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
@@ -1,713 +0,0 @@
1
- #
2
- # Copyright (c) 2023 Airbyte, Inc., all rights reserved.
3
- #
4
-
5
- import json
6
- from typing import Any, Iterator, List, Mapping
7
- from unittest.mock import MagicMock, Mock, patch
8
-
9
- import pytest
10
- from airbyte_cdk.connector_builder.message_grouper import MessageGrouper
11
- from airbyte_cdk.connector_builder.models import HttpRequest, HttpResponse, LogMessage, StreamRead, StreamReadPages
12
- from airbyte_cdk.models import (
13
- AirbyteControlConnectorConfigMessage,
14
- AirbyteControlMessage,
15
- AirbyteLogMessage,
16
- AirbyteMessage,
17
- AirbyteRecordMessage,
18
- Level,
19
- OrchestratorType,
20
- )
21
- from airbyte_cdk.models import Type as MessageType
22
- from unit_tests.connector_builder.utils import create_configured_catalog
23
-
24
- MAX_PAGES_PER_SLICE = 4
25
- MAX_SLICES = 3
26
-
27
- MANIFEST = {
28
- "version": "0.30.0",
29
- "type": "DeclarativeSource",
30
- "definitions": {
31
- "selector": {"extractor": {"field_path": ["items"], "type": "DpathExtractor"}, "type": "RecordSelector"},
32
- "requester": {"url_base": "https://demonslayers.com/api/v1/", "http_method": "GET", "type": "DeclarativeSource"},
33
- "retriever": {
34
- "type": "DeclarativeSource",
35
- "record_selector": {"extractor": {"field_path": ["items"], "type": "DpathExtractor"}, "type": "RecordSelector"},
36
- "paginator": {"type": "NoPagination"},
37
- "requester": {"url_base": "https://demonslayers.com/api/v1/", "http_method": "GET", "type": "HttpRequester"},
38
- },
39
- "hashiras_stream": {
40
- "retriever": {
41
- "type": "DeclarativeSource",
42
- "record_selector": {"extractor": {"field_path": ["items"], "type": "DpathExtractor"}, "type": "RecordSelector"},
43
- "paginator": {"type": "NoPagination"},
44
- "requester": {"url_base": "https://demonslayers.com/api/v1/", "http_method": "GET", "type": "HttpRequester"},
45
- },
46
- "$parameters": {"name": "hashiras", "path": "/hashiras"},
47
- },
48
- "breathing_techniques_stream": {
49
- "retriever": {
50
- "type": "DeclarativeSource",
51
- "record_selector": {"extractor": {"field_path": ["items"], "type": "DpathExtractor"}, "type": "RecordSelector"},
52
- "paginator": {"type": "NoPagination"},
53
- "requester": {"url_base": "https://demonslayers.com/api/v1/", "http_method": "GET", "type": "HttpRequester"},
54
- },
55
- "$parameters": {"name": "breathing-techniques", "path": "/breathing_techniques"},
56
- },
57
- },
58
- "streams": [
59
- {
60
- "type": "DeclarativeStream",
61
- "retriever": {
62
- "type": "SimpleRetriever",
63
- "record_selector": {"extractor": {"field_path": ["items"], "type": "DpathExtractor"}, "type": "RecordSelector"},
64
- "paginator": {"type": "NoPagination"},
65
- "requester": {"url_base": "https://demonslayers.com/api/v1/", "http_method": "GET", "type": "HttpRequester"},
66
- },
67
- "$parameters": {"name": "hashiras", "path": "/hashiras"},
68
- },
69
- {
70
- "type": "DeclarativeStream",
71
- "retriever": {
72
- "type": "SimpleRetriever",
73
- "record_selector": {"extractor": {"field_path": ["items"], "type": "DpathExtractor"}, "type": "RecordSelector"},
74
- "paginator": {"type": "NoPagination"},
75
- "requester": {"url_base": "https://demonslayers.com/api/v1/", "http_method": "GET", "type": "HttpRequester"},
76
- },
77
- "$parameters": {"name": "breathing-techniques", "path": "/breathing_techniques"},
78
- },
79
- ],
80
- "check": {"stream_names": ["hashiras"], "type": "CheckStream"},
81
- }
82
-
83
- CONFIG = {"rank": "upper-six"}
84
-
85
- A_SOURCE = MagicMock()
86
-
87
-
88
- @patch("airbyte_cdk.connector_builder.message_grouper.AirbyteEntrypoint.read")
89
- def test_get_grouped_messages(mock_entrypoint_read: Mock) -> None:
90
- url = "https://demonslayers.com/api/v1/hashiras?era=taisho"
91
- request = {
92
- "headers": {"Content-Type": "application/json"},
93
- "method": "GET",
94
- "body": {"content": '{"custom": "field"}'},
95
- }
96
- response = {"status_code": 200, "headers": {"field": "value"}, "body": {"content": '{"name": "field"}'}}
97
- expected_schema = {
98
- "$schema": "http://json-schema.org/schema#",
99
- "properties": {"name": {"type": "string"}, "date": {"type": "string"}},
100
- "type": "object",
101
- }
102
- expected_datetime_fields = {"date": "%Y-%m-%d"}
103
- expected_pages = [
104
- StreamReadPages(
105
- request=HttpRequest(
106
- url="https://demonslayers.com/api/v1/hashiras",
107
- parameters={"era": ["taisho"]},
108
- headers={"Content-Type": "application/json"},
109
- body='{"custom": "field"}',
110
- http_method="GET",
111
- ),
112
- response=HttpResponse(status=200, headers={"field": "value"}, body='{"name": "field"}'),
113
- records=[{"name": "Shinobu Kocho", "date": "2023-03-03"}, {"name": "Muichiro Tokito", "date": "2023-03-04"}],
114
- ),
115
- StreamReadPages(
116
- request=HttpRequest(
117
- url="https://demonslayers.com/api/v1/hashiras",
118
- parameters={"era": ["taisho"]},
119
- headers={"Content-Type": "application/json"},
120
- body='{"custom": "field"}',
121
- http_method="GET",
122
- ),
123
- response=HttpResponse(status=200, headers={"field": "value"}, body='{"name": "field"}'),
124
- records=[{"name": "Mitsuri Kanroji", "date": "2023-03-05"}],
125
- ),
126
- ]
127
-
128
- mock_source = make_mock_source(
129
- mock_entrypoint_read,
130
- iter(
131
- [
132
- request_response_log_message(request, response, url),
133
- record_message("hashiras", {"name": "Shinobu Kocho", "date": "2023-03-03"}),
134
- record_message("hashiras", {"name": "Muichiro Tokito", "date": "2023-03-04"}),
135
- request_response_log_message(request, response, url),
136
- record_message("hashiras", {"name": "Mitsuri Kanroji", "date": "2023-03-05"}),
137
- ]
138
- ),
139
- )
140
-
141
- connector_builder_handler = MessageGrouper(MAX_PAGES_PER_SLICE, MAX_SLICES)
142
- actual_response: StreamRead = connector_builder_handler.get_message_groups(
143
- source=mock_source, config=CONFIG, configured_catalog=create_configured_catalog("hashiras")
144
- )
145
-
146
- assert actual_response.inferred_schema == expected_schema
147
- assert actual_response.inferred_datetime_formats == expected_datetime_fields
148
-
149
- single_slice = actual_response.slices[0]
150
- for i, actual_page in enumerate(single_slice.pages):
151
- assert actual_page == expected_pages[i]
152
-
153
-
154
- @patch("airbyte_cdk.connector_builder.message_grouper.AirbyteEntrypoint.read")
155
- def test_get_grouped_messages_with_logs(mock_entrypoint_read: Mock) -> None:
156
- url = "https://demonslayers.com/api/v1/hashiras?era=taisho"
157
- request = {
158
- "headers": {"Content-Type": "application/json"},
159
- "method": "GET",
160
- "body": {"content": '{"custom": "field"}'},
161
- }
162
- response = {"status_code": 200, "headers": {"field": "value"}, "body": {"content": '{"name": "field"}'}}
163
- expected_pages = [
164
- StreamReadPages(
165
- request=HttpRequest(
166
- url="https://demonslayers.com/api/v1/hashiras",
167
- parameters={"era": ["taisho"]},
168
- headers={"Content-Type": "application/json"},
169
- body='{"custom": "field"}',
170
- http_method="GET",
171
- ),
172
- response=HttpResponse(status=200, headers={"field": "value"}, body='{"name": "field"}'),
173
- records=[{"name": "Shinobu Kocho"}, {"name": "Muichiro Tokito"}],
174
- ),
175
- StreamReadPages(
176
- request=HttpRequest(
177
- url="https://demonslayers.com/api/v1/hashiras",
178
- parameters={"era": ["taisho"]},
179
- headers={"Content-Type": "application/json"},
180
- body='{"custom": "field"}',
181
- http_method="GET",
182
- ),
183
- response=HttpResponse(status=200, headers={"field": "value"}, body='{"name": "field"}'),
184
- records=[{"name": "Mitsuri Kanroji"}],
185
- ),
186
- ]
187
- expected_logs = [
188
- LogMessage(**{"message": "log message before the request", "level": "INFO"}),
189
- LogMessage(**{"message": "log message during the page", "level": "INFO"}),
190
- LogMessage(**{"message": "log message after the response", "level": "INFO"}),
191
- ]
192
-
193
- mock_source = make_mock_source(
194
- mock_entrypoint_read,
195
- iter(
196
- [
197
- AirbyteMessage(type=MessageType.LOG, log=AirbyteLogMessage(level=Level.INFO, message="log message before the request")),
198
- request_response_log_message(request, response, url),
199
- record_message("hashiras", {"name": "Shinobu Kocho"}),
200
- AirbyteMessage(type=MessageType.LOG, log=AirbyteLogMessage(level=Level.INFO, message="log message during the page")),
201
- record_message("hashiras", {"name": "Muichiro Tokito"}),
202
- AirbyteMessage(type=MessageType.LOG, log=AirbyteLogMessage(level=Level.INFO, message="log message after the response")),
203
- ]
204
- ),
205
- )
206
-
207
- connector_builder_handler = MessageGrouper(MAX_PAGES_PER_SLICE, MAX_SLICES)
208
-
209
- actual_response: StreamRead = connector_builder_handler.get_message_groups(
210
- source=mock_source, config=CONFIG, configured_catalog=create_configured_catalog("hashiras")
211
- )
212
- single_slice = actual_response.slices[0]
213
- for i, actual_page in enumerate(single_slice.pages):
214
- assert actual_page == expected_pages[i]
215
-
216
- for i, actual_log in enumerate(actual_response.logs):
217
- assert actual_log == expected_logs[i]
218
-
219
-
220
- @pytest.mark.parametrize(
221
- "request_record_limit, max_record_limit, should_fail",
222
- [
223
- pytest.param(1, 3, False, id="test_create_request_with_record_limit"),
224
- pytest.param(3, 1, True, id="test_create_request_record_limit_exceeds_max"),
225
- ],
226
- )
227
- @patch("airbyte_cdk.connector_builder.message_grouper.AirbyteEntrypoint.read")
228
- def test_get_grouped_messages_record_limit(mock_entrypoint_read: Mock, request_record_limit: int, max_record_limit: int, should_fail: bool) -> None:
229
- url = "https://demonslayers.com/api/v1/hashiras?era=taisho"
230
- request = {
231
- "headers": {"Content-Type": "application/json"},
232
- "method": "GET",
233
- "body": {"content": '{"custom": "field"}'},
234
- }
235
- response = {"status_code": 200, "headers": {"field": "value"}, "body": {"content": '{"name": "field"}'}}
236
- mock_source = make_mock_source(
237
- mock_entrypoint_read,
238
- iter(
239
- [
240
- request_response_log_message(request, response, url),
241
- record_message("hashiras", {"name": "Shinobu Kocho"}),
242
- record_message("hashiras", {"name": "Muichiro Tokito"}),
243
- request_response_log_message(request, response, url),
244
- record_message("hashiras", {"name": "Mitsuri Kanroji"}),
245
- ]
246
- ),
247
- )
248
- n_records = 2
249
- record_limit = min(request_record_limit, max_record_limit)
250
-
251
- api = MessageGrouper(MAX_PAGES_PER_SLICE, MAX_SLICES, max_record_limit=max_record_limit)
252
- # this is the call we expect to raise an exception
253
- if should_fail:
254
- with pytest.raises(ValueError):
255
- api.get_message_groups(
256
- mock_source, config=CONFIG, configured_catalog=create_configured_catalog("hashiras"), record_limit=request_record_limit
257
- )
258
- else:
259
- actual_response: StreamRead = api.get_message_groups(
260
- mock_source, config=CONFIG, configured_catalog=create_configured_catalog("hashiras"), record_limit=request_record_limit
261
- )
262
- single_slice = actual_response.slices[0]
263
- total_records = 0
264
- for i, actual_page in enumerate(single_slice.pages):
265
- total_records += len(actual_page.records)
266
- assert total_records == min([record_limit, n_records])
267
-
268
- assert (total_records >= max_record_limit) == actual_response.test_read_limit_reached
269
-
270
-
271
- @pytest.mark.parametrize(
272
- "max_record_limit",
273
- [
274
- pytest.param(2, id="test_create_request_no_record_limit"),
275
- pytest.param(1, id="test_create_request_no_record_limit_n_records_exceed_max"),
276
- ],
277
- )
278
- @patch("airbyte_cdk.connector_builder.message_grouper.AirbyteEntrypoint.read")
279
- def test_get_grouped_messages_default_record_limit(mock_entrypoint_read: Mock, max_record_limit: int) -> None:
280
- url = "https://demonslayers.com/api/v1/hashiras?era=taisho"
281
- request = {
282
- "headers": {"Content-Type": "application/json"},
283
- "method": "GET",
284
- "body": {"content": '{"custom": "field"}'},
285
- }
286
- response = {"status_code": 200, "headers": {"field": "value"}, "body": {"content": '{"name": "field"}'}}
287
- mock_source = make_mock_source(
288
- mock_entrypoint_read,
289
- iter(
290
- [
291
- request_response_log_message(request, response, url),
292
- record_message("hashiras", {"name": "Shinobu Kocho"}),
293
- record_message("hashiras", {"name": "Muichiro Tokito"}),
294
- request_response_log_message(request, response, url),
295
- record_message("hashiras", {"name": "Mitsuri Kanroji"}),
296
- ]
297
- ),
298
- )
299
- n_records = 2
300
-
301
- api = MessageGrouper(MAX_PAGES_PER_SLICE, MAX_SLICES, max_record_limit=max_record_limit)
302
- actual_response: StreamRead = api.get_message_groups(
303
- source=mock_source, config=CONFIG, configured_catalog=create_configured_catalog("hashiras")
304
- )
305
- single_slice = actual_response.slices[0]
306
- total_records = 0
307
- for i, actual_page in enumerate(single_slice.pages):
308
- total_records += len(actual_page.records)
309
- assert total_records == min([max_record_limit, n_records])
310
-
311
-
312
- @patch("airbyte_cdk.connector_builder.message_grouper.AirbyteEntrypoint.read")
313
- def test_get_grouped_messages_limit_0(mock_entrypoint_read: Mock) -> None:
314
- url = "https://demonslayers.com/api/v1/hashiras?era=taisho"
315
- request = {
316
- "headers": {"Content-Type": "application/json"},
317
- "method": "GET",
318
- "body": {"content": '{"custom": "field"}'},
319
- }
320
- response = {"status_code": 200, "headers": {"field": "value"}, "body": {"content": '{"name": "field"}'}}
321
- mock_source = make_mock_source(
322
- mock_entrypoint_read,
323
- iter(
324
- [
325
- request_response_log_message(request, response, url),
326
- record_message("hashiras", {"name": "Shinobu Kocho"}),
327
- record_message("hashiras", {"name": "Muichiro Tokito"}),
328
- request_response_log_message(request, response, url),
329
- record_message("hashiras", {"name": "Mitsuri Kanroji"}),
330
- ]
331
- ),
332
- )
333
- api = MessageGrouper(MAX_PAGES_PER_SLICE, MAX_SLICES)
334
-
335
- with pytest.raises(ValueError):
336
- api.get_message_groups(source=mock_source, config=CONFIG, configured_catalog=create_configured_catalog("hashiras"), record_limit=0)
337
-
338
-
339
- @patch("airbyte_cdk.connector_builder.message_grouper.AirbyteEntrypoint.read")
340
- def test_get_grouped_messages_no_records(mock_entrypoint_read: Mock) -> None:
341
- url = "https://demonslayers.com/api/v1/hashiras?era=taisho"
342
- request = {
343
- "headers": {"Content-Type": "application/json"},
344
- "method": "GET",
345
- "body": {"content": '{"custom": "field"}'},
346
- }
347
- response = {"status_code": 200, "headers": {"field": "value"}, "body": {"content": '{"name": "field"}'}}
348
- expected_pages = [
349
- StreamReadPages(
350
- request=HttpRequest(
351
- url="https://demonslayers.com/api/v1/hashiras",
352
- parameters={"era": ["taisho"]},
353
- headers={"Content-Type": "application/json"},
354
- body='{"custom": "field"}',
355
- http_method="GET",
356
- ),
357
- response=HttpResponse(status=200, headers={"field": "value"}, body='{"name": "field"}'),
358
- records=[],
359
- ),
360
- StreamReadPages(
361
- request=HttpRequest(
362
- url="https://demonslayers.com/api/v1/hashiras",
363
- parameters={"era": ["taisho"]},
364
- headers={"Content-Type": "application/json"},
365
- body='{"custom": "field"}',
366
- http_method="GET",
367
- ),
368
- response=HttpResponse(status=200, headers={"field": "value"}, body='{"name": "field"}'),
369
- records=[],
370
- ),
371
- ]
372
-
373
- mock_source = make_mock_source(
374
- mock_entrypoint_read,
375
- iter(
376
- [
377
- request_response_log_message(request, response, url),
378
- request_response_log_message(request, response, url),
379
- ]
380
- ),
381
- )
382
-
383
- message_grouper = MessageGrouper(MAX_PAGES_PER_SLICE, MAX_SLICES)
384
-
385
- actual_response: StreamRead = message_grouper.get_message_groups(
386
- source=mock_source, config=CONFIG, configured_catalog=create_configured_catalog("hashiras")
387
- )
388
-
389
- single_slice = actual_response.slices[0]
390
- for i, actual_page in enumerate(single_slice.pages):
391
- assert actual_page == expected_pages[i]
392
-
393
-
394
- @pytest.mark.parametrize(
395
- "log_message, expected_response",
396
- [
397
- pytest.param(
398
- {
399
- "http": {
400
- "response": {
401
- "status_code": 200,
402
- "headers": {"field": "name"},
403
- "body": {"content": '{"id": "fire", "owner": "kyojuro_rengoku"}'},
404
- }
405
- }
406
- },
407
- HttpResponse(status=200, headers={"field": "name"}, body='{"id": "fire", "owner": "kyojuro_rengoku"}'),
408
- id="test_create_response_with_all_fields",
409
- ),
410
- pytest.param(
411
- {"http": {"response": {"status_code": 200, "headers": {"field": "name"}}}},
412
- HttpResponse(status=200, headers={"field": "name"}, body=""),
413
- id="test_create_response_with_no_body",
414
- ),
415
- pytest.param(
416
- {"http": {"response": {"status_code": 200, "body": {"content": '{"id": "fire", "owner": "kyojuro_rengoku"}'}}}},
417
- HttpResponse(status=200, body='{"id": "fire", "owner": "kyojuro_rengoku"}'),
418
- id="test_create_response_with_no_headers",
419
- ),
420
- pytest.param(
421
- {
422
- "http": {
423
- "response": {
424
- "status_code": 200,
425
- "headers": {"field": "name"},
426
- "body": {"content": '[{"id": "fire", "owner": "kyojuro_rengoku"}, {"id": "mist", "owner": "muichiro_tokito"}]'},
427
- }
428
- }
429
- },
430
- HttpResponse(
431
- status=200,
432
- headers={"field": "name"},
433
- body='[{"id": "fire", "owner": "kyojuro_rengoku"}, {"id": "mist", "owner": "muichiro_tokito"}]',
434
- ),
435
- id="test_create_response_with_array",
436
- ),
437
- pytest.param(
438
- {"http": {"response": {"status_code": 200, "body": {"content": "tomioka"}}}},
439
- HttpResponse(status=200, body="tomioka"),
440
- id="test_create_response_with_string",
441
- ),
442
- ],
443
- )
444
- def test_create_response_from_log_message(log_message: str, expected_response: HttpResponse) -> None:
445
- if isinstance(log_message, str):
446
- response_message = json.loads(log_message)
447
- else:
448
- response_message = log_message
449
-
450
- connector_builder_handler = MessageGrouper(MAX_PAGES_PER_SLICE, MAX_SLICES)
451
- actual_response = connector_builder_handler._create_response_from_log_message(response_message)
452
-
453
- assert actual_response == expected_response
454
-
455
-
456
- @patch("airbyte_cdk.connector_builder.message_grouper.AirbyteEntrypoint.read")
457
- def test_get_grouped_messages_with_many_slices(mock_entrypoint_read: Mock) -> None:
458
- url = "http://a-url.com"
459
- request: Mapping[str, Any] = {}
460
- response = {"status_code": 200}
461
-
462
- mock_source = make_mock_source(
463
- mock_entrypoint_read,
464
- iter(
465
- [
466
- slice_message('{"descriptor": "first_slice"}'),
467
- request_response_log_message(request, response, url),
468
- record_message("hashiras", {"name": "Muichiro Tokito"}),
469
- slice_message('{"descriptor": "second_slice"}'),
470
- request_response_log_message(request, response, url),
471
- record_message("hashiras", {"name": "Shinobu Kocho"}),
472
- record_message("hashiras", {"name": "Mitsuri Kanroji"}),
473
- request_response_log_message(request, response, url),
474
- record_message("hashiras", {"name": "Obanai Iguro"}),
475
- request_response_log_message(request, response, url),
476
- ]
477
- ),
478
- )
479
-
480
- connector_builder_handler = MessageGrouper(MAX_PAGES_PER_SLICE, MAX_SLICES)
481
-
482
- stream_read: StreamRead = connector_builder_handler.get_message_groups(
483
- source=mock_source, config=CONFIG, configured_catalog=create_configured_catalog("hashiras")
484
- )
485
-
486
- assert not stream_read.test_read_limit_reached
487
- assert len(stream_read.slices) == 2
488
-
489
- assert stream_read.slices[0].slice_descriptor == {"descriptor": "first_slice"}
490
- assert len(stream_read.slices[0].pages) == 1
491
- assert len(stream_read.slices[0].pages[0].records) == 1
492
-
493
- assert stream_read.slices[1].slice_descriptor == {"descriptor": "second_slice"}
494
- assert len(stream_read.slices[1].pages) == 3
495
- assert len(stream_read.slices[1].pages[0].records) == 2
496
- assert len(stream_read.slices[1].pages[1].records) == 1
497
- assert len(stream_read.slices[1].pages[2].records) == 0
498
-
499
-
500
- @patch("airbyte_cdk.connector_builder.message_grouper.AirbyteEntrypoint.read")
501
- def test_get_grouped_messages_given_maximum_number_of_slices_then_test_read_limit_reached(mock_entrypoint_read: Mock) -> None:
502
- maximum_number_of_slices = 5
503
- request: Mapping[str, Any] = {}
504
- response = {"status_code": 200}
505
- mock_source = make_mock_source(
506
- mock_entrypoint_read, iter([slice_message(), request_response_log_message(request, response, "a_url")] * maximum_number_of_slices)
507
- )
508
-
509
- api = MessageGrouper(MAX_PAGES_PER_SLICE, MAX_SLICES)
510
-
511
- stream_read: StreamRead = api.get_message_groups(
512
- source=mock_source, config=CONFIG, configured_catalog=create_configured_catalog("hashiras")
513
- )
514
-
515
- assert stream_read.test_read_limit_reached
516
-
517
-
518
- @patch("airbyte_cdk.connector_builder.message_grouper.AirbyteEntrypoint.read")
519
- def test_get_grouped_messages_given_maximum_number_of_pages_then_test_read_limit_reached(mock_entrypoint_read: Mock) -> None:
520
- maximum_number_of_pages_per_slice = 5
521
- request: Mapping[str, Any] = {}
522
- response = {"status_code": 200}
523
- mock_source = make_mock_source(
524
- mock_entrypoint_read,
525
- iter([slice_message()] + [request_response_log_message(request, response, "a_url")] * maximum_number_of_pages_per_slice),
526
- )
527
-
528
- api = MessageGrouper(MAX_PAGES_PER_SLICE, MAX_SLICES)
529
-
530
- stream_read: StreamRead = api.get_message_groups(
531
- source=mock_source, config=CONFIG, configured_catalog=create_configured_catalog("hashiras")
532
- )
533
-
534
- assert stream_read.test_read_limit_reached
535
-
536
-
537
- def test_read_stream_returns_error_if_stream_does_not_exist() -> None:
538
- mock_source = MagicMock()
539
- mock_source.read.side_effect = ValueError("error")
540
-
541
- full_config: Mapping[str, Any] = {**CONFIG, **{"__injected_declarative_manifest": MANIFEST}}
542
-
543
- message_grouper = MessageGrouper(MAX_PAGES_PER_SLICE, MAX_SLICES)
544
- actual_response = message_grouper.get_message_groups(
545
- source=mock_source, config=full_config, configured_catalog=create_configured_catalog("not_in_manifest")
546
- )
547
-
548
- assert 1 == len(actual_response.logs)
549
- assert "Traceback" in actual_response.logs[0].message
550
- assert "ERROR" in actual_response.logs[0].level
551
-
552
-
553
- @patch("airbyte_cdk.connector_builder.message_grouper.AirbyteEntrypoint.read")
554
- def test_given_control_message_then_stream_read_has_config_update(mock_entrypoint_read: Mock) -> None:
555
- updated_config = {"x": 1}
556
- mock_source = make_mock_source(
557
- mock_entrypoint_read, iter(any_request_and_response_with_a_record() + [connector_configuration_control_message(1, updated_config)])
558
- )
559
- connector_builder_handler = MessageGrouper(MAX_PAGES_PER_SLICE, MAX_SLICES)
560
- stream_read: StreamRead = connector_builder_handler.get_message_groups(
561
- source=mock_source, config=CONFIG, configured_catalog=create_configured_catalog("hashiras")
562
- )
563
-
564
- assert stream_read.latest_config_update == updated_config
565
-
566
-
567
- @patch("airbyte_cdk.connector_builder.message_grouper.AirbyteEntrypoint.read")
568
- def test_given_multiple_control_messages_then_stream_read_has_latest_based_on_emitted_at(mock_entrypoint_read: Mock) -> None:
569
- earliest = 0
570
- earliest_config = {"earliest": 0}
571
- latest = 1
572
- latest_config = {"latest": 1}
573
- mock_source = make_mock_source(
574
- mock_entrypoint_read,
575
- iter(
576
- any_request_and_response_with_a_record()
577
- + [
578
- # here, we test that even if messages are emitted in a different order, we still rely on `emitted_at`
579
- connector_configuration_control_message(latest, latest_config),
580
- connector_configuration_control_message(earliest, earliest_config),
581
- ]
582
- ),
583
- )
584
- connector_builder_handler = MessageGrouper(MAX_PAGES_PER_SLICE, MAX_SLICES)
585
- stream_read: StreamRead = connector_builder_handler.get_message_groups(
586
- source=mock_source, config=CONFIG, configured_catalog=create_configured_catalog("hashiras")
587
- )
588
-
589
- assert stream_read.latest_config_update == latest_config
590
-
591
-
592
- @patch("airbyte_cdk.connector_builder.message_grouper.AirbyteEntrypoint.read")
593
- def test_given_multiple_control_messages_with_same_timestamp_then_stream_read_has_latest_based_on_message_order(
594
- mock_entrypoint_read: Mock,
595
- ) -> None:
596
- emitted_at = 0
597
- earliest_config = {"earliest": 0}
598
- latest_config = {"latest": 1}
599
- mock_source = make_mock_source(
600
- mock_entrypoint_read,
601
- iter(
602
- any_request_and_response_with_a_record()
603
- + [
604
- connector_configuration_control_message(emitted_at, earliest_config),
605
- connector_configuration_control_message(emitted_at, latest_config),
606
- ]
607
- ),
608
- )
609
- connector_builder_handler = MessageGrouper(MAX_PAGES_PER_SLICE, MAX_SLICES)
610
- stream_read: StreamRead = connector_builder_handler.get_message_groups(
611
- source=mock_source, config=CONFIG, configured_catalog=create_configured_catalog("hashiras")
612
- )
613
-
614
- assert stream_read.latest_config_update == latest_config
615
-
616
-
617
- @patch("airbyte_cdk.connector_builder.message_grouper.AirbyteEntrypoint.read")
618
- def test_given_auxiliary_requests_then_return_auxiliary_request(mock_entrypoint_read: Mock) -> None:
619
- mock_source = make_mock_source(mock_entrypoint_read, iter(any_request_and_response_with_a_record() + [auxiliary_request_log_message()]))
620
- connector_builder_handler = MessageGrouper(MAX_PAGES_PER_SLICE, MAX_SLICES)
621
- stream_read: StreamRead = connector_builder_handler.get_message_groups(
622
- source=mock_source, config=CONFIG, configured_catalog=create_configured_catalog("hashiras")
623
- )
624
-
625
- assert len(stream_read.auxiliary_requests) == 1
626
-
627
-
628
- @patch("airbyte_cdk.connector_builder.message_grouper.AirbyteEntrypoint.read")
629
- def test_given_no_slices_then_return_empty_slices(mock_entrypoint_read: Mock) -> None:
630
- mock_source = make_mock_source(mock_entrypoint_read, iter([auxiliary_request_log_message()]))
631
- connector_builder_handler = MessageGrouper(MAX_PAGES_PER_SLICE, MAX_SLICES)
632
- stream_read: StreamRead = connector_builder_handler.get_message_groups(
633
- source=mock_source, config=CONFIG, configured_catalog=create_configured_catalog("hashiras")
634
- )
635
-
636
- assert len(stream_read.slices) == 0
637
-
638
-
639
- def make_mock_source(mock_entrypoint_read: Mock, return_value: Iterator[AirbyteMessage]) -> MagicMock:
640
- mock_source = MagicMock()
641
- mock_entrypoint_read.return_value = return_value
642
- return mock_source
643
-
644
-
645
- def request_log_message(request: Mapping[str, Any]) -> AirbyteMessage:
646
- return AirbyteMessage(type=MessageType.LOG, log=AirbyteLogMessage(level=Level.INFO, message=f"request:{json.dumps(request)}"))
647
-
648
-
649
- def response_log_message(response: Mapping[str, Any]) -> AirbyteMessage:
650
- return AirbyteMessage(type=MessageType.LOG, log=AirbyteLogMessage(level=Level.INFO, message=f"response:{json.dumps(response)}"))
651
-
652
-
653
- def record_message(stream: str, data: Mapping[str, Any]) -> AirbyteMessage:
654
- return AirbyteMessage(type=MessageType.RECORD, record=AirbyteRecordMessage(stream=stream, data=data, emitted_at=1234))
655
-
656
-
657
- def slice_message(slice_descriptor: str = '{"key": "value"}') -> AirbyteMessage:
658
- return AirbyteMessage(type=MessageType.LOG, log=AirbyteLogMessage(level=Level.INFO, message="slice:" + slice_descriptor))
659
-
660
-
661
- def connector_configuration_control_message(emitted_at: float, config: Mapping[str, Any]) -> AirbyteMessage:
662
- return AirbyteMessage(
663
- type=MessageType.CONTROL,
664
- control=AirbyteControlMessage(
665
- type=OrchestratorType.CONNECTOR_CONFIG,
666
- emitted_at=emitted_at,
667
- connectorConfig=AirbyteControlConnectorConfigMessage(config=config),
668
- ),
669
- )
670
-
671
-
672
- def auxiliary_request_log_message() -> AirbyteMessage:
673
- return AirbyteMessage(
674
- type=MessageType.LOG,
675
- log=AirbyteLogMessage(
676
- level=Level.INFO,
677
- message=json.dumps(
678
- {
679
- "http": {
680
- "is_auxiliary": True,
681
- "title": "a title",
682
- "description": "a description",
683
- "request": {},
684
- "response": {},
685
- },
686
- "url": {"full": "https://a-url.com"},
687
- }
688
- ),
689
- ),
690
- )
691
-
692
-
693
- def request_response_log_message(request: Mapping[str, Any], response: Mapping[str, Any], url: str) -> AirbyteMessage:
694
- return AirbyteMessage(
695
- type=MessageType.LOG,
696
- log=AirbyteLogMessage(
697
- level=Level.INFO,
698
- message=json.dumps(
699
- {
700
- "airbyte_cdk": {"stream": {"name": "a stream name"}},
701
- "http": {"title": "a title", "description": "a description", "request": request, "response": response},
702
- "url": {"full": url},
703
- }
704
- ),
705
- ),
706
- )
707
-
708
-
709
- def any_request_and_response_with_a_record() -> List[AirbyteMessage]:
710
- return [
711
- request_response_log_message({"request": 1}, {"response": 2}, "http://any_url.com"),
712
- record_message("hashiras", {"name": "Shinobu Kocho"}),
713
- ]