airbyte-cdk 0.72.1__py3-none-any.whl → 6.17.1.dev1__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 +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 +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 +340 -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 +174 -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 +1319 -603
  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 +1759 -225
  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 +313 -48
  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.dev1.dist-info/METADATA +109 -0
  304. airbyte_cdk-6.17.1.dev1.dist-info/RECORD +350 -0
  305. {airbyte_cdk-0.72.1.dist-info → airbyte_cdk-6.17.1.dev1.dist-info}/WHEEL +1 -2
  306. airbyte_cdk-6.17.1.dev1.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-0.72.1.dist-info/METADATA +0 -243
  327. airbyte_cdk-0.72.1.dist-info/RECORD +0 -466
  328. airbyte_cdk-0.72.1.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 -1847
  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-0.72.1.dist-info → airbyte_cdk-6.17.1.dev1.dist-info}/LICENSE.txt +0 -0
@@ -1,726 +0,0 @@
1
- #
2
- # Copyright (c) 2023 Airbyte, Inc., all rights reserved.
3
- #
4
-
5
-
6
- from airbyte_cdk.utils.traced_exception import AirbyteTracedException
7
- from unit_tests.sources.file_based.scenarios.file_based_source_builder import FileBasedSourceBuilder
8
- from unit_tests.sources.file_based.scenarios.scenario_builder import TestScenarioBuilder
9
-
10
- _base_single_stream_scenario = (
11
- TestScenarioBuilder()
12
- .set_source_builder(
13
- FileBasedSourceBuilder()
14
- .set_files(
15
- {
16
- "a.csv": {
17
- "contents": [
18
- ("col1", "col2"),
19
- ("val_a_11", "1"),
20
- ("val_a_12", "2"),
21
- ],
22
- "last_modified": "2023-06-05T03:54:07.000Z",
23
- },
24
- "b.csv": { # The records in this file do not conform to the schema
25
- "contents": [
26
- ("col1", "col2"),
27
- ("val_b_11", "this is text that will trigger validation policy"),
28
- ("val_b_12", "2"),
29
- ],
30
- "last_modified": "2023-06-05T03:54:07.000Z",
31
- },
32
- "c.csv": {
33
- "contents": [
34
- ("col1",),
35
- ("val_c_11",),
36
- ("val_c_12", "val_c_22"), # This record is not parsable
37
- ("val_c_13",),
38
- ],
39
- "last_modified": "2023-06-05T03:54:07.000Z",
40
- },
41
- "d.csv": {
42
- "contents": [
43
- ("col1",),
44
- ("val_d_11",),
45
- ],
46
- "last_modified": "2023-06-05T03:54:07.000Z",
47
- },
48
- }
49
- )
50
- .set_file_type("csv")
51
- )
52
- .set_expected_catalog(
53
- {
54
- "streams": [
55
- {
56
- "default_cursor_field": ["_ab_source_file_last_modified"],
57
- "json_schema": {
58
- "type": "object",
59
- "properties": {
60
- "col1": {
61
- "type": "string",
62
- },
63
- "col2": {
64
- "type": "integer",
65
- },
66
- "_ab_source_file_last_modified": {"type": "string"},
67
- "_ab_source_file_url": {"type": "string"},
68
- },
69
- },
70
- "name": "stream1",
71
- "source_defined_cursor": True,
72
- "supported_sync_modes": ["full_refresh", "incremental"],
73
- }
74
- ]
75
- }
76
- )
77
- )
78
-
79
-
80
- _base_multi_stream_scenario = (
81
- TestScenarioBuilder()
82
- .set_source_builder(
83
- FileBasedSourceBuilder()
84
- .set_files(
85
- {
86
- "a/a1.csv": {
87
- "contents": [
88
- ("col1", "col2"),
89
- ("val_aa1_11", "1"),
90
- ("val_aa1_12", "2"),
91
- ],
92
- "last_modified": "2023-06-05T03:54:07.000Z",
93
- },
94
- "a/a2.csv": {
95
- "contents": [
96
- ("col1", "col2"),
97
- ("val_aa2_11", "this is text that will trigger validation policy"),
98
- ("val_aa2_12", "2"),
99
- ],
100
- "last_modified": "2023-06-05T03:54:07.000Z",
101
- },
102
- "a/a3.csv": {
103
- "contents": [
104
- ("col1",),
105
- ("val_aa3_11",),
106
- ("val_aa3_12", "val_aa3_22"), # This record is not parsable
107
- ("val_aa3_13",),
108
- ],
109
- "last_modified": "2023-06-05T03:54:07.000Z",
110
- },
111
- "a/a4.csv": {
112
- "contents": [
113
- ("col1",),
114
- ("val_aa4_11",),
115
- ],
116
- "last_modified": "2023-06-05T03:54:07.000Z",
117
- },
118
- "b/b1.csv": { # The records in this file do not conform to the schema
119
- "contents": [
120
- ("col1", "col2"),
121
- ("val_bb1_11", "1"),
122
- ("val_bb1_12", "2"),
123
- ],
124
- "last_modified": "2023-06-05T03:54:07.000Z",
125
- },
126
- "b/b2.csv": {
127
- "contents": [
128
- ("col1", "col2"),
129
- ("val_bb2_11", "this is text that will trigger validation policy"),
130
- ("val_bb2_12", "2"),
131
- ],
132
- "last_modified": "2023-06-05T03:54:07.000Z",
133
- },
134
- "b/b3.csv": {
135
- "contents": [
136
- ("col1",),
137
- ("val_bb3_11",),
138
- ("val_bb3_12",),
139
- ],
140
- "last_modified": "2023-06-05T03:54:07.000Z",
141
- },
142
- }
143
- )
144
- .set_file_type("csv")
145
- )
146
- .set_expected_catalog(
147
- {
148
- "streams": [
149
- {
150
- "default_cursor_field": ["_ab_source_file_last_modified"],
151
- "json_schema": {
152
- "type": "object",
153
- "properties": {
154
- "col1": {
155
- "type": "string",
156
- },
157
- "col2": {
158
- "type": "integer",
159
- },
160
- "_ab_source_file_last_modified": {"type": "string"},
161
- "_ab_source_file_url": {"type": "string"},
162
- },
163
- },
164
- "name": "stream1",
165
- "source_defined_cursor": True,
166
- "supported_sync_modes": ["full_refresh", "incremental"],
167
- },
168
- {
169
- "json_schema": {
170
- "default_cursor_field": ["_ab_source_file_last_modified"],
171
- "type": "object",
172
- "properties": {
173
- "col1": {
174
- "type": "string",
175
- },
176
- "col2": {
177
- "type": "integer",
178
- },
179
- "_ab_source_file_last_modified": {"type": "string"},
180
- "_ab_source_file_url": {"type": "string"},
181
- },
182
- },
183
- "name": "stream2",
184
- "source_defined_cursor": True,
185
- "supported_sync_modes": ["full_refresh", "incremental"],
186
- },
187
- ]
188
- }
189
- )
190
- )
191
-
192
-
193
- skip_record_scenario_single_stream = (
194
- _base_single_stream_scenario.copy()
195
- .set_name("skip_record_scenario_single_stream")
196
- .set_config(
197
- {
198
- "streams": [
199
- {
200
- "name": "stream1",
201
- "format": {"filetype": "csv"},
202
- "globs": ["*.csv"],
203
- "validation_policy": "Skip Record",
204
- }
205
- ]
206
- }
207
- )
208
- .set_expected_records(
209
- [
210
- {
211
- "data": {
212
- "col1": "val_a_11",
213
- "col2": 1,
214
- "_ab_source_file_last_modified": "2023-06-05T03:54:07.000000Z",
215
- "_ab_source_file_url": "a.csv",
216
- },
217
- "stream": "stream1",
218
- },
219
- {
220
- "data": {
221
- "col1": "val_a_12",
222
- "col2": 2,
223
- "_ab_source_file_last_modified": "2023-06-05T03:54:07.000000Z",
224
- "_ab_source_file_url": "a.csv",
225
- },
226
- "stream": "stream1",
227
- },
228
- # {"data": {"col1": "val_b_11", "col2": "this is text that will trigger validation policy", "_ab_source_file_last_modified": "2023-06-05T03:54:07.000000Z", "_ab_source_file_url": "b.csv"}, "stream": "stream1"}, # This record is skipped because it does not conform
229
- {
230
- "data": {
231
- "col1": "val_b_12",
232
- "col2": 2,
233
- "_ab_source_file_last_modified": "2023-06-05T03:54:07.000000Z",
234
- "_ab_source_file_url": "b.csv",
235
- },
236
- "stream": "stream1",
237
- },
238
- {
239
- "data": {
240
- "col1": "val_c_11",
241
- "_ab_source_file_last_modified": "2023-06-05T03:54:07.000000Z",
242
- "_ab_source_file_url": "c.csv",
243
- },
244
- "stream": "stream1",
245
- },
246
- # {"data": {"col1": "val_c_12", None: "val_c_22", "_ab_source_file_last_modified": "2023-06-05T03:54:07.000000Z", "_ab_source_file_url": "c.csv"}, "stream": "stream1"}, # This record is malformed so should not be emitted
247
- # {"data": {"col1": "val_c_13", "_ab_source_file_last_modified": "2023-06-05T03:54:07.000000Z", "_ab_source_file_url": "c.csv"}, "stream": "stream1"}, # Skipped since previous record is malformed
248
- {
249
- "data": {
250
- "col1": "val_d_11",
251
- "_ab_source_file_last_modified": "2023-06-05T03:54:07.000000Z",
252
- "_ab_source_file_url": "d.csv",
253
- },
254
- "stream": "stream1",
255
- },
256
- ]
257
- )
258
- .set_expected_logs(
259
- {
260
- "read": [
261
- {
262
- "level": "WARN",
263
- "message": "Records in file did not pass validation policy. stream=stream1 file=b.csv n_skipped=1 validation_policy=skip_record",
264
- },
265
- {
266
- "level": "ERROR",
267
- "message": "Error parsing record. This could be due to a mismatch between the config's file type and the actual file type, or because the file or record is not parseable. stream=stream1 file=c.csv line_no=2 n_skipped=0",
268
- },
269
- {
270
- "level": "WARN",
271
- "message": "Could not cast the value to the expected type.: col2: value=this is text that will trigger validation policy,expected_type=integer",
272
- },
273
- ]
274
- }
275
- )
276
- .set_expected_read_error(
277
- AirbyteTracedException,
278
- "Please check the logged errors for more information.",
279
- )
280
- ).build()
281
-
282
-
283
- skip_record_scenario_multi_stream = (
284
- _base_multi_stream_scenario.copy()
285
- .set_name("skip_record_scenario_multi_stream")
286
- .set_config(
287
- {
288
- "streams": [
289
- {
290
- "name": "stream1",
291
- "format": {"filetype": "csv"},
292
- "globs": ["a/*.csv"],
293
- "validation_policy": "Skip Record",
294
- },
295
- {
296
- "name": "stream2",
297
- "format": {"filetype": "csv"},
298
- "globs": ["b/*.csv"],
299
- "validation_policy": "Skip Record",
300
- },
301
- ]
302
- }
303
- )
304
- .set_expected_records(
305
- [
306
- {
307
- "data": {
308
- "col1": "val_aa1_11",
309
- "col2": 1,
310
- "_ab_source_file_last_modified": "2023-06-05T03:54:07.000000Z",
311
- "_ab_source_file_url": "a/a1.csv",
312
- },
313
- "stream": "stream1",
314
- },
315
- {
316
- "data": {
317
- "col1": "val_aa1_12",
318
- "col2": 2,
319
- "_ab_source_file_last_modified": "2023-06-05T03:54:07.000000Z",
320
- "_ab_source_file_url": "a/a1.csv",
321
- },
322
- "stream": "stream1",
323
- },
324
- # {"data": {"col1": "val_aa2_11", "col2": "this is text that will trigger validation policy", "_ab_source_file_last_modified": "2023-06-05T03:54:07.000000Z", "_ab_source_file_url": "a/a2.csv"}, "stream": "stream1"}, # This record is skipped because it does not conform
325
- {
326
- "data": {
327
- "col1": "val_aa2_12",
328
- "col2": 2,
329
- "_ab_source_file_last_modified": "2023-06-05T03:54:07.000000Z",
330
- "_ab_source_file_url": "a/a2.csv",
331
- },
332
- "stream": "stream1",
333
- },
334
- {
335
- "data": {
336
- "col1": "val_aa3_11",
337
- "_ab_source_file_last_modified": "2023-06-05T03:54:07.000000Z",
338
- "_ab_source_file_url": "a/a3.csv",
339
- },
340
- "stream": "stream1",
341
- },
342
- # {"data": {"col1": "val_aa3_12", None: "val_aa3_22", "_ab_source_file_last_modified": "2023-06-05T03:54:07.000000Z", "_ab_source_file_url": "a/a3.csv"}, "stream": "stream1"}, # This record is malformed so should not be emitted
343
- # {"data": {"col1": "val_aa3_13", "_ab_source_file_last_modified": "2023-06-05T03:54:07.000000Z", "_ab_source_file_url": "a/a3.csv"}, "stream": "stream1"}, # Skipped since previous record is malformed
344
- {
345
- "data": {
346
- "col1": "val_aa4_11",
347
- "_ab_source_file_last_modified": "2023-06-05T03:54:07.000000Z",
348
- "_ab_source_file_url": "a/a4.csv",
349
- },
350
- "stream": "stream1",
351
- },
352
- {
353
- "data": {
354
- "col1": "val_bb1_11",
355
- "col2": 1,
356
- "_ab_source_file_last_modified": "2023-06-05T03:54:07.000000Z",
357
- "_ab_source_file_url": "b/b1.csv",
358
- },
359
- "stream": "stream2",
360
- },
361
- {
362
- "data": {
363
- "col1": "val_bb1_12",
364
- "col2": 2,
365
- "_ab_source_file_last_modified": "2023-06-05T03:54:07.000000Z",
366
- "_ab_source_file_url": "b/b1.csv",
367
- },
368
- "stream": "stream2",
369
- },
370
- # {"data": {"col1": "val_bb2_11", "col2": "val_bb2_21", "_ab_source_file_last_modified": "2023-06-05T03:54:07.000000Z", "_ab_source_file_url": "b/b2.csv"}, "stream": "stream2"}, # This record is skipped because it does not conform
371
- {
372
- "data": {
373
- "col1": "val_bb2_12",
374
- "col2": 2,
375
- "_ab_source_file_last_modified": "2023-06-05T03:54:07.000000Z",
376
- "_ab_source_file_url": "b/b2.csv",
377
- },
378
- "stream": "stream2",
379
- },
380
- {
381
- "data": {
382
- "col1": "val_bb3_11",
383
- "_ab_source_file_last_modified": "2023-06-05T03:54:07.000000Z",
384
- "_ab_source_file_url": "b/b3.csv",
385
- },
386
- "stream": "stream2",
387
- },
388
- {
389
- "data": {
390
- "col1": "val_bb3_12",
391
- "_ab_source_file_last_modified": "2023-06-05T03:54:07.000000Z",
392
- "_ab_source_file_url": "b/b3.csv",
393
- },
394
- "stream": "stream2",
395
- },
396
- ]
397
- )
398
- .set_expected_logs(
399
- {
400
- "read": [
401
- {
402
- "level": "WARN",
403
- "message": "Records in file did not pass validation policy. stream=stream1 file=a/a2.csv n_skipped=1 validation_policy=skip_record",
404
- },
405
- {
406
- "level": "ERROR",
407
- "message": "Error parsing record. This could be due to a mismatch between the config's file type and the actual file type, or because the file or record is not parseable. stream=stream1 file=a/a3.csv line_no=2 n_skipped=0",
408
- },
409
- {
410
- "level": "WARN",
411
- "message": "Records in file did not pass validation policy. stream=stream2 file=b/b2.csv n_skipped=1 validation_policy=skip_record",
412
- },
413
- {
414
- "level": "WARN",
415
- "message": "Could not cast the value to the expected type.: col2: value=this is text that will trigger validation policy,expected_type=integer",
416
- },
417
- {
418
- "level": "WARN",
419
- "message": "Could not cast the value to the expected type.: col2: value=this is text that will trigger validation policy,expected_type=integer",
420
- },
421
- ]
422
- }
423
- )
424
- .set_expected_read_error(
425
- AirbyteTracedException,
426
- "Please check the logged errors for more information.",
427
- )
428
- ).build()
429
-
430
-
431
- emit_record_scenario_single_stream = (
432
- _base_single_stream_scenario.copy()
433
- .set_name("emit_record_scenario_single_stream")
434
- .set_config(
435
- {
436
- "streams": [
437
- {
438
- "name": "stream1",
439
- "format": {"filetype": "csv"},
440
- "globs": ["*.csv"],
441
- "validation_policy": "Emit Record",
442
- }
443
- ]
444
- }
445
- )
446
- .set_expected_records(
447
- [
448
- {
449
- "data": {
450
- "col1": "val_a_11",
451
- "col2": 1,
452
- "_ab_source_file_last_modified": "2023-06-05T03:54:07.000000Z",
453
- "_ab_source_file_url": "a.csv",
454
- },
455
- "stream": "stream1",
456
- },
457
- {
458
- "data": {
459
- "col1": "val_a_12",
460
- "col2": 2,
461
- "_ab_source_file_last_modified": "2023-06-05T03:54:07.000000Z",
462
- "_ab_source_file_url": "a.csv",
463
- },
464
- "stream": "stream1",
465
- },
466
- {
467
- "data": {
468
- "col1": "val_b_11",
469
- "col2": "this is text that will trigger validation policy",
470
- "_ab_source_file_last_modified": "2023-06-05T03:54:07.000000Z",
471
- "_ab_source_file_url": "b.csv",
472
- },
473
- "stream": "stream1",
474
- }, # This record is skipped because it does not conform
475
- {
476
- "data": {
477
- "col1": "val_b_12",
478
- "col2": 2,
479
- "_ab_source_file_last_modified": "2023-06-05T03:54:07.000000Z",
480
- "_ab_source_file_url": "b.csv",
481
- },
482
- "stream": "stream1",
483
- },
484
- {
485
- "data": {
486
- "col1": "val_c_11",
487
- "_ab_source_file_last_modified": "2023-06-05T03:54:07.000000Z",
488
- "_ab_source_file_url": "c.csv",
489
- },
490
- "stream": "stream1",
491
- },
492
- # {"data": {"col1": "val_c_12", None: "val_c_22", "_ab_source_file_last_modified": "2023-06-05T03:54:07.000000Z", "_ab_source_file_url": "c.csv"}, "stream": "stream1"}, # This record is malformed so should not be emitted
493
- # {"data": {"col1": "val_c_13", "_ab_source_file_last_modified": "2023-06-05T03:54:07.000000Z", "_ab_source_file_url": "c.csv"}, "stream": "stream1"}, # No more records from this stream are emitted after we hit a parse error
494
- {
495
- "data": {
496
- "col1": "val_d_11",
497
- "_ab_source_file_last_modified": "2023-06-05T03:54:07.000000Z",
498
- "_ab_source_file_url": "d.csv",
499
- },
500
- "stream": "stream1",
501
- },
502
- ]
503
- )
504
- .set_expected_read_error(
505
- AirbyteTracedException,
506
- "Please check the logged errors for more information.",
507
- )
508
- ).build()
509
-
510
-
511
- emit_record_scenario_multi_stream = (
512
- _base_multi_stream_scenario.copy()
513
- .set_name("emit_record_scenario_multi_stream")
514
- .set_config(
515
- {
516
- "streams": [
517
- {
518
- "name": "stream1",
519
- "format": {"filetype": "csv"},
520
- "globs": ["a/*.csv"],
521
- "validation_policy": "Emit Record",
522
- },
523
- {
524
- "name": "stream2",
525
- "format": {"filetype": "csv"},
526
- "globs": ["b/*.csv"],
527
- "validation_policy": "Emit Record",
528
- },
529
- ]
530
- }
531
- )
532
- .set_expected_records(
533
- [
534
- {
535
- "data": {
536
- "col1": "val_aa1_11",
537
- "col2": 1,
538
- "_ab_source_file_last_modified": "2023-06-05T03:54:07.000000Z",
539
- "_ab_source_file_url": "a/a1.csv",
540
- },
541
- "stream": "stream1",
542
- },
543
- {
544
- "data": {
545
- "col1": "val_aa1_12",
546
- "col2": 2,
547
- "_ab_source_file_last_modified": "2023-06-05T03:54:07.000000Z",
548
- "_ab_source_file_url": "a/a1.csv",
549
- },
550
- "stream": "stream1",
551
- },
552
- {
553
- "data": {
554
- "col1": "val_aa2_11",
555
- "col2": "this is text that will trigger validation policy",
556
- "_ab_source_file_last_modified": "2023-06-05T03:54:07.000000Z",
557
- "_ab_source_file_url": "a/a2.csv",
558
- },
559
- "stream": "stream1",
560
- },
561
- {
562
- "data": {
563
- "col1": "val_aa2_12",
564
- "col2": 2,
565
- "_ab_source_file_last_modified": "2023-06-05T03:54:07.000000Z",
566
- "_ab_source_file_url": "a/a2.csv",
567
- },
568
- "stream": "stream1",
569
- },
570
- {
571
- "data": {
572
- "col1": "val_aa3_11",
573
- "_ab_source_file_last_modified": "2023-06-05T03:54:07.000000Z",
574
- "_ab_source_file_url": "a/a3.csv",
575
- },
576
- "stream": "stream1",
577
- },
578
- # {"data": {"col1": "val_aa3_12", None: "val_aa3_22", "_ab_source_file_last_modified": "2023-06-05T03:54:07.000000Z", "_ab_source_file_url": "a/a3.csv"}, "stream": "stream1"}, # This record is malformed so should not be emitted
579
- # {"data": {"col1": "val_aa3_13", "_ab_source_file_last_modified": "2023-06-05T03:54:07.000000Z", "_ab_source_file_url": "a/a3.csv"}, "stream": "stream1"}, # Skipped since previous record is malformed
580
- {
581
- "data": {
582
- "col1": "val_aa4_11",
583
- "_ab_source_file_last_modified": "2023-06-05T03:54:07.000000Z",
584
- "_ab_source_file_url": "a/a4.csv",
585
- },
586
- "stream": "stream1",
587
- },
588
- {
589
- "data": {
590
- "col1": "val_bb1_11",
591
- "col2": 1,
592
- "_ab_source_file_last_modified": "2023-06-05T03:54:07.000000Z",
593
- "_ab_source_file_url": "b/b1.csv",
594
- },
595
- "stream": "stream2",
596
- },
597
- {
598
- "data": {
599
- "col1": "val_bb1_12",
600
- "col2": 2,
601
- "_ab_source_file_last_modified": "2023-06-05T03:54:07.000000Z",
602
- "_ab_source_file_url": "b/b1.csv",
603
- },
604
- "stream": "stream2",
605
- },
606
- {
607
- "data": {
608
- "col1": "val_bb2_11",
609
- "col2": "this is text that will trigger validation policy",
610
- "_ab_source_file_last_modified": "2023-06-05T03:54:07.000000Z",
611
- "_ab_source_file_url": "b/b2.csv",
612
- },
613
- "stream": "stream2",
614
- },
615
- {
616
- "data": {
617
- "col1": "val_bb2_12",
618
- "col2": 2,
619
- "_ab_source_file_last_modified": "2023-06-05T03:54:07.000000Z",
620
- "_ab_source_file_url": "b/b2.csv",
621
- },
622
- "stream": "stream2",
623
- },
624
- {
625
- "data": {
626
- "col1": "val_bb3_11",
627
- "_ab_source_file_last_modified": "2023-06-05T03:54:07.000000Z",
628
- "_ab_source_file_url": "b/b3.csv",
629
- },
630
- "stream": "stream2",
631
- },
632
- {
633
- "data": {
634
- "col1": "val_bb3_12",
635
- "_ab_source_file_last_modified": "2023-06-05T03:54:07.000000Z",
636
- "_ab_source_file_url": "b/b3.csv",
637
- },
638
- "stream": "stream2",
639
- },
640
- ]
641
- )
642
- .set_expected_read_error(
643
- AirbyteTracedException,
644
- "Please check the logged errors for more information.",
645
- )
646
- ).build()
647
-
648
-
649
- wait_for_rediscovery_scenario_single_stream = (
650
- _base_single_stream_scenario.copy()
651
- .set_name("wait_for_rediscovery_scenario_single_stream")
652
- .set_config(
653
- {
654
- "streams": [
655
- {
656
- "name": "stream1",
657
- "format": {"filetype": "csv"},
658
- "globs": ["*.csv"],
659
- "validation_policy": "Wait for Discover",
660
- }
661
- ]
662
- }
663
- )
664
- .set_expected_records(None) # When syncing streams concurrently we don't know how many records will be emitted before the sync stops
665
- .set_expected_logs(
666
- {
667
- "read": [
668
- {
669
- "level": "WARN",
670
- "message": "Stopping sync in accordance with the configured validation policy. Records in file did not conform to the schema. stream=stream1 file=b.csv validation_policy=Wait for Discover n_skipped=0",
671
- },
672
- {
673
- "level": "WARN",
674
- "message": "Could not cast the value to the expected type.: col2: value=this is text that will trigger validation policy,expected_type=integer",
675
- },
676
- ]
677
- }
678
- )
679
- ).build()
680
-
681
-
682
- wait_for_rediscovery_scenario_multi_stream = (
683
- _base_multi_stream_scenario.copy()
684
- .set_name("wait_for_rediscovery_scenario_multi_stream")
685
- .set_config(
686
- {
687
- "streams": [
688
- {
689
- "name": "stream1",
690
- "format": {"filetype": "csv"},
691
- "globs": ["a/*.csv"],
692
- "validation_policy": "Wait for Discover",
693
- },
694
- {
695
- "name": "stream2",
696
- "format": {"filetype": "csv"},
697
- "globs": ["b/*.csv"],
698
- "validation_policy": "Wait for Discover",
699
- },
700
- ]
701
- }
702
- )
703
- .set_expected_records(None) # When syncing streams concurrently we don't know how many records will be emitted before the sync stops
704
- .set_expected_logs(
705
- {
706
- "read": [
707
- {
708
- "level": "WARN",
709
- "message": "Stopping sync in accordance with the configured validation policy. Records in file did not conform to the schema. stream=stream1 file=a/a2.csv validation_policy=Wait for Discover n_skipped=0",
710
- },
711
- {
712
- "level": "WARN",
713
- "message": "Stopping sync in accordance with the configured validation policy. Records in file did not conform to the schema. stream=stream2 file=b/b2.csv validation_policy=Wait for Discover n_skipped=0",
714
- },
715
- {
716
- "level": "WARN",
717
- "message": "Could not cast the value to the expected type.: col2: value=this is text that will trigger validation policy,expected_type=integer",
718
- },
719
- {
720
- "level": "WARN",
721
- "message": "Could not cast the value to the expected type.: col2: value=this is text that will trigger validation policy,expected_type=integer",
722
- },
723
- ]
724
- }
725
- )
726
- ).build()