airbyte-cdk 0.72.1__py3-none-any.whl → 6.13.1.dev4107__py3-none-any.whl

Sign up to get free protection for your applications and to get access to all the features.
Files changed (517) hide show
  1. airbyte_cdk/__init__.py +355 -6
  2. airbyte_cdk/cli/__init__.py +1 -0
  3. airbyte_cdk/cli/source_declarative_manifest/__init__.py +5 -0
  4. airbyte_cdk/cli/source_declarative_manifest/_run.py +230 -0
  5. airbyte_cdk/cli/source_declarative_manifest/spec.json +17 -0
  6. airbyte_cdk/config_observation.py +29 -10
  7. airbyte_cdk/connector.py +24 -24
  8. airbyte_cdk/connector_builder/README.md +53 -0
  9. airbyte_cdk/connector_builder/connector_builder_handler.py +37 -11
  10. airbyte_cdk/connector_builder/main.py +45 -13
  11. airbyte_cdk/connector_builder/message_grouper.py +189 -50
  12. airbyte_cdk/connector_builder/models.py +3 -2
  13. airbyte_cdk/destinations/__init__.py +4 -3
  14. airbyte_cdk/destinations/destination.py +54 -20
  15. airbyte_cdk/destinations/vector_db_based/README.md +37 -0
  16. airbyte_cdk/destinations/vector_db_based/config.py +40 -17
  17. airbyte_cdk/destinations/vector_db_based/document_processor.py +56 -17
  18. airbyte_cdk/destinations/vector_db_based/embedder.py +57 -15
  19. airbyte_cdk/destinations/vector_db_based/test_utils.py +14 -4
  20. airbyte_cdk/destinations/vector_db_based/utils.py +8 -2
  21. airbyte_cdk/destinations/vector_db_based/writer.py +24 -5
  22. airbyte_cdk/entrypoint.py +153 -44
  23. airbyte_cdk/exception_handler.py +21 -3
  24. airbyte_cdk/logger.py +30 -44
  25. airbyte_cdk/models/__init__.py +13 -2
  26. airbyte_cdk/models/airbyte_protocol.py +86 -1
  27. airbyte_cdk/models/airbyte_protocol_serializers.py +44 -0
  28. airbyte_cdk/models/file_transfer_record_message.py +13 -0
  29. airbyte_cdk/models/well_known_types.py +1 -1
  30. airbyte_cdk/sources/__init__.py +5 -1
  31. airbyte_cdk/sources/abstract_source.py +125 -79
  32. airbyte_cdk/sources/concurrent_source/__init__.py +7 -2
  33. airbyte_cdk/sources/concurrent_source/concurrent_read_processor.py +102 -36
  34. airbyte_cdk/sources/concurrent_source/concurrent_source.py +29 -36
  35. airbyte_cdk/sources/concurrent_source/concurrent_source_adapter.py +94 -10
  36. airbyte_cdk/sources/concurrent_source/stream_thread_exception.py +25 -0
  37. airbyte_cdk/sources/concurrent_source/thread_pool_manager.py +20 -14
  38. airbyte_cdk/sources/config.py +3 -2
  39. airbyte_cdk/sources/connector_state_manager.py +49 -83
  40. airbyte_cdk/sources/declarative/async_job/job.py +52 -0
  41. airbyte_cdk/sources/declarative/async_job/job_orchestrator.py +497 -0
  42. airbyte_cdk/sources/declarative/async_job/job_tracker.py +75 -0
  43. airbyte_cdk/sources/declarative/async_job/repository.py +35 -0
  44. airbyte_cdk/sources/declarative/async_job/status.py +24 -0
  45. airbyte_cdk/sources/declarative/async_job/timer.py +39 -0
  46. airbyte_cdk/sources/declarative/auth/__init__.py +2 -3
  47. airbyte_cdk/sources/declarative/auth/declarative_authenticator.py +3 -1
  48. airbyte_cdk/sources/declarative/auth/jwt.py +191 -0
  49. airbyte_cdk/sources/declarative/auth/oauth.py +60 -20
  50. airbyte_cdk/sources/declarative/auth/selective_authenticator.py +10 -2
  51. airbyte_cdk/sources/declarative/auth/token.py +28 -10
  52. airbyte_cdk/sources/declarative/auth/token_provider.py +9 -8
  53. airbyte_cdk/sources/declarative/checks/check_stream.py +16 -8
  54. airbyte_cdk/sources/declarative/checks/connection_checker.py +4 -2
  55. airbyte_cdk/sources/declarative/concurrency_level/__init__.py +7 -0
  56. airbyte_cdk/sources/declarative/concurrency_level/concurrency_level.py +50 -0
  57. airbyte_cdk/sources/declarative/concurrent_declarative_source.py +421 -0
  58. airbyte_cdk/sources/declarative/datetime/datetime_parser.py +4 -0
  59. airbyte_cdk/sources/declarative/datetime/min_max_datetime.py +26 -6
  60. airbyte_cdk/sources/declarative/declarative_component_schema.yaml +1185 -85
  61. airbyte_cdk/sources/declarative/declarative_source.py +5 -2
  62. airbyte_cdk/sources/declarative/declarative_stream.py +95 -9
  63. airbyte_cdk/sources/declarative/decoders/__init__.py +23 -2
  64. airbyte_cdk/sources/declarative/decoders/composite_raw_decoder.py +97 -0
  65. airbyte_cdk/sources/declarative/decoders/decoder.py +11 -4
  66. airbyte_cdk/sources/declarative/decoders/json_decoder.py +92 -5
  67. airbyte_cdk/sources/declarative/decoders/noop_decoder.py +21 -0
  68. airbyte_cdk/sources/declarative/decoders/pagination_decoder_decorator.py +39 -0
  69. airbyte_cdk/sources/declarative/decoders/xml_decoder.py +98 -0
  70. airbyte_cdk/sources/declarative/extractors/__init__.py +12 -1
  71. airbyte_cdk/sources/declarative/extractors/dpath_extractor.py +29 -24
  72. airbyte_cdk/sources/declarative/extractors/http_selector.py +4 -5
  73. airbyte_cdk/sources/declarative/extractors/record_extractor.py +2 -3
  74. airbyte_cdk/sources/declarative/extractors/record_filter.py +65 -8
  75. airbyte_cdk/sources/declarative/extractors/record_selector.py +85 -26
  76. airbyte_cdk/sources/declarative/extractors/response_to_file_extractor.py +177 -0
  77. airbyte_cdk/sources/declarative/extractors/type_transformer.py +55 -0
  78. airbyte_cdk/sources/declarative/incremental/__init__.py +25 -3
  79. airbyte_cdk/sources/declarative/incremental/datetime_based_cursor.py +156 -48
  80. airbyte_cdk/sources/declarative/incremental/declarative_cursor.py +13 -0
  81. airbyte_cdk/sources/declarative/incremental/global_substream_cursor.py +350 -0
  82. airbyte_cdk/sources/declarative/incremental/per_partition_cursor.py +159 -74
  83. airbyte_cdk/sources/declarative/incremental/per_partition_with_global.py +200 -0
  84. airbyte_cdk/sources/declarative/incremental/resumable_full_refresh_cursor.py +122 -0
  85. airbyte_cdk/sources/declarative/interpolation/filters.py +27 -1
  86. airbyte_cdk/sources/declarative/interpolation/interpolated_boolean.py +23 -5
  87. airbyte_cdk/sources/declarative/interpolation/interpolated_mapping.py +12 -8
  88. airbyte_cdk/sources/declarative/interpolation/interpolated_nested_mapping.py +13 -6
  89. airbyte_cdk/sources/declarative/interpolation/interpolated_string.py +21 -6
  90. airbyte_cdk/sources/declarative/interpolation/interpolation.py +9 -3
  91. airbyte_cdk/sources/declarative/interpolation/jinja.py +72 -37
  92. airbyte_cdk/sources/declarative/interpolation/macros.py +72 -17
  93. airbyte_cdk/sources/declarative/manifest_declarative_source.py +193 -52
  94. airbyte_cdk/sources/declarative/migrations/legacy_to_per_partition_state_migration.py +98 -0
  95. airbyte_cdk/sources/declarative/migrations/state_migration.py +24 -0
  96. airbyte_cdk/sources/declarative/models/__init__.py +1 -1
  97. airbyte_cdk/sources/declarative/models/declarative_component_schema.py +1319 -603
  98. airbyte_cdk/sources/declarative/parsers/custom_exceptions.py +2 -2
  99. airbyte_cdk/sources/declarative/parsers/manifest_component_transformer.py +26 -4
  100. airbyte_cdk/sources/declarative/parsers/manifest_reference_resolver.py +26 -15
  101. airbyte_cdk/sources/declarative/parsers/model_to_component_factory.py +1695 -225
  102. airbyte_cdk/sources/declarative/partition_routers/__init__.py +24 -4
  103. airbyte_cdk/sources/declarative/partition_routers/async_job_partition_router.py +65 -0
  104. airbyte_cdk/sources/declarative/partition_routers/cartesian_product_stream_slicer.py +176 -0
  105. airbyte_cdk/sources/declarative/partition_routers/list_partition_router.py +39 -9
  106. airbyte_cdk/sources/declarative/partition_routers/partition_router.py +62 -0
  107. airbyte_cdk/sources/declarative/partition_routers/single_partition_router.py +15 -3
  108. airbyte_cdk/sources/declarative/partition_routers/substream_partition_router.py +222 -39
  109. airbyte_cdk/sources/declarative/requesters/error_handlers/__init__.py +19 -5
  110. airbyte_cdk/sources/declarative/requesters/error_handlers/backoff_strategies/__init__.py +3 -1
  111. airbyte_cdk/sources/declarative/requesters/error_handlers/backoff_strategies/constant_backoff_strategy.py +19 -7
  112. airbyte_cdk/sources/declarative/requesters/error_handlers/backoff_strategies/exponential_backoff_strategy.py +19 -7
  113. airbyte_cdk/sources/declarative/requesters/error_handlers/backoff_strategies/header_helper.py +4 -2
  114. airbyte_cdk/sources/declarative/requesters/error_handlers/backoff_strategies/wait_time_from_header_backoff_strategy.py +41 -9
  115. airbyte_cdk/sources/declarative/requesters/error_handlers/backoff_strategies/wait_until_time_from_header_backoff_strategy.py +29 -14
  116. airbyte_cdk/sources/declarative/requesters/error_handlers/backoff_strategy.py +5 -13
  117. airbyte_cdk/sources/declarative/requesters/error_handlers/composite_error_handler.py +32 -16
  118. airbyte_cdk/sources/declarative/requesters/error_handlers/default_error_handler.py +46 -56
  119. airbyte_cdk/sources/declarative/requesters/error_handlers/default_http_response_filter.py +40 -0
  120. airbyte_cdk/sources/declarative/requesters/error_handlers/error_handler.py +6 -32
  121. airbyte_cdk/sources/declarative/requesters/error_handlers/http_response_filter.py +119 -41
  122. airbyte_cdk/sources/declarative/requesters/http_job_repository.py +228 -0
  123. airbyte_cdk/sources/declarative/requesters/http_requester.py +98 -344
  124. airbyte_cdk/sources/declarative/requesters/paginators/__init__.py +14 -3
  125. airbyte_cdk/sources/declarative/requesters/paginators/default_paginator.py +105 -46
  126. airbyte_cdk/sources/declarative/requesters/paginators/no_pagination.py +14 -8
  127. airbyte_cdk/sources/declarative/requesters/paginators/paginator.py +19 -8
  128. airbyte_cdk/sources/declarative/requesters/paginators/strategies/__init__.py +9 -3
  129. airbyte_cdk/sources/declarative/requesters/paginators/strategies/cursor_pagination_strategy.py +53 -21
  130. airbyte_cdk/sources/declarative/requesters/paginators/strategies/offset_increment.py +42 -19
  131. airbyte_cdk/sources/declarative/requesters/paginators/strategies/page_increment.py +25 -12
  132. airbyte_cdk/sources/declarative/requesters/paginators/strategies/pagination_strategy.py +13 -10
  133. airbyte_cdk/sources/declarative/requesters/paginators/strategies/stop_condition.py +26 -13
  134. airbyte_cdk/sources/declarative/requesters/request_options/__init__.py +15 -2
  135. airbyte_cdk/sources/declarative/requesters/request_options/datetime_based_request_options_provider.py +91 -0
  136. airbyte_cdk/sources/declarative/requesters/request_options/default_request_options_provider.py +60 -0
  137. airbyte_cdk/sources/declarative/requesters/request_options/interpolated_nested_request_input_provider.py +31 -14
  138. airbyte_cdk/sources/declarative/requesters/request_options/interpolated_request_input_provider.py +27 -15
  139. airbyte_cdk/sources/declarative/requesters/request_options/interpolated_request_options_provider.py +63 -10
  140. airbyte_cdk/sources/declarative/requesters/request_options/request_options_provider.py +1 -1
  141. airbyte_cdk/sources/declarative/requesters/requester.py +9 -17
  142. airbyte_cdk/sources/declarative/resolvers/__init__.py +41 -0
  143. airbyte_cdk/sources/declarative/resolvers/components_resolver.py +55 -0
  144. airbyte_cdk/sources/declarative/resolvers/config_components_resolver.py +136 -0
  145. airbyte_cdk/sources/declarative/resolvers/http_components_resolver.py +112 -0
  146. airbyte_cdk/sources/declarative/retrievers/__init__.py +6 -2
  147. airbyte_cdk/sources/declarative/retrievers/async_retriever.py +100 -0
  148. airbyte_cdk/sources/declarative/retrievers/retriever.py +1 -3
  149. airbyte_cdk/sources/declarative/retrievers/simple_retriever.py +228 -72
  150. airbyte_cdk/sources/declarative/schema/__init__.py +14 -1
  151. airbyte_cdk/sources/declarative/schema/default_schema_loader.py +5 -3
  152. airbyte_cdk/sources/declarative/schema/dynamic_schema_loader.py +236 -0
  153. airbyte_cdk/sources/declarative/schema/json_file_schema_loader.py +8 -8
  154. airbyte_cdk/sources/declarative/spec/spec.py +12 -5
  155. airbyte_cdk/sources/declarative/stream_slicers/__init__.py +1 -2
  156. airbyte_cdk/sources/declarative/stream_slicers/declarative_partition_generator.py +88 -0
  157. airbyte_cdk/sources/declarative/stream_slicers/stream_slicer.py +9 -14
  158. airbyte_cdk/sources/declarative/transformations/add_fields.py +19 -11
  159. airbyte_cdk/sources/declarative/transformations/flatten_fields.py +52 -0
  160. airbyte_cdk/sources/declarative/transformations/keys_replace_transformation.py +61 -0
  161. airbyte_cdk/sources/declarative/transformations/keys_to_lower_transformation.py +22 -0
  162. airbyte_cdk/sources/declarative/transformations/keys_to_snake_transformation.py +68 -0
  163. airbyte_cdk/sources/declarative/transformations/remove_fields.py +13 -10
  164. airbyte_cdk/sources/declarative/transformations/transformation.py +5 -5
  165. airbyte_cdk/sources/declarative/types.py +19 -110
  166. airbyte_cdk/sources/declarative/yaml_declarative_source.py +31 -10
  167. airbyte_cdk/sources/embedded/base_integration.py +16 -5
  168. airbyte_cdk/sources/embedded/catalog.py +16 -4
  169. airbyte_cdk/sources/embedded/runner.py +19 -3
  170. airbyte_cdk/sources/embedded/tools.py +5 -2
  171. airbyte_cdk/sources/file_based/README.md +152 -0
  172. airbyte_cdk/sources/file_based/__init__.py +24 -0
  173. airbyte_cdk/sources/file_based/availability_strategy/__init__.py +9 -2
  174. airbyte_cdk/sources/file_based/availability_strategy/abstract_file_based_availability_strategy.py +22 -6
  175. airbyte_cdk/sources/file_based/availability_strategy/default_file_based_availability_strategy.py +46 -10
  176. airbyte_cdk/sources/file_based/config/abstract_file_based_spec.py +58 -10
  177. airbyte_cdk/sources/file_based/config/avro_format.py +2 -1
  178. airbyte_cdk/sources/file_based/config/csv_format.py +29 -10
  179. airbyte_cdk/sources/file_based/config/excel_format.py +18 -0
  180. airbyte_cdk/sources/file_based/config/file_based_stream_config.py +16 -4
  181. airbyte_cdk/sources/file_based/config/jsonl_format.py +2 -1
  182. airbyte_cdk/sources/file_based/config/parquet_format.py +2 -1
  183. airbyte_cdk/sources/file_based/config/unstructured_format.py +13 -5
  184. airbyte_cdk/sources/file_based/discovery_policy/__init__.py +6 -2
  185. airbyte_cdk/sources/file_based/discovery_policy/abstract_discovery_policy.py +2 -4
  186. airbyte_cdk/sources/file_based/discovery_policy/default_discovery_policy.py +7 -2
  187. airbyte_cdk/sources/file_based/exceptions.py +52 -15
  188. airbyte_cdk/sources/file_based/file_based_source.py +163 -33
  189. airbyte_cdk/sources/file_based/file_based_stream_reader.py +83 -5
  190. airbyte_cdk/sources/file_based/file_types/__init__.py +14 -1
  191. airbyte_cdk/sources/file_based/file_types/avro_parser.py +75 -24
  192. airbyte_cdk/sources/file_based/file_types/csv_parser.py +116 -34
  193. airbyte_cdk/sources/file_based/file_types/excel_parser.py +196 -0
  194. airbyte_cdk/sources/file_based/file_types/file_transfer.py +37 -0
  195. airbyte_cdk/sources/file_based/file_types/file_type_parser.py +4 -1
  196. airbyte_cdk/sources/file_based/file_types/jsonl_parser.py +24 -8
  197. airbyte_cdk/sources/file_based/file_types/parquet_parser.py +60 -18
  198. airbyte_cdk/sources/file_based/file_types/unstructured_parser.py +147 -41
  199. airbyte_cdk/sources/file_based/remote_file.py +1 -1
  200. airbyte_cdk/sources/file_based/schema_helpers.py +38 -10
  201. airbyte_cdk/sources/file_based/schema_validation_policies/__init__.py +3 -1
  202. airbyte_cdk/sources/file_based/schema_validation_policies/abstract_schema_validation_policy.py +3 -1
  203. airbyte_cdk/sources/file_based/schema_validation_policies/default_schema_validation_policies.py +16 -5
  204. airbyte_cdk/sources/file_based/stream/abstract_file_based_stream.py +50 -13
  205. airbyte_cdk/sources/file_based/stream/concurrent/adapters.py +67 -27
  206. airbyte_cdk/sources/file_based/stream/concurrent/cursor/__init__.py +5 -1
  207. airbyte_cdk/sources/file_based/stream/concurrent/cursor/abstract_concurrent_file_based_cursor.py +14 -23
  208. airbyte_cdk/sources/file_based/stream/concurrent/cursor/file_based_concurrent_cursor.py +54 -18
  209. airbyte_cdk/sources/file_based/stream/concurrent/cursor/file_based_final_state_cursor.py +21 -9
  210. airbyte_cdk/sources/file_based/stream/cursor/abstract_file_based_cursor.py +3 -1
  211. airbyte_cdk/sources/file_based/stream/cursor/default_file_based_cursor.py +27 -10
  212. airbyte_cdk/sources/file_based/stream/default_file_based_stream.py +175 -45
  213. airbyte_cdk/sources/http_logger.py +8 -3
  214. airbyte_cdk/sources/message/__init__.py +7 -1
  215. airbyte_cdk/sources/message/repository.py +18 -4
  216. airbyte_cdk/sources/source.py +42 -38
  217. airbyte_cdk/sources/streams/__init__.py +2 -2
  218. airbyte_cdk/sources/streams/availability_strategy.py +54 -3
  219. airbyte_cdk/sources/streams/call_rate.py +64 -21
  220. airbyte_cdk/sources/streams/checkpoint/__init__.py +26 -0
  221. airbyte_cdk/sources/streams/checkpoint/checkpoint_reader.py +335 -0
  222. airbyte_cdk/sources/{declarative/incremental → streams/checkpoint}/cursor.py +17 -14
  223. airbyte_cdk/sources/streams/checkpoint/per_partition_key_serializer.py +22 -0
  224. airbyte_cdk/sources/streams/checkpoint/resumable_full_refresh_cursor.py +51 -0
  225. airbyte_cdk/sources/streams/checkpoint/substream_resumable_full_refresh_cursor.py +110 -0
  226. airbyte_cdk/sources/streams/concurrent/README.md +7 -0
  227. airbyte_cdk/sources/streams/concurrent/abstract_stream.py +7 -2
  228. airbyte_cdk/sources/streams/concurrent/adapters.py +84 -75
  229. airbyte_cdk/sources/streams/concurrent/availability_strategy.py +30 -2
  230. airbyte_cdk/sources/streams/concurrent/cursor.py +298 -42
  231. airbyte_cdk/sources/streams/concurrent/default_stream.py +12 -3
  232. airbyte_cdk/sources/streams/concurrent/exceptions.py +3 -0
  233. airbyte_cdk/sources/streams/concurrent/helpers.py +14 -3
  234. airbyte_cdk/sources/streams/concurrent/partition_enqueuer.py +12 -3
  235. airbyte_cdk/sources/streams/concurrent/partition_reader.py +10 -3
  236. airbyte_cdk/sources/streams/concurrent/partitions/partition.py +1 -16
  237. airbyte_cdk/sources/streams/concurrent/partitions/stream_slicer.py +21 -0
  238. airbyte_cdk/sources/streams/concurrent/partitions/types.py +15 -5
  239. airbyte_cdk/sources/streams/concurrent/state_converters/abstract_stream_state_converter.py +109 -17
  240. airbyte_cdk/sources/streams/concurrent/state_converters/datetime_stream_state_converter.py +90 -72
  241. airbyte_cdk/sources/streams/core.py +412 -87
  242. airbyte_cdk/sources/streams/http/__init__.py +2 -1
  243. airbyte_cdk/sources/streams/http/availability_strategy.py +12 -101
  244. airbyte_cdk/sources/streams/http/error_handlers/__init__.py +22 -0
  245. airbyte_cdk/sources/streams/http/error_handlers/backoff_strategy.py +28 -0
  246. airbyte_cdk/sources/streams/http/error_handlers/default_backoff_strategy.py +17 -0
  247. airbyte_cdk/sources/streams/http/error_handlers/default_error_mapping.py +86 -0
  248. airbyte_cdk/sources/streams/http/error_handlers/error_handler.py +42 -0
  249. airbyte_cdk/sources/streams/http/error_handlers/error_message_parser.py +19 -0
  250. airbyte_cdk/sources/streams/http/error_handlers/http_status_error_handler.py +110 -0
  251. airbyte_cdk/sources/streams/http/error_handlers/json_error_message_parser.py +52 -0
  252. airbyte_cdk/sources/streams/http/error_handlers/response_models.py +65 -0
  253. airbyte_cdk/sources/streams/http/exceptions.py +27 -7
  254. airbyte_cdk/sources/streams/http/http.py +369 -246
  255. airbyte_cdk/sources/streams/http/http_client.py +531 -0
  256. airbyte_cdk/sources/streams/http/rate_limiting.py +76 -12
  257. airbyte_cdk/sources/streams/http/requests_native_auth/abstract_oauth.py +28 -9
  258. airbyte_cdk/sources/streams/http/requests_native_auth/abstract_token.py +2 -1
  259. airbyte_cdk/sources/streams/http/requests_native_auth/oauth.py +90 -35
  260. airbyte_cdk/sources/streams/http/requests_native_auth/token.py +13 -3
  261. airbyte_cdk/sources/types.py +154 -0
  262. airbyte_cdk/sources/utils/record_helper.py +36 -21
  263. airbyte_cdk/sources/utils/schema_helpers.py +13 -6
  264. airbyte_cdk/sources/utils/slice_logger.py +4 -1
  265. airbyte_cdk/sources/utils/transform.py +54 -20
  266. airbyte_cdk/sql/_util/hashing.py +34 -0
  267. airbyte_cdk/sql/_util/name_normalizers.py +92 -0
  268. airbyte_cdk/sql/constants.py +32 -0
  269. airbyte_cdk/sql/exceptions.py +235 -0
  270. airbyte_cdk/sql/secrets.py +123 -0
  271. airbyte_cdk/sql/shared/__init__.py +15 -0
  272. airbyte_cdk/sql/shared/catalog_providers.py +145 -0
  273. airbyte_cdk/sql/shared/sql_processor.py +786 -0
  274. airbyte_cdk/sql/types.py +160 -0
  275. airbyte_cdk/test/catalog_builder.py +70 -18
  276. airbyte_cdk/test/entrypoint_wrapper.py +117 -42
  277. airbyte_cdk/test/mock_http/__init__.py +1 -1
  278. airbyte_cdk/test/mock_http/matcher.py +6 -0
  279. airbyte_cdk/test/mock_http/mocker.py +57 -10
  280. airbyte_cdk/test/mock_http/request.py +19 -3
  281. airbyte_cdk/test/mock_http/response.py +3 -1
  282. airbyte_cdk/test/mock_http/response_builder.py +32 -16
  283. airbyte_cdk/test/state_builder.py +18 -10
  284. airbyte_cdk/test/utils/__init__.py +1 -0
  285. airbyte_cdk/test/utils/data.py +24 -0
  286. airbyte_cdk/test/utils/http_mocking.py +16 -0
  287. airbyte_cdk/test/utils/manifest_only_fixtures.py +60 -0
  288. airbyte_cdk/test/utils/reading.py +26 -0
  289. airbyte_cdk/utils/__init__.py +2 -1
  290. airbyte_cdk/utils/airbyte_secrets_utils.py +5 -3
  291. airbyte_cdk/utils/analytics_message.py +10 -2
  292. airbyte_cdk/utils/datetime_format_inferrer.py +4 -1
  293. airbyte_cdk/utils/event_timing.py +10 -10
  294. airbyte_cdk/utils/mapping_helpers.py +3 -1
  295. airbyte_cdk/utils/message_utils.py +20 -11
  296. airbyte_cdk/utils/print_buffer.py +75 -0
  297. airbyte_cdk/utils/schema_inferrer.py +198 -28
  298. airbyte_cdk/utils/slice_hasher.py +30 -0
  299. airbyte_cdk/utils/spec_schema_transformations.py +6 -3
  300. airbyte_cdk/utils/stream_status_utils.py +8 -1
  301. airbyte_cdk/utils/traced_exception.py +61 -21
  302. airbyte_cdk-6.13.1.dev4107.dist-info/METADATA +109 -0
  303. airbyte_cdk-6.13.1.dev4107.dist-info/RECORD +349 -0
  304. {airbyte_cdk-0.72.1.dist-info → airbyte_cdk-6.13.1.dev4107.dist-info}/WHEEL +1 -2
  305. airbyte_cdk-6.13.1.dev4107.dist-info/entry_points.txt +3 -0
  306. airbyte_cdk/sources/declarative/create_partial.py +0 -92
  307. airbyte_cdk/sources/declarative/parsers/class_types_registry.py +0 -102
  308. airbyte_cdk/sources/declarative/parsers/default_implementation_registry.py +0 -64
  309. airbyte_cdk/sources/declarative/requesters/error_handlers/response_action.py +0 -16
  310. airbyte_cdk/sources/declarative/requesters/error_handlers/response_status.py +0 -68
  311. airbyte_cdk/sources/declarative/stream_slicers/cartesian_product_stream_slicer.py +0 -114
  312. airbyte_cdk/sources/deprecated/base_source.py +0 -94
  313. airbyte_cdk/sources/deprecated/client.py +0 -99
  314. airbyte_cdk/sources/singer/__init__.py +0 -8
  315. airbyte_cdk/sources/singer/singer_helpers.py +0 -304
  316. airbyte_cdk/sources/singer/source.py +0 -186
  317. airbyte_cdk/sources/streams/concurrent/partitions/record.py +0 -23
  318. airbyte_cdk/sources/streams/http/auth/__init__.py +0 -17
  319. airbyte_cdk/sources/streams/http/auth/core.py +0 -29
  320. airbyte_cdk/sources/streams/http/auth/oauth.py +0 -113
  321. airbyte_cdk/sources/streams/http/auth/token.py +0 -47
  322. airbyte_cdk/sources/streams/utils/stream_helper.py +0 -40
  323. airbyte_cdk/sources/utils/catalog_helpers.py +0 -22
  324. airbyte_cdk/sources/utils/schema_models.py +0 -84
  325. airbyte_cdk-0.72.1.dist-info/METADATA +0 -243
  326. airbyte_cdk-0.72.1.dist-info/RECORD +0 -466
  327. airbyte_cdk-0.72.1.dist-info/top_level.txt +0 -3
  328. source_declarative_manifest/main.py +0 -29
  329. unit_tests/connector_builder/__init__.py +0 -3
  330. unit_tests/connector_builder/test_connector_builder_handler.py +0 -871
  331. unit_tests/connector_builder/test_message_grouper.py +0 -713
  332. unit_tests/connector_builder/utils.py +0 -27
  333. unit_tests/destinations/test_destination.py +0 -243
  334. unit_tests/singer/test_singer_helpers.py +0 -56
  335. unit_tests/singer/test_singer_source.py +0 -112
  336. unit_tests/sources/__init__.py +0 -0
  337. unit_tests/sources/concurrent_source/__init__.py +0 -3
  338. unit_tests/sources/concurrent_source/test_concurrent_source_adapter.py +0 -106
  339. unit_tests/sources/declarative/__init__.py +0 -3
  340. unit_tests/sources/declarative/auth/__init__.py +0 -3
  341. unit_tests/sources/declarative/auth/test_oauth.py +0 -331
  342. unit_tests/sources/declarative/auth/test_selective_authenticator.py +0 -39
  343. unit_tests/sources/declarative/auth/test_session_token_auth.py +0 -182
  344. unit_tests/sources/declarative/auth/test_token_auth.py +0 -200
  345. unit_tests/sources/declarative/auth/test_token_provider.py +0 -73
  346. unit_tests/sources/declarative/checks/__init__.py +0 -3
  347. unit_tests/sources/declarative/checks/test_check_stream.py +0 -146
  348. unit_tests/sources/declarative/decoders/__init__.py +0 -0
  349. unit_tests/sources/declarative/decoders/test_json_decoder.py +0 -16
  350. unit_tests/sources/declarative/external_component.py +0 -13
  351. unit_tests/sources/declarative/extractors/__init__.py +0 -3
  352. unit_tests/sources/declarative/extractors/test_dpath_extractor.py +0 -55
  353. unit_tests/sources/declarative/extractors/test_record_filter.py +0 -55
  354. unit_tests/sources/declarative/extractors/test_record_selector.py +0 -179
  355. unit_tests/sources/declarative/incremental/__init__.py +0 -0
  356. unit_tests/sources/declarative/incremental/test_datetime_based_cursor.py +0 -860
  357. unit_tests/sources/declarative/incremental/test_per_partition_cursor.py +0 -406
  358. unit_tests/sources/declarative/incremental/test_per_partition_cursor_integration.py +0 -332
  359. unit_tests/sources/declarative/interpolation/__init__.py +0 -3
  360. unit_tests/sources/declarative/interpolation/test_filters.py +0 -80
  361. unit_tests/sources/declarative/interpolation/test_interpolated_boolean.py +0 -40
  362. unit_tests/sources/declarative/interpolation/test_interpolated_mapping.py +0 -35
  363. unit_tests/sources/declarative/interpolation/test_interpolated_nested_mapping.py +0 -45
  364. unit_tests/sources/declarative/interpolation/test_interpolated_string.py +0 -25
  365. unit_tests/sources/declarative/interpolation/test_jinja.py +0 -240
  366. unit_tests/sources/declarative/interpolation/test_macros.py +0 -73
  367. unit_tests/sources/declarative/parsers/__init__.py +0 -3
  368. unit_tests/sources/declarative/parsers/test_manifest_component_transformer.py +0 -406
  369. unit_tests/sources/declarative/parsers/test_manifest_reference_resolver.py +0 -139
  370. unit_tests/sources/declarative/parsers/test_model_to_component_factory.py +0 -1847
  371. unit_tests/sources/declarative/parsers/testing_components.py +0 -36
  372. unit_tests/sources/declarative/partition_routers/__init__.py +0 -3
  373. unit_tests/sources/declarative/partition_routers/test_list_partition_router.py +0 -155
  374. unit_tests/sources/declarative/partition_routers/test_single_partition_router.py +0 -14
  375. unit_tests/sources/declarative/partition_routers/test_substream_partition_router.py +0 -404
  376. unit_tests/sources/declarative/requesters/__init__.py +0 -3
  377. unit_tests/sources/declarative/requesters/error_handlers/__init__.py +0 -3
  378. unit_tests/sources/declarative/requesters/error_handlers/backoff_strategies/__init__.py +0 -3
  379. unit_tests/sources/declarative/requesters/error_handlers/backoff_strategies/test_constant_backoff.py +0 -34
  380. unit_tests/sources/declarative/requesters/error_handlers/backoff_strategies/test_exponential_backoff.py +0 -36
  381. unit_tests/sources/declarative/requesters/error_handlers/backoff_strategies/test_header_helper.py +0 -38
  382. unit_tests/sources/declarative/requesters/error_handlers/backoff_strategies/test_wait_time_from_header.py +0 -35
  383. unit_tests/sources/declarative/requesters/error_handlers/backoff_strategies/test_wait_until_time_from_header.py +0 -64
  384. unit_tests/sources/declarative/requesters/error_handlers/test_composite_error_handler.py +0 -213
  385. unit_tests/sources/declarative/requesters/error_handlers/test_default_error_handler.py +0 -178
  386. unit_tests/sources/declarative/requesters/error_handlers/test_http_response_filter.py +0 -121
  387. unit_tests/sources/declarative/requesters/error_handlers/test_response_status.py +0 -44
  388. unit_tests/sources/declarative/requesters/paginators/__init__.py +0 -3
  389. unit_tests/sources/declarative/requesters/paginators/test_cursor_pagination_strategy.py +0 -64
  390. unit_tests/sources/declarative/requesters/paginators/test_default_paginator.py +0 -313
  391. unit_tests/sources/declarative/requesters/paginators/test_no_paginator.py +0 -12
  392. unit_tests/sources/declarative/requesters/paginators/test_offset_increment.py +0 -58
  393. unit_tests/sources/declarative/requesters/paginators/test_page_increment.py +0 -70
  394. unit_tests/sources/declarative/requesters/paginators/test_request_option.py +0 -43
  395. unit_tests/sources/declarative/requesters/paginators/test_stop_condition.py +0 -105
  396. unit_tests/sources/declarative/requesters/request_options/__init__.py +0 -3
  397. unit_tests/sources/declarative/requesters/request_options/test_interpolated_request_options_provider.py +0 -101
  398. unit_tests/sources/declarative/requesters/test_http_requester.py +0 -974
  399. unit_tests/sources/declarative/requesters/test_interpolated_request_input_provider.py +0 -32
  400. unit_tests/sources/declarative/retrievers/__init__.py +0 -3
  401. unit_tests/sources/declarative/retrievers/test_simple_retriever.py +0 -542
  402. unit_tests/sources/declarative/schema/__init__.py +0 -6
  403. unit_tests/sources/declarative/schema/source_test/SourceTest.py +0 -8
  404. unit_tests/sources/declarative/schema/source_test/__init__.py +0 -3
  405. unit_tests/sources/declarative/schema/test_default_schema_loader.py +0 -32
  406. unit_tests/sources/declarative/schema/test_inline_schema_loader.py +0 -19
  407. unit_tests/sources/declarative/schema/test_json_file_schema_loader.py +0 -26
  408. unit_tests/sources/declarative/states/__init__.py +0 -3
  409. unit_tests/sources/declarative/stream_slicers/__init__.py +0 -3
  410. unit_tests/sources/declarative/stream_slicers/test_cartesian_product_stream_slicer.py +0 -225
  411. unit_tests/sources/declarative/test_create_partial.py +0 -83
  412. unit_tests/sources/declarative/test_declarative_stream.py +0 -103
  413. unit_tests/sources/declarative/test_manifest_declarative_source.py +0 -1260
  414. unit_tests/sources/declarative/test_types.py +0 -39
  415. unit_tests/sources/declarative/test_yaml_declarative_source.py +0 -148
  416. unit_tests/sources/file_based/__init__.py +0 -0
  417. unit_tests/sources/file_based/availability_strategy/__init__.py +0 -0
  418. unit_tests/sources/file_based/availability_strategy/test_default_file_based_availability_strategy.py +0 -100
  419. unit_tests/sources/file_based/config/__init__.py +0 -0
  420. unit_tests/sources/file_based/config/test_abstract_file_based_spec.py +0 -28
  421. unit_tests/sources/file_based/config/test_csv_format.py +0 -34
  422. unit_tests/sources/file_based/config/test_file_based_stream_config.py +0 -84
  423. unit_tests/sources/file_based/discovery_policy/__init__.py +0 -0
  424. unit_tests/sources/file_based/discovery_policy/test_default_discovery_policy.py +0 -31
  425. unit_tests/sources/file_based/file_types/__init__.py +0 -0
  426. unit_tests/sources/file_based/file_types/test_avro_parser.py +0 -243
  427. unit_tests/sources/file_based/file_types/test_csv_parser.py +0 -546
  428. unit_tests/sources/file_based/file_types/test_jsonl_parser.py +0 -158
  429. unit_tests/sources/file_based/file_types/test_parquet_parser.py +0 -274
  430. unit_tests/sources/file_based/file_types/test_unstructured_parser.py +0 -593
  431. unit_tests/sources/file_based/helpers.py +0 -70
  432. unit_tests/sources/file_based/in_memory_files_source.py +0 -211
  433. unit_tests/sources/file_based/scenarios/__init__.py +0 -0
  434. unit_tests/sources/file_based/scenarios/avro_scenarios.py +0 -744
  435. unit_tests/sources/file_based/scenarios/check_scenarios.py +0 -220
  436. unit_tests/sources/file_based/scenarios/concurrent_incremental_scenarios.py +0 -2844
  437. unit_tests/sources/file_based/scenarios/csv_scenarios.py +0 -3105
  438. unit_tests/sources/file_based/scenarios/file_based_source_builder.py +0 -91
  439. unit_tests/sources/file_based/scenarios/incremental_scenarios.py +0 -1926
  440. unit_tests/sources/file_based/scenarios/jsonl_scenarios.py +0 -930
  441. unit_tests/sources/file_based/scenarios/parquet_scenarios.py +0 -754
  442. unit_tests/sources/file_based/scenarios/scenario_builder.py +0 -234
  443. unit_tests/sources/file_based/scenarios/unstructured_scenarios.py +0 -608
  444. unit_tests/sources/file_based/scenarios/user_input_schema_scenarios.py +0 -746
  445. unit_tests/sources/file_based/scenarios/validation_policy_scenarios.py +0 -726
  446. unit_tests/sources/file_based/stream/__init__.py +0 -0
  447. unit_tests/sources/file_based/stream/concurrent/__init__.py +0 -0
  448. unit_tests/sources/file_based/stream/concurrent/test_adapters.py +0 -362
  449. unit_tests/sources/file_based/stream/concurrent/test_file_based_concurrent_cursor.py +0 -458
  450. unit_tests/sources/file_based/stream/test_default_file_based_cursor.py +0 -310
  451. unit_tests/sources/file_based/stream/test_default_file_based_stream.py +0 -244
  452. unit_tests/sources/file_based/test_file_based_scenarios.py +0 -320
  453. unit_tests/sources/file_based/test_file_based_stream_reader.py +0 -272
  454. unit_tests/sources/file_based/test_scenarios.py +0 -253
  455. unit_tests/sources/file_based/test_schema_helpers.py +0 -346
  456. unit_tests/sources/fixtures/__init__.py +0 -3
  457. unit_tests/sources/fixtures/source_test_fixture.py +0 -153
  458. unit_tests/sources/message/__init__.py +0 -0
  459. unit_tests/sources/message/test_repository.py +0 -153
  460. unit_tests/sources/streams/__init__.py +0 -0
  461. unit_tests/sources/streams/concurrent/__init__.py +0 -3
  462. unit_tests/sources/streams/concurrent/scenarios/__init__.py +0 -3
  463. unit_tests/sources/streams/concurrent/scenarios/incremental_scenarios.py +0 -250
  464. unit_tests/sources/streams/concurrent/scenarios/stream_facade_builder.py +0 -140
  465. unit_tests/sources/streams/concurrent/scenarios/stream_facade_scenarios.py +0 -452
  466. unit_tests/sources/streams/concurrent/scenarios/test_concurrent_scenarios.py +0 -76
  467. unit_tests/sources/streams/concurrent/scenarios/thread_based_concurrent_stream_scenarios.py +0 -418
  468. unit_tests/sources/streams/concurrent/scenarios/thread_based_concurrent_stream_source_builder.py +0 -142
  469. unit_tests/sources/streams/concurrent/scenarios/utils.py +0 -55
  470. unit_tests/sources/streams/concurrent/test_adapters.py +0 -380
  471. unit_tests/sources/streams/concurrent/test_concurrent_read_processor.py +0 -684
  472. unit_tests/sources/streams/concurrent/test_cursor.py +0 -139
  473. unit_tests/sources/streams/concurrent/test_datetime_state_converter.py +0 -369
  474. unit_tests/sources/streams/concurrent/test_default_stream.py +0 -197
  475. unit_tests/sources/streams/concurrent/test_partition_enqueuer.py +0 -90
  476. unit_tests/sources/streams/concurrent/test_partition_reader.py +0 -67
  477. unit_tests/sources/streams/concurrent/test_thread_pool_manager.py +0 -106
  478. unit_tests/sources/streams/http/__init__.py +0 -0
  479. unit_tests/sources/streams/http/auth/__init__.py +0 -0
  480. unit_tests/sources/streams/http/auth/test_auth.py +0 -173
  481. unit_tests/sources/streams/http/requests_native_auth/__init__.py +0 -0
  482. unit_tests/sources/streams/http/requests_native_auth/test_requests_native_auth.py +0 -423
  483. unit_tests/sources/streams/http/test_availability_strategy.py +0 -180
  484. unit_tests/sources/streams/http/test_http.py +0 -635
  485. unit_tests/sources/streams/test_availability_strategy.py +0 -70
  486. unit_tests/sources/streams/test_call_rate.py +0 -300
  487. unit_tests/sources/streams/test_stream_read.py +0 -405
  488. unit_tests/sources/streams/test_streams_core.py +0 -184
  489. unit_tests/sources/test_abstract_source.py +0 -1442
  490. unit_tests/sources/test_concurrent_source.py +0 -112
  491. unit_tests/sources/test_config.py +0 -92
  492. unit_tests/sources/test_connector_state_manager.py +0 -482
  493. unit_tests/sources/test_http_logger.py +0 -252
  494. unit_tests/sources/test_integration_source.py +0 -86
  495. unit_tests/sources/test_source.py +0 -684
  496. unit_tests/sources/test_source_read.py +0 -460
  497. unit_tests/test/__init__.py +0 -0
  498. unit_tests/test/mock_http/__init__.py +0 -0
  499. unit_tests/test/mock_http/test_matcher.py +0 -53
  500. unit_tests/test/mock_http/test_mocker.py +0 -214
  501. unit_tests/test/mock_http/test_request.py +0 -117
  502. unit_tests/test/mock_http/test_response_builder.py +0 -177
  503. unit_tests/test/test_entrypoint_wrapper.py +0 -240
  504. unit_tests/utils/__init__.py +0 -0
  505. unit_tests/utils/test_datetime_format_inferrer.py +0 -60
  506. unit_tests/utils/test_mapping_helpers.py +0 -54
  507. unit_tests/utils/test_message_utils.py +0 -91
  508. unit_tests/utils/test_rate_limiting.py +0 -26
  509. unit_tests/utils/test_schema_inferrer.py +0 -202
  510. unit_tests/utils/test_secret_utils.py +0 -135
  511. unit_tests/utils/test_stream_status_utils.py +0 -61
  512. unit_tests/utils/test_traced_exception.py +0 -107
  513. /airbyte_cdk/sources/{deprecated → declarative/async_job}/__init__.py +0 -0
  514. {source_declarative_manifest → airbyte_cdk/sources/declarative/migrations}/__init__.py +0 -0
  515. {unit_tests/destinations → airbyte_cdk/sql}/__init__.py +0 -0
  516. {unit_tests/singer → airbyte_cdk/sql/_util}/__init__.py +0 -0
  517. {airbyte_cdk-0.72.1.dist-info → airbyte_cdk-6.13.1.dev4107.dist-info}/LICENSE.txt +0 -0
@@ -1,754 +0,0 @@
1
- #
2
- # Copyright (c) 2023 Airbyte, Inc., all rights reserved.
3
- #
4
-
5
- import datetime
6
- import decimal
7
-
8
- import pyarrow as pa
9
- from airbyte_cdk.utils.traced_exception import AirbyteTracedException
10
- from unit_tests.sources.file_based.in_memory_files_source import TemporaryParquetFilesStreamReader
11
- from unit_tests.sources.file_based.scenarios.file_based_source_builder import FileBasedSourceBuilder
12
- from unit_tests.sources.file_based.scenarios.scenario_builder import TestScenarioBuilder
13
-
14
- _single_parquet_file = {
15
- "a.parquet": {
16
- "contents": [
17
- ("col1", "col2"),
18
- ("val11", "val12"),
19
- ("val21", "val22"),
20
- ],
21
- "last_modified": "2023-06-05T03:54:07.000Z",
22
- }
23
- }
24
-
25
- _single_partitioned_parquet_file = {
26
- "path_prefix/partition1=1/partition2=2/a.parquet": {
27
- "contents": [
28
- ("col1", "col2"),
29
- ("val11", "val12"),
30
- ("val21", "val22"),
31
- ],
32
- "last_modified": "2023-06-05T03:54:07.000Z",
33
- }
34
- }
35
-
36
- _parquet_file_with_decimal = {
37
- "a.parquet": {
38
- "contents": [
39
- ("col1",),
40
- (decimal.Decimal("13.00"),),
41
- ],
42
- "schema": pa.schema(
43
- [
44
- pa.field("col1", pa.decimal128(5, 2)),
45
- ]
46
- ),
47
- "last_modified": "2023-06-05T03:54:07.000Z",
48
- }
49
- }
50
-
51
- _multiple_parquet_file = {
52
- "a.parquet": {
53
- "contents": [
54
- ("col1", "col2"),
55
- ("val11a", "val12a"),
56
- ("val21a", "val22a"),
57
- ],
58
- "last_modified": "2023-06-05T03:54:07.000Z",
59
- },
60
- "b.parquet": {
61
- "contents": [
62
- ("col1", "col2", "col3"),
63
- ("val11b", "val12b", "val13b"),
64
- ("val21b", "val22b", "val23b"),
65
- ],
66
- "last_modified": "2023-06-05T03:54:07.000Z",
67
- },
68
- }
69
-
70
- _parquet_file_with_various_types = {
71
- "a.parquet": {
72
- "contents": [
73
- (
74
- "col_bool",
75
- "col_int8",
76
- "col_int16",
77
- "col_int32",
78
- "col_uint8",
79
- "col_uint16",
80
- "col_uint32",
81
- "col_uint64",
82
- "col_float32",
83
- "col_float64",
84
- "col_string",
85
- "col_date32",
86
- "col_date64",
87
- "col_timestamp_without_tz",
88
- "col_timestamp_with_tz",
89
- "col_time32s",
90
- "col_time32ms",
91
- "col_time64us",
92
- "col_struct",
93
- "col_list",
94
- "col_duration",
95
- "col_binary",
96
- ),
97
- (
98
- True,
99
- -1,
100
- 1,
101
- 2,
102
- 2,
103
- 3,
104
- 4,
105
- 5,
106
- 3.14,
107
- 5.0,
108
- "2020-01-01",
109
- datetime.date(2021, 1, 1),
110
- datetime.date(2022, 1, 1),
111
- datetime.datetime(2023, 1, 1, 1, 2, 3),
112
- datetime.datetime(2024, 3, 4, 5, 6, 7, tzinfo=datetime.timezone.utc),
113
- datetime.time(1, 2, 3),
114
- datetime.time(2, 3, 4),
115
- datetime.time(1, 2, 3, 4),
116
- {"struct_key": "struct_value"},
117
- [1, 2, 3, 4],
118
- 12345,
119
- b"binary string. Hello world!",
120
- ),
121
- ],
122
- "schema": pa.schema(
123
- [
124
- pa.field("col_bool", pa.bool_()),
125
- pa.field("col_int8", pa.int8()),
126
- pa.field("col_int16", pa.int16()),
127
- pa.field("col_int32", pa.int32()),
128
- pa.field("col_uint8", pa.uint8()),
129
- pa.field("col_uint16", pa.uint16()),
130
- pa.field("col_uint32", pa.uint32()),
131
- pa.field("col_uint64", pa.uint64()),
132
- pa.field("col_float32", pa.float32()),
133
- pa.field("col_float64", pa.float64()),
134
- pa.field("col_string", pa.string()),
135
- pa.field("col_date32", pa.date32()),
136
- pa.field("col_date64", pa.date64()),
137
- pa.field("col_timestamp_without_tz", pa.timestamp("s")),
138
- pa.field("col_timestamp_with_tz", pa.timestamp("s", tz="UTC")),
139
- pa.field("col_time32s", pa.time32("s")),
140
- pa.field("col_time32ms", pa.time32("ms")),
141
- pa.field("col_time64us", pa.time64("us")),
142
- pa.field("col_struct", pa.struct([pa.field("struct_key", pa.string())])),
143
- pa.field("col_list", pa.list_(pa.int32())),
144
- pa.field("col_duration", pa.duration("s")),
145
- pa.field("col_binary", pa.binary()),
146
- ]
147
- ),
148
- "last_modified": "2023-06-05T03:54:07.000Z",
149
- }
150
- }
151
-
152
- single_parquet_scenario = (
153
- TestScenarioBuilder()
154
- .set_name("single_parquet_stream")
155
- .set_config(
156
- {
157
- "streams": [
158
- {
159
- "name": "stream1",
160
- "format": {"filetype": "parquet"},
161
- "globs": ["*"],
162
- "validation_policy": "Emit Record",
163
- }
164
- ]
165
- }
166
- )
167
- .set_source_builder(
168
- FileBasedSourceBuilder()
169
- .set_stream_reader(TemporaryParquetFilesStreamReader(files=_single_parquet_file, file_type="parquet"))
170
- .set_file_type("parquet")
171
- )
172
- .set_expected_records(
173
- [
174
- {
175
- "data": {
176
- "col1": "val11",
177
- "col2": "val12",
178
- "_ab_source_file_last_modified": "2023-06-05T03:54:07.000000Z",
179
- "_ab_source_file_url": "a.parquet",
180
- },
181
- "stream": "stream1",
182
- },
183
- {
184
- "data": {
185
- "col1": "val21",
186
- "col2": "val22",
187
- "_ab_source_file_last_modified": "2023-06-05T03:54:07.000000Z",
188
- "_ab_source_file_url": "a.parquet",
189
- },
190
- "stream": "stream1",
191
- },
192
- ]
193
- )
194
- .set_expected_catalog(
195
- {
196
- "streams": [
197
- {
198
- "default_cursor_field": ["_ab_source_file_last_modified"],
199
- "json_schema": {
200
- "type": "object",
201
- "properties": {
202
- "col1": {"type": ["null", "string"]},
203
- "col2": {"type": ["null", "string"]},
204
- "_ab_source_file_last_modified": {"type": "string"},
205
- "_ab_source_file_url": {"type": "string"},
206
- },
207
- },
208
- "name": "stream1",
209
- "source_defined_cursor": True,
210
- "supported_sync_modes": ["full_refresh", "incremental"],
211
- }
212
- ]
213
- }
214
- )
215
- ).build()
216
-
217
- single_partitioned_parquet_scenario = (
218
- TestScenarioBuilder()
219
- .set_name("single_partitioned_parquet_stream")
220
- .set_config(
221
- {
222
- "streams": [
223
- {
224
- "name": "stream1",
225
- "format": {"filetype": "parquet"},
226
- "globs": ["path_prefix/**/*"],
227
- "validation_policy": "Emit Record",
228
- }
229
- ]
230
- }
231
- )
232
- .set_source_builder(
233
- FileBasedSourceBuilder()
234
- .set_stream_reader(TemporaryParquetFilesStreamReader(files=_single_partitioned_parquet_file, file_type="parquet"))
235
- .set_file_type("parquet")
236
- )
237
- .set_expected_records(
238
- [
239
- {
240
- "data": {
241
- "col1": "val11",
242
- "col2": "val12",
243
- "partition1": "1",
244
- "partition2": "2",
245
- "_ab_source_file_last_modified": "2023-06-05T03:54:07.000000Z",
246
- "_ab_source_file_url": "path_prefix/partition1=1/partition2=2/a.parquet",
247
- },
248
- "stream": "stream1",
249
- },
250
- {
251
- "data": {
252
- "col1": "val21",
253
- "col2": "val22",
254
- "partition1": "1",
255
- "partition2": "2",
256
- "_ab_source_file_last_modified": "2023-06-05T03:54:07.000000Z",
257
- "_ab_source_file_url": "path_prefix/partition1=1/partition2=2/a.parquet",
258
- },
259
- "stream": "stream1",
260
- },
261
- ]
262
- )
263
- .set_expected_catalog(
264
- {
265
- "streams": [
266
- {
267
- "default_cursor_field": ["_ab_source_file_last_modified"],
268
- "json_schema": {
269
- "type": "object",
270
- "properties": {
271
- "col1": {"type": ["null", "string"]},
272
- "col2": {"type": ["null", "string"]},
273
- "partition1": {"type": ["null", "string"]},
274
- "partition2": {"type": ["null", "string"]},
275
- "_ab_source_file_last_modified": {"type": "string"},
276
- "_ab_source_file_url": {"type": "string"},
277
- },
278
- },
279
- "name": "stream1",
280
- "source_defined_cursor": True,
281
- "supported_sync_modes": ["full_refresh", "incremental"],
282
- }
283
- ]
284
- }
285
- )
286
- ).build()
287
-
288
- multi_parquet_scenario = (
289
- TestScenarioBuilder()
290
- .set_name("multi_parquet_stream")
291
- .set_config(
292
- {
293
- "streams": [
294
- {
295
- "name": "stream1",
296
- "format": {"filetype": "parquet"},
297
- "globs": ["*"],
298
- "validation_policy": "Emit Record",
299
- }
300
- ]
301
- }
302
- )
303
- .set_source_builder(
304
- FileBasedSourceBuilder()
305
- .set_file_type("parquet")
306
- .set_stream_reader(TemporaryParquetFilesStreamReader(files=_multiple_parquet_file, file_type="parquet"))
307
- )
308
- .set_expected_catalog(
309
- {
310
- "streams": [
311
- {
312
- "default_cursor_field": ["_ab_source_file_last_modified"],
313
- "json_schema": {
314
- "type": "object",
315
- "properties": {
316
- "col1": {"type": ["null", "string"]},
317
- "col2": {"type": ["null", "string"]},
318
- "col3": {"type": ["null", "string"]},
319
- "_ab_source_file_last_modified": {"type": "string"},
320
- "_ab_source_file_url": {"type": "string"},
321
- },
322
- },
323
- "name": "stream1",
324
- "source_defined_cursor": True,
325
- "supported_sync_modes": ["full_refresh", "incremental"],
326
- }
327
- ]
328
- }
329
- )
330
- .set_expected_records(
331
- [
332
- {
333
- "data": {
334
- "col1": "val11a",
335
- "col2": "val12a",
336
- "_ab_source_file_last_modified": "2023-06-05T03:54:07.000000Z",
337
- "_ab_source_file_url": "a.parquet",
338
- },
339
- "stream": "stream1",
340
- },
341
- {
342
- "data": {
343
- "col1": "val21a",
344
- "col2": "val22a",
345
- "_ab_source_file_last_modified": "2023-06-05T03:54:07.000000Z",
346
- "_ab_source_file_url": "a.parquet",
347
- },
348
- "stream": "stream1",
349
- },
350
- {
351
- "data": {
352
- "col1": "val11b",
353
- "col2": "val12b",
354
- "col3": "val13b",
355
- "_ab_source_file_last_modified": "2023-06-05T03:54:07.000000Z",
356
- "_ab_source_file_url": "b.parquet",
357
- },
358
- "stream": "stream1",
359
- },
360
- {
361
- "data": {
362
- "col1": "val21b",
363
- "col2": "val22b",
364
- "col3": "val23b",
365
- "_ab_source_file_last_modified": "2023-06-05T03:54:07.000000Z",
366
- "_ab_source_file_url": "b.parquet",
367
- },
368
- "stream": "stream1",
369
- },
370
- ]
371
- )
372
- ).build()
373
-
374
- parquet_various_types_scenario = (
375
- TestScenarioBuilder()
376
- .set_name("parquet_various_types")
377
- .set_config(
378
- {
379
- "streams": [
380
- {
381
- "name": "stream1",
382
- "format": {"filetype": "parquet"},
383
- "globs": ["*"],
384
- "validation_policy": "Emit Record",
385
- }
386
- ]
387
- }
388
- )
389
- .set_source_builder(
390
- FileBasedSourceBuilder()
391
- .set_stream_reader(TemporaryParquetFilesStreamReader(files=_parquet_file_with_various_types, file_type="parquet"))
392
- .set_file_type("parquet")
393
- )
394
- .set_expected_catalog(
395
- {
396
- "streams": [
397
- {
398
- "default_cursor_field": ["_ab_source_file_last_modified"],
399
- "json_schema": {
400
- "type": "object",
401
- "properties": {
402
- "col_bool": {
403
- "type": ["null", "boolean"],
404
- },
405
- "col_int8": {
406
- "type": ["null", "integer"],
407
- },
408
- "col_int16": {
409
- "type": ["null", "integer"],
410
- },
411
- "col_int32": {
412
- "type": ["null", "integer"],
413
- },
414
- "col_uint8": {
415
- "type": ["null", "integer"],
416
- },
417
- "col_uint16": {
418
- "type": ["null", "integer"],
419
- },
420
- "col_uint32": {
421
- "type": ["null", "integer"],
422
- },
423
- "col_uint64": {
424
- "type": ["null", "integer"],
425
- },
426
- "col_float32": {
427
- "type": ["null", "number"],
428
- },
429
- "col_float64": {
430
- "type": ["null", "number"],
431
- },
432
- "col_string": {
433
- "type": ["null", "string"],
434
- },
435
- "col_date32": {"type": ["null", "string"], "format": "date"},
436
- "col_date64": {"type": ["null", "string"], "format": "date"},
437
- "col_timestamp_without_tz": {"type": ["null", "string"], "format": "date-time"},
438
- "col_timestamp_with_tz": {"type": ["null", "string"], "format": "date-time"},
439
- "col_time32s": {
440
- "type": ["null", "string"],
441
- },
442
- "col_time32ms": {
443
- "type": ["null", "string"],
444
- },
445
- "col_time64us": {
446
- "type": ["null", "string"],
447
- },
448
- "col_struct": {
449
- "type": ["null", "object"],
450
- },
451
- "col_list": {
452
- "type": ["null", "array"],
453
- },
454
- "col_duration": {
455
- "type": ["null", "integer"],
456
- },
457
- "col_binary": {
458
- "type": ["null", "string"],
459
- },
460
- "_ab_source_file_last_modified": {
461
- "type": "string",
462
- },
463
- "_ab_source_file_url": {
464
- "type": "string",
465
- },
466
- },
467
- },
468
- "name": "stream1",
469
- "source_defined_cursor": True,
470
- "supported_sync_modes": ["full_refresh", "incremental"],
471
- }
472
- ]
473
- }
474
- )
475
- .set_expected_records(
476
- [
477
- {
478
- "data": {
479
- "col_bool": True,
480
- "col_int8": -1,
481
- "col_int16": 1,
482
- "col_int32": 2,
483
- "col_uint8": 2,
484
- "col_uint16": 3,
485
- "col_uint32": 4,
486
- "col_uint64": 5,
487
- "col_float32": 3.14,
488
- "col_float64": 5.0,
489
- "col_string": "2020-01-01",
490
- "col_date32": "2021-01-01",
491
- "col_date64": "2022-01-01",
492
- "col_timestamp_without_tz": "2023-01-01T01:02:03",
493
- "col_timestamp_with_tz": "2024-03-04T05:06:07+00:00",
494
- "col_time32s": "01:02:03",
495
- "col_time32ms": "02:03:04",
496
- "col_time64us": "01:02:03.000004",
497
- "col_struct": {"struct_key": "struct_value"},
498
- "col_list": [1, 2, 3, 4],
499
- "col_duration": 12345,
500
- "col_binary": "binary string. Hello world!",
501
- "_ab_source_file_last_modified": "2023-06-05T03:54:07.000000Z",
502
- "_ab_source_file_url": "a.parquet",
503
- },
504
- "stream": "stream1",
505
- },
506
- ]
507
- )
508
- ).build()
509
-
510
- parquet_file_with_decimal_no_config_scenario = (
511
- TestScenarioBuilder()
512
- .set_name("parquet_file_with_decimal_no_config")
513
- .set_config(
514
- {
515
- "streams": [
516
- {
517
- "name": "stream1",
518
- "format": {"filetype": "parquet"},
519
- "globs": ["*"],
520
- "validation_policy": "Emit Record",
521
- }
522
- ]
523
- }
524
- )
525
- .set_source_builder(
526
- FileBasedSourceBuilder()
527
- .set_stream_reader(TemporaryParquetFilesStreamReader(files=_parquet_file_with_decimal, file_type="parquet"))
528
- .set_file_type("parquet")
529
- )
530
- .set_expected_records(
531
- [
532
- {
533
- "data": {
534
- "col1": "13.00",
535
- "_ab_source_file_last_modified": "2023-06-05T03:54:07.000000Z",
536
- "_ab_source_file_url": "a.parquet",
537
- },
538
- "stream": "stream1",
539
- },
540
- ]
541
- )
542
- .set_expected_catalog(
543
- {
544
- "streams": [
545
- {
546
- "default_cursor_field": ["_ab_source_file_last_modified"],
547
- "json_schema": {
548
- "type": "object",
549
- "properties": {
550
- "col1": {"type": ["null", "string"]},
551
- "_ab_source_file_last_modified": {"type": "string"},
552
- "_ab_source_file_url": {"type": "string"},
553
- },
554
- },
555
- "name": "stream1",
556
- "source_defined_cursor": True,
557
- "supported_sync_modes": ["full_refresh", "incremental"],
558
- }
559
- ]
560
- }
561
- )
562
- ).build()
563
-
564
- parquet_file_with_decimal_as_string_scenario = (
565
- TestScenarioBuilder()
566
- .set_name("parquet_file_with_decimal_as_string")
567
- .set_config(
568
- {
569
- "streams": [
570
- {
571
- "name": "stream1",
572
- "globs": ["*"],
573
- "validation_policy": "Emit Record",
574
- "format": {"filetype": "parquet", "decimal_as_float": False},
575
- }
576
- ]
577
- }
578
- )
579
- .set_source_builder(
580
- FileBasedSourceBuilder()
581
- .set_stream_reader(TemporaryParquetFilesStreamReader(files=_parquet_file_with_decimal, file_type="parquet"))
582
- .set_file_type("parquet")
583
- )
584
- .set_expected_records(
585
- [
586
- {
587
- "data": {
588
- "col1": "13.00",
589
- "_ab_source_file_last_modified": "2023-06-05T03:54:07.000000Z",
590
- "_ab_source_file_url": "a.parquet",
591
- },
592
- "stream": "stream1",
593
- },
594
- ]
595
- )
596
- .set_expected_catalog(
597
- {
598
- "streams": [
599
- {
600
- "default_cursor_field": ["_ab_source_file_last_modified"],
601
- "json_schema": {
602
- "type": "object",
603
- "properties": {
604
- "col1": {"type": ["null", "string"]},
605
- "_ab_source_file_last_modified": {"type": "string"},
606
- "_ab_source_file_url": {"type": "string"},
607
- },
608
- },
609
- "name": "stream1",
610
- "source_defined_cursor": True,
611
- "supported_sync_modes": ["full_refresh", "incremental"],
612
- }
613
- ]
614
- }
615
- )
616
- ).build()
617
-
618
- parquet_file_with_decimal_as_float_scenario = (
619
- TestScenarioBuilder()
620
- .set_name("parquet_file_with_decimal_as_float")
621
- .set_config(
622
- {
623
- "streams": [
624
- {
625
- "name": "stream1",
626
- "globs": ["*"],
627
- "validation_policy": "Emit Record",
628
- "format": {"filetype": "parquet", "decimal_as_float": True},
629
- }
630
- ]
631
- }
632
- )
633
- .set_source_builder(
634
- FileBasedSourceBuilder()
635
- .set_stream_reader(TemporaryParquetFilesStreamReader(files=_parquet_file_with_decimal, file_type="parquet"))
636
- .set_file_type("parquet")
637
- )
638
- .set_expected_records(
639
- [
640
- {
641
- "data": {"col1": 13.00, "_ab_source_file_last_modified": "2023-06-05T03:54:07.000000Z", "_ab_source_file_url": "a.parquet"},
642
- "stream": "stream1",
643
- },
644
- ]
645
- )
646
- .set_expected_catalog(
647
- {
648
- "streams": [
649
- {
650
- "default_cursor_field": ["_ab_source_file_last_modified"],
651
- "json_schema": {
652
- "type": "object",
653
- "properties": {
654
- "col1": {"type": ["null", "number"]},
655
- "_ab_source_file_last_modified": {"type": "string"},
656
- "_ab_source_file_url": {"type": "string"},
657
- },
658
- },
659
- "name": "stream1",
660
- "source_defined_cursor": True,
661
- "supported_sync_modes": ["full_refresh", "incremental"],
662
- }
663
- ]
664
- }
665
- )
666
- ).build()
667
-
668
- parquet_file_with_decimal_legacy_config_scenario = (
669
- TestScenarioBuilder()
670
- .set_name("parquet_file_with_decimal_legacy_config")
671
- .set_config(
672
- {
673
- "streams": [
674
- {
675
- "name": "stream1",
676
- "format": {
677
- "filetype": "parquet",
678
- },
679
- "globs": ["*"],
680
- "validation_policy": "emit_record",
681
- }
682
- ]
683
- }
684
- )
685
- .set_source_builder(
686
- FileBasedSourceBuilder()
687
- .set_stream_reader(TemporaryParquetFilesStreamReader(files=_parquet_file_with_decimal, file_type="parquet"))
688
- .set_file_type("parquet")
689
- )
690
- .set_expected_records(
691
- [
692
- {
693
- "data": {"col1": 13.00, "_ab_source_file_last_modified": "2023-06-05T03:54:07.000000Z", "_ab_source_file_url": "a.parquet"},
694
- "stream": "stream1",
695
- },
696
- ]
697
- )
698
- .set_expected_catalog(
699
- {
700
- "streams": [
701
- {
702
- "default_cursor_field": ["_ab_source_file_last_modified"],
703
- "json_schema": {
704
- "type": "object",
705
- "properties": {
706
- "col1": {"type": ["null", "number"]},
707
- "_ab_source_file_last_modified": {"type": "string"},
708
- "_ab_source_file_url": {"type": "string"},
709
- },
710
- },
711
- "name": "stream1",
712
- "source_defined_cursor": True,
713
- "supported_sync_modes": ["full_refresh", "incremental"],
714
- }
715
- ]
716
- }
717
- )
718
- ).build()
719
-
720
- parquet_with_invalid_config_scenario = (
721
- TestScenarioBuilder()
722
- .set_name("parquet_with_invalid_config")
723
- .set_config({"streams": [{"name": "stream1", "globs": ["*"], "validation_policy": "Emit Record", "format": {"filetype": "csv"}}]})
724
- .set_source_builder(
725
- FileBasedSourceBuilder()
726
- .set_stream_reader(TemporaryParquetFilesStreamReader(files=_single_parquet_file, file_type="parquet"))
727
- .set_file_type("parquet")
728
- )
729
- .set_expected_records([])
730
- .set_expected_logs({"read": [{"level": "ERROR", "message": "Error parsing record"}]})
731
- .set_expected_discover_error(AirbyteTracedException, "Error inferring schema from files")
732
- .set_expected_read_error(AirbyteTracedException, "Please check the logged errors for more information.")
733
- .set_expected_catalog(
734
- {
735
- "streams": [
736
- {
737
- "default_cursor_field": ["_ab_source_file_last_modified"],
738
- "json_schema": {
739
- "type": "object",
740
- "properties": {
741
- "col1": {"type": ["null", "string"]},
742
- "col2": {"type": ["null", "string"]},
743
- "_ab_source_file_last_modified": {"type": "string"},
744
- "_ab_source_file_url": {"type": "string"},
745
- },
746
- },
747
- "name": "stream1",
748
- "source_defined_cursor": True,
749
- "supported_sync_modes": ["full_refresh", "incremental"],
750
- }
751
- ]
752
- }
753
- )
754
- ).build()