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
@@ -1,635 +0,0 @@
1
- #
2
- # Copyright (c) 2023 Airbyte, Inc., all rights reserved.
3
- #
4
-
5
-
6
- import json
7
- from http import HTTPStatus
8
- from typing import Any, Iterable, Mapping, Optional
9
- from unittest.mock import ANY, MagicMock, patch
10
-
11
- import pytest
12
- import requests
13
- from airbyte_cdk.models import SyncMode
14
- from airbyte_cdk.sources.streams.http import HttpStream, HttpSubStream
15
- from airbyte_cdk.sources.streams.http.auth import NoAuth
16
- from airbyte_cdk.sources.streams.http.auth import TokenAuthenticator as HttpTokenAuthenticator
17
- from airbyte_cdk.sources.streams.http.exceptions import DefaultBackoffException, RequestBodyException, UserDefinedBackoffException
18
- from airbyte_cdk.sources.streams.http.requests_native_auth import TokenAuthenticator
19
-
20
-
21
- class StubBasicReadHttpStream(HttpStream):
22
- url_base = "https://test_base_url.com"
23
- primary_key = ""
24
-
25
- def __init__(self, deduplicate_query_params: bool = False, **kwargs):
26
- super().__init__(**kwargs)
27
- self.resp_counter = 1
28
- self._deduplicate_query_params = deduplicate_query_params
29
-
30
- def next_page_token(self, response: requests.Response) -> Optional[Mapping[str, Any]]:
31
- return None
32
-
33
- def path(self, **kwargs) -> str:
34
- return ""
35
-
36
- def parse_response(self, response: requests.Response, **kwargs) -> Iterable[Mapping]:
37
- stubResp = {"data": self.resp_counter}
38
- self.resp_counter += 1
39
- yield stubResp
40
-
41
- def must_deduplicate_query_params(self) -> bool:
42
- return self._deduplicate_query_params
43
-
44
-
45
- def test_default_authenticator():
46
- stream = StubBasicReadHttpStream()
47
- assert isinstance(stream.authenticator, NoAuth)
48
- assert stream._session.auth is None
49
-
50
-
51
- def test_requests_native_token_authenticator():
52
- stream = StubBasicReadHttpStream(authenticator=TokenAuthenticator("test-token"))
53
- assert isinstance(stream.authenticator, NoAuth)
54
- assert isinstance(stream._session.auth, TokenAuthenticator)
55
-
56
-
57
- def test_http_token_authenticator():
58
- stream = StubBasicReadHttpStream(authenticator=HttpTokenAuthenticator("test-token"))
59
- assert isinstance(stream.authenticator, HttpTokenAuthenticator)
60
- assert stream._session.auth is None
61
-
62
-
63
- def test_request_kwargs_used(mocker, requests_mock):
64
- stream = StubBasicReadHttpStream()
65
- request_kwargs = {"cert": None, "proxies": "google.com"}
66
- mocker.patch.object(stream, "request_kwargs", return_value=request_kwargs)
67
- send_mock = mocker.patch.object(stream._session, "send", wraps=stream._session.send)
68
- requests_mock.register_uri("GET", stream.url_base)
69
-
70
- list(stream.read_records(sync_mode=SyncMode.full_refresh))
71
-
72
- stream._session.send.assert_any_call(ANY, **request_kwargs)
73
- assert send_mock.call_count == 1
74
-
75
-
76
- def test_stub_basic_read_http_stream_read_records(mocker):
77
- stream = StubBasicReadHttpStream()
78
- blank_response = {} # Send a blank response is fine as we ignore the response in `parse_response anyway.
79
- mocker.patch.object(StubBasicReadHttpStream, "_send_request", return_value=blank_response)
80
-
81
- records = list(stream.read_records(SyncMode.full_refresh))
82
-
83
- assert [{"data": 1}] == records
84
-
85
-
86
- class StubNextPageTokenHttpStream(StubBasicReadHttpStream):
87
- current_page = 0
88
-
89
- def __init__(self, pages: int = 5):
90
- super().__init__()
91
- self._pages = pages
92
-
93
- def next_page_token(self, response: requests.Response) -> Optional[Mapping[str, Any]]:
94
- while self.current_page < self._pages:
95
- page_token = {"page": self.current_page}
96
- self.current_page += 1
97
- return page_token
98
- return None
99
-
100
-
101
- def test_next_page_token_is_input_to_other_methods(mocker):
102
- """Validates that the return value from next_page_token is passed into other methods that need it like request_params, headers, body, etc.."""
103
- pages = 5
104
- stream = StubNextPageTokenHttpStream(pages=pages)
105
- blank_response = {} # Send a blank response is fine as we ignore the response in `parse_response anyway.
106
- mocker.patch.object(StubNextPageTokenHttpStream, "_send_request", return_value=blank_response)
107
-
108
- methods = ["request_params", "request_headers", "request_body_json"]
109
- for method in methods:
110
- # Wrap all methods we're interested in testing with mocked objects so we can later spy on their input args and verify they were what we expect
111
- mocker.patch.object(stream, method, wraps=getattr(stream, method))
112
-
113
- records = list(stream.read_records(SyncMode.full_refresh))
114
-
115
- # Since we have 5 pages, we expect 5 tokens which are {"page":1}, {"page":2}, etc...
116
- expected_next_page_tokens = [{"page": i} for i in range(pages)]
117
- for method in methods:
118
- # First assert that they were called with no next_page_token. This is the first call in the pagination loop.
119
- getattr(stream, method).assert_any_call(next_page_token=None, stream_slice=None, stream_state={})
120
- for token in expected_next_page_tokens:
121
- # Then verify that each method
122
- getattr(stream, method).assert_any_call(next_page_token=token, stream_slice=None, stream_state={})
123
-
124
- expected = [{"data": 1}, {"data": 2}, {"data": 3}, {"data": 4}, {"data": 5}, {"data": 6}]
125
-
126
- assert expected == records
127
-
128
-
129
- class StubBadUrlHttpStream(StubBasicReadHttpStream):
130
- url_base = "bad_url"
131
-
132
-
133
- def test_stub_bad_url_http_stream_read_records(mocker):
134
- stream = StubBadUrlHttpStream()
135
-
136
- with pytest.raises(requests.exceptions.RequestException):
137
- list(stream.read_records(SyncMode.full_refresh))
138
-
139
-
140
- class StubCustomBackoffHttpStream(StubBasicReadHttpStream):
141
- def backoff_time(self, response: requests.Response) -> Optional[float]:
142
- return 0.5
143
-
144
-
145
- def test_stub_custom_backoff_http_stream(mocker):
146
- mocker.patch("time.sleep", lambda x: None)
147
- stream = StubCustomBackoffHttpStream()
148
- req = requests.Response()
149
- req.status_code = 429
150
-
151
- send_mock = mocker.patch.object(requests.Session, "send", return_value=req)
152
-
153
- with pytest.raises(UserDefinedBackoffException):
154
- list(stream.read_records(SyncMode.full_refresh))
155
- assert send_mock.call_count == stream.max_retries + 1
156
-
157
- # TODO(davin): Figure out how to assert calls.
158
-
159
-
160
- @pytest.mark.parametrize("retries", [-20, -1, 0, 1, 2, 10])
161
- def test_stub_custom_backoff_http_stream_retries(mocker, retries):
162
- mocker.patch("time.sleep", lambda x: None)
163
-
164
- class StubCustomBackoffHttpStreamRetries(StubCustomBackoffHttpStream):
165
- @property
166
- def max_retries(self):
167
- return retries
168
-
169
- stream = StubCustomBackoffHttpStreamRetries()
170
- req = requests.Response()
171
- req.status_code = HTTPStatus.TOO_MANY_REQUESTS
172
- send_mock = mocker.patch.object(requests.Session, "send", return_value=req)
173
-
174
- with pytest.raises(UserDefinedBackoffException, match="Request URL: https://test_base_url.com/, Response Code: 429") as excinfo:
175
- list(stream.read_records(SyncMode.full_refresh))
176
- assert isinstance(excinfo.value.request, requests.PreparedRequest)
177
- assert isinstance(excinfo.value.response, requests.Response)
178
- if retries <= 0:
179
- assert send_mock.call_count == 1
180
- else:
181
- assert send_mock.call_count == stream.max_retries + 1
182
-
183
-
184
- def test_stub_custom_backoff_http_stream_endless_retries(mocker):
185
- mocker.patch("time.sleep", lambda x: None)
186
-
187
- class StubCustomBackoffHttpStreamRetries(StubCustomBackoffHttpStream):
188
- @property
189
- def max_retries(self):
190
- return None
191
-
192
- infinite_number = 20
193
-
194
- stream = StubCustomBackoffHttpStreamRetries()
195
- req = requests.Response()
196
- req.status_code = HTTPStatus.TOO_MANY_REQUESTS
197
- send_mock = mocker.patch.object(requests.Session, "send", side_effect=[req] * infinite_number)
198
-
199
- # Expecting mock object to raise a RuntimeError when the end of side_effect list parameter reached.
200
- with pytest.raises(RuntimeError):
201
- list(stream.read_records(SyncMode.full_refresh))
202
- assert send_mock.call_count == infinite_number + 1
203
-
204
-
205
- @pytest.mark.parametrize("http_code", [400, 401, 403])
206
- def test_4xx_error_codes_http_stream(mocker, http_code):
207
- stream = StubCustomBackoffHttpStream()
208
- req = requests.Response()
209
- req.status_code = http_code
210
- mocker.patch.object(requests.Session, "send", return_value=req)
211
-
212
- with pytest.raises(requests.exceptions.HTTPError):
213
- list(stream.read_records(SyncMode.full_refresh))
214
-
215
-
216
- class AutoFailFalseHttpStream(StubBasicReadHttpStream):
217
- raise_on_http_errors = False
218
- max_retries = 3
219
- retry_factor = 0.01
220
-
221
-
222
- def test_raise_on_http_errors_off_429(mocker):
223
- stream = AutoFailFalseHttpStream()
224
- req = requests.Response()
225
- req.status_code = 429
226
-
227
- mocker.patch.object(requests.Session, "send", return_value=req)
228
- with pytest.raises(DefaultBackoffException, match="Request URL: https://test_base_url.com/, Response Code: 429"):
229
- list(stream.read_records(SyncMode.full_refresh))
230
-
231
-
232
- @pytest.mark.parametrize("status_code", [500, 501, 503, 504])
233
- def test_raise_on_http_errors_off_5xx(mocker, status_code):
234
- stream = AutoFailFalseHttpStream()
235
- req = requests.Response()
236
- req.status_code = status_code
237
-
238
- send_mock = mocker.patch.object(requests.Session, "send", return_value=req)
239
- with pytest.raises(DefaultBackoffException):
240
- list(stream.read_records(SyncMode.full_refresh))
241
- assert send_mock.call_count == stream.max_retries + 1
242
-
243
-
244
- @pytest.mark.parametrize("status_code", [400, 401, 402, 403, 416])
245
- def test_raise_on_http_errors_off_non_retryable_4xx(mocker, status_code):
246
- stream = AutoFailFalseHttpStream()
247
- req = requests.PreparedRequest()
248
- res = requests.Response()
249
- res.status_code = status_code
250
-
251
- mocker.patch.object(requests.Session, "send", return_value=res)
252
- response = stream._send_request(req, {})
253
- assert response.status_code == status_code
254
-
255
-
256
- @pytest.mark.parametrize(
257
- "error",
258
- (
259
- requests.exceptions.ConnectTimeout,
260
- requests.exceptions.ConnectionError,
261
- requests.exceptions.ChunkedEncodingError,
262
- requests.exceptions.ReadTimeout,
263
- ),
264
- )
265
- def test_raise_on_http_errors(mocker, error):
266
- stream = AutoFailFalseHttpStream()
267
- send_mock = mocker.patch.object(requests.Session, "send", side_effect=error())
268
-
269
- with pytest.raises(error):
270
- list(stream.read_records(SyncMode.full_refresh))
271
- assert send_mock.call_count == stream.max_retries + 1
272
-
273
-
274
- class PostHttpStream(StubBasicReadHttpStream):
275
- http_method = "POST"
276
-
277
- def parse_response(self, response: requests.Response, **kwargs) -> Iterable[Mapping]:
278
- """Returns response data as is"""
279
- yield response.json()
280
-
281
-
282
- class TestRequestBody:
283
- """Suite of different tests for request bodies"""
284
-
285
- json_body = {"key": "value"}
286
- data_body = "key:value"
287
- form_body = {"key1": "value1", "key2": 1234}
288
- urlencoded_form_body = "key1=value1&key2=1234"
289
-
290
- def request2response(self, request, context):
291
- return json.dumps({"body": request.text, "content_type": request.headers.get("Content-Type")})
292
-
293
- def test_json_body(self, mocker, requests_mock):
294
-
295
- stream = PostHttpStream()
296
- mocker.patch.object(stream, "request_body_json", return_value=self.json_body)
297
-
298
- requests_mock.register_uri("POST", stream.url_base, text=self.request2response)
299
- response = list(stream.read_records(sync_mode=SyncMode.full_refresh))[0]
300
-
301
- assert response["content_type"] == "application/json"
302
- assert json.loads(response["body"]) == self.json_body
303
-
304
- def test_text_body(self, mocker, requests_mock):
305
-
306
- stream = PostHttpStream()
307
- mocker.patch.object(stream, "request_body_data", return_value=self.data_body)
308
-
309
- requests_mock.register_uri("POST", stream.url_base, text=self.request2response)
310
- response = list(stream.read_records(sync_mode=SyncMode.full_refresh))[0]
311
-
312
- assert response["content_type"] is None
313
- assert response["body"] == self.data_body
314
-
315
- def test_form_body(self, mocker, requests_mock):
316
-
317
- stream = PostHttpStream()
318
- mocker.patch.object(stream, "request_body_data", return_value=self.form_body)
319
-
320
- requests_mock.register_uri("POST", stream.url_base, text=self.request2response)
321
- response = list(stream.read_records(sync_mode=SyncMode.full_refresh))[0]
322
-
323
- assert response["content_type"] == "application/x-www-form-urlencoded"
324
- assert response["body"] == self.urlencoded_form_body
325
-
326
- def test_text_json_body(self, mocker, requests_mock):
327
- """checks a exception if both functions were overridden"""
328
- stream = PostHttpStream()
329
- mocker.patch.object(stream, "request_body_data", return_value=self.data_body)
330
- mocker.patch.object(stream, "request_body_json", return_value=self.json_body)
331
- requests_mock.register_uri("POST", stream.url_base, text=self.request2response)
332
- with pytest.raises(RequestBodyException):
333
- list(stream.read_records(sync_mode=SyncMode.full_refresh))
334
-
335
- def test_body_for_all_methods(self, mocker, requests_mock):
336
- """Stream must send a body for GET/POST/PATCH/PUT methods only"""
337
- stream = PostHttpStream()
338
- methods = {
339
- "POST": True,
340
- "PUT": True,
341
- "PATCH": True,
342
- "GET": True,
343
- "DELETE": False,
344
- "OPTIONS": False,
345
- }
346
- for method, with_body in methods.items():
347
- stream.http_method = method
348
- mocker.patch.object(stream, "request_body_data", return_value=self.data_body)
349
- requests_mock.register_uri(method, stream.url_base, text=self.request2response)
350
- response = list(stream.read_records(sync_mode=SyncMode.full_refresh))[0]
351
- if with_body:
352
- assert response["body"] == self.data_body
353
- else:
354
- assert response["body"] is None
355
-
356
-
357
- class CacheHttpStream(StubBasicReadHttpStream):
358
- use_cache = True
359
-
360
-
361
- class CacheHttpSubStream(HttpSubStream):
362
- url_base = "https://example.com"
363
- primary_key = ""
364
-
365
- def __init__(self, parent):
366
- super().__init__(parent=parent)
367
-
368
- def parse_response(self, response: requests.Response, **kwargs) -> Iterable[Mapping]:
369
- return []
370
-
371
- def next_page_token(self, response: requests.Response) -> Optional[Mapping[str, Any]]:
372
- return None
373
-
374
- def path(self, **kwargs) -> str:
375
- return ""
376
-
377
-
378
- def test_caching_filename():
379
- stream = CacheHttpStream()
380
- assert stream.cache_filename == f"{stream.name}.sqlite"
381
-
382
-
383
- def test_caching_sessions_are_different():
384
- stream_1 = CacheHttpStream()
385
- stream_2 = CacheHttpStream()
386
-
387
- assert stream_1._session != stream_2._session
388
- assert stream_1.cache_filename == stream_2.cache_filename
389
-
390
-
391
- # def test_cached_streams_wortk_when_request_path_is_not_set(mocker, requests_mock):
392
- # This test verifies that HttpStreams with a cached session work even if the path is not set
393
- # For instance, when running in a unit test
394
- # stream = CacheHttpStream()
395
- # with mocker.patch.object(stream._session, "send", wraps=stream._session.send):
396
- # requests_mock.register_uri("GET", stream.url_base)
397
- # records = list(stream.read_records(sync_mode=SyncMode.full_refresh))
398
- # assert records == [{"data": 1}]
399
- # ""
400
-
401
-
402
- def test_parent_attribute_exist():
403
- parent_stream = CacheHttpStream()
404
- child_stream = CacheHttpSubStream(parent=parent_stream)
405
-
406
- assert child_stream.parent == parent_stream
407
-
408
-
409
- def test_that_response_was_cached(mocker, requests_mock):
410
- requests_mock.register_uri("GET", "https://google.com/", text="text")
411
- stream = CacheHttpStream()
412
- stream.clear_cache()
413
- mocker.patch.object(stream, "url_base", "https://google.com/")
414
- records = list(stream.read_records(sync_mode=SyncMode.full_refresh))
415
-
416
- assert requests_mock.called
417
-
418
- requests_mock.reset_mock()
419
- new_records = list(stream.read_records(sync_mode=SyncMode.full_refresh))
420
-
421
- assert len(records) == len(new_records)
422
- assert not requests_mock.called
423
-
424
-
425
- class CacheHttpStreamWithSlices(CacheHttpStream):
426
- paths = ["", "search"]
427
-
428
- def path(self, stream_slice: Mapping[str, Any] = None, **kwargs) -> str:
429
- return f'{stream_slice["path"]}' if stream_slice else ""
430
-
431
- def stream_slices(self, **kwargs) -> Iterable[Optional[Mapping[str, Any]]]:
432
- for path in self.paths:
433
- yield {"path": path}
434
-
435
- def parse_response(self, response: requests.Response, **kwargs) -> Iterable[Mapping]:
436
- yield {"value": len(response.text)}
437
-
438
-
439
- @patch("airbyte_cdk.sources.streams.core.logging", MagicMock())
440
- def test_using_cache(mocker, requests_mock):
441
- requests_mock.register_uri("GET", "https://google.com/", text="text")
442
- requests_mock.register_uri("GET", "https://google.com/search", text="text")
443
-
444
- parent_stream = CacheHttpStreamWithSlices()
445
- mocker.patch.object(parent_stream, "url_base", "https://google.com/")
446
- parent_stream.clear_cache()
447
-
448
- assert requests_mock.call_count == 0
449
- assert len(parent_stream._session.cache.responses) == 0
450
-
451
- for _slice in parent_stream.stream_slices():
452
- list(parent_stream.read_records(sync_mode=SyncMode.full_refresh, stream_slice=_slice))
453
-
454
- assert requests_mock.call_count == 2
455
- assert len(parent_stream._session.cache.responses) == 2
456
-
457
- child_stream = CacheHttpSubStream(parent=parent_stream)
458
-
459
- for _slice in child_stream.stream_slices(sync_mode=SyncMode.full_refresh):
460
- pass
461
-
462
- assert requests_mock.call_count == 2
463
- assert len(parent_stream._session.cache.responses) == 2
464
- assert parent_stream._session.cache.contains(url="https://google.com/")
465
- assert parent_stream._session.cache.contains(url="https://google.com/search")
466
-
467
-
468
- class AutoFailTrueHttpStream(StubBasicReadHttpStream):
469
- raise_on_http_errors = True
470
-
471
-
472
- @pytest.mark.parametrize("status_code", range(400, 600))
473
- def test_send_raise_on_http_errors_logs(mocker, status_code):
474
- mocker.patch.object(AutoFailTrueHttpStream, "logger")
475
- mocker.patch.object(AutoFailTrueHttpStream, "should_retry", mocker.Mock(return_value=False))
476
- stream = AutoFailTrueHttpStream()
477
- req = requests.PreparedRequest()
478
- res = requests.Response()
479
- res.status_code = status_code
480
- mocker.patch.object(requests.Session, "send", return_value=res)
481
- with pytest.raises(requests.exceptions.HTTPError):
482
- response = stream._send_request(req, {})
483
- stream.logger.error.assert_called_with(response.text)
484
- assert response.status_code == status_code
485
-
486
-
487
- @pytest.mark.parametrize(
488
- "api_response, expected_message",
489
- [
490
- ({"error": "something broke"}, "something broke"),
491
- ({"error": {"message": "something broke"}}, "something broke"),
492
- ({"error": "err-001", "message": "something broke"}, "something broke"),
493
- ({"failure": {"message": "something broke"}}, "something broke"),
494
- ({"error": {"errors": [{"message": "one"}, {"message": "two"}, {"message": "three"}]}}, "one, two, three"),
495
- ({"errors": ["one", "two", "three"]}, "one, two, three"),
496
- ({"messages": ["one", "two", "three"]}, "one, two, three"),
497
- ({"errors": [{"message": "one"}, {"message": "two"}, {"message": "three"}]}, "one, two, three"),
498
- ({"error": [{"message": "one"}, {"message": "two"}, {"message": "three"}]}, "one, two, three"),
499
- ({"errors": [{"error": "one"}, {"error": "two"}, {"error": "three"}]}, "one, two, three"),
500
- ({"failures": [{"message": "one"}, {"message": "two"}, {"message": "three"}]}, "one, two, three"),
501
- (["one", "two", "three"], "one, two, three"),
502
- ([{"error": "one"}, {"error": "two"}, {"error": "three"}], "one, two, three"),
503
- ({"error": True}, None),
504
- ({"something_else": "hi"}, None),
505
- ({}, None),
506
- ],
507
- )
508
- def test_default_parse_response_error_message(api_response: dict, expected_message: Optional[str]):
509
- stream = StubBasicReadHttpStream()
510
- response = MagicMock()
511
- response.json.return_value = api_response
512
-
513
- message = stream.parse_response_error_message(response)
514
- assert message == expected_message
515
-
516
-
517
- def test_default_parse_response_error_message_not_json(requests_mock):
518
- stream = StubBasicReadHttpStream()
519
- requests_mock.register_uri("GET", "mock://test.com/not_json", text="this is not json")
520
- response = requests.get("mock://test.com/not_json")
521
-
522
- message = stream.parse_response_error_message(response)
523
- assert message is None
524
-
525
-
526
- def test_default_get_error_display_message_handles_http_error(mocker):
527
- stream = StubBasicReadHttpStream()
528
- mocker.patch.object(stream, "parse_response_error_message", return_value="my custom message")
529
-
530
- non_http_err_msg = stream.get_error_display_message(RuntimeError("not me"))
531
- assert non_http_err_msg is None
532
-
533
- response = requests.Response()
534
- http_exception = requests.HTTPError(response=response)
535
- http_err_msg = stream.get_error_display_message(http_exception)
536
- assert http_err_msg == "my custom message"
537
-
538
-
539
- @pytest.mark.parametrize(
540
- "test_name, base_url, path, expected_full_url",
541
- [
542
- ("test_no_slashes", "https://airbyte.io", "my_endpoint", "https://airbyte.io/my_endpoint"),
543
- ("test_trailing_slash_on_base_url", "https://airbyte.io/", "my_endpoint", "https://airbyte.io/my_endpoint"),
544
- (
545
- "test_trailing_slash_on_base_url_and_leading_slash_on_path",
546
- "https://airbyte.io/",
547
- "/my_endpoint",
548
- "https://airbyte.io/my_endpoint",
549
- ),
550
- ("test_leading_slash_on_path", "https://airbyte.io", "/my_endpoint", "https://airbyte.io/my_endpoint"),
551
- ("test_trailing_slash_on_path", "https://airbyte.io", "/my_endpoint/", "https://airbyte.io/my_endpoint/"),
552
- ("test_nested_path_no_leading_slash", "https://airbyte.io", "v1/my_endpoint", "https://airbyte.io/v1/my_endpoint"),
553
- ("test_nested_path_with_leading_slash", "https://airbyte.io", "/v1/my_endpoint", "https://airbyte.io/v1/my_endpoint"),
554
- ],
555
- )
556
- def test_join_url(test_name, base_url, path, expected_full_url):
557
- actual_url = HttpStream._join_url(base_url, path)
558
- assert actual_url == expected_full_url
559
-
560
-
561
- @pytest.mark.parametrize(
562
- "deduplicate_query_params, path, params, expected_url",
563
- [
564
- pytest.param(
565
- True, "v1/endpoint?param1=value1", {}, "https://test_base_url.com/v1/endpoint?param1=value1", id="test_params_only_in_path"
566
- ),
567
- pytest.param(
568
- True, "v1/endpoint", {"param1": "value1"}, "https://test_base_url.com/v1/endpoint?param1=value1", id="test_params_only_in_path"
569
- ),
570
- pytest.param(True, "v1/endpoint", None, "https://test_base_url.com/v1/endpoint", id="test_params_is_none_and_no_params_in_path"),
571
- pytest.param(
572
- True,
573
- "v1/endpoint?param1=value1",
574
- None,
575
- "https://test_base_url.com/v1/endpoint?param1=value1",
576
- id="test_params_is_none_and_no_params_in_path",
577
- ),
578
- pytest.param(
579
- True,
580
- "v1/endpoint?param1=value1",
581
- {"param2": "value2"},
582
- "https://test_base_url.com/v1/endpoint?param1=value1&param2=value2",
583
- id="test_no_duplicate_params",
584
- ),
585
- pytest.param(
586
- True,
587
- "v1/endpoint?param1=value1",
588
- {"param1": "value1"},
589
- "https://test_base_url.com/v1/endpoint?param1=value1",
590
- id="test_duplicate_params_same_value",
591
- ),
592
- pytest.param(
593
- True,
594
- "v1/endpoint?param1=1",
595
- {"param1": 1},
596
- "https://test_base_url.com/v1/endpoint?param1=1",
597
- id="test_duplicate_params_same_value_not_string",
598
- ),
599
- pytest.param(
600
- True,
601
- "v1/endpoint?param1=value1",
602
- {"param1": "value2"},
603
- "https://test_base_url.com/v1/endpoint?param1=value1&param1=value2",
604
- id="test_duplicate_params_different_value",
605
- ),
606
- pytest.param(
607
- False,
608
- "v1/endpoint?param1=value1",
609
- {"param1": "value2"},
610
- "https://test_base_url.com/v1/endpoint?param1=value1&param1=value2",
611
- id="test_same_params_different_value_no_deduplication",
612
- ),
613
- pytest.param(
614
- False,
615
- "v1/endpoint?param1=value1",
616
- {"param1": "value1"},
617
- "https://test_base_url.com/v1/endpoint?param1=value1&param1=value1",
618
- id="test_same_params_same_value_no_deduplication",
619
- ),
620
- ],
621
- )
622
- def test_duplicate_request_params_are_deduped(deduplicate_query_params, path, params, expected_url):
623
- stream = StubBasicReadHttpStream(deduplicate_query_params)
624
-
625
- if expected_url is None:
626
- with pytest.raises(ValueError):
627
- stream._create_prepared_request(path=path, params=params)
628
- else:
629
- prepared_request = stream._create_prepared_request(path=path, params=params)
630
- assert prepared_request.url == expected_url
631
-
632
-
633
- def test_connection_pool():
634
- stream = StubBasicReadHttpStream(authenticator=HttpTokenAuthenticator("test-token"))
635
- assert stream._session.adapters["https://"]._pool_connections == 20