airbyte-cdk 0.72.0__py3-none-any.whl → 6.13.1.dev4106__py3-none-any.whl

Sign up to get free protection for your applications and to get access to all the features.
Files changed (517) hide show
  1. airbyte_cdk/__init__.py +355 -6
  2. airbyte_cdk/cli/__init__.py +1 -0
  3. airbyte_cdk/cli/source_declarative_manifest/__init__.py +5 -0
  4. airbyte_cdk/cli/source_declarative_manifest/_run.py +230 -0
  5. airbyte_cdk/cli/source_declarative_manifest/spec.json +17 -0
  6. airbyte_cdk/config_observation.py +29 -10
  7. airbyte_cdk/connector.py +24 -24
  8. airbyte_cdk/connector_builder/README.md +53 -0
  9. airbyte_cdk/connector_builder/connector_builder_handler.py +37 -11
  10. airbyte_cdk/connector_builder/main.py +45 -13
  11. airbyte_cdk/connector_builder/message_grouper.py +189 -50
  12. airbyte_cdk/connector_builder/models.py +3 -2
  13. airbyte_cdk/destinations/__init__.py +4 -3
  14. airbyte_cdk/destinations/destination.py +54 -20
  15. airbyte_cdk/destinations/vector_db_based/README.md +37 -0
  16. airbyte_cdk/destinations/vector_db_based/config.py +40 -17
  17. airbyte_cdk/destinations/vector_db_based/document_processor.py +56 -17
  18. airbyte_cdk/destinations/vector_db_based/embedder.py +57 -15
  19. airbyte_cdk/destinations/vector_db_based/test_utils.py +14 -4
  20. airbyte_cdk/destinations/vector_db_based/utils.py +8 -2
  21. airbyte_cdk/destinations/vector_db_based/writer.py +24 -5
  22. airbyte_cdk/entrypoint.py +153 -44
  23. airbyte_cdk/exception_handler.py +21 -3
  24. airbyte_cdk/logger.py +30 -44
  25. airbyte_cdk/models/__init__.py +13 -2
  26. airbyte_cdk/models/airbyte_protocol.py +86 -1
  27. airbyte_cdk/models/airbyte_protocol_serializers.py +44 -0
  28. airbyte_cdk/models/file_transfer_record_message.py +13 -0
  29. airbyte_cdk/models/well_known_types.py +1 -1
  30. airbyte_cdk/sources/__init__.py +5 -1
  31. airbyte_cdk/sources/abstract_source.py +125 -79
  32. airbyte_cdk/sources/concurrent_source/__init__.py +7 -2
  33. airbyte_cdk/sources/concurrent_source/concurrent_read_processor.py +102 -36
  34. airbyte_cdk/sources/concurrent_source/concurrent_source.py +29 -36
  35. airbyte_cdk/sources/concurrent_source/concurrent_source_adapter.py +94 -10
  36. airbyte_cdk/sources/concurrent_source/stream_thread_exception.py +25 -0
  37. airbyte_cdk/sources/concurrent_source/thread_pool_manager.py +20 -14
  38. airbyte_cdk/sources/config.py +3 -2
  39. airbyte_cdk/sources/connector_state_manager.py +49 -83
  40. airbyte_cdk/sources/declarative/async_job/job.py +52 -0
  41. airbyte_cdk/sources/declarative/async_job/job_orchestrator.py +497 -0
  42. airbyte_cdk/sources/declarative/async_job/job_tracker.py +75 -0
  43. airbyte_cdk/sources/declarative/async_job/repository.py +35 -0
  44. airbyte_cdk/sources/declarative/async_job/status.py +24 -0
  45. airbyte_cdk/sources/declarative/async_job/timer.py +39 -0
  46. airbyte_cdk/sources/declarative/auth/__init__.py +2 -3
  47. airbyte_cdk/sources/declarative/auth/declarative_authenticator.py +3 -1
  48. airbyte_cdk/sources/declarative/auth/jwt.py +191 -0
  49. airbyte_cdk/sources/declarative/auth/oauth.py +60 -20
  50. airbyte_cdk/sources/declarative/auth/selective_authenticator.py +10 -2
  51. airbyte_cdk/sources/declarative/auth/token.py +28 -10
  52. airbyte_cdk/sources/declarative/auth/token_provider.py +9 -8
  53. airbyte_cdk/sources/declarative/checks/check_stream.py +16 -8
  54. airbyte_cdk/sources/declarative/checks/connection_checker.py +4 -2
  55. airbyte_cdk/sources/declarative/concurrency_level/__init__.py +7 -0
  56. airbyte_cdk/sources/declarative/concurrency_level/concurrency_level.py +50 -0
  57. airbyte_cdk/sources/declarative/concurrent_declarative_source.py +421 -0
  58. airbyte_cdk/sources/declarative/datetime/datetime_parser.py +4 -0
  59. airbyte_cdk/sources/declarative/datetime/min_max_datetime.py +26 -6
  60. airbyte_cdk/sources/declarative/declarative_component_schema.yaml +1213 -88
  61. airbyte_cdk/sources/declarative/declarative_source.py +5 -2
  62. airbyte_cdk/sources/declarative/declarative_stream.py +95 -9
  63. airbyte_cdk/sources/declarative/decoders/__init__.py +23 -2
  64. airbyte_cdk/sources/declarative/decoders/composite_raw_decoder.py +97 -0
  65. airbyte_cdk/sources/declarative/decoders/decoder.py +11 -4
  66. airbyte_cdk/sources/declarative/decoders/json_decoder.py +92 -5
  67. airbyte_cdk/sources/declarative/decoders/noop_decoder.py +21 -0
  68. airbyte_cdk/sources/declarative/decoders/pagination_decoder_decorator.py +39 -0
  69. airbyte_cdk/sources/declarative/decoders/xml_decoder.py +98 -0
  70. airbyte_cdk/sources/declarative/extractors/__init__.py +12 -1
  71. airbyte_cdk/sources/declarative/extractors/dpath_extractor.py +29 -24
  72. airbyte_cdk/sources/declarative/extractors/http_selector.py +4 -5
  73. airbyte_cdk/sources/declarative/extractors/record_extractor.py +2 -3
  74. airbyte_cdk/sources/declarative/extractors/record_filter.py +65 -8
  75. airbyte_cdk/sources/declarative/extractors/record_selector.py +85 -26
  76. airbyte_cdk/sources/declarative/extractors/response_to_file_extractor.py +177 -0
  77. airbyte_cdk/sources/declarative/extractors/type_transformer.py +55 -0
  78. airbyte_cdk/sources/declarative/incremental/__init__.py +25 -3
  79. airbyte_cdk/sources/declarative/incremental/datetime_based_cursor.py +156 -48
  80. airbyte_cdk/sources/declarative/incremental/declarative_cursor.py +13 -0
  81. airbyte_cdk/sources/declarative/incremental/global_substream_cursor.py +350 -0
  82. airbyte_cdk/sources/declarative/incremental/per_partition_cursor.py +159 -74
  83. airbyte_cdk/sources/declarative/incremental/per_partition_with_global.py +200 -0
  84. airbyte_cdk/sources/declarative/incremental/resumable_full_refresh_cursor.py +122 -0
  85. airbyte_cdk/sources/declarative/interpolation/filters.py +27 -1
  86. airbyte_cdk/sources/declarative/interpolation/interpolated_boolean.py +23 -5
  87. airbyte_cdk/sources/declarative/interpolation/interpolated_mapping.py +12 -8
  88. airbyte_cdk/sources/declarative/interpolation/interpolated_nested_mapping.py +13 -6
  89. airbyte_cdk/sources/declarative/interpolation/interpolated_string.py +21 -6
  90. airbyte_cdk/sources/declarative/interpolation/interpolation.py +9 -3
  91. airbyte_cdk/sources/declarative/interpolation/jinja.py +72 -37
  92. airbyte_cdk/sources/declarative/interpolation/macros.py +72 -17
  93. airbyte_cdk/sources/declarative/manifest_declarative_source.py +193 -52
  94. airbyte_cdk/sources/declarative/migrations/legacy_to_per_partition_state_migration.py +98 -0
  95. airbyte_cdk/sources/declarative/migrations/state_migration.py +24 -0
  96. airbyte_cdk/sources/declarative/models/__init__.py +1 -1
  97. airbyte_cdk/sources/declarative/models/declarative_component_schema.py +1329 -595
  98. airbyte_cdk/sources/declarative/parsers/custom_exceptions.py +2 -2
  99. airbyte_cdk/sources/declarative/parsers/manifest_component_transformer.py +26 -4
  100. airbyte_cdk/sources/declarative/parsers/manifest_reference_resolver.py +26 -15
  101. airbyte_cdk/sources/declarative/parsers/model_to_component_factory.py +1699 -226
  102. airbyte_cdk/sources/declarative/partition_routers/__init__.py +24 -4
  103. airbyte_cdk/sources/declarative/partition_routers/async_job_partition_router.py +65 -0
  104. airbyte_cdk/sources/declarative/partition_routers/cartesian_product_stream_slicer.py +176 -0
  105. airbyte_cdk/sources/declarative/partition_routers/list_partition_router.py +39 -9
  106. airbyte_cdk/sources/declarative/partition_routers/partition_router.py +62 -0
  107. airbyte_cdk/sources/declarative/partition_routers/single_partition_router.py +15 -3
  108. airbyte_cdk/sources/declarative/partition_routers/substream_partition_router.py +222 -39
  109. airbyte_cdk/sources/declarative/requesters/error_handlers/__init__.py +19 -5
  110. airbyte_cdk/sources/declarative/requesters/error_handlers/backoff_strategies/__init__.py +3 -1
  111. airbyte_cdk/sources/declarative/requesters/error_handlers/backoff_strategies/constant_backoff_strategy.py +19 -7
  112. airbyte_cdk/sources/declarative/requesters/error_handlers/backoff_strategies/exponential_backoff_strategy.py +19 -7
  113. airbyte_cdk/sources/declarative/requesters/error_handlers/backoff_strategies/header_helper.py +4 -2
  114. airbyte_cdk/sources/declarative/requesters/error_handlers/backoff_strategies/wait_time_from_header_backoff_strategy.py +41 -9
  115. airbyte_cdk/sources/declarative/requesters/error_handlers/backoff_strategies/wait_until_time_from_header_backoff_strategy.py +29 -14
  116. airbyte_cdk/sources/declarative/requesters/error_handlers/backoff_strategy.py +5 -13
  117. airbyte_cdk/sources/declarative/requesters/error_handlers/composite_error_handler.py +32 -16
  118. airbyte_cdk/sources/declarative/requesters/error_handlers/default_error_handler.py +46 -56
  119. airbyte_cdk/sources/declarative/requesters/error_handlers/default_http_response_filter.py +40 -0
  120. airbyte_cdk/sources/declarative/requesters/error_handlers/error_handler.py +6 -32
  121. airbyte_cdk/sources/declarative/requesters/error_handlers/http_response_filter.py +119 -41
  122. airbyte_cdk/sources/declarative/requesters/http_job_repository.py +228 -0
  123. airbyte_cdk/sources/declarative/requesters/http_requester.py +98 -344
  124. airbyte_cdk/sources/declarative/requesters/paginators/__init__.py +14 -3
  125. airbyte_cdk/sources/declarative/requesters/paginators/default_paginator.py +105 -46
  126. airbyte_cdk/sources/declarative/requesters/paginators/no_pagination.py +14 -8
  127. airbyte_cdk/sources/declarative/requesters/paginators/paginator.py +19 -8
  128. airbyte_cdk/sources/declarative/requesters/paginators/strategies/__init__.py +9 -3
  129. airbyte_cdk/sources/declarative/requesters/paginators/strategies/cursor_pagination_strategy.py +53 -21
  130. airbyte_cdk/sources/declarative/requesters/paginators/strategies/offset_increment.py +42 -19
  131. airbyte_cdk/sources/declarative/requesters/paginators/strategies/page_increment.py +25 -12
  132. airbyte_cdk/sources/declarative/requesters/paginators/strategies/pagination_strategy.py +13 -10
  133. airbyte_cdk/sources/declarative/requesters/paginators/strategies/stop_condition.py +26 -13
  134. airbyte_cdk/sources/declarative/requesters/request_options/__init__.py +15 -2
  135. airbyte_cdk/sources/declarative/requesters/request_options/datetime_based_request_options_provider.py +91 -0
  136. airbyte_cdk/sources/declarative/requesters/request_options/default_request_options_provider.py +60 -0
  137. airbyte_cdk/sources/declarative/requesters/request_options/interpolated_nested_request_input_provider.py +31 -14
  138. airbyte_cdk/sources/declarative/requesters/request_options/interpolated_request_input_provider.py +27 -15
  139. airbyte_cdk/sources/declarative/requesters/request_options/interpolated_request_options_provider.py +63 -10
  140. airbyte_cdk/sources/declarative/requesters/request_options/request_options_provider.py +1 -1
  141. airbyte_cdk/sources/declarative/requesters/requester.py +9 -17
  142. airbyte_cdk/sources/declarative/resolvers/__init__.py +41 -0
  143. airbyte_cdk/sources/declarative/resolvers/components_resolver.py +55 -0
  144. airbyte_cdk/sources/declarative/resolvers/config_components_resolver.py +136 -0
  145. airbyte_cdk/sources/declarative/resolvers/http_components_resolver.py +112 -0
  146. airbyte_cdk/sources/declarative/retrievers/__init__.py +6 -2
  147. airbyte_cdk/sources/declarative/retrievers/async_retriever.py +100 -0
  148. airbyte_cdk/sources/declarative/retrievers/retriever.py +1 -3
  149. airbyte_cdk/sources/declarative/retrievers/simple_retriever.py +228 -72
  150. airbyte_cdk/sources/declarative/schema/__init__.py +14 -1
  151. airbyte_cdk/sources/declarative/schema/default_schema_loader.py +5 -3
  152. airbyte_cdk/sources/declarative/schema/dynamic_schema_loader.py +236 -0
  153. airbyte_cdk/sources/declarative/schema/json_file_schema_loader.py +8 -8
  154. airbyte_cdk/sources/declarative/spec/spec.py +12 -5
  155. airbyte_cdk/sources/declarative/stream_slicers/__init__.py +1 -2
  156. airbyte_cdk/sources/declarative/stream_slicers/declarative_partition_generator.py +88 -0
  157. airbyte_cdk/sources/declarative/stream_slicers/stream_slicer.py +9 -14
  158. airbyte_cdk/sources/declarative/transformations/add_fields.py +19 -11
  159. airbyte_cdk/sources/declarative/transformations/flatten_fields.py +52 -0
  160. airbyte_cdk/sources/declarative/transformations/keys_replace_transformation.py +61 -0
  161. airbyte_cdk/sources/declarative/transformations/keys_to_lower_transformation.py +22 -0
  162. airbyte_cdk/sources/declarative/transformations/keys_to_snake_transformation.py +68 -0
  163. airbyte_cdk/sources/declarative/transformations/remove_fields.py +13 -10
  164. airbyte_cdk/sources/declarative/transformations/transformation.py +5 -5
  165. airbyte_cdk/sources/declarative/types.py +19 -110
  166. airbyte_cdk/sources/declarative/yaml_declarative_source.py +31 -10
  167. airbyte_cdk/sources/embedded/base_integration.py +16 -5
  168. airbyte_cdk/sources/embedded/catalog.py +16 -4
  169. airbyte_cdk/sources/embedded/runner.py +19 -3
  170. airbyte_cdk/sources/embedded/tools.py +5 -2
  171. airbyte_cdk/sources/file_based/README.md +152 -0
  172. airbyte_cdk/sources/file_based/__init__.py +24 -0
  173. airbyte_cdk/sources/file_based/availability_strategy/__init__.py +9 -2
  174. airbyte_cdk/sources/file_based/availability_strategy/abstract_file_based_availability_strategy.py +22 -6
  175. airbyte_cdk/sources/file_based/availability_strategy/default_file_based_availability_strategy.py +46 -10
  176. airbyte_cdk/sources/file_based/config/abstract_file_based_spec.py +58 -10
  177. airbyte_cdk/sources/file_based/config/avro_format.py +2 -1
  178. airbyte_cdk/sources/file_based/config/csv_format.py +29 -10
  179. airbyte_cdk/sources/file_based/config/excel_format.py +18 -0
  180. airbyte_cdk/sources/file_based/config/file_based_stream_config.py +16 -4
  181. airbyte_cdk/sources/file_based/config/jsonl_format.py +2 -1
  182. airbyte_cdk/sources/file_based/config/parquet_format.py +2 -1
  183. airbyte_cdk/sources/file_based/config/unstructured_format.py +13 -5
  184. airbyte_cdk/sources/file_based/discovery_policy/__init__.py +6 -2
  185. airbyte_cdk/sources/file_based/discovery_policy/abstract_discovery_policy.py +2 -4
  186. airbyte_cdk/sources/file_based/discovery_policy/default_discovery_policy.py +7 -2
  187. airbyte_cdk/sources/file_based/exceptions.py +52 -15
  188. airbyte_cdk/sources/file_based/file_based_source.py +163 -33
  189. airbyte_cdk/sources/file_based/file_based_stream_reader.py +83 -5
  190. airbyte_cdk/sources/file_based/file_types/__init__.py +14 -1
  191. airbyte_cdk/sources/file_based/file_types/avro_parser.py +75 -24
  192. airbyte_cdk/sources/file_based/file_types/csv_parser.py +116 -34
  193. airbyte_cdk/sources/file_based/file_types/excel_parser.py +196 -0
  194. airbyte_cdk/sources/file_based/file_types/file_transfer.py +37 -0
  195. airbyte_cdk/sources/file_based/file_types/file_type_parser.py +4 -1
  196. airbyte_cdk/sources/file_based/file_types/jsonl_parser.py +24 -8
  197. airbyte_cdk/sources/file_based/file_types/parquet_parser.py +60 -18
  198. airbyte_cdk/sources/file_based/file_types/unstructured_parser.py +145 -41
  199. airbyte_cdk/sources/file_based/remote_file.py +1 -1
  200. airbyte_cdk/sources/file_based/schema_helpers.py +38 -10
  201. airbyte_cdk/sources/file_based/schema_validation_policies/__init__.py +3 -1
  202. airbyte_cdk/sources/file_based/schema_validation_policies/abstract_schema_validation_policy.py +3 -1
  203. airbyte_cdk/sources/file_based/schema_validation_policies/default_schema_validation_policies.py +16 -5
  204. airbyte_cdk/sources/file_based/stream/abstract_file_based_stream.py +50 -13
  205. airbyte_cdk/sources/file_based/stream/concurrent/adapters.py +67 -27
  206. airbyte_cdk/sources/file_based/stream/concurrent/cursor/__init__.py +5 -1
  207. airbyte_cdk/sources/file_based/stream/concurrent/cursor/abstract_concurrent_file_based_cursor.py +14 -23
  208. airbyte_cdk/sources/file_based/stream/concurrent/cursor/file_based_concurrent_cursor.py +54 -18
  209. airbyte_cdk/sources/file_based/stream/concurrent/cursor/file_based_final_state_cursor.py +21 -9
  210. airbyte_cdk/sources/file_based/stream/cursor/abstract_file_based_cursor.py +3 -1
  211. airbyte_cdk/sources/file_based/stream/cursor/default_file_based_cursor.py +27 -10
  212. airbyte_cdk/sources/file_based/stream/default_file_based_stream.py +175 -45
  213. airbyte_cdk/sources/http_logger.py +8 -3
  214. airbyte_cdk/sources/message/__init__.py +7 -1
  215. airbyte_cdk/sources/message/repository.py +18 -4
  216. airbyte_cdk/sources/source.py +42 -38
  217. airbyte_cdk/sources/streams/__init__.py +2 -2
  218. airbyte_cdk/sources/streams/availability_strategy.py +54 -3
  219. airbyte_cdk/sources/streams/call_rate.py +64 -21
  220. airbyte_cdk/sources/streams/checkpoint/__init__.py +26 -0
  221. airbyte_cdk/sources/streams/checkpoint/checkpoint_reader.py +335 -0
  222. airbyte_cdk/sources/{declarative/incremental → streams/checkpoint}/cursor.py +17 -14
  223. airbyte_cdk/sources/streams/checkpoint/per_partition_key_serializer.py +22 -0
  224. airbyte_cdk/sources/streams/checkpoint/resumable_full_refresh_cursor.py +51 -0
  225. airbyte_cdk/sources/streams/checkpoint/substream_resumable_full_refresh_cursor.py +110 -0
  226. airbyte_cdk/sources/streams/concurrent/README.md +7 -0
  227. airbyte_cdk/sources/streams/concurrent/abstract_stream.py +7 -2
  228. airbyte_cdk/sources/streams/concurrent/adapters.py +84 -75
  229. airbyte_cdk/sources/streams/concurrent/availability_strategy.py +30 -2
  230. airbyte_cdk/sources/streams/concurrent/cursor.py +298 -42
  231. airbyte_cdk/sources/streams/concurrent/default_stream.py +12 -3
  232. airbyte_cdk/sources/streams/concurrent/exceptions.py +3 -0
  233. airbyte_cdk/sources/streams/concurrent/helpers.py +14 -3
  234. airbyte_cdk/sources/streams/concurrent/partition_enqueuer.py +12 -3
  235. airbyte_cdk/sources/streams/concurrent/partition_reader.py +10 -3
  236. airbyte_cdk/sources/streams/concurrent/partitions/partition.py +1 -16
  237. airbyte_cdk/sources/streams/concurrent/partitions/stream_slicer.py +21 -0
  238. airbyte_cdk/sources/streams/concurrent/partitions/types.py +15 -5
  239. airbyte_cdk/sources/streams/concurrent/state_converters/abstract_stream_state_converter.py +109 -17
  240. airbyte_cdk/sources/streams/concurrent/state_converters/datetime_stream_state_converter.py +90 -72
  241. airbyte_cdk/sources/streams/core.py +412 -87
  242. airbyte_cdk/sources/streams/http/__init__.py +2 -1
  243. airbyte_cdk/sources/streams/http/availability_strategy.py +12 -101
  244. airbyte_cdk/sources/streams/http/error_handlers/__init__.py +22 -0
  245. airbyte_cdk/sources/streams/http/error_handlers/backoff_strategy.py +28 -0
  246. airbyte_cdk/sources/streams/http/error_handlers/default_backoff_strategy.py +17 -0
  247. airbyte_cdk/sources/streams/http/error_handlers/default_error_mapping.py +86 -0
  248. airbyte_cdk/sources/streams/http/error_handlers/error_handler.py +42 -0
  249. airbyte_cdk/sources/streams/http/error_handlers/error_message_parser.py +19 -0
  250. airbyte_cdk/sources/streams/http/error_handlers/http_status_error_handler.py +110 -0
  251. airbyte_cdk/sources/streams/http/error_handlers/json_error_message_parser.py +52 -0
  252. airbyte_cdk/sources/streams/http/error_handlers/response_models.py +65 -0
  253. airbyte_cdk/sources/streams/http/exceptions.py +27 -7
  254. airbyte_cdk/sources/streams/http/http.py +369 -246
  255. airbyte_cdk/sources/streams/http/http_client.py +531 -0
  256. airbyte_cdk/sources/streams/http/rate_limiting.py +76 -12
  257. airbyte_cdk/sources/streams/http/requests_native_auth/abstract_oauth.py +28 -9
  258. airbyte_cdk/sources/streams/http/requests_native_auth/abstract_token.py +2 -1
  259. airbyte_cdk/sources/streams/http/requests_native_auth/oauth.py +90 -35
  260. airbyte_cdk/sources/streams/http/requests_native_auth/token.py +13 -3
  261. airbyte_cdk/sources/types.py +154 -0
  262. airbyte_cdk/sources/utils/record_helper.py +36 -21
  263. airbyte_cdk/sources/utils/schema_helpers.py +13 -6
  264. airbyte_cdk/sources/utils/slice_logger.py +4 -1
  265. airbyte_cdk/sources/utils/transform.py +54 -20
  266. airbyte_cdk/sql/_util/hashing.py +34 -0
  267. airbyte_cdk/sql/_util/name_normalizers.py +92 -0
  268. airbyte_cdk/sql/constants.py +32 -0
  269. airbyte_cdk/sql/exceptions.py +235 -0
  270. airbyte_cdk/sql/secrets.py +123 -0
  271. airbyte_cdk/sql/shared/__init__.py +15 -0
  272. airbyte_cdk/sql/shared/catalog_providers.py +145 -0
  273. airbyte_cdk/sql/shared/sql_processor.py +786 -0
  274. airbyte_cdk/sql/types.py +160 -0
  275. airbyte_cdk/test/catalog_builder.py +70 -18
  276. airbyte_cdk/test/entrypoint_wrapper.py +117 -42
  277. airbyte_cdk/test/mock_http/__init__.py +1 -1
  278. airbyte_cdk/test/mock_http/matcher.py +6 -0
  279. airbyte_cdk/test/mock_http/mocker.py +57 -10
  280. airbyte_cdk/test/mock_http/request.py +19 -3
  281. airbyte_cdk/test/mock_http/response.py +3 -1
  282. airbyte_cdk/test/mock_http/response_builder.py +32 -16
  283. airbyte_cdk/test/state_builder.py +18 -10
  284. airbyte_cdk/test/utils/__init__.py +1 -0
  285. airbyte_cdk/test/utils/data.py +24 -0
  286. airbyte_cdk/test/utils/http_mocking.py +16 -0
  287. airbyte_cdk/test/utils/manifest_only_fixtures.py +60 -0
  288. airbyte_cdk/test/utils/reading.py +26 -0
  289. airbyte_cdk/utils/__init__.py +2 -1
  290. airbyte_cdk/utils/airbyte_secrets_utils.py +5 -3
  291. airbyte_cdk/utils/analytics_message.py +10 -2
  292. airbyte_cdk/utils/datetime_format_inferrer.py +4 -1
  293. airbyte_cdk/utils/event_timing.py +10 -10
  294. airbyte_cdk/utils/mapping_helpers.py +3 -1
  295. airbyte_cdk/utils/message_utils.py +20 -11
  296. airbyte_cdk/utils/print_buffer.py +75 -0
  297. airbyte_cdk/utils/schema_inferrer.py +198 -28
  298. airbyte_cdk/utils/slice_hasher.py +30 -0
  299. airbyte_cdk/utils/spec_schema_transformations.py +6 -3
  300. airbyte_cdk/utils/stream_status_utils.py +8 -1
  301. airbyte_cdk/utils/traced_exception.py +61 -21
  302. airbyte_cdk-6.13.1.dev4106.dist-info/METADATA +109 -0
  303. airbyte_cdk-6.13.1.dev4106.dist-info/RECORD +349 -0
  304. {airbyte_cdk-0.72.0.dist-info → airbyte_cdk-6.13.1.dev4106.dist-info}/WHEEL +1 -2
  305. airbyte_cdk-6.13.1.dev4106.dist-info/entry_points.txt +3 -0
  306. airbyte_cdk/sources/declarative/create_partial.py +0 -92
  307. airbyte_cdk/sources/declarative/parsers/class_types_registry.py +0 -102
  308. airbyte_cdk/sources/declarative/parsers/default_implementation_registry.py +0 -64
  309. airbyte_cdk/sources/declarative/requesters/error_handlers/response_action.py +0 -16
  310. airbyte_cdk/sources/declarative/requesters/error_handlers/response_status.py +0 -68
  311. airbyte_cdk/sources/declarative/stream_slicers/cartesian_product_stream_slicer.py +0 -114
  312. airbyte_cdk/sources/deprecated/base_source.py +0 -94
  313. airbyte_cdk/sources/deprecated/client.py +0 -99
  314. airbyte_cdk/sources/singer/__init__.py +0 -8
  315. airbyte_cdk/sources/singer/singer_helpers.py +0 -304
  316. airbyte_cdk/sources/singer/source.py +0 -186
  317. airbyte_cdk/sources/streams/concurrent/partitions/record.py +0 -23
  318. airbyte_cdk/sources/streams/http/auth/__init__.py +0 -17
  319. airbyte_cdk/sources/streams/http/auth/core.py +0 -29
  320. airbyte_cdk/sources/streams/http/auth/oauth.py +0 -113
  321. airbyte_cdk/sources/streams/http/auth/token.py +0 -47
  322. airbyte_cdk/sources/streams/utils/stream_helper.py +0 -40
  323. airbyte_cdk/sources/utils/catalog_helpers.py +0 -22
  324. airbyte_cdk/sources/utils/schema_models.py +0 -84
  325. airbyte_cdk-0.72.0.dist-info/METADATA +0 -243
  326. airbyte_cdk-0.72.0.dist-info/RECORD +0 -466
  327. airbyte_cdk-0.72.0.dist-info/top_level.txt +0 -3
  328. source_declarative_manifest/main.py +0 -29
  329. unit_tests/connector_builder/__init__.py +0 -3
  330. unit_tests/connector_builder/test_connector_builder_handler.py +0 -871
  331. unit_tests/connector_builder/test_message_grouper.py +0 -713
  332. unit_tests/connector_builder/utils.py +0 -27
  333. unit_tests/destinations/test_destination.py +0 -243
  334. unit_tests/singer/test_singer_helpers.py +0 -56
  335. unit_tests/singer/test_singer_source.py +0 -112
  336. unit_tests/sources/__init__.py +0 -0
  337. unit_tests/sources/concurrent_source/__init__.py +0 -3
  338. unit_tests/sources/concurrent_source/test_concurrent_source_adapter.py +0 -106
  339. unit_tests/sources/declarative/__init__.py +0 -3
  340. unit_tests/sources/declarative/auth/__init__.py +0 -3
  341. unit_tests/sources/declarative/auth/test_oauth.py +0 -331
  342. unit_tests/sources/declarative/auth/test_selective_authenticator.py +0 -39
  343. unit_tests/sources/declarative/auth/test_session_token_auth.py +0 -182
  344. unit_tests/sources/declarative/auth/test_token_auth.py +0 -200
  345. unit_tests/sources/declarative/auth/test_token_provider.py +0 -73
  346. unit_tests/sources/declarative/checks/__init__.py +0 -3
  347. unit_tests/sources/declarative/checks/test_check_stream.py +0 -146
  348. unit_tests/sources/declarative/decoders/__init__.py +0 -0
  349. unit_tests/sources/declarative/decoders/test_json_decoder.py +0 -16
  350. unit_tests/sources/declarative/external_component.py +0 -13
  351. unit_tests/sources/declarative/extractors/__init__.py +0 -3
  352. unit_tests/sources/declarative/extractors/test_dpath_extractor.py +0 -55
  353. unit_tests/sources/declarative/extractors/test_record_filter.py +0 -55
  354. unit_tests/sources/declarative/extractors/test_record_selector.py +0 -179
  355. unit_tests/sources/declarative/incremental/__init__.py +0 -0
  356. unit_tests/sources/declarative/incremental/test_datetime_based_cursor.py +0 -860
  357. unit_tests/sources/declarative/incremental/test_per_partition_cursor.py +0 -406
  358. unit_tests/sources/declarative/incremental/test_per_partition_cursor_integration.py +0 -332
  359. unit_tests/sources/declarative/interpolation/__init__.py +0 -3
  360. unit_tests/sources/declarative/interpolation/test_filters.py +0 -80
  361. unit_tests/sources/declarative/interpolation/test_interpolated_boolean.py +0 -40
  362. unit_tests/sources/declarative/interpolation/test_interpolated_mapping.py +0 -35
  363. unit_tests/sources/declarative/interpolation/test_interpolated_nested_mapping.py +0 -45
  364. unit_tests/sources/declarative/interpolation/test_interpolated_string.py +0 -25
  365. unit_tests/sources/declarative/interpolation/test_jinja.py +0 -240
  366. unit_tests/sources/declarative/interpolation/test_macros.py +0 -73
  367. unit_tests/sources/declarative/parsers/__init__.py +0 -3
  368. unit_tests/sources/declarative/parsers/test_manifest_component_transformer.py +0 -406
  369. unit_tests/sources/declarative/parsers/test_manifest_reference_resolver.py +0 -139
  370. unit_tests/sources/declarative/parsers/test_model_to_component_factory.py +0 -1841
  371. unit_tests/sources/declarative/parsers/testing_components.py +0 -36
  372. unit_tests/sources/declarative/partition_routers/__init__.py +0 -3
  373. unit_tests/sources/declarative/partition_routers/test_list_partition_router.py +0 -155
  374. unit_tests/sources/declarative/partition_routers/test_single_partition_router.py +0 -14
  375. unit_tests/sources/declarative/partition_routers/test_substream_partition_router.py +0 -404
  376. unit_tests/sources/declarative/requesters/__init__.py +0 -3
  377. unit_tests/sources/declarative/requesters/error_handlers/__init__.py +0 -3
  378. unit_tests/sources/declarative/requesters/error_handlers/backoff_strategies/__init__.py +0 -3
  379. unit_tests/sources/declarative/requesters/error_handlers/backoff_strategies/test_constant_backoff.py +0 -34
  380. unit_tests/sources/declarative/requesters/error_handlers/backoff_strategies/test_exponential_backoff.py +0 -36
  381. unit_tests/sources/declarative/requesters/error_handlers/backoff_strategies/test_header_helper.py +0 -38
  382. unit_tests/sources/declarative/requesters/error_handlers/backoff_strategies/test_wait_time_from_header.py +0 -35
  383. unit_tests/sources/declarative/requesters/error_handlers/backoff_strategies/test_wait_until_time_from_header.py +0 -64
  384. unit_tests/sources/declarative/requesters/error_handlers/test_composite_error_handler.py +0 -213
  385. unit_tests/sources/declarative/requesters/error_handlers/test_default_error_handler.py +0 -178
  386. unit_tests/sources/declarative/requesters/error_handlers/test_http_response_filter.py +0 -121
  387. unit_tests/sources/declarative/requesters/error_handlers/test_response_status.py +0 -44
  388. unit_tests/sources/declarative/requesters/paginators/__init__.py +0 -3
  389. unit_tests/sources/declarative/requesters/paginators/test_cursor_pagination_strategy.py +0 -64
  390. unit_tests/sources/declarative/requesters/paginators/test_default_paginator.py +0 -313
  391. unit_tests/sources/declarative/requesters/paginators/test_no_paginator.py +0 -12
  392. unit_tests/sources/declarative/requesters/paginators/test_offset_increment.py +0 -58
  393. unit_tests/sources/declarative/requesters/paginators/test_page_increment.py +0 -70
  394. unit_tests/sources/declarative/requesters/paginators/test_request_option.py +0 -43
  395. unit_tests/sources/declarative/requesters/paginators/test_stop_condition.py +0 -105
  396. unit_tests/sources/declarative/requesters/request_options/__init__.py +0 -3
  397. unit_tests/sources/declarative/requesters/request_options/test_interpolated_request_options_provider.py +0 -101
  398. unit_tests/sources/declarative/requesters/test_http_requester.py +0 -974
  399. unit_tests/sources/declarative/requesters/test_interpolated_request_input_provider.py +0 -32
  400. unit_tests/sources/declarative/retrievers/__init__.py +0 -3
  401. unit_tests/sources/declarative/retrievers/test_simple_retriever.py +0 -542
  402. unit_tests/sources/declarative/schema/__init__.py +0 -6
  403. unit_tests/sources/declarative/schema/source_test/SourceTest.py +0 -8
  404. unit_tests/sources/declarative/schema/source_test/__init__.py +0 -3
  405. unit_tests/sources/declarative/schema/test_default_schema_loader.py +0 -32
  406. unit_tests/sources/declarative/schema/test_inline_schema_loader.py +0 -19
  407. unit_tests/sources/declarative/schema/test_json_file_schema_loader.py +0 -26
  408. unit_tests/sources/declarative/states/__init__.py +0 -3
  409. unit_tests/sources/declarative/stream_slicers/__init__.py +0 -3
  410. unit_tests/sources/declarative/stream_slicers/test_cartesian_product_stream_slicer.py +0 -225
  411. unit_tests/sources/declarative/test_create_partial.py +0 -83
  412. unit_tests/sources/declarative/test_declarative_stream.py +0 -103
  413. unit_tests/sources/declarative/test_manifest_declarative_source.py +0 -1260
  414. unit_tests/sources/declarative/test_types.py +0 -39
  415. unit_tests/sources/declarative/test_yaml_declarative_source.py +0 -148
  416. unit_tests/sources/file_based/__init__.py +0 -0
  417. unit_tests/sources/file_based/availability_strategy/__init__.py +0 -0
  418. unit_tests/sources/file_based/availability_strategy/test_default_file_based_availability_strategy.py +0 -100
  419. unit_tests/sources/file_based/config/__init__.py +0 -0
  420. unit_tests/sources/file_based/config/test_abstract_file_based_spec.py +0 -28
  421. unit_tests/sources/file_based/config/test_csv_format.py +0 -34
  422. unit_tests/sources/file_based/config/test_file_based_stream_config.py +0 -84
  423. unit_tests/sources/file_based/discovery_policy/__init__.py +0 -0
  424. unit_tests/sources/file_based/discovery_policy/test_default_discovery_policy.py +0 -31
  425. unit_tests/sources/file_based/file_types/__init__.py +0 -0
  426. unit_tests/sources/file_based/file_types/test_avro_parser.py +0 -243
  427. unit_tests/sources/file_based/file_types/test_csv_parser.py +0 -546
  428. unit_tests/sources/file_based/file_types/test_jsonl_parser.py +0 -158
  429. unit_tests/sources/file_based/file_types/test_parquet_parser.py +0 -274
  430. unit_tests/sources/file_based/file_types/test_unstructured_parser.py +0 -593
  431. unit_tests/sources/file_based/helpers.py +0 -70
  432. unit_tests/sources/file_based/in_memory_files_source.py +0 -211
  433. unit_tests/sources/file_based/scenarios/__init__.py +0 -0
  434. unit_tests/sources/file_based/scenarios/avro_scenarios.py +0 -744
  435. unit_tests/sources/file_based/scenarios/check_scenarios.py +0 -220
  436. unit_tests/sources/file_based/scenarios/concurrent_incremental_scenarios.py +0 -2844
  437. unit_tests/sources/file_based/scenarios/csv_scenarios.py +0 -3105
  438. unit_tests/sources/file_based/scenarios/file_based_source_builder.py +0 -91
  439. unit_tests/sources/file_based/scenarios/incremental_scenarios.py +0 -1926
  440. unit_tests/sources/file_based/scenarios/jsonl_scenarios.py +0 -930
  441. unit_tests/sources/file_based/scenarios/parquet_scenarios.py +0 -754
  442. unit_tests/sources/file_based/scenarios/scenario_builder.py +0 -234
  443. unit_tests/sources/file_based/scenarios/unstructured_scenarios.py +0 -608
  444. unit_tests/sources/file_based/scenarios/user_input_schema_scenarios.py +0 -746
  445. unit_tests/sources/file_based/scenarios/validation_policy_scenarios.py +0 -726
  446. unit_tests/sources/file_based/stream/__init__.py +0 -0
  447. unit_tests/sources/file_based/stream/concurrent/__init__.py +0 -0
  448. unit_tests/sources/file_based/stream/concurrent/test_adapters.py +0 -362
  449. unit_tests/sources/file_based/stream/concurrent/test_file_based_concurrent_cursor.py +0 -458
  450. unit_tests/sources/file_based/stream/test_default_file_based_cursor.py +0 -310
  451. unit_tests/sources/file_based/stream/test_default_file_based_stream.py +0 -244
  452. unit_tests/sources/file_based/test_file_based_scenarios.py +0 -320
  453. unit_tests/sources/file_based/test_file_based_stream_reader.py +0 -272
  454. unit_tests/sources/file_based/test_scenarios.py +0 -253
  455. unit_tests/sources/file_based/test_schema_helpers.py +0 -346
  456. unit_tests/sources/fixtures/__init__.py +0 -3
  457. unit_tests/sources/fixtures/source_test_fixture.py +0 -153
  458. unit_tests/sources/message/__init__.py +0 -0
  459. unit_tests/sources/message/test_repository.py +0 -153
  460. unit_tests/sources/streams/__init__.py +0 -0
  461. unit_tests/sources/streams/concurrent/__init__.py +0 -3
  462. unit_tests/sources/streams/concurrent/scenarios/__init__.py +0 -3
  463. unit_tests/sources/streams/concurrent/scenarios/incremental_scenarios.py +0 -250
  464. unit_tests/sources/streams/concurrent/scenarios/stream_facade_builder.py +0 -140
  465. unit_tests/sources/streams/concurrent/scenarios/stream_facade_scenarios.py +0 -452
  466. unit_tests/sources/streams/concurrent/scenarios/test_concurrent_scenarios.py +0 -76
  467. unit_tests/sources/streams/concurrent/scenarios/thread_based_concurrent_stream_scenarios.py +0 -418
  468. unit_tests/sources/streams/concurrent/scenarios/thread_based_concurrent_stream_source_builder.py +0 -142
  469. unit_tests/sources/streams/concurrent/scenarios/utils.py +0 -55
  470. unit_tests/sources/streams/concurrent/test_adapters.py +0 -380
  471. unit_tests/sources/streams/concurrent/test_concurrent_read_processor.py +0 -684
  472. unit_tests/sources/streams/concurrent/test_cursor.py +0 -139
  473. unit_tests/sources/streams/concurrent/test_datetime_state_converter.py +0 -369
  474. unit_tests/sources/streams/concurrent/test_default_stream.py +0 -197
  475. unit_tests/sources/streams/concurrent/test_partition_enqueuer.py +0 -90
  476. unit_tests/sources/streams/concurrent/test_partition_reader.py +0 -67
  477. unit_tests/sources/streams/concurrent/test_thread_pool_manager.py +0 -106
  478. unit_tests/sources/streams/http/__init__.py +0 -0
  479. unit_tests/sources/streams/http/auth/__init__.py +0 -0
  480. unit_tests/sources/streams/http/auth/test_auth.py +0 -173
  481. unit_tests/sources/streams/http/requests_native_auth/__init__.py +0 -0
  482. unit_tests/sources/streams/http/requests_native_auth/test_requests_native_auth.py +0 -423
  483. unit_tests/sources/streams/http/test_availability_strategy.py +0 -180
  484. unit_tests/sources/streams/http/test_http.py +0 -635
  485. unit_tests/sources/streams/test_availability_strategy.py +0 -70
  486. unit_tests/sources/streams/test_call_rate.py +0 -300
  487. unit_tests/sources/streams/test_stream_read.py +0 -405
  488. unit_tests/sources/streams/test_streams_core.py +0 -184
  489. unit_tests/sources/test_abstract_source.py +0 -1442
  490. unit_tests/sources/test_concurrent_source.py +0 -112
  491. unit_tests/sources/test_config.py +0 -92
  492. unit_tests/sources/test_connector_state_manager.py +0 -482
  493. unit_tests/sources/test_http_logger.py +0 -252
  494. unit_tests/sources/test_integration_source.py +0 -86
  495. unit_tests/sources/test_source.py +0 -684
  496. unit_tests/sources/test_source_read.py +0 -460
  497. unit_tests/test/__init__.py +0 -0
  498. unit_tests/test/mock_http/__init__.py +0 -0
  499. unit_tests/test/mock_http/test_matcher.py +0 -53
  500. unit_tests/test/mock_http/test_mocker.py +0 -214
  501. unit_tests/test/mock_http/test_request.py +0 -117
  502. unit_tests/test/mock_http/test_response_builder.py +0 -177
  503. unit_tests/test/test_entrypoint_wrapper.py +0 -240
  504. unit_tests/utils/__init__.py +0 -0
  505. unit_tests/utils/test_datetime_format_inferrer.py +0 -60
  506. unit_tests/utils/test_mapping_helpers.py +0 -54
  507. unit_tests/utils/test_message_utils.py +0 -91
  508. unit_tests/utils/test_rate_limiting.py +0 -26
  509. unit_tests/utils/test_schema_inferrer.py +0 -202
  510. unit_tests/utils/test_secret_utils.py +0 -135
  511. unit_tests/utils/test_stream_status_utils.py +0 -61
  512. unit_tests/utils/test_traced_exception.py +0 -107
  513. /airbyte_cdk/sources/{deprecated → declarative/async_job}/__init__.py +0 -0
  514. {source_declarative_manifest → airbyte_cdk/sources/declarative/migrations}/__init__.py +0 -0
  515. {unit_tests/destinations → airbyte_cdk/sql}/__init__.py +0 -0
  516. {unit_tests/singer → airbyte_cdk/sql/_util}/__init__.py +0 -0
  517. {airbyte_cdk-0.72.0.dist-info → airbyte_cdk-6.13.1.dev4106.dist-info}/LICENSE.txt +0 -0
@@ -1,1260 +0,0 @@
1
- #
2
- # Copyright (c) 2023 Airbyte, Inc., all rights reserved.
3
- #
4
-
5
- import json
6
- import logging
7
- import os
8
- import sys
9
- from copy import deepcopy
10
- from typing import Any, List, Mapping
11
- from unittest.mock import call, patch
12
-
13
- import pytest
14
- import requests
15
- import yaml
16
- from airbyte_cdk.models import (
17
- AirbyteLogMessage,
18
- AirbyteMessage,
19
- AirbyteStream,
20
- ConfiguredAirbyteCatalog,
21
- ConfiguredAirbyteStream,
22
- DestinationSyncMode,
23
- Level,
24
- SyncMode,
25
- Type,
26
- )
27
- from airbyte_cdk.sources.declarative.declarative_stream import DeclarativeStream
28
- from airbyte_cdk.sources.declarative.manifest_declarative_source import ManifestDeclarativeSource
29
- from airbyte_cdk.sources.declarative.retrievers.simple_retriever import SimpleRetriever
30
- from jsonschema.exceptions import ValidationError
31
-
32
- logger = logging.getLogger("airbyte")
33
-
34
- EXTERNAL_CONNECTION_SPECIFICATION = {
35
- "type": "object",
36
- "required": ["api_token"],
37
- "additionalProperties": False,
38
- "properties": {"api_token": {"type": "string"}},
39
- }
40
-
41
-
42
- class MockManifestDeclarativeSource(ManifestDeclarativeSource):
43
- """
44
- Mock test class that is needed to monkey patch how we read from various files that make up a declarative source because of how our
45
- tests write configuration files during testing. It is also used to properly namespace where files get written in specific
46
- cases like when we temporarily write files like spec.yaml to the package unit_tests, which is the directory where it will
47
- be read in during the tests.
48
- """
49
-
50
-
51
- class TestManifestDeclarativeSource:
52
- @pytest.fixture
53
- def use_external_yaml_spec(self):
54
- # Our way of resolving the absolute path to root of the airbyte-cdk unit test directory where spec.yaml files should
55
- # be written to (i.e. ~/airbyte/airbyte-cdk/python/unit-tests) because that is where they are read from during testing.
56
- module = sys.modules[__name__]
57
- module_path = os.path.abspath(module.__file__)
58
- test_path = os.path.dirname(module_path)
59
- spec_root = test_path.split("/sources/declarative")[0]
60
-
61
- spec = {"documentationUrl": "https://airbyte.com/#yaml-from-external", "connectionSpecification": EXTERNAL_CONNECTION_SPECIFICATION}
62
-
63
- yaml_path = os.path.join(spec_root, "spec.yaml")
64
- with open(yaml_path, "w") as f:
65
- f.write(yaml.dump(spec))
66
- yield
67
- os.remove(yaml_path)
68
-
69
- def test_valid_manifest(self):
70
- manifest = {
71
- "version": "0.29.3",
72
- "definitions": {},
73
- "streams": [
74
- {
75
- "type": "DeclarativeStream",
76
- "$parameters": {"name": "lists", "primary_key": "id", "url_base": "https://api.sendgrid.com"},
77
- "schema_loader": {
78
- "name": "{{ parameters.stream_name }}",
79
- "file_path": "./source_sendgrid/schemas/{{ parameters.name }}.yaml",
80
- },
81
- "retriever": {
82
- "paginator": {
83
- "type": "DefaultPaginator",
84
- "page_size": 10,
85
- "page_size_option": {"type": "RequestOption", "inject_into": "request_parameter", "field_name": "page_size"},
86
- "page_token_option": {"type": "RequestPath"},
87
- "pagination_strategy": {
88
- "type": "CursorPagination",
89
- "cursor_value": "{{ response._metadata.next }}",
90
- "page_size": 10,
91
- },
92
- },
93
- "requester": {
94
- "path": "/v3/marketing/lists",
95
- "authenticator": {"type": "BearerAuthenticator", "api_token": "{{ config.apikey }}"},
96
- "request_parameters": {"page_size": "{{ 10 }}"},
97
- },
98
- "record_selector": {"extractor": {"field_path": ["result"]}},
99
- },
100
- },
101
- {
102
- "type": "DeclarativeStream",
103
- "$parameters": {"name": "stream_with_custom_requester", "primary_key": "id", "url_base": "https://api.sendgrid.com"},
104
- "schema_loader": {
105
- "name": "{{ parameters.stream_name }}",
106
- "file_path": "./source_sendgrid/schemas/{{ parameters.name }}.yaml",
107
- },
108
- "retriever": {
109
- "paginator": {
110
- "type": "DefaultPaginator",
111
- "page_size": 10,
112
- "page_size_option": {"type": "RequestOption", "inject_into": "request_parameter", "field_name": "page_size"},
113
- "page_token_option": {"type": "RequestPath"},
114
- "pagination_strategy": {
115
- "type": "CursorPagination",
116
- "cursor_value": "{{ response._metadata.next }}",
117
- "page_size": 10,
118
- },
119
- },
120
- "requester": {
121
- "type": "CustomRequester",
122
- "class_name": "unit_tests.sources.declarative.external_component.SampleCustomComponent",
123
- "path": "/v3/marketing/lists",
124
- "custom_request_parameters": {"page_size": 10},
125
- },
126
- "record_selector": {"extractor": {"field_path": ["result"]}},
127
- },
128
- },
129
- ],
130
- "check": {"type": "CheckStream", "stream_names": ["lists"]},
131
- }
132
- source = ManifestDeclarativeSource(source_config=manifest)
133
-
134
- check_stream = source.connection_checker
135
- check_stream.check_connection(source, logging.getLogger(""), {})
136
-
137
- streams = source.streams({})
138
- assert len(streams) == 2
139
- assert isinstance(streams[0], DeclarativeStream)
140
- assert isinstance(streams[1], DeclarativeStream)
141
-
142
- def test_manifest_with_spec(self):
143
- manifest = {
144
- "version": "0.29.3",
145
- "definitions": {
146
- "schema_loader": {
147
- "name": "{{ parameters.stream_name }}",
148
- "file_path": "./source_sendgrid/schemas/{{ parameters.name }}.yaml",
149
- },
150
- "retriever": {
151
- "paginator": {
152
- "type": "DefaultPaginator",
153
- "page_size": 10,
154
- "page_size_option": {"type": "RequestOption", "inject_into": "request_parameter", "field_name": "page_size"},
155
- "page_token_option": {"type": "RequestPath"},
156
- "pagination_strategy": {"type": "CursorPagination", "cursor_value": "{{ response._metadata.next }}"},
157
- },
158
- "requester": {
159
- "path": "/v3/marketing/lists",
160
- "authenticator": {"type": "BearerAuthenticator", "api_token": "{{ config.apikey }}"},
161
- "request_parameters": {"page_size": "{{ 10 }}"},
162
- },
163
- "record_selector": {"extractor": {"field_path": ["result"]}},
164
- },
165
- },
166
- "streams": [
167
- {
168
- "type": "DeclarativeStream",
169
- "$parameters": {"name": "lists", "primary_key": "id", "url_base": "https://api.sendgrid.com"},
170
- "schema_loader": {
171
- "name": "{{ parameters.stream_name }}",
172
- "file_path": "./source_sendgrid/schemas/{{ parameters.name }}.yaml",
173
- },
174
- "retriever": {
175
- "paginator": {
176
- "type": "DefaultPaginator",
177
- "page_size": 10,
178
- "page_size_option": {"type": "RequestOption", "inject_into": "request_parameter", "field_name": "page_size"},
179
- "page_token_option": {"type": "RequestPath"},
180
- "pagination_strategy": {"type": "CursorPagination", "cursor_value": "{{ response._metadata.next }}"},
181
- },
182
- "requester": {
183
- "path": "/v3/marketing/lists",
184
- "authenticator": {"type": "BearerAuthenticator", "api_token": "{{ config.apikey }}"},
185
- "request_parameters": {"page_size": "{{ 10 }}"},
186
- },
187
- "record_selector": {"extractor": {"field_path": ["result"]}},
188
- },
189
- }
190
- ],
191
- "check": {"type": "CheckStream", "stream_names": ["lists"]},
192
- "spec": {
193
- "type": "Spec",
194
- "documentation_url": "https://airbyte.com/#yaml-from-manifest",
195
- "connection_specification": {
196
- "title": "Test Spec",
197
- "type": "object",
198
- "required": ["api_key"],
199
- "additionalProperties": False,
200
- "properties": {
201
- "api_key": {"type": "string", "airbyte_secret": True, "title": "API Key", "description": "Test API Key", "order": 0}
202
- },
203
- },
204
- },
205
- }
206
- source = ManifestDeclarativeSource(source_config=manifest)
207
- connector_specification = source.spec(logger)
208
- assert connector_specification is not None
209
- assert connector_specification.documentationUrl == "https://airbyte.com/#yaml-from-manifest"
210
- assert connector_specification.connectionSpecification["title"] == "Test Spec"
211
- assert connector_specification.connectionSpecification["required"][0] == "api_key"
212
- assert connector_specification.connectionSpecification["additionalProperties"] is False
213
- assert connector_specification.connectionSpecification["properties"]["api_key"] == {
214
- "type": "string",
215
- "airbyte_secret": True,
216
- "title": "API Key",
217
- "description": "Test API Key",
218
- "order": 0,
219
- }
220
-
221
- def test_manifest_with_external_spec(self, use_external_yaml_spec):
222
- manifest = {
223
- "version": "0.29.3",
224
- "definitions": {
225
- "schema_loader": {
226
- "name": "{{ parameters.stream_name }}",
227
- "file_path": "./source_sendgrid/schemas/{{ parameters.name }}.yaml",
228
- },
229
- "retriever": {
230
- "paginator": {
231
- "type": "DefaultPaginator",
232
- "page_size": 10,
233
- "page_size_option": {"type": "RequestOption", "inject_into": "request_parameter", "field_name": "page_size"},
234
- "page_token_option": {"type": "RequestPath"},
235
- "pagination_strategy": {"type": "CursorPagination", "cursor_value": "{{ response._metadata.next }}"},
236
- },
237
- "requester": {
238
- "path": "/v3/marketing/lists",
239
- "authenticator": {"type": "BearerAuthenticator", "api_token": "{{ config.apikey }}"},
240
- "request_parameters": {"page_size": "{{ 10 }}"},
241
- },
242
- "record_selector": {"extractor": {"field_path": ["result"]}},
243
- },
244
- },
245
- "streams": [
246
- {
247
- "type": "DeclarativeStream",
248
- "$parameters": {"name": "lists", "primary_key": "id", "url_base": "https://api.sendgrid.com"},
249
- "schema_loader": {
250
- "name": "{{ parameters.stream_name }}",
251
- "file_path": "./source_sendgrid/schemas/{{ parameters.name }}.yaml",
252
- },
253
- "retriever": {
254
- "paginator": {
255
- "type": "DefaultPaginator",
256
- "page_size": 10,
257
- "page_size_option": {"type": "RequestOption", "inject_into": "request_parameter", "field_name": "page_size"},
258
- "page_token_option": {"type": "RequestPath"},
259
- "pagination_strategy": {"type": "CursorPagination", "cursor_value": "{{ response._metadata.next }}"},
260
- },
261
- "requester": {
262
- "path": "/v3/marketing/lists",
263
- "authenticator": {"type": "BearerAuthenticator", "api_token": "{{ config.apikey }}"},
264
- "request_parameters": {"page_size": "{{ 10 }}"},
265
- },
266
- "record_selector": {"extractor": {"field_path": ["result"]}},
267
- },
268
- }
269
- ],
270
- "check": {"type": "CheckStream", "stream_names": ["lists"]},
271
- }
272
- source = MockManifestDeclarativeSource(source_config=manifest)
273
-
274
- connector_specification = source.spec(logger)
275
-
276
- assert connector_specification.documentationUrl == "https://airbyte.com/#yaml-from-external"
277
- assert connector_specification.connectionSpecification == EXTERNAL_CONNECTION_SPECIFICATION
278
-
279
- def test_source_is_not_created_if_toplevel_fields_are_unknown(self):
280
- manifest = {
281
- "version": "0.29.3",
282
- "definitions": {
283
- "schema_loader": {
284
- "name": "{{ parameters.stream_name }}",
285
- "file_path": "./source_sendgrid/schemas/{{ parameters.name }}.yaml",
286
- },
287
- "retriever": {
288
- "paginator": {
289
- "type": "DefaultPaginator",
290
- "page_size": 10,
291
- "page_size_option": {"type": "RequestOption", "inject_into": "request_parameter", "field_name": "page_size"},
292
- "page_token_option": {"type": "RequestPath"},
293
- "pagination_strategy": {"type": "CursorPagination", "cursor_value": "{{ response._metadata.next }}"},
294
- },
295
- "requester": {
296
- "path": "/v3/marketing/lists",
297
- "authenticator": {"type": "BearerAuthenticator", "api_token": "{{ config.apikey }}"},
298
- "request_parameters": {"page_size": 10},
299
- },
300
- "record_selector": {"extractor": {"field_path": ["result"]}},
301
- },
302
- },
303
- "streams": [
304
- {
305
- "type": "DeclarativeStream",
306
- "$parameters": {"name": "lists", "primary_key": "id", "url_base": "https://api.sendgrid.com"},
307
- "schema_loader": {
308
- "name": "{{ parameters.stream_name }}",
309
- "file_path": "./source_sendgrid/schemas/{{ parameters.name }}.yaml",
310
- },
311
- "retriever": {
312
- "paginator": {
313
- "type": "DefaultPaginator",
314
- "page_size": 10,
315
- "page_size_option": {"type": "RequestOption", "inject_into": "request_parameter", "field_name": "page_size"},
316
- "page_token_option": {"type": "RequestPath"},
317
- "pagination_strategy": {"type": "CursorPagination", "cursor_value": "{{ response._metadata.next }}"},
318
- },
319
- "requester": {
320
- "path": "/v3/marketing/lists",
321
- "authenticator": {"type": "BearerAuthenticator", "api_token": "{{ config.apikey }}"},
322
- "request_parameters": {"page_size": 10},
323
- },
324
- "record_selector": {"extractor": {"field_path": ["result"]}},
325
- },
326
- }
327
- ],
328
- "check": {"type": "CheckStream", "stream_names": ["lists"]},
329
- "not_a_valid_field": "error",
330
- }
331
- with pytest.raises(ValidationError):
332
- ManifestDeclarativeSource(source_config=manifest)
333
-
334
- def test_source_missing_checker_fails_validation(self):
335
- manifest = {
336
- "version": "0.29.3",
337
- "definitions": {
338
- "schema_loader": {
339
- "name": "{{ parameters.stream_name }}",
340
- "file_path": "./source_sendgrid/schemas/{{ parameters.name }}.yaml",
341
- },
342
- "retriever": {
343
- "paginator": {
344
- "type": "DefaultPaginator",
345
- "page_size": 10,
346
- "page_size_option": {"type": "RequestOption", "inject_into": "request_parameter", "field_name": "page_size"},
347
- "page_token_option": {"type": "RequestPath"},
348
- "pagination_strategy": {"type": "CursorPagination", "cursor_value": "{{ response._metadata.next }}"},
349
- },
350
- "requester": {
351
- "path": "/v3/marketing/lists",
352
- "authenticator": {"type": "BearerAuthenticator", "api_token": "{{ config.apikey }}"},
353
- "request_parameters": {"page_size": 10},
354
- },
355
- "record_selector": {"extractor": {"field_path": ["result"]}},
356
- },
357
- },
358
- "streams": [
359
- {
360
- "type": "DeclarativeStream",
361
- "$parameters": {"name": "lists", "primary_key": "id", "url_base": "https://api.sendgrid.com"},
362
- "schema_loader": {
363
- "name": "{{ parameters.stream_name }}",
364
- "file_path": "./source_sendgrid/schemas/{{ parameters.name }}.yaml",
365
- },
366
- "retriever": {
367
- "paginator": {
368
- "type": "DefaultPaginator",
369
- "page_size": 10,
370
- "page_size_option": {"type": "RequestOption", "inject_into": "request_parameter", "field_name": "page_size"},
371
- "page_token_option": {"type": "RequestPath"},
372
- "pagination_strategy": {"type": "CursorPagination", "cursor_value": "{{ response._metadata.next }}"},
373
- },
374
- "requester": {
375
- "path": "/v3/marketing/lists",
376
- "authenticator": {"type": "BearerAuthenticator", "api_token": "{{ config.apikey }}"},
377
- "request_parameters": {"page_size": 10},
378
- },
379
- "record_selector": {"extractor": {"field_path": ["result"]}},
380
- },
381
- }
382
- ],
383
- "check": {"type": "CheckStream"},
384
- }
385
- with pytest.raises(ValidationError):
386
- ManifestDeclarativeSource(source_config=manifest)
387
-
388
- def test_source_with_missing_streams_fails(self):
389
- manifest = {"version": "0.29.3", "definitions": None, "check": {"type": "CheckStream", "stream_names": ["lists"]}}
390
- with pytest.raises(ValidationError):
391
- ManifestDeclarativeSource(source_config=manifest)
392
-
393
- def test_source_with_missing_version_fails(self):
394
- manifest = {
395
- "definitions": {
396
- "schema_loader": {
397
- "name": "{{ parameters.stream_name }}",
398
- "file_path": "./source_sendgrid/schemas/{{ parameters.name }}.yaml",
399
- },
400
- "retriever": {
401
- "paginator": {
402
- "type": "DefaultPaginator",
403
- "page_size": 10,
404
- "page_size_option": {"type": "RequestOption", "inject_into": "request_parameter", "field_name": "page_size"},
405
- "page_token_option": {"type": "RequestPath"},
406
- "pagination_strategy": {"type": "CursorPagination", "cursor_value": "{{ response._metadata.next }}"},
407
- },
408
- "requester": {
409
- "path": "/v3/marketing/lists",
410
- "authenticator": {"type": "BearerAuthenticator", "api_token": "{{ config.apikey }}"},
411
- "request_parameters": {"page_size": 10},
412
- },
413
- "record_selector": {"extractor": {"field_path": ["result"]}},
414
- },
415
- },
416
- "streams": [
417
- {
418
- "type": "DeclarativeStream",
419
- "$parameters": {"name": "lists", "primary_key": "id", "url_base": "https://api.sendgrid.com"},
420
- "schema_loader": {
421
- "name": "{{ parameters.stream_name }}",
422
- "file_path": "./source_sendgrid/schemas/{{ parameters.name }}.yaml",
423
- },
424
- "retriever": {
425
- "paginator": {
426
- "type": "DefaultPaginator",
427
- "page_size": 10,
428
- "page_size_option": {"type": "RequestOption", "inject_into": "request_parameter", "field_name": "page_size"},
429
- "page_token_option": {"type": "RequestPath"},
430
- "pagination_strategy": {"type": "CursorPagination", "cursor_value": "{{ response._metadata.next }}"},
431
- },
432
- "requester": {
433
- "path": "/v3/marketing/lists",
434
- "authenticator": {"type": "BearerAuthenticator", "api_token": "{{ config.apikey }}"},
435
- "request_parameters": {"page_size": 10},
436
- },
437
- "record_selector": {"extractor": {"field_path": ["result"]}},
438
- },
439
- }
440
- ],
441
- "check": {"type": "CheckStream", "stream_names": ["lists"]},
442
- }
443
- with pytest.raises(ValidationError):
444
- ManifestDeclarativeSource(source_config=manifest)
445
-
446
- @pytest.mark.parametrize(
447
- "cdk_version, manifest_version, expected_error",
448
- [
449
- pytest.param("0.35.0", "0.30.0", None, id="manifest_version_less_than_cdk_package_should_run"),
450
- pytest.param("1.5.0", "0.29.0", None, id="manifest_version_less_than_cdk_major_package_should_run"),
451
- pytest.param("0.29.0", "0.29.0", None, id="manifest_version_matching_cdk_package_should_run"),
452
- pytest.param(
453
- "0.29.0",
454
- "0.25.0",
455
- ValidationError,
456
- id="manifest_version_before_beta_that_uses_the_beta_0.29.0_cdk_package_should_throw_error",
457
- ),
458
- pytest.param(
459
- "1.5.0",
460
- "0.25.0",
461
- ValidationError,
462
- id="manifest_version_before_beta_that_uses_package_later_major_version_than_beta_0.29.0_cdk_package_should_throw_error",
463
- ),
464
- pytest.param("0.34.0", "0.35.0", ValidationError, id="manifest_version_greater_than_cdk_package_should_throw_error"),
465
- pytest.param("0.29.0", "-1.5.0", ValidationError, id="manifest_version_has_invalid_major_format"),
466
- pytest.param("0.29.0", "0.invalid.0", ValidationError, id="manifest_version_has_invalid_minor_format"),
467
- pytest.param("0.29.0", "0.29.0.1", ValidationError, id="manifest_version_has_extra_version_parts"),
468
- pytest.param("0.29.0", "5.0", ValidationError, id="manifest_version_has_too_few_version_parts"),
469
- pytest.param("0.29.0:dev", "0.29.0", ValidationError, id="manifest_version_has_extra_release"),
470
- ],
471
- )
472
- @patch("importlib.metadata.version")
473
- def test_manifest_versions(self, version, cdk_version, manifest_version, expected_error):
474
- # Used to mock the metadata.version() for test scenarios which normally returns the actual version of the airbyte-cdk package
475
- version.return_value = cdk_version
476
-
477
- manifest = {
478
- "version": manifest_version,
479
- "definitions": {},
480
- "streams": [
481
- {
482
- "type": "DeclarativeStream",
483
- "$parameters": {"name": "lists", "primary_key": "id", "url_base": "https://api.sendgrid.com"},
484
- "schema_loader": {
485
- "name": "{{ parameters.stream_name }}",
486
- "file_path": "./source_sendgrid/schemas/{{ parameters.name }}.yaml",
487
- },
488
- "retriever": {
489
- "paginator": {
490
- "type": "DefaultPaginator",
491
- "page_size": 10,
492
- "page_size_option": {"type": "RequestOption", "inject_into": "request_parameter", "field_name": "page_size"},
493
- "page_token_option": {"type": "RequestPath"},
494
- "pagination_strategy": {
495
- "type": "CursorPagination",
496
- "cursor_value": "{{ response._metadata.next }}",
497
- "page_size": 10,
498
- },
499
- },
500
- "requester": {
501
- "path": "/v3/marketing/lists",
502
- "authenticator": {"type": "BearerAuthenticator", "api_token": "{{ config.apikey }}"},
503
- "request_parameters": {"page_size": "{{ 10 }}"},
504
- },
505
- "record_selector": {"extractor": {"field_path": ["result"]}},
506
- },
507
- },
508
- {
509
- "type": "DeclarativeStream",
510
- "$parameters": {"name": "stream_with_custom_requester", "primary_key": "id", "url_base": "https://api.sendgrid.com"},
511
- "schema_loader": {
512
- "name": "{{ parameters.stream_name }}",
513
- "file_path": "./source_sendgrid/schemas/{{ parameters.name }}.yaml",
514
- },
515
- "retriever": {
516
- "paginator": {
517
- "type": "DefaultPaginator",
518
- "page_size": 10,
519
- "page_size_option": {"type": "RequestOption", "inject_into": "request_parameter", "field_name": "page_size"},
520
- "page_token_option": {"type": "RequestPath"},
521
- "pagination_strategy": {
522
- "type": "CursorPagination",
523
- "cursor_value": "{{ response._metadata.next }}",
524
- "page_size": 10,
525
- },
526
- },
527
- "requester": {
528
- "type": "CustomRequester",
529
- "class_name": "unit_tests.sources.declarative.external_component.SampleCustomComponent",
530
- "path": "/v3/marketing/lists",
531
- "custom_request_parameters": {"page_size": 10},
532
- },
533
- "record_selector": {"extractor": {"field_path": ["result"]}},
534
- },
535
- },
536
- ],
537
- "check": {"type": "CheckStream", "stream_names": ["lists"]},
538
- }
539
- if expected_error:
540
- with pytest.raises(expected_error):
541
- ManifestDeclarativeSource(source_config=manifest)
542
- else:
543
- ManifestDeclarativeSource(source_config=manifest)
544
-
545
- def test_source_with_invalid_stream_config_fails_validation(self):
546
- manifest = {
547
- "version": "0.29.3",
548
- "definitions": {
549
- "schema_loader": {
550
- "name": "{{ parameters.stream_name }}",
551
- "file_path": "./source_sendgrid/schemas/{{ parameters.name }}.yaml",
552
- }
553
- },
554
- "streams": [
555
- {
556
- "type": "DeclarativeStream",
557
- "$parameters": {"name": "lists", "primary_key": "id", "url_base": "https://api.sendgrid.com"},
558
- "schema_loader": {
559
- "name": "{{ parameters.stream_name }}",
560
- "file_path": "./source_sendgrid/schemas/{{ parameters.name }}.yaml",
561
- },
562
- }
563
- ],
564
- "check": {"type": "CheckStream", "stream_names": ["lists"]},
565
- }
566
- with pytest.raises(ValidationError):
567
- ManifestDeclarativeSource(source_config=manifest)
568
-
569
- def test_source_with_no_external_spec_and_no_in_yaml_spec_fails(self):
570
- manifest = {
571
- "version": "0.29.3",
572
- "definitions": {
573
- "schema_loader": {
574
- "name": "{{ parameters.stream_name }}",
575
- "file_path": "./source_sendgrid/schemas/{{ parameters.name }}.yaml",
576
- },
577
- "retriever": {
578
- "paginator": {
579
- "type": "DefaultPaginator",
580
- "page_size": 10,
581
- "page_size_option": {"type": "RequestOption", "inject_into": "request_parameter", "field_name": "page_size"},
582
- "page_token_option": {"type": "RequestPath"},
583
- "pagination_strategy": {"type": "CursorPagination", "cursor_value": "{{ response._metadata.next }}"},
584
- },
585
- "requester": {
586
- "path": "/v3/marketing/lists",
587
- "authenticator": {"type": "BearerAuthenticator", "api_token": "{{ config.apikey }}"},
588
- "request_parameters": {"page_size": "{{ 10 }}"},
589
- },
590
- "record_selector": {"extractor": {"field_path": ["result"]}},
591
- },
592
- },
593
- "streams": [
594
- {
595
- "type": "DeclarativeStream",
596
- "$parameters": {"name": "lists", "primary_key": "id", "url_base": "https://api.sendgrid.com"},
597
- "schema_loader": {
598
- "name": "{{ parameters.stream_name }}",
599
- "file_path": "./source_sendgrid/schemas/{{ parameters.name }}.yaml",
600
- },
601
- "retriever": {
602
- "paginator": {
603
- "type": "DefaultPaginator",
604
- "page_size": 10,
605
- "page_size_option": {"type": "RequestOption", "inject_into": "request_parameter", "field_name": "page_size"},
606
- "page_token_option": {"type": "RequestPath"},
607
- "pagination_strategy": {"type": "CursorPagination", "cursor_value": "{{ response._metadata.next }}"},
608
- },
609
- "requester": {
610
- "path": "/v3/marketing/lists",
611
- "authenticator": {"type": "BearerAuthenticator", "api_token": "{{ config.apikey }}"},
612
- "request_parameters": {"page_size": "{{ 10 }}"},
613
- },
614
- "record_selector": {"extractor": {"field_path": ["result"]}},
615
- },
616
- }
617
- ],
618
- "check": {"type": "CheckStream", "stream_names": ["lists"]},
619
- }
620
- source = ManifestDeclarativeSource(source_config=manifest)
621
-
622
- # We expect to fail here because we have not created a temporary spec.yaml file
623
- with pytest.raises(FileNotFoundError):
624
- source.spec(logger)
625
-
626
- def test_manifest_without_at_least_one_stream(self):
627
- manifest = {
628
- "version": "0.29.3",
629
- "definitions": {
630
- "schema_loader": {
631
- "name": "{{ parameters.stream_name }}",
632
- "file_path": "./source_sendgrid/schemas/{{ parameters.name }}.yaml",
633
- },
634
- "retriever": {
635
- "paginator": {
636
- "type": "DefaultPaginator",
637
- "page_size": 10,
638
- "page_size_option": {"type": "RequestOption", "inject_into": "request_parameter", "field_name": "page_size"},
639
- "page_token_option": {"type": "RequestPath"},
640
- "pagination_strategy": {"type": "CursorPagination", "cursor_value": "{{ response._metadata.next }}"},
641
- },
642
- "requester": {
643
- "path": "/v3/marketing/lists",
644
- "authenticator": {"type": "BearerAuthenticator", "api_token": "{{ config.apikey }}"},
645
- "request_parameters": {"page_size": 10},
646
- },
647
- "record_selector": {"extractor": {"field_path": ["result"]}},
648
- },
649
- },
650
- "streams": [],
651
- "check": {"type": "CheckStream", "stream_names": ["lists"]},
652
- }
653
- with pytest.raises(ValidationError):
654
- ManifestDeclarativeSource(source_config=manifest)
655
-
656
- @patch("airbyte_cdk.sources.declarative.declarative_source.DeclarativeSource.read")
657
- def test_given_debug_when_read_then_set_log_level(self, declarative_source_read):
658
- any_valid_manifest = {
659
- "version": "0.29.3",
660
- "definitions": {
661
- "schema_loader": {
662
- "name": "{{ parameters.stream_name }}",
663
- "file_path": "./source_sendgrid/schemas/{{ parameters.name }}.yaml",
664
- },
665
- "retriever": {
666
- "paginator": {
667
- "type": "DefaultPaginator",
668
- "page_size": 10,
669
- "page_size_option": {"type": "RequestOption", "inject_into": "request_parameter", "field_name": "page_size"},
670
- "page_token_option": {"type": "RequestPath"},
671
- "pagination_strategy": {"type": "CursorPagination", "cursor_value": "{{ response._metadata.next }}"},
672
- },
673
- "requester": {
674
- "path": "/v3/marketing/lists",
675
- "authenticator": {"type": "BearerAuthenticator", "api_token": "{{ config.apikey }}"},
676
- "request_parameters": {"page_size": "10"},
677
- },
678
- "record_selector": {"extractor": {"field_path": ["result"]}},
679
- },
680
- },
681
- "streams": [
682
- {
683
- "type": "DeclarativeStream",
684
- "$parameters": {"name": "lists", "primary_key": "id", "url_base": "https://api.sendgrid.com"},
685
- "schema_loader": {
686
- "name": "{{ parameters.stream_name }}",
687
- "file_path": "./source_sendgrid/schemas/{{ parameters.name }}.yaml",
688
- },
689
- "retriever": {
690
- "paginator": {
691
- "type": "DefaultPaginator",
692
- "page_size": 10,
693
- "page_size_option": {"type": "RequestOption", "inject_into": "request_parameter", "field_name": "page_size"},
694
- "page_token_option": {"type": "RequestPath"},
695
- "pagination_strategy": {
696
- "type": "CursorPagination",
697
- "cursor_value": "{{ response._metadata.next }}",
698
- "page_size": 10,
699
- },
700
- },
701
- "requester": {
702
- "path": "/v3/marketing/lists",
703
- "authenticator": {"type": "BearerAuthenticator", "api_token": "{{ config.apikey }}"},
704
- "request_parameters": {"page_size": "{{ 10 }}"},
705
- },
706
- "record_selector": {"extractor": {"field_path": ["result"]}},
707
- },
708
- },
709
- {
710
- "type": "DeclarativeStream",
711
- "$parameters": {"name": "stream_with_custom_requester", "primary_key": "id", "url_base": "https://api.sendgrid.com"},
712
- "schema_loader": {
713
- "name": "{{ parameters.stream_name }}",
714
- "file_path": "./source_sendgrid/schemas/{{ parameters.name }}.yaml",
715
- },
716
- "retriever": {
717
- "paginator": {
718
- "type": "DefaultPaginator",
719
- "page_size": 10,
720
- "page_size_option": {"type": "RequestOption", "inject_into": "request_parameter", "field_name": "page_size"},
721
- "page_token_option": {"type": "RequestPath"},
722
- "pagination_strategy": {
723
- "type": "CursorPagination",
724
- "cursor_value": "{{ response._metadata.next }}",
725
- "page_size": 10,
726
- },
727
- },
728
- "requester": {
729
- "type": "CustomRequester",
730
- "class_name": "unit_tests.sources.declarative.external_component.SampleCustomComponent",
731
- "path": "/v3/marketing/lists",
732
- "custom_request_parameters": {"page_size": 10},
733
- },
734
- "record_selector": {"extractor": {"field_path": ["result"]}},
735
- },
736
- },
737
- ],
738
- "check": {"type": "CheckStream", "stream_names": ["lists"]},
739
- }
740
- source = ManifestDeclarativeSource(source_config=any_valid_manifest, debug=True)
741
-
742
- debug_logger = logging.getLogger("logger.debug")
743
- list(source.read(debug_logger, {}, {}, {}))
744
-
745
- assert debug_logger.isEnabledFor(logging.DEBUG)
746
-
747
-
748
- def request_log_message(request: dict) -> AirbyteMessage:
749
- return AirbyteMessage(type=Type.LOG, log=AirbyteLogMessage(level=Level.INFO, message=f"request:{json.dumps(request)}"))
750
-
751
-
752
- def response_log_message(response: dict) -> AirbyteMessage:
753
- return AirbyteMessage(type=Type.LOG, log=AirbyteLogMessage(level=Level.INFO, message=f"response:{json.dumps(response)}"))
754
-
755
-
756
- def _create_request():
757
- url = "https://example.com/api"
758
- headers = {"Content-Type": "application/json"}
759
- return requests.Request("POST", url, headers=headers, json={"key": "value"}).prepare()
760
-
761
-
762
- def _create_response(body):
763
- response = requests.Response()
764
- response.status_code = 200
765
- response._content = bytes(json.dumps(body), "utf-8")
766
- response.headers["Content-Type"] = "application/json"
767
- return response
768
-
769
-
770
- def _create_page(response_body):
771
- response = _create_response(response_body)
772
- response.request = _create_request()
773
- return response
774
-
775
-
776
- @pytest.mark.parametrize(
777
- "test_name, manifest, pages, expected_records, expected_calls",
778
- [
779
- (
780
- "test_read_manifest_no_pagination_no_partitions",
781
- {
782
- "version": "0.34.2",
783
- "type": "DeclarativeSource",
784
- "check": {"type": "CheckStream", "stream_names": ["Rates"]},
785
- "streams": [
786
- {
787
- "type": "DeclarativeStream",
788
- "name": "Rates",
789
- "primary_key": [],
790
- "schema_loader": {
791
- "type": "InlineSchemaLoader",
792
- "schema": {
793
- "$schema": "http://json-schema.org/schema#",
794
- "properties": {
795
- "ABC": {"type": "number"},
796
- "AED": {"type": "number"},
797
- },
798
- "type": "object",
799
- },
800
- },
801
- "retriever": {
802
- "type": "SimpleRetriever",
803
- "requester": {
804
- "type": "HttpRequester",
805
- "url_base": "https://api.apilayer.com",
806
- "path": "/exchangerates_data/latest",
807
- "http_method": "GET",
808
- "request_parameters": {},
809
- "request_headers": {},
810
- "request_body_json": {},
811
- "authenticator": {
812
- "type": "ApiKeyAuthenticator",
813
- "header": "apikey",
814
- "api_token": "{{ config['api_key'] }}",
815
- },
816
- },
817
- "record_selector": {"type": "RecordSelector", "extractor": {"type": "DpathExtractor", "field_path": ["rates"]}},
818
- "paginator": {"type": "NoPagination"},
819
- },
820
- }
821
- ],
822
- "spec": {
823
- "connection_specification": {
824
- "$schema": "http://json-schema.org/draft-07/schema#",
825
- "type": "object",
826
- "required": ["api_key"],
827
- "properties": {"api_key": {"type": "string", "title": "API Key", "airbyte_secret": True}},
828
- "additionalProperties": True,
829
- },
830
- "documentation_url": "https://example.org",
831
- "type": "Spec",
832
- },
833
- },
834
- (
835
- _create_page({"rates": [{"ABC": 0}, {"AED": 1}], "_metadata": {"next": "next"}}),
836
- _create_page({"rates": [{"USD": 2}], "_metadata": {"next": "next"}}),
837
- )
838
- * 10,
839
- [{"ABC": 0}, {"AED": 1}],
840
- [call({}, {}, None)],
841
- ),
842
- (
843
- "test_read_manifest_with_added_fields",
844
- {
845
- "version": "0.34.2",
846
- "type": "DeclarativeSource",
847
- "check": {"type": "CheckStream", "stream_names": ["Rates"]},
848
- "streams": [
849
- {
850
- "type": "DeclarativeStream",
851
- "name": "Rates",
852
- "primary_key": [],
853
- "schema_loader": {
854
- "type": "InlineSchemaLoader",
855
- "schema": {
856
- "$schema": "http://json-schema.org/schema#",
857
- "properties": {
858
- "ABC": {"type": "number"},
859
- "AED": {"type": "number"},
860
- },
861
- "type": "object",
862
- },
863
- },
864
- "transformations": [
865
- {
866
- "type": "AddFields",
867
- "fields": [{"type": "AddedFieldDefinition", "path": ["added_field_key"], "value": "added_field_value"}],
868
- }
869
- ],
870
- "retriever": {
871
- "type": "SimpleRetriever",
872
- "requester": {
873
- "type": "HttpRequester",
874
- "url_base": "https://api.apilayer.com",
875
- "path": "/exchangerates_data/latest",
876
- "http_method": "GET",
877
- "request_parameters": {},
878
- "request_headers": {},
879
- "request_body_json": {},
880
- "authenticator": {
881
- "type": "ApiKeyAuthenticator",
882
- "header": "apikey",
883
- "api_token": "{{ config['api_key'] }}",
884
- },
885
- },
886
- "record_selector": {"type": "RecordSelector", "extractor": {"type": "DpathExtractor", "field_path": ["rates"]}},
887
- "paginator": {"type": "NoPagination"},
888
- },
889
- }
890
- ],
891
- "spec": {
892
- "connection_specification": {
893
- "$schema": "http://json-schema.org/draft-07/schema#",
894
- "type": "object",
895
- "required": ["api_key"],
896
- "properties": {"api_key": {"type": "string", "title": "API Key", "airbyte_secret": True}},
897
- "additionalProperties": True,
898
- },
899
- "documentation_url": "https://example.org",
900
- "type": "Spec",
901
- },
902
- },
903
- (
904
- _create_page({"rates": [{"ABC": 0}, {"AED": 1}], "_metadata": {"next": "next"}}),
905
- _create_page({"rates": [{"USD": 2}], "_metadata": {"next": "next"}}),
906
- )
907
- * 10,
908
- [{"ABC": 0, "added_field_key": "added_field_value"}, {"AED": 1, "added_field_key": "added_field_value"}],
909
- [call({}, {}, None)],
910
- ),
911
- (
912
- "test_read_with_pagination_no_partitions",
913
- {
914
- "version": "0.34.2",
915
- "type": "DeclarativeSource",
916
- "check": {"type": "CheckStream", "stream_names": ["Rates"]},
917
- "streams": [
918
- {
919
- "type": "DeclarativeStream",
920
- "name": "Rates",
921
- "primary_key": [],
922
- "schema_loader": {
923
- "type": "InlineSchemaLoader",
924
- "schema": {
925
- "$schema": "http://json-schema.org/schema#",
926
- "properties": {
927
- "ABC": {"type": "number"},
928
- "AED": {"type": "number"},
929
- "USD": {"type": "number"},
930
- },
931
- "type": "object",
932
- },
933
- },
934
- "retriever": {
935
- "type": "SimpleRetriever",
936
- "requester": {
937
- "type": "HttpRequester",
938
- "url_base": "https://api.apilayer.com",
939
- "path": "/exchangerates_data/latest",
940
- "http_method": "GET",
941
- "request_parameters": {},
942
- "request_headers": {},
943
- "request_body_json": {},
944
- "authenticator": {
945
- "type": "ApiKeyAuthenticator",
946
- "header": "apikey",
947
- "api_token": "{{ config['api_key'] }}",
948
- },
949
- },
950
- "record_selector": {"type": "RecordSelector", "extractor": {"type": "DpathExtractor", "field_path": ["rates"]}},
951
- "paginator": {
952
- "type": "DefaultPaginator",
953
- "page_size": 2,
954
- "page_size_option": {"inject_into": "request_parameter", "field_name": "page_size"},
955
- "page_token_option": {"inject_into": "path", "type": "RequestPath"},
956
- "pagination_strategy": {
957
- "type": "CursorPagination",
958
- "cursor_value": "{{ response._metadata.next }}",
959
- "page_size": 2,
960
- },
961
- },
962
- },
963
- }
964
- ],
965
- "spec": {
966
- "connection_specification": {
967
- "$schema": "http://json-schema.org/draft-07/schema#",
968
- "type": "object",
969
- "required": ["api_key"],
970
- "properties": {"api_key": {"type": "string", "title": "API Key", "airbyte_secret": True}},
971
- "additionalProperties": True,
972
- },
973
- "documentation_url": "https://example.org",
974
- "type": "Spec",
975
- },
976
- },
977
- (
978
- _create_page({"rates": [{"ABC": 0}, {"AED": 1}], "_metadata": {"next": "next"}}),
979
- _create_page({"rates": [{"USD": 2}], "_metadata": {}}),
980
- )
981
- * 10,
982
- [{"ABC": 0}, {"AED": 1}, {"USD": 2}],
983
- [call({}, {}, None), call({}, {}, {"next_page_token": "next"})],
984
- ),
985
- (
986
- "test_no_pagination_with_partition_router",
987
- {
988
- "version": "0.34.2",
989
- "type": "DeclarativeSource",
990
- "check": {"type": "CheckStream", "stream_names": ["Rates"]},
991
- "streams": [
992
- {
993
- "type": "DeclarativeStream",
994
- "name": "Rates",
995
- "primary_key": [],
996
- "schema_loader": {
997
- "type": "InlineSchemaLoader",
998
- "schema": {
999
- "$schema": "http://json-schema.org/schema#",
1000
- "properties": {"ABC": {"type": "number"}, "AED": {"type": "number"}, "partition": {"type": "number"}},
1001
- "type": "object",
1002
- },
1003
- },
1004
- "retriever": {
1005
- "type": "SimpleRetriever",
1006
- "requester": {
1007
- "type": "HttpRequester",
1008
- "url_base": "https://api.apilayer.com",
1009
- "path": "/exchangerates_data/latest",
1010
- "http_method": "GET",
1011
- "request_parameters": {},
1012
- "request_headers": {},
1013
- "request_body_json": {},
1014
- "authenticator": {
1015
- "type": "ApiKeyAuthenticator",
1016
- "header": "apikey",
1017
- "api_token": "{{ config['api_key'] }}",
1018
- },
1019
- },
1020
- "partition_router": {"type": "ListPartitionRouter", "values": ["0", "1"], "cursor_field": "partition"},
1021
- "record_selector": {"type": "RecordSelector", "extractor": {"type": "DpathExtractor", "field_path": ["rates"]}},
1022
- "paginator": {"type": "NoPagination"},
1023
- },
1024
- }
1025
- ],
1026
- "spec": {
1027
- "connection_specification": {
1028
- "$schema": "http://json-schema.org/draft-07/schema#",
1029
- "type": "object",
1030
- "required": ["api_key"],
1031
- "properties": {"api_key": {"type": "string", "title": "API Key", "airbyte_secret": True}},
1032
- "additionalProperties": True,
1033
- },
1034
- "documentation_url": "https://example.org",
1035
- "type": "Spec",
1036
- },
1037
- },
1038
- (
1039
- _create_page({"rates": [{"ABC": 0, "partition": 0}, {"AED": 1, "partition": 0}], "_metadata": {"next": "next"}}),
1040
- _create_page({"rates": [{"ABC": 2, "partition": 1}], "_metadata": {"next": "next"}}),
1041
- ),
1042
- [{"ABC": 0, "partition": 0}, {"AED": 1, "partition": 0}, {"ABC": 2, "partition": 1}],
1043
- [call({}, {"partition": "0"}, None), call({}, {"partition": "1"}, None)],
1044
- ),
1045
- (
1046
- "test_with_pagination_and_partition_router",
1047
- {
1048
- "version": "0.34.2",
1049
- "type": "DeclarativeSource",
1050
- "check": {"type": "CheckStream", "stream_names": ["Rates"]},
1051
- "streams": [
1052
- {
1053
- "type": "DeclarativeStream",
1054
- "name": "Rates",
1055
- "primary_key": [],
1056
- "schema_loader": {
1057
- "type": "InlineSchemaLoader",
1058
- "schema": {
1059
- "$schema": "http://json-schema.org/schema#",
1060
- "properties": {"ABC": {"type": "number"}, "AED": {"type": "number"}, "partition": {"type": "number"}},
1061
- "type": "object",
1062
- },
1063
- },
1064
- "retriever": {
1065
- "type": "SimpleRetriever",
1066
- "requester": {
1067
- "type": "HttpRequester",
1068
- "url_base": "https://api.apilayer.com",
1069
- "path": "/exchangerates_data/latest",
1070
- "http_method": "GET",
1071
- "request_parameters": {},
1072
- "request_headers": {},
1073
- "request_body_json": {},
1074
- "authenticator": {
1075
- "type": "ApiKeyAuthenticator",
1076
- "header": "apikey",
1077
- "api_token": "{{ config['api_key'] }}",
1078
- },
1079
- },
1080
- "partition_router": {"type": "ListPartitionRouter", "values": ["0", "1"], "cursor_field": "partition"},
1081
- "record_selector": {"type": "RecordSelector", "extractor": {"type": "DpathExtractor", "field_path": ["rates"]}},
1082
- "paginator": {
1083
- "type": "DefaultPaginator",
1084
- "page_size": 2,
1085
- "page_size_option": {"inject_into": "request_parameter", "field_name": "page_size"},
1086
- "page_token_option": {"inject_into": "path", "type": "RequestPath"},
1087
- "pagination_strategy": {
1088
- "type": "CursorPagination",
1089
- "cursor_value": "{{ response._metadata.next }}",
1090
- "page_size": 2,
1091
- },
1092
- },
1093
- },
1094
- }
1095
- ],
1096
- "spec": {
1097
- "connection_specification": {
1098
- "$schema": "http://json-schema.org/draft-07/schema#",
1099
- "type": "object",
1100
- "required": ["api_key"],
1101
- "properties": {"api_key": {"type": "string", "title": "API Key", "airbyte_secret": True}},
1102
- "additionalProperties": True,
1103
- },
1104
- "documentation_url": "https://example.org",
1105
- "type": "Spec",
1106
- },
1107
- },
1108
- (
1109
- _create_page({"rates": [{"ABC": 0, "partition": 0}, {"AED": 1, "partition": 0}], "_metadata": {"next": "next"}}),
1110
- _create_page({"rates": [{"USD": 3, "partition": 0}], "_metadata": {}}),
1111
- _create_page({"rates": [{"ABC": 2, "partition": 1}], "_metadata": {}}),
1112
- ),
1113
- [{"ABC": 0, "partition": 0}, {"AED": 1, "partition": 0}, {"USD": 3, "partition": 0}, {"ABC": 2, "partition": 1}],
1114
- [
1115
- call({}, {"partition": "0"}, None),
1116
- call({}, {"partition": "0"}, {"next_page_token": "next"}),
1117
- call({}, {"partition": "1"}, None),
1118
- ],
1119
- ),
1120
- ],
1121
- )
1122
- def test_read_manifest_declarative_source(test_name, manifest, pages, expected_records, expected_calls):
1123
- _stream_name = "Rates"
1124
- with patch.object(SimpleRetriever, "_fetch_next_page", side_effect=pages) as mock_retriever:
1125
- output_data = [message.record.data for message in _run_read(manifest, _stream_name) if message.record]
1126
- assert expected_records == output_data
1127
- mock_retriever.assert_has_calls(expected_calls)
1128
-
1129
-
1130
- def test_only_parent_streams_use_cache():
1131
- applications_stream = {
1132
- "type": "DeclarativeStream",
1133
- "$parameters": {"name": "applications", "primary_key": "id", "url_base": "https://harvest.greenhouse.io/v1/"},
1134
- "schema_loader": {
1135
- "name": "{{ parameters.stream_name }}",
1136
- "file_path": "./source_sendgrid/schemas/{{ parameters.name }}.yaml",
1137
- },
1138
- "retriever": {
1139
- "paginator": {
1140
- "type": "DefaultPaginator",
1141
- "page_size": 10,
1142
- "page_size_option": {"type": "RequestOption", "inject_into": "request_parameter", "field_name": "per_page"},
1143
- "page_token_option": {"type": "RequestPath"},
1144
- "pagination_strategy": {
1145
- "type": "CursorPagination",
1146
- "cursor_value": "{{ headers['link']['next']['url'] }}",
1147
- "stop_condition": "{{ 'next' not in headers['link'] }}",
1148
- "page_size": 100,
1149
- },
1150
- },
1151
- "requester": {
1152
- "path": "applications",
1153
- "authenticator": {"type": "BasicHttpAuthenticator", "username": "{{ config['api_key'] }}"},
1154
- },
1155
- "record_selector": {"extractor": {"type": "DpathExtractor", "field_path": []}},
1156
- },
1157
- }
1158
-
1159
- manifest = {
1160
- "version": "0.29.3",
1161
- "definitions": {},
1162
- "streams": [
1163
- deepcopy(applications_stream),
1164
- {
1165
- "type": "DeclarativeStream",
1166
- "$parameters": {"name": "applications_interviews", "primary_key": "id", "url_base": "https://harvest.greenhouse.io/v1/"},
1167
- "schema_loader": {
1168
- "name": "{{ parameters.stream_name }}",
1169
- "file_path": "./source_sendgrid/schemas/{{ parameters.name }}.yaml",
1170
- },
1171
- "retriever": {
1172
- "paginator": {
1173
- "type": "DefaultPaginator",
1174
- "page_size": 10,
1175
- "page_size_option": {"type": "RequestOption", "inject_into": "request_parameter", "field_name": "per_page"},
1176
- "page_token_option": {"type": "RequestPath"},
1177
- "pagination_strategy": {
1178
- "type": "CursorPagination",
1179
- "cursor_value": "{{ headers['link']['next']['url'] }}",
1180
- "stop_condition": "{{ 'next' not in headers['link'] }}",
1181
- "page_size": 100,
1182
- },
1183
- },
1184
- "requester": {
1185
- "path": "applications_interviews",
1186
- "authenticator": {"type": "BasicHttpAuthenticator", "username": "{{ config['api_key'] }}"},
1187
- },
1188
- "record_selector": {"extractor": {"type": "DpathExtractor", "field_path": []}},
1189
- "partition_router": {
1190
- "parent_stream_configs": [
1191
- {"parent_key": "id", "partition_field": "parent_id", "stream": deepcopy(applications_stream)}
1192
- ],
1193
- "type": "SubstreamPartitionRouter",
1194
- },
1195
- },
1196
- },
1197
- {
1198
- "type": "DeclarativeStream",
1199
- "$parameters": {"name": "jobs", "primary_key": "id", "url_base": "https://harvest.greenhouse.io/v1/"},
1200
- "schema_loader": {
1201
- "name": "{{ parameters.stream_name }}",
1202
- "file_path": "./source_sendgrid/schemas/{{ parameters.name }}.yaml",
1203
- },
1204
- "retriever": {
1205
- "paginator": {
1206
- "type": "DefaultPaginator",
1207
- "page_size": 10,
1208
- "page_size_option": {"type": "RequestOption", "inject_into": "request_parameter", "field_name": "per_page"},
1209
- "page_token_option": {"type": "RequestPath"},
1210
- "pagination_strategy": {
1211
- "type": "CursorPagination",
1212
- "cursor_value": "{{ headers['link']['next']['url'] }}",
1213
- "stop_condition": "{{ 'next' not in headers['link'] }}",
1214
- "page_size": 100,
1215
- },
1216
- },
1217
- "requester": {
1218
- "path": "jobs",
1219
- "authenticator": {"type": "BasicHttpAuthenticator", "username": "{{ config['api_key'] }}"},
1220
- },
1221
- "record_selector": {"extractor": {"type": "DpathExtractor", "field_path": []}},
1222
- },
1223
- },
1224
- ],
1225
- "check": {"type": "CheckStream", "stream_names": ["applications"]},
1226
- }
1227
- source = ManifestDeclarativeSource(source_config=manifest)
1228
-
1229
- streams = source.streams({})
1230
- assert len(streams) == 3
1231
-
1232
- # Main stream with caching (parent for substream `applications_interviews`)
1233
- assert streams[0].name == "applications"
1234
- assert streams[0].retriever.requester.use_cache
1235
-
1236
- # Substream
1237
- assert streams[1].name == "applications_interviews"
1238
- assert not streams[1].retriever.requester.use_cache
1239
-
1240
- # Parent stream created for substream
1241
- assert streams[1].retriever.stream_slicer.parent_stream_configs[0].stream.name == "applications"
1242
- assert streams[1].retriever.stream_slicer.parent_stream_configs[0].stream.retriever.requester.use_cache
1243
-
1244
- # Main stream without caching
1245
- assert streams[2].name == "jobs"
1246
- assert not streams[2].retriever.requester.use_cache
1247
-
1248
-
1249
- def _run_read(manifest: Mapping[str, Any], stream_name: str) -> List[AirbyteMessage]:
1250
- source = ManifestDeclarativeSource(source_config=manifest)
1251
- catalog = ConfiguredAirbyteCatalog(
1252
- streams=[
1253
- ConfiguredAirbyteStream(
1254
- stream=AirbyteStream(name=stream_name, json_schema={}, supported_sync_modes=[SyncMode.full_refresh]),
1255
- sync_mode=SyncMode.full_refresh,
1256
- destination_sync_mode=DestinationSyncMode.append,
1257
- )
1258
- ]
1259
- )
1260
- return list(source.read(logger, {}, catalog, {}))