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
@@ -1,253 +0,0 @@
1
- #
2
- # Copyright (c) 2023 Airbyte, Inc., all rights reserved.
3
- #
4
-
5
- import json
6
- import math
7
- from pathlib import Path, PosixPath
8
- from typing import Any, Dict, List, Mapping, Optional, Union
9
-
10
- import pytest
11
- from _pytest.capture import CaptureFixture
12
- from _pytest.reports import ExceptionInfo
13
- from airbyte_cdk.entrypoint import launch
14
- from airbyte_cdk.models import AirbyteAnalyticsTraceMessage, SyncMode
15
- from airbyte_cdk.sources import AbstractSource
16
- from airbyte_cdk.sources.file_based.stream.concurrent.cursor import AbstractConcurrentFileBasedCursor
17
- from airbyte_cdk.test.entrypoint_wrapper import EntrypointOutput
18
- from airbyte_cdk.test.entrypoint_wrapper import read as entrypoint_read
19
- from airbyte_cdk.utils import message_utils
20
- from airbyte_cdk.utils.traced_exception import AirbyteTracedException
21
- from airbyte_protocol.models import AirbyteLogMessage, AirbyteMessage, ConfiguredAirbyteCatalog
22
- from unit_tests.sources.file_based.scenarios.scenario_builder import TestScenario
23
-
24
-
25
- def verify_discover(capsys: CaptureFixture[str], tmp_path: PosixPath, scenario: TestScenario[AbstractSource]) -> None:
26
- expected_exc, expected_msg = scenario.expected_discover_error
27
- expected_logs = scenario.expected_logs
28
- if expected_exc:
29
- with pytest.raises(expected_exc) as exc:
30
- discover(capsys, tmp_path, scenario)
31
- if expected_msg:
32
- assert expected_msg in get_error_message_from_exc(exc)
33
- elif scenario.expected_catalog:
34
- output = discover(capsys, tmp_path, scenario)
35
- catalog, logs = output["catalog"], output["logs"]
36
- assert catalog == scenario.expected_catalog
37
- if expected_logs:
38
- discover_logs = expected_logs.get("discover")
39
- logs = [log for log in logs if log.get("log", {}).get("level") in ("ERROR", "WARN")]
40
- _verify_expected_logs(logs, discover_logs)
41
-
42
-
43
- def verify_read(scenario: TestScenario[AbstractSource]) -> None:
44
- if scenario.incremental_scenario_config:
45
- run_test_read_incremental(scenario)
46
- else:
47
- run_test_read_full_refresh(scenario)
48
-
49
-
50
- def run_test_read_full_refresh(scenario: TestScenario[AbstractSource]) -> None:
51
- expected_exc, expected_msg = scenario.expected_read_error
52
- output = read(scenario)
53
- if expected_exc:
54
- assert_exception(expected_exc, output)
55
- if expected_msg:
56
- assert expected_msg in output.errors[-1].trace.error.internal_message
57
- else:
58
- _verify_read_output(output, scenario)
59
-
60
-
61
- def run_test_read_incremental(scenario: TestScenario[AbstractSource]) -> None:
62
- expected_exc, expected_msg = scenario.expected_read_error
63
- output = read_with_state(scenario)
64
- if expected_exc:
65
- assert_exception(expected_exc, output)
66
- else:
67
- _verify_read_output(output, scenario)
68
-
69
-
70
- def assert_exception(expected_exception: type[BaseException], output: EntrypointOutput) -> None:
71
- assert expected_exception.__name__ in output.errors[-1].trace.error.stack_trace
72
-
73
-
74
- def _verify_read_output(output: EntrypointOutput, scenario: TestScenario[AbstractSource]) -> None:
75
- records_and_state_messages, log_messages = output.records_and_state_messages, output.logs
76
- logs = [message.log for message in log_messages if message.log.level.value in scenario.log_levels]
77
- if scenario.expected_records is None:
78
- return
79
-
80
- expected_records = [r for r in scenario.expected_records] if scenario.expected_records else []
81
-
82
- sorted_expected_records = sorted(
83
- filter(lambda e: "data" in e, expected_records),
84
- key=lambda record: ",".join(
85
- f"{k}={v}" for k, v in sorted(record["data"].items(), key=lambda items: (items[0], items[1])) if k != "emitted_at"
86
- ),
87
- )
88
- sorted_records = sorted(
89
- filter(lambda r: r.record, records_and_state_messages),
90
- key=lambda record: ",".join(
91
- f"{k}={v}" for k, v in sorted(record.record.data.items(), key=lambda items: (items[0], items[1])) if k != "emitted_at"
92
- ),
93
- )
94
-
95
- assert len(sorted_records) == len(sorted_expected_records)
96
-
97
- for actual, expected in zip(sorted_records, sorted_expected_records):
98
- if actual.record:
99
- assert len(actual.record.data) == len(expected["data"])
100
- for key, value in actual.record.data.items():
101
- if isinstance(value, float):
102
- assert math.isclose(value, expected["data"][key], abs_tol=1e-04)
103
- else:
104
- assert value == expected["data"][key]
105
- assert actual.record.stream == expected["stream"]
106
-
107
- expected_states = list(filter(lambda e: "data" not in e, expected_records))
108
- states = list(filter(lambda r: r.state, records_and_state_messages))
109
- assert len(states) > 0, "No state messages emitted. Successful syncs should emit at least one stream state."
110
- _verify_state_record_counts(sorted_records, states)
111
-
112
- if hasattr(scenario.source, "cursor_cls") and issubclass(scenario.source.cursor_cls, AbstractConcurrentFileBasedCursor):
113
- # Only check the last state emitted because we don't know the order the others will be in.
114
- # This may be needed for non-file-based concurrent scenarios too.
115
- assert states[-1].state.stream.stream_state.dict() == expected_states[-1]
116
- else:
117
- for actual, expected in zip(states, expected_states): # states should be emitted in sorted order
118
- assert actual.state.stream.stream_state.dict() == expected
119
-
120
- if scenario.expected_logs:
121
- read_logs = scenario.expected_logs.get("read")
122
- assert len(logs) == (len(read_logs) if read_logs else 0)
123
- _verify_expected_logs(logs, read_logs)
124
-
125
- if scenario.expected_analytics:
126
- analytics = output.analytics_messages
127
-
128
- _verify_analytics(analytics, scenario.expected_analytics)
129
-
130
-
131
- def _verify_state_record_counts(records: List[AirbyteMessage], states: List[AirbyteMessage]) -> None:
132
- actual_record_counts = {}
133
- for record in records:
134
- stream_descriptor = message_utils.get_stream_descriptor(record)
135
- actual_record_counts[stream_descriptor] = actual_record_counts.get(stream_descriptor, 0) + 1
136
-
137
- state_record_count_sums = {}
138
- for state_message in states:
139
- stream_descriptor = message_utils.get_stream_descriptor(state_message)
140
- state_record_count_sums[stream_descriptor] = (
141
- state_record_count_sums.get(stream_descriptor, 0)
142
- + state_message.state.sourceStats.recordCount
143
- )
144
-
145
- for stream, actual_count in actual_record_counts.items():
146
- assert state_record_count_sums.get(stream) == actual_count
147
-
148
- # We can have extra keys in state_record_count_sums if we processed a stream and reported 0 records
149
- extra_keys = state_record_count_sums.keys() - actual_record_counts.keys()
150
- for stream in extra_keys:
151
- assert state_record_count_sums[stream] == 0
152
-
153
-
154
- def _verify_analytics(analytics: List[AirbyteMessage], expected_analytics: Optional[List[AirbyteAnalyticsTraceMessage]]) -> None:
155
- if expected_analytics:
156
- assert len(analytics) == len(
157
- expected_analytics), \
158
- f"Number of actual analytics messages ({len(analytics)}) did not match expected ({len(expected_analytics)})"
159
- for actual, expected in zip(analytics, expected_analytics):
160
- actual_type, actual_value = actual.trace.analytics.type, actual.trace.analytics.value
161
- expected_type = expected.type
162
- expected_value = expected.value
163
- assert actual_type == expected_type
164
- assert actual_value == expected_value
165
-
166
-
167
- def _verify_expected_logs(logs: List[AirbyteLogMessage], expected_logs: Optional[List[Mapping[str, Any]]]) -> None:
168
- if expected_logs:
169
- for actual, expected in zip(logs, expected_logs):
170
- actual_level, actual_message = actual.level.value, actual.message
171
- expected_level = expected["level"]
172
- expected_message = expected["message"]
173
- assert actual_level == expected_level
174
- assert expected_message in actual_message
175
-
176
-
177
- def verify_spec(capsys: CaptureFixture[str], scenario: TestScenario[AbstractSource]) -> None:
178
- assert spec(capsys, scenario) == scenario.expected_spec
179
-
180
-
181
- def verify_check(capsys: CaptureFixture[str], tmp_path: PosixPath, scenario: TestScenario[AbstractSource]) -> None:
182
- expected_exc, expected_msg = scenario.expected_check_error
183
-
184
- if expected_exc:
185
- with pytest.raises(expected_exc):
186
- output = check(capsys, tmp_path, scenario)
187
- if expected_msg:
188
- # expected_msg is a string. what's the expected value field?
189
- assert expected_msg in output["message"] # type: ignore
190
- assert output["status"] == scenario.expected_check_status
191
-
192
- else:
193
- output = check(capsys, tmp_path, scenario)
194
- assert output["status"] == scenario.expected_check_status
195
-
196
-
197
- def spec(capsys: CaptureFixture[str], scenario: TestScenario[AbstractSource]) -> Mapping[str, Any]:
198
- launch(
199
- scenario.source,
200
- ["spec"],
201
- )
202
- captured = capsys.readouterr()
203
- return json.loads(captured.out.splitlines()[0])["spec"] # type: ignore
204
-
205
-
206
- def check(capsys: CaptureFixture[str], tmp_path: PosixPath, scenario: TestScenario[AbstractSource]) -> Dict[str, Any]:
207
- launch(
208
- scenario.source,
209
- ["check", "--config", make_file(tmp_path / "config.json", scenario.config)],
210
- )
211
- captured = capsys.readouterr()
212
- return json.loads(captured.out.splitlines()[0])["connectionStatus"] # type: ignore
213
-
214
-
215
- def discover(capsys: CaptureFixture[str], tmp_path: PosixPath, scenario: TestScenario[AbstractSource]) -> Dict[str, Any]:
216
- launch(
217
- scenario.source,
218
- ["discover", "--config", make_file(tmp_path / "config.json", scenario.config)],
219
- )
220
- output = [json.loads(line) for line in capsys.readouterr().out.splitlines()]
221
- [catalog] = [o["catalog"] for o in output if o.get("catalog")] # type: ignore
222
- return {
223
- "catalog": catalog,
224
- "logs": [o["log"] for o in output if o.get("log")],
225
- }
226
-
227
-
228
- def read(scenario: TestScenario[AbstractSource]) -> EntrypointOutput:
229
- return entrypoint_read(
230
- scenario.source,
231
- scenario.config,
232
- ConfiguredAirbyteCatalog.parse_obj(scenario.configured_catalog(SyncMode.full_refresh)),
233
- )
234
-
235
-
236
- def read_with_state(scenario: TestScenario[AbstractSource]) -> EntrypointOutput:
237
- return entrypoint_read(
238
- scenario.source,
239
- scenario.config,
240
- ConfiguredAirbyteCatalog.parse_obj(scenario.configured_catalog(SyncMode.incremental)),
241
- scenario.input_state(),
242
- )
243
-
244
-
245
- def make_file(path: Path, file_contents: Optional[Union[Mapping[str, Any], List[Mapping[str, Any]]]]) -> str:
246
- path.write_text(json.dumps(file_contents))
247
- return str(path)
248
-
249
-
250
- def get_error_message_from_exc(exc: ExceptionInfo[Any]) -> str:
251
- if isinstance(exc.value, AirbyteTracedException):
252
- return exc.value.message
253
- return str(exc.value.args[0])
@@ -1,346 +0,0 @@
1
- #
2
- # Copyright (c) 2023 Airbyte, Inc., all rights reserved.
3
- #
4
-
5
- from typing import Any, Mapping, Optional
6
-
7
- import pytest
8
- from airbyte_cdk.sources.file_based.exceptions import ConfigValidationError, SchemaInferenceError
9
- from airbyte_cdk.sources.file_based.schema_helpers import (
10
- ComparableType,
11
- SchemaType,
12
- conforms_to_schema,
13
- merge_schemas,
14
- type_mapping_to_jsonschema,
15
- )
16
-
17
- COMPLETE_CONFORMING_RECORD = {
18
- "null_field": None,
19
- "boolean_field": True,
20
- "integer_field": 1,
21
- "number_field": 1.5,
22
- "string_field": "val1",
23
- "array_field": [1.1, 2.2],
24
- "object_field": {"col": "val"},
25
- }
26
-
27
-
28
- NONCONFORMING_EXTRA_COLUMN_RECORD = {
29
- "null_field": None,
30
- "boolean_field": True,
31
- "integer_field": 1,
32
- "number_field": 1.5,
33
- "string_field": "val1",
34
- "array_field": [1.1, 2.2],
35
- "object_field": {"col": "val"},
36
- "column_x": "extra",
37
- }
38
-
39
- CONFORMING_WITH_MISSING_COLUMN_RECORD = {
40
- "null_field": None,
41
- "boolean_field": True,
42
- "integer_field": 1,
43
- "number_field": 1.5,
44
- "string_field": "val1",
45
- "array_field": [1.1, 2.2],
46
- }
47
-
48
- CONFORMING_WITH_NARROWER_TYPE_RECORD = {
49
- "null_field": None,
50
- "boolean_field": True,
51
- "integer_field": True,
52
- "number_field": True,
53
- "string_field": True,
54
- "array_field": [1.1, 2.2],
55
- "object_field": {"col": "val"},
56
- }
57
-
58
- NONCONFORMING_WIDER_TYPE_RECORD = {
59
- "null_field": "not None",
60
- "boolean_field": True,
61
- "integer_field": 1,
62
- "number_field": 1.5,
63
- "string_field": "val1",
64
- "array_field": [1.1, 2.2],
65
- "object_field": {"col": "val"},
66
- }
67
-
68
- NONCONFORMING_NON_OBJECT_RECORD = {
69
- "null_field": None,
70
- "boolean_field": True,
71
- "integer_field": 1,
72
- "number_field": 1.5,
73
- "string_field": "val1",
74
- "array_field": [1.1, 2.2],
75
- "object_field": "not an object",
76
- }
77
-
78
- NONCONFORMING_NON_ARRAY_RECORD = {
79
- "null_field": None,
80
- "boolean_field": True,
81
- "integer_field": 1,
82
- "number_field": 1.5,
83
- "string_field": "val1",
84
- "array_field": "not an array",
85
- "object_field": {"col": "val"},
86
- }
87
-
88
- CONFORMING_MIXED_TYPE_NARROWER_RECORD = {
89
- "null_field": None,
90
- "boolean_field": True,
91
- "integer_field": 1,
92
- "number_field": 1.5,
93
- "string_field": "val1",
94
- "array_field": [1.1, 2.2],
95
- "object_field": {"col": "val"},
96
- }
97
-
98
- NONCONFORMING_MIXED_TYPE_WIDER_RECORD = {
99
- "null_field": None,
100
- "boolean_field": True,
101
- "integer_field": 1,
102
- "number_field": 1.5,
103
- "string_field": "val1",
104
- "array_field": [1.1, 2.2],
105
- "object_field": {"col": "val"},
106
- }
107
-
108
- CONFORMING_MIXED_TYPE_WITHIN_TYPE_RANGE_RECORD = {
109
- "null_field": None,
110
- "boolean_field": True,
111
- "integer_field": 1,
112
- "number_field": 1.5,
113
- "string_field": "val1",
114
- "array_field": [1.1, 2.2],
115
- "object_field": {"col": "val"},
116
- }
117
-
118
- NONCONFORMING_INVALID_ARRAY_RECORD = {
119
- "null_field": None,
120
- "boolean_field": True,
121
- "integer_field": 1,
122
- "number_field": 1.5,
123
- "string_field": ["this should not be an array"],
124
- "array_field": [1.1, 2.2],
125
- "object_field": {"col": "val"},
126
- }
127
-
128
- NONCONFORMING_TOO_WIDE_ARRAY_RECORD = {
129
- "null_field": None,
130
- "boolean_field": True,
131
- "integer_field": 1,
132
- "number_field": 1.5,
133
- "string_field": "okay",
134
- "array_field": ["val1", "val2"],
135
- "object_field": {"col": "val"},
136
- }
137
-
138
-
139
- CONFORMING_NARROWER_ARRAY_RECORD = {
140
- "null_field": None,
141
- "boolean_field": True,
142
- "integer_field": 1,
143
- "number_field": 1.5,
144
- "string_field": "okay",
145
- "array_field": [1, 2],
146
- "object_field": {"col": "val"},
147
- }
148
-
149
-
150
- NONCONFORMING_INVALID_OBJECT_RECORD = {
151
- "null_field": None,
152
- "boolean_field": True,
153
- "integer_field": 1,
154
- "number_field": 1.5,
155
- "string_field": {"this": "should not be an object"},
156
- "array_field": [1.1, 2.2],
157
- "object_field": {"col": "val"},
158
- }
159
-
160
-
161
- SCHEMA = {
162
- "type": "object",
163
- "properties": {
164
- "null_field": {"type": "null"},
165
- "boolean_field": {"type": "boolean"},
166
- "integer_field": {"type": "integer"},
167
- "number_field": {"type": "number"},
168
- "string_field": {"type": "string"},
169
- "array_field": {
170
- "type": "array",
171
- "items": {
172
- "type": "number",
173
- },
174
- },
175
- "object_field": {"type": "object"},
176
- },
177
- }
178
-
179
-
180
- @pytest.mark.parametrize(
181
- "record,schema,expected_result",
182
- [
183
- pytest.param(COMPLETE_CONFORMING_RECORD, SCHEMA, True, id="record-conforms"),
184
- pytest.param(NONCONFORMING_EXTRA_COLUMN_RECORD, SCHEMA, False, id="nonconforming-extra-column"),
185
- pytest.param(CONFORMING_WITH_MISSING_COLUMN_RECORD, SCHEMA, True, id="record-conforms-with-missing-column"),
186
- pytest.param(CONFORMING_WITH_NARROWER_TYPE_RECORD, SCHEMA, True, id="record-conforms-with-narrower-type"),
187
- pytest.param(NONCONFORMING_WIDER_TYPE_RECORD, SCHEMA, False, id="nonconforming-wider-type"),
188
- pytest.param(NONCONFORMING_NON_OBJECT_RECORD, SCHEMA, False, id="nonconforming-string-is-not-an-object"),
189
- pytest.param(NONCONFORMING_NON_ARRAY_RECORD, SCHEMA, False, id="nonconforming-string-is-not-an-array"),
190
- pytest.param(NONCONFORMING_TOO_WIDE_ARRAY_RECORD, SCHEMA, False, id="nonconforming-array-values-too-wide"),
191
- pytest.param(CONFORMING_NARROWER_ARRAY_RECORD, SCHEMA, True, id="conforming-array-values-narrower-than-schema"),
192
- pytest.param(NONCONFORMING_INVALID_ARRAY_RECORD, SCHEMA, False, id="nonconforming-array-is-not-a-string"),
193
- pytest.param(NONCONFORMING_INVALID_OBJECT_RECORD, SCHEMA, False, id="nonconforming-object-is-not-a-string"),
194
- ],
195
- )
196
- def test_conforms_to_schema(record: Mapping[str, Any], schema: Mapping[str, Any], expected_result: bool) -> None:
197
- assert conforms_to_schema(record, schema) == expected_result
198
-
199
-
200
- def test_comparable_types() -> None:
201
- assert ComparableType.OBJECT > ComparableType.STRING
202
- assert ComparableType.STRING > ComparableType.NUMBER
203
- assert ComparableType.NUMBER > ComparableType.INTEGER
204
- assert ComparableType.INTEGER > ComparableType.BOOLEAN
205
- assert ComparableType["OBJECT"] == ComparableType.OBJECT
206
-
207
-
208
- @pytest.mark.parametrize(
209
- "schema1,schema2,expected_result",
210
- [
211
- pytest.param({}, {}, {}, id="empty-schemas"),
212
- pytest.param({"a": None}, {}, None, id="null-value-in-schema"),
213
- pytest.param({"a": {"type": "integer"}}, {}, {"a": {"type": "integer"}}, id="single-key-schema1"),
214
- pytest.param({}, {"a": {"type": "integer"}}, {"a": {"type": "integer"}}, id="single-key-schema2"),
215
- pytest.param({"a": {"type": "integer"}}, {"a": {"type": "integer"}}, {"a": {"type": "integer"}}, id="single-key-both-schemas"),
216
- pytest.param({"a": {"type": "integer"}}, {"a": {"type": "number"}}, {"a": {"type": "number"}}, id="single-key-schema2-is-wider"),
217
- pytest.param({"a": {"type": "number"}}, {"a": {"type": "integer"}}, {"a": {"type": "number"}}, id="single-key-schema1-is-wider"),
218
- pytest.param({"a": {"type": "array"}}, {"a": {"type": "integer"}}, None, id="single-key-with-array-schema1"),
219
- pytest.param({"a": {"type": "integer"}}, {"a": {"type": "array"}}, None, id="single-key-with-array-schema2"),
220
- pytest.param(
221
- {"a": {"type": "object", "properties": {"b": {"type": "integer"}}}},
222
- {"a": {"type": "object", "properties": {"b": {"type": "integer"}}}},
223
- {"a": {"type": "object", "properties": {"b": {"type": "integer"}}}},
224
- id="single-key-same-object",
225
- ),
226
- pytest.param(
227
- {"a": {"type": "object", "properties": {"b": {"type": "integer"}}}},
228
- {"a": {"type": "object", "properties": {"b": {"type": "string"}}}},
229
- None,
230
- id="single-key-different-objects",
231
- ),
232
- pytest.param(
233
- {"a": {"type": "object", "properties": {"b": {"type": "integer"}}}},
234
- {"a": {"type": "number"}},
235
- None,
236
- id="single-key-with-object-schema1",
237
- ),
238
- pytest.param(
239
- {"a": {"type": "number"}},
240
- {"a": {"type": "object", "properties": {"b": {"type": "integer"}}}},
241
- None,
242
- id="single-key-with-object-schema2",
243
- ),
244
- pytest.param(
245
- {"a": {"type": "array", "items": {"type": "number"}}},
246
- {"a": {"type": "array", "items": {"type": "number"}}},
247
- {"a": {"type": "array", "items": {"type": "number"}}},
248
- id="equal-arrays-in-both-schemas",
249
- ),
250
- pytest.param(
251
- {"a": {"type": "array", "items": {"type": "integer"}}},
252
- {"a": {"type": "array", "items": {"type": "number"}}},
253
- None,
254
- id="different-arrays-in-both-schemas",
255
- ),
256
- pytest.param(
257
- {"a": {"type": "integer"}, "b": {"type": "string"}},
258
- {"c": {"type": "number"}},
259
- {"a": {"type": "integer"}, "b": {"type": "string"}, "c": {"type": "number"}},
260
- id="",
261
- ),
262
- pytest.param({"a": {"type": "invalid_type"}}, {"b": {"type": "integer"}}, None, id="invalid-type"),
263
- ],
264
- )
265
- def test_merge_schemas(schema1: SchemaType, schema2: SchemaType, expected_result: Optional[SchemaType]) -> None:
266
- if expected_result is not None:
267
- assert merge_schemas(schema1, schema2) == expected_result
268
- else:
269
- with pytest.raises(SchemaInferenceError):
270
- merge_schemas(schema1, schema2)
271
-
272
-
273
- @pytest.mark.parametrize(
274
- "type_mapping,expected_schema,expected_exc_msg",
275
- [
276
- pytest.param(
277
- '{"col1": "null", "col2": "array", "col3": "boolean", "col4": "float", "col5": "integer", "col6": "number", "col7": "object", "col8": "string"}',
278
- {
279
- "type": "object",
280
- "properties": {
281
- "col1": {"type": "null"},
282
- "col2": {"type": "array"},
283
- "col3": {"type": "boolean"},
284
- "col4": {"type": "number"},
285
- "col5": {"type": "integer"},
286
- "col6": {"type": "number"},
287
- "col7": {"type": "object"},
288
- "col8": {"type": "string"},
289
- },
290
- },
291
- None,
292
- id="valid_all_types",
293
- ),
294
- pytest.param(
295
- '{"col1 ": " string", "col2": " integer"}',
296
- {"type": "object", "properties": {"col1": {"type": "string"}, "col2": {"type": "integer"}}},
297
- None,
298
- id="valid_extra_spaces",
299
- ),
300
- pytest.param(
301
- "",
302
- None,
303
- None,
304
- id="valid_empty_string",
305
- ),
306
- pytest.param(
307
- '{"col1": "x", "col2": "integer"}',
308
- None,
309
- "Invalid type 'x' for property 'col1'",
310
- id="invalid_type",
311
- ),
312
- pytest.param(
313
- '{"col1": "", "col2": "integer"}',
314
- None,
315
- "Invalid input schema",
316
- id="invalid_missing_type",
317
- ),
318
- pytest.param(
319
- '{"": "string", "col2": "integer"}',
320
- None,
321
- "Invalid input schema",
322
- id="invalid_missing_name",
323
- ),
324
- pytest.param(
325
- '{"type": "object", "properties": {"col1": {"type": "string"}, "col2": {"type": "integer"}}}',
326
- None,
327
- "Invalid input schema; nested schemas are not supported.",
328
- id="invalid_nested_input_string",
329
- ),
330
- pytest.param(
331
- '{"type": "object", "properties": {"col1": {"type": "string"}, "col2": {"type": "integer"}}}',
332
- None,
333
- "Invalid input schema; nested schemas are not supported.",
334
- id="invalid_nested_input_json",
335
- ),
336
- ],
337
- )
338
- def test_type_mapping_to_jsonschema(
339
- type_mapping: Mapping[str, Any], expected_schema: Optional[Mapping[str, Any]], expected_exc_msg: Optional[str]
340
- ) -> None:
341
- if expected_exc_msg:
342
- with pytest.raises(ConfigValidationError) as exc:
343
- type_mapping_to_jsonschema(type_mapping)
344
- assert expected_exc_msg in exc.value.args[0]
345
- else:
346
- assert type_mapping_to_jsonschema(type_mapping) == expected_schema
@@ -1,3 +0,0 @@
1
- #
2
- # Copyright (c) 2021 Airbyte, Inc., all rights reserved.
3
- #