_cdk/sources/declarative/transformations/keys_to_lower_transformation.py +22 -0
  163. airbyte_cdk/sources/declarative/transformations/keys_to_snake_transformation.py +68 -0
  164. airbyte_cdk/sources/declarative/transformations/remove_fields.py +13 -10
  165. airbyte_cdk/sources/declarative/transformations/transformation.py +5 -5
  166. airbyte_cdk/sources/declarative/types.py +19 -110
  167. airbyte_cdk/sources/declarative/yaml_declarative_source.py +31 -10
  168. airbyte_cdk/sources/embedded/base_integration.py +16 -5
  169. airbyte_cdk/sources/embedded/catalog.py +16 -4
  170. airbyte_cdk/sources/embedded/runner.py +19 -3
  171. airbyte_cdk/sources/embedded/tools.py +5 -2
  172. airbyte_cdk/sources/file_based/README.md +152 -0
  173. airbyte_cdk/sources/file_based/__init__.py +24 -0
  174. airbyte_cdk/sources/file_based/availability_strategy/__init__.py +9 -2
  175. airbyte_cdk/sources/file_based/availability_strategy/abstract_file_based_availability_strategy.py +22 -6
  176. airbyte_cdk/sources/file_based/availability_strategy/default_file_based_availability_strategy.py +46 -10
  177. airbyte_cdk/sources/file_based/config/abstract_file_based_spec.py +47 -10
  178. airbyte_cdk/sources/file_based/config/avro_format.py +2 -1
  179. airbyte_cdk/sources/file_based/config/csv_format.py +29 -10
  180. airbyte_cdk/sources/file_based/config/excel_format.py +18 -0
  181. airbyte_cdk/sources/file_based/config/file_based_stream_config.py +16 -4
  182. airbyte_cdk/sources/file_based/config/jsonl_format.py +2 -1
  183. airbyte_cdk/sources/file_based/config/parquet_format.py +2 -1
  184. airbyte_cdk/sources/file_based/config/unstructured_format.py +13 -5
  185. airbyte_cdk/sources/file_based/discovery_policy/__init__.py +6 -2
  186. airbyte_cdk/sources/file_based/discovery_policy/abstract_discovery_policy.py +2 -4
  187. airbyte_cdk/sources/file_based/discovery_policy/default_discovery_policy.py +7 -2
  188. airbyte_cdk/sources/file_based/exceptions.py +18 -15
  189. airbyte_cdk/sources/file_based/file_based_source.py +140 -33
  190. airbyte_cdk/sources/file_based/file_based_stream_reader.py +69 -5
  191. airbyte_cdk/sources/file_based/file_types/__init__.py +14 -1
  192. airbyte_cdk/sources/file_based/file_types/avro_parser.py +75 -24
  193. airbyte_cdk/sources/file_based/file_types/csv_parser.py +116 -34
  194. airbyte_cdk/sources/file_based/file_types/excel_parser.py +196 -0
  195. airbyte_cdk/sources/file_based/file_types/file_transfer.py +37 -0
  196. airbyte_cdk/sources/file_based/file_types/file_type_parser.py +4 -1
  197. airbyte_cdk/sources/file_based/file_types/jsonl_parser.py +24 -8
  198. airbyte_cdk/sources/file_based/file_types/parquet_parser.py +60 -18
  199. airbyte_cdk/sources/file_based/file_types/unstructured_parser.py +141 -41
  200. airbyte_cdk/sources/file_based/remote_file.py +1 -1
  201. airbyte_cdk/sources/file_based/schema_helpers.py +38 -10
  202. airbyte_cdk/sources/file_based/schema_validation_policies/__init__.py +3 -1
  203. airbyte_cdk/sources/file_based/schema_validation_policies/abstract_schema_validation_policy.py +3 -1
  204. airbyte_cdk/sources/file_based/schema_validation_policies/default_schema_validation_policies.py +16 -5
  205. airbyte_cdk/sources/file_based/stream/abstract_file_based_stream.py +50 -13
  206. airbyte_cdk/sources/file_based/stream/concurrent/adapters.py +67 -27
  207. airbyte_cdk/sources/file_based/stream/concurrent/cursor/__init__.py +5 -1
  208. airbyte_cdk/sources/file_based/stream/concurrent/cursor/abstract_concurrent_file_based_cursor.py +14 -23
  209. airbyte_cdk/sources/file_based/stream/concurrent/cursor/file_based_concurrent_cursor.py +54 -18
  210. airbyte_cdk/sources/file_based/stream/concurrent/cursor/file_based_final_state_cursor.py +21 -9
  211. airbyte_cdk/sources/file_based/stream/cursor/abstract_file_based_cursor.py +3 -1
  212. airbyte_cdk/sources/file_based/stream/cursor/default_file_based_cursor.py +27 -10
  213. airbyte_cdk/sources/file_based/stream/default_file_based_stream.py +147 -45
  214. airbyte_cdk/sources/http_logger.py +8 -3
  215. airbyte_cdk/sources/message/__init__.py +7 -1
  216. airbyte_cdk/sources/message/repository.py +18 -4
  217. airbyte_cdk/sources/source.py +42 -38
  218. airbyte_cdk/sources/streams/__init__.py +2 -2
  219. airbyte_cdk/sources/streams/availability_strategy.py +54 -3
  220. airbyte_cdk/sources/streams/call_rate.py +64 -21
  221. airbyte_cdk/sources/streams/checkpoint/__init__.py +26 -0
  222. airbyte_cdk/sources/streams/checkpoint/checkpoint_reader.py +335 -0
  223. airbyte_cdk/sources/{declarative/incremental → streams/checkpoint}/cursor.py +17 -14
  224. airbyte_cdk/sources/streams/checkpoint/per_partition_key_serializer.py +22 -0
  225. airbyte_cdk/sources/streams/checkpoint/resumable_full_refresh_cursor.py +51 -0
  226. airbyte_cdk/sources/streams/checkpoint/substream_resumable_full_refresh_cursor.py +110 -0
  227. airbyte_cdk/sources/streams/concurrent/README.md +7 -0
  228. airbyte_cdk/sources/streams/concurrent/abstract_stream.py +7 -2
  229. airbyte_cdk/sources/streams/concurrent/adapters.py +84 -75
  230. airbyte_cdk/sources/streams/concurrent/availability_strategy.py +30 -2
  231. airbyte_cdk/sources/streams/concurrent/cursor.py +298 -42
  232. airbyte_cdk/sources/streams/concurrent/default_stream.py +12 -3
  233. airbyte_cdk/sources/streams/concurrent/exceptions.py +3 -0
  234. airbyte_cdk/sources/streams/concurrent/helpers.py +14 -3
  235. airbyte_cdk/sources/streams/concurrent/partition_enqueuer.py +12 -3
  236. airbyte_cdk/sources/streams/concurrent/partition_reader.py +10 -3
  237. airbyte_cdk/sources/streams/concurrent/partitions/partition.py +1 -16
  238. airbyte_cdk/sources/streams/concurrent/partitions/stream_slicer.py +21 -0
  239. airbyte_cdk/sources/streams/concurrent/partitions/types.py +15 -5
  240. airbyte_cdk/sources/streams/concurrent/state_converters/abstract_stream_state_converter.py +109 -17
  241. airbyte_cdk/sources/streams/concurrent/state_converters/datetime_stream_state_converter.py +90 -72
  242. airbyte_cdk/sources/streams/core.py +412 -87
  243. airbyte_cdk/sources/streams/http/__init__.py +2 -1
  244. airbyte_cdk/sources/streams/http/availability_strategy.py +12 -101
  245. airbyte_cdk/sources/streams/http/error_handlers/__init__.py +22 -0
  246. airbyte_cdk/sources/streams/http/error_handlers/backoff_strategy.py +28 -0
  247. airbyte_cdk/sources/streams/http/error_handlers/default_backoff_strategy.py +17 -0
  248. airbyte_cdk/sources/streams/http/error_handlers/default_error_mapping.py +86 -0
  249. airbyte_cdk/sources/streams/http/error_handlers/error_handler.py +42 -0
  250. airbyte_cdk/sources/streams/http/error_handlers/error_message_parser.py +19 -0
  251. airbyte_cdk/sources/streams/http/error_handlers/http_status_error_handler.py +110 -0
  252. airbyte_cdk/sources/streams/http/error_handlers/json_error_message_parser.py +52 -0
  253. airbyte_cdk/sources/streams/http/error_handlers/response_models.py +65 -0
  254. airbyte_cdk/sources/streams/http/exceptions.py +27 -7
  255. airbyte_cdk/sources/streams/http/http.py +369 -246
  256. airbyte_cdk/sources/streams/http/http_client.py +531 -0
  257. airbyte_cdk/sources/streams/http/rate_limiting.py +76 -12
  258. airbyte_cdk/sources/streams/http/requests_native_auth/abstract_oauth.py +28 -9
  259. airbyte_cdk/sources/streams/http/requests_native_auth/abstract_token.py +2 -1
  260. airbyte_cdk/sources/streams/http/requests_native_auth/oauth.py +90 -35
  261. airbyte_cdk/sources/streams/http/requests_native_auth/token.py +13 -3
  262. airbyte_cdk/sources/types.py +154 -0
  263. airbyte_cdk/sources/utils/record_helper.py +36 -21
  264. airbyte_cdk/sources/utils/schema_helpers.py +13 -6
  265. airbyte_cdk/sources/utils/slice_logger.py +4 -1
  266. airbyte_cdk/sources/utils/transform.py +54 -20
  267. airbyte_cdk/sql/_util/hashing.py +34 -0
  268. airbyte_cdk/sql/_util/name_normalizers.py +92 -0
  269. airbyte_cdk/sql/constants.py +32 -0
  270. airbyte_cdk/sql/exceptions.py +235 -0
  271. airbyte_cdk/sql/secrets.py +123 -0
  272. airbyte_cdk/sql/shared/__init__.py +15 -0
  273. airbyte_cdk/sql/shared/catalog_providers.py +145 -0
  274. airbyte_cdk/sql/shared/sql_processor.py +786 -0
  275. airbyte_cdk/sql/types.py +160 -0
  276. airbyte_cdk/test/catalog_builder.py +70 -18
  277. airbyte_cdk/test/entrypoint_wrapper.py +117 -42
  278. airbyte_cdk/test/mock_http/__init__.py +1 -1
  279. airbyte_cdk/test/mock_http/matcher.py +6 -0
  280. airbyte_cdk/test/mock_http/mocker.py +57 -10
  281. airbyte_cdk/test/mock_http/request.py +19 -3
  282. airbyte_cdk/test/mock_http/response.py +3 -1
  283. airbyte_cdk/test/mock_http/response_builder.py +32 -16
  284. airbyte_cdk/test/state_builder.py +18 -10
  285. airbyte_cdk/test/utils/__init__.py +1 -0
  286. airbyte_cdk/test/utils/data.py +24 -0
  287. airbyte_cdk/test/utils/http_mocking.py +16 -0
  288. airbyte_cdk/test/utils/manifest_only_fixtures.py +60 -0
  289. airbyte_cdk/test/utils/reading.py +26 -0
  290. airbyte_cdk/utils/__init__.py +2 -1
  291. airbyte_cdk/utils/airbyte_secrets_utils.py +5 -3
  292. airbyte_cdk/utils/analytics_message.py +10 -2
  293. airbyte_cdk/utils/datetime_format_inferrer.py +4 -1
  294. airbyte_cdk/utils/event_timing.py +10 -10
  295. airbyte_cdk/utils/mapping_helpers.py +3 -1
  296. airbyte_cdk/utils/message_utils.py +20 -11
  297. airbyte_cdk/utils/print_buffer.py +75 -0
  298. airbyte_cdk/utils/schema_inferrer.py +198 -28
  299. airbyte_cdk/utils/slice_hasher.py +30 -0
  300. airbyte_cdk/utils/spec_schema_transformations.py +6 -3
  301. airbyte_cdk/utils/stream_status_utils.py +8 -1
  302. airbyte_cdk/utils/traced_exception.py +61 -21
  303. airbyte_cdk-6.17.1.dev0.dist-info/METADATA +109 -0
  304. airbyte_cdk-6.17.1.dev0.dist-info/RECORD +350 -0
  305. {airbyte_cdk-0.72.0.dist-info → airbyte_cdk-6.17.1.dev0.dist-info}/WHEEL +1 -2
  306. airbyte_cdk-6.17.1.dev0.dist-info/entry_points.txt +3 -0
  307. airbyte_cdk/sources/declarative/create_partial.py +0 -92
  308. airbyte_cdk/sources/declarative/parsers/class_types_registry.py +0 -102
  309. airbyte_cdk/sources/declarative/parsers/default_implementation_registry.py +0 -64
  310. airbyte_cdk/sources/declarative/requesters/error_handlers/response_action.py +0 -16
  311. airbyte_cdk/sources/declarative/requesters/error_handlers/response_status.py +0 -68
  312. airbyte_cdk/sources/declarative/stream_slicers/cartesian_product_stream_slicer.py +0 -114
  313. airbyte_cdk/sources/deprecated/base_source.py +0 -94
  314. airbyte_cdk/sources/deprecated/client.py +0 -99
  315. airbyte_cdk/sources/singer/__init__.py +0 -8
  316. airbyte_cdk/sources/singer/singer_helpers.py +0 -304
  317. airbyte_cdk/sources/singer/source.py +0 -186
  318. airbyte_cdk/sources/streams/concurrent/partitions/record.py +0 -23
  319. airbyte_cdk/sources/streams/http/auth/__init__.py +0 -17
  320. airbyte_cdk/sources/streams/http/auth/core.py +0 -29
  321. airbyte_cdk/sources/streams/http/auth/oauth.py +0 -113
  322. airbyte_cdk/sources/streams/http/auth/token.py +0 -47
  323. airbyte_cdk/sources/streams/utils/stream_helper.py +0 -40
  324. airbyte_cdk/sources/utils/catalog_helpers.py +0 -22
  325. airbyte_cdk/sources/utils/schema_models.py +0 -84
  326. airbyte_cdk-0.72.0.dist-info/METADATA +0 -243
  327. airbyte_cdk-0.72.0.dist-info/RECORD +0 -466
  328. airbyte_cdk-0.72.0.dist-info/top_level.txt +0 -3
  329. source_declarative_manifest/main.py +0 -29
  330. unit_tests/connector_builder/__init__.py +0 -3
  331. unit_tests/connector_builder/test_connector_builder_handler.py +0 -871
  332. unit_tests/connector_builder/test_message_grouper.py +0 -713
  333. unit_tests/connector_builder/utils.py +0 -27
  334. unit_tests/destinations/test_destination.py +0 -243
  335. unit_tests/singer/test_singer_helpers.py +0 -56
  336. unit_tests/singer/test_singer_source.py +0 -112
  337. unit_tests/sources/__init__.py +0 -0
  338. unit_tests/sources/concurrent_source/__init__.py +0 -3
  339. unit_tests/sources/concurrent_source/test_concurrent_source_adapter.py +0 -106
  340. unit_tests/sources/declarative/__init__.py +0 -3
  341. unit_tests/sources/declarative/auth/__init__.py +0 -3
  342. unit_tests/sources/declarative/auth/test_oauth.py +0 -331
  343. unit_tests/sources/declarative/auth/test_selective_authenticator.py +0 -39
  344. unit_tests/sources/declarative/auth/test_session_token_auth.py +0 -182
  345. unit_tests/sources/declarative/auth/test_token_auth.py +0 -200
  346. unit_tests/sources/declarative/auth/test_token_provider.py +0 -73
  347. unit_tests/sources/declarative/checks/__init__.py +0 -3
  348. unit_tests/sources/declarative/checks/test_check_stream.py +0 -146
  349. unit_tests/sources/declarative/decoders/__init__.py +0 -0
  350. unit_tests/sources/declarative/decoders/test_json_decoder.py +0 -16
  351. unit_tests/sources/declarative/external_component.py +0 -13
  352. unit_tests/sources/declarative/extractors/__init__.py +0 -3
  353. unit_tests/sources/declarative/extractors/test_dpath_extractor.py +0 -55
  354. unit_tests/sources/declarative/extractors/test_record_filter.py +0 -55
  355. unit_tests/sources/declarative/extractors/test_record_selector.py +0 -179
  356. unit_tests/sources/declarative/incremental/__init__.py +0 -0
  357. unit_tests/sources/declarative/incremental/test_datetime_based_cursor.py +0 -860
  358. unit_tests/sources/declarative/incremental/test_per_partition_cursor.py +0 -406
  359. unit_tests/sources/declarative/incremental/test_per_partition_cursor_integration.py +0 -332
  360. unit_tests/sources/declarative/interpolation/__init__.py +0 -3
  361. unit_tests/sources/declarative/interpolation/test_filters.py +0 -80
  362. unit_tests/sources/declarative/interpolation/test_interpolated_boolean.py +0 -40
  363. unit_tests/sources/declarative/interpolation/test_interpolated_mapping.py +0 -35
  364. unit_tests/sources/declarative/interpolation/test_interpolated_nested_mapping.py +0 -45
  365. unit_tests/sources/declarative/interpolation/test_interpolated_string.py +0 -25
  366. unit_tests/sources/declarative/interpolation/test_jinja.py +0 -240
  367. unit_tests/sources/declarative/interpolation/test_macros.py +0 -73
  368. unit_tests/sources/declarative/parsers/__init__.py +0 -3
  369. unit_tests/sources/declarative/parsers/test_manifest_component_transformer.py +0 -406
  370. unit_tests/sources/declarative/parsers/test_manifest_reference_resolver.py +0 -139
  371. unit_tests/sources/declarative/parsers/test_model_to_component_factory.py +0 -1841
  372. unit_tests/sources/declarative/parsers/testing_components.py +0 -36
  373. unit_tests/sources/declarative/partition_routers/__init__.py +0 -3
  374. unit_tests/sources/declarative/partition_routers/test_list_partition_router.py +0 -155
  375. unit_tests/sources/declarative/partition_routers/test_single_partition_router.py +0 -14
  376. unit_tests/sources/declarative/partition_routers/test_substream_partition_router.py +0 -404
  377. unit_tests/sources/declarative/requesters/__init__.py +0 -3
  378. unit_tests/sources/declarative/requesters/error_handlers/__init__.py +0 -3
  379. unit_tests/sources/declarative/requesters/error_handlers/backoff_strategies/__init__.py +0 -3
  380. unit_tests/sources/declarative/requesters/error_handlers/backoff_strategies/test_constant_backoff.py +0 -34
  381. unit_tests/sources/declarative/requesters/error_handlers/backoff_strategies/test_exponential_backoff.py +0 -36
  382. unit_tests/sources/declarative/requesters/error_handlers/backoff_strategies/test_header_helper.py +0 -38
  383. unit_tests/sources/declarative/requesters/error_handlers/backoff_strategies/test_wait_time_from_header.py +0 -35
  384. unit_tests/sources/declarative/requesters/error_handlers/backoff_strategies/test_wait_until_time_from_header.py +0 -64
  385. unit_tests/sources/declarative/requesters/error_handlers/test_composite_error_handler.py +0 -213
  386. unit_tests/sources/declarative/requesters/error_handlers/test_default_error_handler.py +0 -178
  387. unit_tests/sources/declarative/requesters/error_handlers/test_http_response_filter.py +0 -121
  388. unit_tests/sources/declarative/requesters/error_handlers/test_response_status.py +0 -44
  389. unit_tests/sources/declarative/requesters/paginators/__init__.py +0 -3
  390. unit_tests/sources/declarative/requesters/paginators/test_cursor_pagination_strategy.py +0 -64
  391. unit_tests/sources/declarative/requesters/paginators/test_default_paginator.py +0 -313
  392. unit_tests/sources/declarative/requesters/paginators/test_no_paginator.py +0 -12
  393. unit_tests/sources/declarative/requesters/paginators/test_offset_increment.py +0 -58
  394. unit_tests/sources/declarative/requesters/paginators/test_page_increment.py +0 -70
  395. unit_tests/sources/declarative/requesters/paginators/test_request_option.py +0 -43
  396. unit_tests/sources/declarative/requesters/paginators/test_stop_condition.py +0 -105
  397. unit_tests/sources/declarative/requesters/request_options/__init__.py +0 -3
  398. unit_tests/sources/declarative/requesters/request_options/test_interpolated_request_options_provider.py +0 -101
  399. unit_tests/sources/declarative/requesters/test_http_requester.py +0 -974
  400. unit_tests/sources/declarative/requesters/test_interpolated_request_input_provider.py +0 -32
  401. unit_tests/sources/declarative/retrievers/__init__.py +0 -3
  402. unit_tests/sources/declarative/retrievers/test_simple_retriever.py +0 -542
  403. unit_tests/sources/declarative/schema/__init__.py +0 -6
  404. unit_tests/sources/declarative/schema/source_test/SourceTest.py +0 -8
  405. unit_tests/sources/declarative/schema/source_test/__init__.py +0 -3
  406. unit_tests/sources/declarative/schema/test_default_schema_loader.py +0 -32
  407. unit_tests/sources/declarative/schema/test_inline_schema_loader.py +0 -19
  408. unit_tests/sources/declarative/schema/test_json_file_schema_loader.py +0 -26
  409. unit_tests/sources/declarative/states/__init__.py +0 -3
  410. unit_tests/sources/declarative/stream_slicers/__init__.py +0 -3
  411. unit_tests/sources/declarative/stream_slicers/test_cartesian_product_stream_slicer.py +0 -225
  412. unit_tests/sources/declarative/test_create_partial.py +0 -83
  413. unit_tests/sources/declarative/test_declarative_stream.py +0 -103
  414. unit_tests/sources/declarative/test_manifest_declarative_source.py +0 -1260
  415. unit_tests/sources/declarative/test_types.py +0 -39
  416. unit_tests/sources/declarative/test_yaml_declarative_source.py +0 -148
  417. unit_tests/sources/file_based/__init__.py +0 -0
  418. unit_tests/sources/file_based/availability_strategy/__init__.py +0 -0
  419. unit_tests/sources/file_based/availability_strategy/test_default_file_based_availability_strategy.py +0 -100
  420. unit_tests/sources/file_based/config/__init__.py +0 -0
  421. unit_tests/sources/file_based/config/test_abstract_file_based_spec.py +0 -28
  422. unit_tests/sources/file_based/config/test_csv_format.py +0 -34
  423. unit_tests/sources/file_based/config/test_file_based_stream_config.py +0 -84
  424. unit_tests/sources/file_based/discovery_policy/__init__.py +0 -0
  425. unit_tests/sources/file_based/discovery_policy/test_default_discovery_policy.py +0 -31
  426. unit_tests/sources/file_based/file_types/__init__.py +0 -0
  427. unit_tests/sources/file_based/file_types/test_avro_parser.py +0 -243
  428. unit_tests/sources/file_based/file_types/test_csv_parser.py +0 -546
  429. unit_tests/sources/file_based/file_types/test_jsonl_parser.py +0 -158
  430. unit_tests/sources/file_based/file_types/test_parquet_parser.py +0 -274
  431. unit_tests/sources/file_based/file_types/test_unstructured_parser.py +0 -593
  432. unit_tests/sources/file_based/helpers.py +0 -70
  433. unit_tests/sources/file_based/in_memory_files_source.py +0 -211
  434. unit_tests/sources/file_based/scenarios/__init__.py +0 -0
  435. unit_tests/sources/file_based/scenarios/avro_scenarios.py +0 -744
  436. unit_tests/sources/file_based/scenarios/check_scenarios.py +0 -220
  437. unit_tests/sources/file_based/scenarios/concurrent_incremental_scenarios.py +0 -2844
  438. unit_tests/sources/file_based/scenarios/csv_scenarios.py +0 -3105
  439. unit_tests/sources/file_based/scenarios/file_based_source_builder.py +0 -91
  440. unit_tests/sources/file_based/scenarios/incremental_scenarios.py +0 -1926
  441. unit_tests/sources/file_based/scenarios/jsonl_scenarios.py +0 -930
  442. unit_tests/sources/file_based/scenarios/parquet_scenarios.py +0 -754
  443. unit_tests/sources/file_based/scenarios/scenario_builder.py +0 -234
  444. unit_tests/sources/file_based/scenarios/unstructured_scenarios.py +0 -608
  445. unit_tests/sources/file_based/scenarios/user_input_schema_scenarios.py +0 -746
  446. unit_tests/sources/file_based/scenarios/validation_policy_scenarios.py +0 -726
  447. unit_tests/sources/file_based/stream/__init__.py +0 -0
  448. unit_tests/sources/file_based/stream/concurrent/__init__.py +0 -0
  449. unit_tests/sources/file_based/stream/concurrent/test_adapters.py +0 -362
  450. unit_tests/sources/file_based/stream/concurrent/test_file_based_concurrent_cursor.py +0 -458
  451. unit_tests/sources/file_based/stream/test_default_file_based_cursor.py +0 -310
  452. unit_tests/sources/file_based/stream/test_default_file_based_stream.py +0 -244
  453. unit_tests/sources/file_based/test_file_based_scenarios.py +0 -320
  454. unit_tests/sources/file_based/test_file_based_stream_reader.py +0 -272
  455. unit_tests/sources/file_based/test_scenarios.py +0 -253
  456. unit_tests/sources/file_based/test_schema_helpers.py +0 -346
  457. unit_tests/sources/fixtures/__init__.py +0 -3
  458. unit_tests/sources/fixtures/source_test_fixture.py +0 -153
  459. unit_tests/sources/message/__init__.py +0 -0
  460. unit_tests/sources/message/test_repository.py +0 -153
  461. unit_tests/sources/streams/__init__.py +0 -0
  462. unit_tests/sources/streams/concurrent/__init__.py +0 -3
  463. unit_tests/sources/streams/concurrent/scenarios/__init__.py +0 -3
  464. unit_tests/sources/streams/concurrent/scenarios/incremental_scenarios.py +0 -250
  465. unit_tests/sources/streams/concurrent/scenarios/stream_facade_builder.py +0 -140
  466. unit_tests/sources/streams/concurrent/scenarios/stream_facade_scenarios.py +0 -452
  467. unit_tests/sources/streams/concurrent/scenarios/test_concurrent_scenarios.py +0 -76
  468. unit_tests/sources/streams/concurrent/scenarios/thread_based_concurrent_stream_scenarios.py +0 -418
  469. unit_tests/sources/streams/concurrent/scenarios/thread_based_concurrent_stream_source_builder.py +0 -142
  470. unit_tests/sources/streams/concurrent/scenarios/utils.py +0 -55
  471. unit_tests/sources/streams/concurrent/test_adapters.py +0 -380
  472. unit_tests/sources/streams/concurrent/test_concurrent_read_processor.py +0 -684
  473. unit_tests/sources/streams/concurrent/test_cursor.py +0 -139
  474. unit_tests/sources/streams/concurrent/test_datetime_state_converter.py +0 -369
  475. unit_tests/sources/streams/concurrent/test_default_stream.py +0 -197
  476. unit_tests/sources/streams/concurrent/test_partition_enqueuer.py +0 -90
  477. unit_tests/sources/streams/concurrent/test_partition_reader.py +0 -67
  478. unit_tests/sources/streams/concurrent/test_thread_pool_manager.py +0 -106
  479. unit_tests/sources/streams/http/__init__.py +0 -0
  480. unit_tests/sources/streams/http/auth/__init__.py +0 -0
  481. unit_tests/sources/streams/http/auth/test_auth.py +0 -173
  482. unit_tests/sources/streams/http/requests_native_auth/__init__.py +0 -0
  483. unit_tests/sources/streams/http/requests_native_auth/test_requests_native_auth.py +0 -423
  484. unit_tests/sources/streams/http/test_availability_strategy.py +0 -180
  485. unit_tests/sources/streams/http/test_http.py +0 -635
  486. unit_tests/sources/streams/test_availability_strategy.py +0 -70
  487. unit_tests/sources/streams/test_call_rate.py +0 -300
  488. unit_tests/sources/streams/test_stream_read.py +0 -405
  489. unit_tests/sources/streams/test_streams_core.py +0 -184
  490. unit_tests/sources/test_abstract_source.py +0 -1442
  491. unit_tests/sources/test_concurrent_source.py +0 -112
  492. unit_tests/sources/test_config.py +0 -92
  493. unit_tests/sources/test_connector_state_manager.py +0 -482
  494. unit_tests/sources/test_http_logger.py +0 -252
  495. unit_tests/sources/test_integration_source.py +0 -86
  496. unit_tests/sources/test_source.py +0 -684
  497. unit_tests/sources/test_source_read.py +0 -460
  498. unit_tests/test/__init__.py +0 -0
  499. unit_tests/test/mock_http/__init__.py +0 -0
  500. unit_tests/test/mock_http/test_matcher.py +0 -53
  501. unit_tests/test/mock_http/test_mocker.py +0 -214
  502. unit_tests/test/mock_http/test_request.py +0 -117
  503. unit_tests/test/mock_http/test_response_builder.py +0 -177
  504. unit_tests/test/test_entrypoint_wrapper.py +0 -240
  505. unit_tests/utils/__init__.py +0 -0
  506. unit_tests/utils/test_datetime_format_inferrer.py +0 -60
  507. unit_tests/utils/test_mapping_helpers.py +0 -54
  508. unit_tests/utils/test_message_utils.py +0 -91
  509. unit_tests/utils/test_rate_limiting.py +0 -26
  510. unit_tests/utils/test_schema_inferrer.py +0 -202
  511. unit_tests/utils/test_secret_utils.py +0 -135
  512. unit_tests/utils/test_stream_status_utils.py +0 -61
  513. unit_tests/utils/test_traced_exception.py +0 -107
  514. /airbyte_cdk/sources/{deprecated → declarative/async_job}/__init__.py +0 -0
  515. {source_declarative_manifest → airbyte_cdk/sources/declarative/migrations}/__init__.py +0 -0
  516. {unit_tests/destinations → airbyte_cdk/sql}/__init__.py +0 -0
  517. {unit_tests/singer → airbyte_cdk/sql/_util}/__init__.py +0 -0
  518. {airbyte_cdk-0.72.0.dist-info → airbyte_cdk-6.17.1.dev0.dist-info}/LICENSE.txt +0 -0
@@ -1,112 +0,0 @@
1
- #
2
- # Copyright (c) 2023 Airbyte, Inc., all rights reserved.
3
- #
4
- import concurrent
5
- import logging
6
- from typing import Any, Callable, Dict, Iterable, Mapping, Optional, Tuple
7
- from unittest.mock import Mock
8
-
9
- from airbyte_cdk.models import SyncMode
10
- from airbyte_cdk.sources.concurrent_source.concurrent_source import ConcurrentSource
11
- from airbyte_cdk.sources.concurrent_source.thread_pool_manager import ThreadPoolManager
12
- from airbyte_cdk.sources.message import InMemoryMessageRepository, MessageRepository
13
- from airbyte_cdk.sources.streams.concurrent.abstract_stream import AbstractStream
14
- from airbyte_cdk.sources.streams.concurrent.availability_strategy import StreamAvailability, StreamAvailable, StreamUnavailable
15
- from airbyte_cdk.sources.streams.concurrent.cursor import Cursor, FinalStateCursor
16
- from airbyte_cdk.sources.streams.concurrent.partitions.partition import Partition
17
- from airbyte_cdk.sources.streams.concurrent.partitions.record import Record
18
- from airbyte_protocol.models import AirbyteStream
19
-
20
- logger = logging.getLogger("airbyte")
21
-
22
-
23
- class _MockSource(ConcurrentSource):
24
- def __init__(
25
- self,
26
- check_lambda: Callable[[], Tuple[bool, Optional[Any]]] = None,
27
- per_stream: bool = True,
28
- message_repository: MessageRepository = InMemoryMessageRepository(),
29
- threadpool: ThreadPoolManager = ThreadPoolManager(
30
- concurrent.futures.ThreadPoolExecutor(max_workers=1, thread_name_prefix="workerpool"), logger
31
- ),
32
- exception_on_missing_stream: bool = True,
33
- ):
34
- super().__init__(threadpool, Mock(), Mock(), message_repository)
35
- self.check_lambda = check_lambda
36
- self.per_stream = per_stream
37
- self.exception_on_missing_stream = exception_on_missing_stream
38
- self._message_repository = message_repository
39
-
40
-
41
- MESSAGE_FROM_REPOSITORY = Mock()
42
-
43
-
44
- class _MockStream(AbstractStream):
45
- def __init__(self, name: str, message_repository: MessageRepository, available: bool = True, json_schema: Dict[str, Any] = {}):
46
- self._name = name
47
- self._available = available
48
- self._json_schema = json_schema
49
- self._message_repository = message_repository
50
-
51
- def generate_partitions(self) -> Iterable[Partition]:
52
- yield _MockPartition(self._name)
53
-
54
- @property
55
- def name(self) -> str:
56
- return self._name
57
-
58
- @property
59
- def cursor_field(self) -> Optional[str]:
60
- raise NotImplementedError
61
-
62
- def check_availability(self) -> StreamAvailability:
63
- if self._available:
64
- return StreamAvailable()
65
- else:
66
- return StreamUnavailable("stream is unavailable")
67
-
68
- def get_json_schema(self) -> Mapping[str, Any]:
69
- return self._json_schema
70
-
71
- def as_airbyte_stream(self) -> AirbyteStream:
72
- return AirbyteStream(name=self.name, json_schema=self.get_json_schema(), supported_sync_modes=[SyncMode.full_refresh])
73
-
74
- def log_stream_sync_configuration(self) -> None:
75
- raise NotImplementedError
76
-
77
- @property
78
- def cursor(self) -> Cursor:
79
- return FinalStateCursor(stream_name=self._name, stream_namespace=None, message_repository=self._message_repository)
80
-
81
-
82
- class _MockPartition(Partition):
83
- def __init__(self, name: str):
84
- self._name = name
85
- self._closed = False
86
-
87
- def read(self) -> Iterable[Record]:
88
- yield from [Record({"key": "value"}, self._name)]
89
-
90
- def to_slice(self) -> Optional[Mapping[str, Any]]:
91
- return {}
92
-
93
- def stream_name(self) -> str:
94
- return self._name
95
-
96
- def close(self) -> None:
97
- self._closed = True
98
-
99
- def is_closed(self) -> bool:
100
- return self._closed
101
-
102
- def __hash__(self) -> int:
103
- return hash(self._name)
104
-
105
-
106
- def test_concurrent_source_reading_from_no_streams():
107
- message_repository = InMemoryMessageRepository()
108
- stream = _MockStream("my_stream", message_repository,False, {})
109
- source = _MockSource(message_repository=message_repository)
110
- messages = []
111
- for m in source.read([stream]):
112
- messages.append(m)
@@ -1,92 +0,0 @@
1
- #
2
- # Copyright (c) 2023 Airbyte, Inc., all rights reserved.
3
- #
4
-
5
- from typing import List, Union
6
-
7
- from airbyte_cdk.sources.config import BaseConfig
8
- from pydantic import BaseModel, Field
9
-
10
-
11
- class InnerClass(BaseModel):
12
- field1: str
13
- field2: int
14
-
15
-
16
- class Choice1(BaseModel):
17
- selected_strategy = Field("option1", const=True)
18
-
19
- name: str
20
- count: int
21
-
22
-
23
- class Choice2(BaseModel):
24
- selected_strategy = Field("option2", const=True)
25
-
26
- sequence: List[str]
27
-
28
-
29
- class SomeSourceConfig(BaseConfig):
30
- class Config:
31
- title = "Some Source"
32
-
33
- items: List[InnerClass]
34
- choice: Union[Choice1, Choice2]
35
-
36
-
37
- class TestBaseConfig:
38
- EXPECTED_SCHEMA = {
39
- "properties": {
40
- "choice": {
41
- "oneOf": [
42
- {
43
- "properties": {
44
- "count": {"title": "Count", "type": "integer"},
45
- "name": {"title": "Name", "type": "string"},
46
- "selected_strategy": {
47
- "const": "option1",
48
- "title": "Selected " "Strategy",
49
- "type": "string",
50
- "default": "option1",
51
- },
52
- },
53
- "required": ["name", "count"],
54
- "title": "Choice1",
55
- "type": "object",
56
- },
57
- {
58
- "properties": {
59
- "selected_strategy": {
60
- "const": "option2",
61
- "title": "Selected " "Strategy",
62
- "type": "string",
63
- "default": "option2",
64
- },
65
- "sequence": {"items": {"type": "string"}, "title": "Sequence", "type": "array"},
66
- },
67
- "required": ["sequence"],
68
- "title": "Choice2",
69
- "type": "object",
70
- },
71
- ],
72
- "title": "Choice",
73
- },
74
- "items": {
75
- "items": {
76
- "properties": {"field1": {"title": "Field1", "type": "string"}, "field2": {"title": "Field2", "type": "integer"}},
77
- "required": ["field1", "field2"],
78
- "title": "InnerClass",
79
- "type": "object",
80
- },
81
- "title": "Items",
82
- "type": "array",
83
- },
84
- },
85
- "required": ["items", "choice"],
86
- "title": "Some Source",
87
- "type": "object",
88
- }
89
-
90
- def test_schema_postprocessing(self):
91
- schema = SomeSourceConfig.schema()
92
- assert schema == self.EXPECTED_SCHEMA
@@ -1,482 +0,0 @@
1
- #
2
- # Copyright (c) 2023 Airbyte, Inc., all rights reserved.
3
- #
4
-
5
- from contextlib import nullcontext as does_not_raise
6
- from typing import List
7
-
8
- import pytest
9
- from airbyte_cdk.models import (
10
- AirbyteMessage,
11
- AirbyteStateBlob,
12
- AirbyteStateMessage,
13
- AirbyteStateType,
14
- AirbyteStream,
15
- AirbyteStreamState,
16
- StreamDescriptor,
17
- SyncMode,
18
- )
19
- from airbyte_cdk.models import Type as MessageType
20
- from airbyte_cdk.sources.connector_state_manager import ConnectorStateManager, HashableStreamDescriptor
21
-
22
-
23
- @pytest.mark.parametrize(
24
- "input_stream_state, expected_stream_state, expected_error",
25
- (
26
- pytest.param(
27
- [
28
- {
29
- "type": AirbyteStateType.STREAM,
30
- "stream": {"stream_descriptor": {"name": "actors", "namespace": "public"}, "stream_state": {"id": "mando_michael"}},
31
- },
32
- {
33
- "type": AirbyteStateType.STREAM,
34
- "stream": {"stream_descriptor": {"name": "actresses", "namespace": "public"}, "stream_state": {"id": "seehorn_rhea"}},
35
- },
36
- ],
37
- {
38
- HashableStreamDescriptor(name="actors", namespace="public"): AirbyteStateBlob.parse_obj({"id": "mando_michael"}),
39
- HashableStreamDescriptor(name="actresses", namespace="public"): AirbyteStateBlob.parse_obj({"id": "seehorn_rhea"}),
40
- },
41
- does_not_raise(),
42
- id="test_incoming_per_stream_state",
43
- ),
44
- pytest.param(
45
- [
46
- {
47
- "type": AirbyteStateType.LEGACY,
48
- "data": {"actors": {"id": "fabian_patrick"}, "actresses": {"id": "seehorn_rhea"}, "writers": {"id": "gilligan_vince"}},
49
- }
50
- ],
51
- {
52
- HashableStreamDescriptor(name="actors", namespace="public"): AirbyteStateBlob.parse_obj({"id": "fabian_patrick"}),
53
- HashableStreamDescriptor(name="actresses"): AirbyteStateBlob.parse_obj({"id": "seehorn_rhea"}),
54
- HashableStreamDescriptor(name="writers"): AirbyteStateBlob.parse_obj({"id": "gilligan_vince"}),
55
- },
56
- does_not_raise(),
57
- id="test_incoming_legacy_state_and_uses_stream_namespace",
58
- ),
59
- pytest.param([], {}, does_not_raise(), id="test_incoming_empty_stream_state"),
60
- pytest.param(
61
- [{"type": AirbyteStateType.STREAM, "stream": {"stream_descriptor": {"name": "actresses", "namespace": "public"}}}],
62
- {HashableStreamDescriptor(name="actresses", namespace="public"): None},
63
- does_not_raise(),
64
- id="test_stream_states_that_have_none_state_blob",
65
- ),
66
- pytest.param(
67
- [
68
- {
69
- "type": AirbyteStateType.GLOBAL,
70
- "global": {
71
- "shared_state": {"television": "better_call_saul"},
72
- "stream_states": [
73
- {
74
- "stream_descriptor": StreamDescriptor(name="actors", namespace="public"),
75
- "stream_state": AirbyteStateBlob.parse_obj({"id": "mando_michael"}),
76
- },
77
- {
78
- "stream_descriptor": StreamDescriptor(name="actresses", namespace="public"),
79
- "stream_state": AirbyteStateBlob.parse_obj({"id": "seehorn_rhea"}),
80
- },
81
- ],
82
- },
83
- },
84
- ],
85
- {
86
- HashableStreamDescriptor(name="actors", namespace="public"): AirbyteStateBlob.parse_obj({"id": "mando_michael"}),
87
- HashableStreamDescriptor(name="actresses", namespace="public"): AirbyteStateBlob.parse_obj({"id": "seehorn_rhea"}),
88
- },
89
- pytest.raises(ValueError),
90
- id="test_incoming_global_state_with_shared_state_throws_error",
91
- ),
92
- pytest.param(
93
- [
94
- {
95
- "type": AirbyteStateType.GLOBAL,
96
- "global": {
97
- "stream_states": [
98
- {"stream_descriptor": {"name": "actors", "namespace": "public"}, "stream_state": {"id": "mando_michael"}},
99
- ],
100
- },
101
- },
102
- ],
103
- {
104
- HashableStreamDescriptor(name="actors", namespace="public"): AirbyteStateBlob.parse_obj({"id": "mando_michael"}),
105
- },
106
- does_not_raise(),
107
- id="test_incoming_global_state_without_shared",
108
- ),
109
- pytest.param(
110
- [
111
- {
112
- "type": AirbyteStateType.GLOBAL,
113
- "global": {
114
- "shared_state": None,
115
- "stream_states": [
116
- {
117
- "stream_descriptor": StreamDescriptor(name="actors", namespace="public"),
118
- "stream_state": AirbyteStateBlob.parse_obj({"id": "mando_michael"}),
119
- },
120
- ],
121
- },
122
- },
123
- ],
124
- {
125
- HashableStreamDescriptor(name="actors", namespace="public"): AirbyteStateBlob.parse_obj({"id": "mando_michael"}),
126
- },
127
- does_not_raise(),
128
- id="test_incoming_global_state_with_none_shared",
129
- ),
130
- pytest.param(
131
- [
132
- {
133
- "type": AirbyteStateType.GLOBAL,
134
- "global": {
135
- "stream_states": [
136
- {"stream_descriptor": {"name": "actresses", "namespace": "public"}},
137
- ],
138
- },
139
- },
140
- ],
141
- {HashableStreamDescriptor(name="actresses", namespace="public"): None},
142
- does_not_raise(),
143
- id="test_incoming_global_state_without_stream_state",
144
- ),
145
- pytest.param(
146
- {"actors": {"id": "esposito_giancarlo"}, "actresses": {"id": "seehorn_rhea"}},
147
- {
148
- HashableStreamDescriptor(name="actors", namespace="public"): AirbyteStateBlob.parse_obj({"id": "esposito_giancarlo"}),
149
- HashableStreamDescriptor(name="actresses"): AirbyteStateBlob.parse_obj({"id": "seehorn_rhea"}),
150
- },
151
- does_not_raise(),
152
- id="test_incoming_legacy_json_blob_and_uses_stream_namespace",
153
- ),
154
- pytest.param({}, {}, does_not_raise(), id="test_legacy_state_empty_object"),
155
- pytest.param("strings_are_not_allowed", {}, pytest.raises(ValueError), id="test_value_error_is_raised_on_invalid_state_input"),
156
- ),
157
- )
158
- def test_initialize_state_manager(input_stream_state, expected_stream_state, expected_error):
159
- stream_to_instance_map = {
160
- "actors": AirbyteStream(
161
- name="actors",
162
- namespace="public",
163
- json_schema={},
164
- supported_sync_modes=[SyncMode.full_refresh, SyncMode.incremental],
165
- )
166
- }
167
-
168
- if isinstance(input_stream_state, List):
169
- input_stream_state = [AirbyteStateMessage.parse_obj(state_obj) for state_obj in list(input_stream_state)]
170
-
171
- with expected_error:
172
- state_manager = ConnectorStateManager(stream_to_instance_map, input_stream_state)
173
-
174
- assert state_manager.per_stream_states == expected_stream_state
175
-
176
-
177
- @pytest.mark.parametrize(
178
- "input_state, stream_name, namespace, expected_state",
179
- [
180
- pytest.param(
181
- [
182
- {
183
- "type": AirbyteStateType.STREAM,
184
- "stream": {"stream_descriptor": {"name": "users", "namespace": "public"}, "stream_state": {"created_at": 12345}},
185
- },
186
- {
187
- "type": AirbyteStateType.STREAM,
188
- "stream": {"stream_descriptor": {"name": "accounts", "namespace": "public"}, "stream_state": {"id": "abc"}},
189
- },
190
- ],
191
- "users",
192
- "public",
193
- {"created_at": 12345},
194
- id="test_get_stream_only",
195
- ),
196
- pytest.param(
197
- [
198
- {
199
- "type": AirbyteStateType.STREAM,
200
- "stream": {"stream_descriptor": {"name": "users"}, "stream_state": {"created_at": 12345}},
201
- },
202
- {"type": AirbyteStateType.STREAM, "stream": {"stream_descriptor": {"name": "accounts"}, "stream_state": {"id": "abc"}}},
203
- ],
204
- "users",
205
- None,
206
- {"created_at": 12345},
207
- id="test_get_stream_without_namespace",
208
- ),
209
- pytest.param(
210
- [
211
- {"type": AirbyteStateType.STREAM, "stream": {"stream_descriptor": {"name": "users"}}},
212
- {"type": AirbyteStateType.STREAM, "stream": {"stream_descriptor": {"name": "accounts"}, "stream_state": {"id": "abc"}}},
213
- ],
214
- "users",
215
- None,
216
- {},
217
- id="test_get_stream_without_stream_state",
218
- ),
219
- pytest.param(
220
- [{"type": AirbyteStateType.LEGACY, "data": {"users": {"created_at": 12345}, "accounts": {"id": "abc"}}}],
221
- "users",
222
- "public",
223
- {"created_at": 12345},
224
- id="test_get_stream_from_legacy_state",
225
- ),
226
- pytest.param(
227
- [
228
- {
229
- "type": AirbyteStateType.STREAM,
230
- "stream": {"stream_descriptor": {"name": "users", "namespace": "public"}, "stream_state": {"created_at": 12345}},
231
- },
232
- {
233
- "type": AirbyteStateType.STREAM,
234
- "stream": {"stream_descriptor": {"name": "accounts", "namespace": "public"}, "stream_state": {"id": "abc"}},
235
- },
236
- ],
237
- "missing",
238
- "public",
239
- {},
240
- id="test_get_missing_stream",
241
- ),
242
- pytest.param(
243
- [
244
- {
245
- "type": AirbyteStateType.STREAM,
246
- "stream": {"stream_descriptor": {"name": "users", "namespace": "public"}, "stream_state": {"created_at": 12345}},
247
- },
248
- {
249
- "type": AirbyteStateType.STREAM,
250
- "stream": {"stream_descriptor": {"name": "accounts", "namespace": "public"}, "stream_state": {"id": "abc"}},
251
- },
252
- ],
253
- "users",
254
- "wrong_namespace",
255
- {},
256
- id="test_get_stream_wrong_namespace",
257
- ),
258
- pytest.param([], "users", "public", {}, id="test_get_empty_stream_state_defaults_to_empty_dictionary"),
259
- pytest.param(
260
- [
261
- {
262
- "type": AirbyteStateType.STREAM,
263
- "stream": {"stream_descriptor": {"name": "users", "namespace": "public"}, "stream_state": None},
264
- },
265
- ],
266
- "users",
267
- "public",
268
- {},
269
- id="test_get_stream_with_stream_state_none_returns_empty_map",
270
- ),
271
- ],
272
- )
273
- def test_get_stream_state(input_state, stream_name, namespace, expected_state):
274
- stream_to_instance_map = {
275
- stream_name: AirbyteStream(
276
- name=stream_name, namespace=namespace, json_schema={}, supported_sync_modes=[SyncMode.full_refresh, SyncMode.incremental]
277
- )
278
- }
279
- state_messages = [AirbyteStateMessage.parse_obj(state_obj) for state_obj in list(input_state)]
280
- state_manager = ConnectorStateManager(stream_to_instance_map, state_messages)
281
-
282
- actual_state = state_manager.get_stream_state(stream_name, namespace)
283
-
284
- assert actual_state == expected_state
285
-
286
-
287
- def test_get_state_returns_deep_copy():
288
- input_state = [
289
- AirbyteStateMessage(
290
- type=AirbyteStateType.STREAM,
291
- stream=AirbyteStreamState(
292
- stream_descriptor=StreamDescriptor(name="episodes", namespace="public"),
293
- stream_state=AirbyteStateBlob.parse_obj({"id": [109]}),
294
- ),
295
- )
296
- ]
297
- state_manager = ConnectorStateManager({}, input_state)
298
-
299
- per_stream_state = state_manager.get_stream_state("episodes", "public")
300
- per_stream_state["id"].append(309)
301
-
302
- assert state_manager.get_stream_state("episodes", "public") == {"id": [109]}
303
-
304
-
305
- @pytest.mark.parametrize(
306
- "start_state, update_name, update_namespace, update_value, expected_legacy_state",
307
- [
308
- pytest.param(
309
- [
310
- {
311
- "type": AirbyteStateType.STREAM,
312
- "stream": {"stream_descriptor": {"name": "actors", "namespace": "public"}, "stream_state": {"id": "mckean_michael"}},
313
- },
314
- {
315
- "type": AirbyteStateType.STREAM,
316
- "stream": {"stream_descriptor": {"name": "actresses", "namespace": "public"}, "stream_state": {"id": "seehorn_rhea"}},
317
- },
318
- ],
319
- "actors",
320
- "public",
321
- {"id": "fabian_patrick"},
322
- {"actors": {"id": "fabian_patrick"}, "actresses": {"id": "seehorn_rhea"}},
323
- id="test_update_existing_stream_state",
324
- ),
325
- pytest.param(
326
- [],
327
- "actresses",
328
- None,
329
- {"id": "seehorn_rhea"},
330
- {"actresses": {"id": "seehorn_rhea"}},
331
- id="test_update_first_time_sync_without_namespace",
332
- ),
333
- pytest.param(
334
- [
335
- {
336
- "type": AirbyteStateType.STREAM,
337
- "stream": {"stream_descriptor": {"name": "actresses", "namespace": "public"}, "stream_state": {"id": "seehorn_rhea"}},
338
- }
339
- ],
340
- "actors",
341
- "public",
342
- {"id": "banks_jonathan"},
343
- {"actors": {"id": "banks_jonathan"}, "actresses": {"id": "seehorn_rhea"}},
344
- id="test_update_missing_state",
345
- ),
346
- pytest.param(
347
- [
348
- {
349
- "type": AirbyteStateType.STREAM,
350
- "stream": {"stream_descriptor": {"name": "actresses", "namespace": "public"}, "stream_state": {"id": "seehorn_rhea"}},
351
- }
352
- ],
353
- "actors",
354
- "public",
355
- {"id": "banks_jonathan"},
356
- {"actors": {"id": "banks_jonathan"}, "actresses": {"id": "seehorn_rhea"}},
357
- id="test_ignore_when_per_stream_state_value_is_none",
358
- ),
359
- ],
360
- )
361
- def test_update_state_for_stream(start_state, update_name, update_namespace, update_value, expected_legacy_state):
362
- state_messages = [AirbyteStateMessage.parse_obj(state_obj) for state_obj in list(start_state)]
363
- state_manager = ConnectorStateManager({}, state_messages)
364
-
365
- state_manager.update_state_for_stream(update_name, update_namespace, update_value)
366
-
367
- assert state_manager.per_stream_states[
368
- HashableStreamDescriptor(name=update_name, namespace=update_namespace)
369
- ] == AirbyteStateBlob.parse_obj(update_value)
370
-
371
-
372
- @pytest.mark.parametrize(
373
- "start_state, update_name, update_namespace, expected_state_message",
374
- [
375
- pytest.param(
376
- [
377
- AirbyteStateMessage(
378
- type=AirbyteStateType.STREAM,
379
- stream=AirbyteStreamState(
380
- stream_descriptor=StreamDescriptor(name="episodes", namespace="public"),
381
- stream_state=AirbyteStateBlob.parse_obj({"created_at": "2022_05_22"}),
382
- ),
383
- ),
384
- AirbyteStateMessage(
385
- type=AirbyteStateType.STREAM,
386
- stream=AirbyteStreamState(
387
- stream_descriptor=StreamDescriptor(name="seasons", namespace="public"),
388
- stream_state=AirbyteStateBlob.parse_obj({"id": 1}),
389
- ),
390
- ),
391
- ],
392
- "episodes",
393
- "public",
394
- AirbyteMessage(
395
- type=MessageType.STATE,
396
- state=AirbyteStateMessage(
397
- type=AirbyteStateType.STREAM,
398
- stream=AirbyteStreamState(
399
- stream_descriptor=StreamDescriptor(name="episodes", namespace="public"),
400
- stream_state=AirbyteStateBlob.parse_obj({"created_at": "2022_05_22"}),
401
- ),
402
- ),
403
- ),
404
- id="test_emit_state_message_with_stream_and_legacy",
405
- ),
406
- pytest.param(
407
- [
408
- AirbyteStateMessage(
409
- type=AirbyteStateType.STREAM,
410
- stream=AirbyteStreamState(
411
- stream_descriptor=StreamDescriptor(name="episodes", namespace="public"),
412
- stream_state=None,
413
- ),
414
- ),
415
- ],
416
- "episodes",
417
- "public",
418
- AirbyteMessage(
419
- type=MessageType.STATE,
420
- state=AirbyteStateMessage(
421
- type=AirbyteStateType.STREAM,
422
- stream=AirbyteStreamState(
423
- stream_descriptor=StreamDescriptor(name="episodes", namespace="public"),
424
- stream_state=AirbyteStateBlob(),
425
- ),
426
- ),
427
- ),
428
- id="test_always_emit_message_with_stream_state_blob",
429
- ),
430
- pytest.param(
431
- [
432
- AirbyteStateMessage(
433
- type=AirbyteStateType.STREAM,
434
- stream=AirbyteStreamState(
435
- stream_descriptor=StreamDescriptor(name="episodes", namespace="public"),
436
- stream_state=AirbyteStateBlob.parse_obj({"id": 507}),
437
- ),
438
- )
439
- ],
440
- "missing",
441
- "public",
442
- AirbyteMessage(
443
- type=MessageType.STATE,
444
- state=AirbyteStateMessage(
445
- type=AirbyteStateType.STREAM,
446
- stream=AirbyteStreamState(
447
- stream_descriptor=StreamDescriptor(name="missing", namespace="public"), stream_state=AirbyteStateBlob()
448
- ),
449
- ),
450
- ),
451
- id="test_emit_state_nonexistent_stream_name",
452
- ),
453
- pytest.param(
454
- [
455
- AirbyteStateMessage(
456
- type=AirbyteStateType.STREAM,
457
- stream=AirbyteStreamState(
458
- stream_descriptor=StreamDescriptor(name="episodes", namespace="public"),
459
- stream_state=AirbyteStateBlob.parse_obj({"id": 507}),
460
- ),
461
- )
462
- ],
463
- "episodes",
464
- "nonexistent",
465
- AirbyteMessage(
466
- type=MessageType.STATE,
467
- state=AirbyteStateMessage(
468
- type=AirbyteStateType.STREAM,
469
- stream=AirbyteStreamState(
470
- stream_descriptor=StreamDescriptor(name="episodes", namespace="nonexistent"), stream_state=AirbyteStateBlob()
471
- ),
472
- ),
473
- ),
474
- id="test_emit_state_wrong_namespace",
475
- ),
476
- ],
477
- )
478
- def test_create_state_message(start_state, update_name, update_namespace, expected_state_message):
479
- state_manager = ConnectorStateManager({}, start_state)
480
-
481
- actual_state_message = state_manager.create_state_message(stream_name=update_name, namespace=update_namespace)
482
- assert actual_state_message == expected_state_message