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
@@ -1,871 +0,0 @@
1
- #
2
- # Copyright (c) 2023 Airbyte, Inc., all rights reserved.
3
- #
4
-
5
- import copy
6
- import dataclasses
7
- import json
8
- import logging
9
- import os
10
- from unittest import mock
11
- from unittest.mock import MagicMock, patch
12
-
13
- import pytest
14
- import requests
15
- from airbyte_cdk import connector_builder
16
- from airbyte_cdk.connector_builder.connector_builder_handler import (
17
- DEFAULT_MAXIMUM_NUMBER_OF_PAGES_PER_SLICE,
18
- DEFAULT_MAXIMUM_NUMBER_OF_SLICES,
19
- DEFAULT_MAXIMUM_RECORDS,
20
- TestReadLimits,
21
- create_source,
22
- get_limits,
23
- resolve_manifest,
24
- )
25
- from airbyte_cdk.connector_builder.main import handle_connector_builder_request, handle_request, read_stream
26
- from airbyte_cdk.connector_builder.models import LogMessage, StreamRead, StreamReadPages, StreamReadSlices
27
- from airbyte_cdk.models import (
28
- AirbyteLogMessage,
29
- AirbyteMessage,
30
- AirbyteRecordMessage,
31
- AirbyteStream,
32
- ConfiguredAirbyteCatalog,
33
- ConfiguredAirbyteStream,
34
- ConnectorSpecification,
35
- DestinationSyncMode,
36
- Level,
37
- SyncMode,
38
- )
39
- from airbyte_cdk.models import Type
40
- from airbyte_cdk.models import Type as MessageType
41
- from airbyte_cdk.sources.declarative.declarative_stream import DeclarativeStream
42
- from airbyte_cdk.sources.declarative.manifest_declarative_source import ManifestDeclarativeSource
43
- from airbyte_cdk.sources.declarative.retrievers import SimpleRetrieverTestReadDecorator
44
- from airbyte_cdk.sources.declarative.retrievers.simple_retriever import SimpleRetriever
45
- from unit_tests.connector_builder.utils import create_configured_catalog
46
-
47
- _stream_name = "stream_with_custom_requester"
48
- _stream_primary_key = "id"
49
- _stream_url_base = "https://api.sendgrid.com"
50
- _stream_options = {"name": _stream_name, "primary_key": _stream_primary_key, "url_base": _stream_url_base}
51
- _page_size = 2
52
-
53
- MANIFEST = {
54
- "version": "0.30.3",
55
- "definitions": {
56
- "retriever": {
57
- "paginator": {
58
- "type": "DefaultPaginator",
59
- "page_size": _page_size,
60
- "page_size_option": {"inject_into": "request_parameter", "field_name": "page_size"},
61
- "page_token_option": {"inject_into": "path", "type": "RequestPath"},
62
- "pagination_strategy": {
63
- "type": "CursorPagination",
64
- "cursor_value": "{{ response._metadata.next }}",
65
- "page_size": _page_size,
66
- },
67
- },
68
- "partition_router": {
69
- "type": "ListPartitionRouter",
70
- "values": ["0", "1", "2", "3", "4", "5", "6", "7"],
71
- "cursor_field": "item_id",
72
- },
73
- ""
74
- "requester": {
75
- "path": "/v3/marketing/lists",
76
- "authenticator": {"type": "BearerAuthenticator", "api_token": "{{ config.apikey }}"},
77
- "request_parameters": {"a_param": "10"},
78
- },
79
- "record_selector": {"extractor": {"field_path": ["result"]}},
80
- },
81
- },
82
- "streams": [
83
- {
84
- "type": "DeclarativeStream",
85
- "$parameters": _stream_options,
86
- "retriever": "#/definitions/retriever",
87
- },
88
- ],
89
- "check": {"type": "CheckStream", "stream_names": ["lists"]},
90
- "spec": {
91
- "connection_specification": {
92
- "$schema": "http://json-schema.org/draft-07/schema#",
93
- "type": "object",
94
- "required": [],
95
- "properties": {},
96
- "additionalProperties": True,
97
- },
98
- "type": "Spec",
99
- },
100
- }
101
-
102
- OAUTH_MANIFEST = {
103
- "version": "0.30.3",
104
- "definitions": {
105
- "retriever": {
106
- "paginator": {
107
- "type": "DefaultPaginator",
108
- "page_size": _page_size,
109
- "page_size_option": {"inject_into": "request_parameter", "field_name": "page_size"},
110
- "page_token_option": {"inject_into": "path", "type": "RequestPath"},
111
- "pagination_strategy": {
112
- "type": "CursorPagination",
113
- "cursor_value": "{{ response._metadata.next }}",
114
- "page_size": _page_size,
115
- },
116
- },
117
- "partition_router": {
118
- "type": "ListPartitionRouter",
119
- "values": ["0", "1", "2", "3", "4", "5", "6", "7"],
120
- "cursor_field": "item_id",
121
- },
122
- ""
123
- "requester": {
124
- "path": "/v3/marketing/lists",
125
- "authenticator": {"type": "OAuthAuthenticator", "api_token": "{{ config.apikey }}"},
126
- "request_parameters": {"a_param": "10"},
127
- },
128
- "record_selector": {"extractor": {"field_path": ["result"]}},
129
- },
130
- },
131
- "streams": [
132
- {
133
- "type": "DeclarativeStream",
134
- "$parameters": _stream_options,
135
- "retriever": "#/definitions/retriever",
136
- },
137
- ],
138
- "check": {"type": "CheckStream", "stream_names": ["lists"]},
139
- "spec": {
140
- "connection_specification": {
141
- "$schema": "http://json-schema.org/draft-07/schema#",
142
- "type": "object",
143
- "required": [],
144
- "properties": {},
145
- "additionalProperties": True,
146
- },
147
- "type": "Spec",
148
- },
149
- }
150
-
151
- RESOLVE_MANIFEST_CONFIG = {
152
- "__injected_declarative_manifest": MANIFEST,
153
- "__command": "resolve_manifest",
154
- }
155
-
156
- TEST_READ_CONFIG = {
157
- "__injected_declarative_manifest": MANIFEST,
158
- "__command": "test_read",
159
- "__test_read_config": {"max_pages_per_slice": 2, "max_slices": 5, "max_records": 10},
160
- }
161
-
162
- DUMMY_CATALOG = {
163
- "streams": [
164
- {
165
- "stream": {
166
- "name": "dummy_stream",
167
- "json_schema": {"$schema": "http://json-schema.org/draft-07/schema#", "type": "object", "properties": {}},
168
- "supported_sync_modes": ["full_refresh"],
169
- "source_defined_cursor": False,
170
- },
171
- "sync_mode": "full_refresh",
172
- "destination_sync_mode": "overwrite",
173
- }
174
- ]
175
- }
176
-
177
- CONFIGURED_CATALOG = {
178
- "streams": [
179
- {
180
- "stream": {
181
- "name": _stream_name,
182
- "json_schema": {"$schema": "http://json-schema.org/draft-07/schema#", "type": "object", "properties": {}},
183
- "supported_sync_modes": ["full_refresh"],
184
- "source_defined_cursor": False,
185
- },
186
- "sync_mode": "full_refresh",
187
- "destination_sync_mode": "overwrite",
188
- }
189
- ]
190
- }
191
-
192
- MOCK_RESPONSE = {
193
- "result": [
194
- {"id": 1, "name": "Nora Moon", "position": "director"},
195
- {"id": 2, "name": "Hae Sung Jung", "position": "cinematographer"},
196
- {"id": 3, "name": "Arthur Zenneranski", "position": "composer"},
197
- ]
198
- }
199
-
200
-
201
- @pytest.fixture
202
- def valid_resolve_manifest_config_file(tmp_path):
203
- config_file = tmp_path / "config.json"
204
- config_file.write_text(json.dumps(RESOLVE_MANIFEST_CONFIG))
205
- return config_file
206
-
207
-
208
- @pytest.fixture
209
- def valid_read_config_file(tmp_path):
210
- config_file = tmp_path / "config.json"
211
- config_file.write_text(json.dumps(TEST_READ_CONFIG))
212
- return config_file
213
-
214
-
215
- @pytest.fixture
216
- def dummy_catalog(tmp_path):
217
- config_file = tmp_path / "catalog.json"
218
- config_file.write_text(json.dumps(DUMMY_CATALOG))
219
- return config_file
220
-
221
-
222
- @pytest.fixture
223
- def configured_catalog(tmp_path):
224
- config_file = tmp_path / "catalog.json"
225
- config_file.write_text(json.dumps(CONFIGURED_CATALOG))
226
- return config_file
227
-
228
-
229
- @pytest.fixture
230
- def invalid_config_file(tmp_path):
231
- invalid_config = copy.deepcopy(RESOLVE_MANIFEST_CONFIG)
232
- invalid_config["__command"] = "bad_command"
233
- config_file = tmp_path / "config.json"
234
- config_file.write_text(json.dumps(invalid_config))
235
- return config_file
236
-
237
-
238
- def _mocked_send(self, request, **kwargs) -> requests.Response:
239
- """
240
- Mocks the outbound send operation to provide faster and more reliable responses compared to actual API requests
241
- """
242
- response = requests.Response()
243
- response.request = request
244
- response.status_code = 200
245
- response.headers = {"header": "value"}
246
- response_body = MOCK_RESPONSE
247
- response._content = json.dumps(response_body).encode("utf-8")
248
- return response
249
-
250
-
251
- def test_handle_resolve_manifest(valid_resolve_manifest_config_file, dummy_catalog):
252
- with mock.patch.object(connector_builder.main, "handle_connector_builder_request") as patched_handle:
253
- handle_request(["read", "--config", str(valid_resolve_manifest_config_file), "--catalog", str(dummy_catalog)])
254
- assert patched_handle.call_count == 1
255
-
256
-
257
- def test_handle_test_read(valid_read_config_file, configured_catalog):
258
- with mock.patch.object(connector_builder.main, "handle_connector_builder_request") as patch:
259
- handle_request(["read", "--config", str(valid_read_config_file), "--catalog", str(configured_catalog)])
260
- assert patch.call_count == 1
261
-
262
-
263
- def test_resolve_manifest(valid_resolve_manifest_config_file):
264
- config = copy.deepcopy(RESOLVE_MANIFEST_CONFIG)
265
- command = "resolve_manifest"
266
- config["__command"] = command
267
- source = ManifestDeclarativeSource(MANIFEST)
268
- limits = TestReadLimits()
269
- resolved_manifest = handle_connector_builder_request(source, command, config, create_configured_catalog("dummy_stream"), limits)
270
-
271
- expected_resolved_manifest = {
272
- "type": "DeclarativeSource",
273
- "version": "0.30.3",
274
- "definitions": {
275
- "retriever": {
276
- "paginator": {
277
- "type": "DefaultPaginator",
278
- "page_size": _page_size,
279
- "page_size_option": {"inject_into": "request_parameter", "field_name": "page_size"},
280
- "page_token_option": {"inject_into": "path", "type": "RequestPath"},
281
- "pagination_strategy": {
282
- "type": "CursorPagination",
283
- "cursor_value": "{{ response._metadata.next }}",
284
- "page_size": _page_size,
285
- },
286
- },
287
- "partition_router": {
288
- "type": "ListPartitionRouter",
289
- "values": ["0", "1", "2", "3", "4", "5", "6", "7"],
290
- "cursor_field": "item_id",
291
- },
292
- "requester": {
293
- "path": "/v3/marketing/lists",
294
- "authenticator": {"type": "BearerAuthenticator", "api_token": "{{ config.apikey }}"},
295
- "request_parameters": {"a_param": "10"},
296
- },
297
- "record_selector": {"extractor": {"field_path": ["result"]}},
298
- },
299
- },
300
- "streams": [
301
- {
302
- "type": "DeclarativeStream",
303
- "retriever": {
304
- "type": "SimpleRetriever",
305
- "paginator": {
306
- "type": "DefaultPaginator",
307
- "page_size": _page_size,
308
- "page_size_option": {
309
- "type": "RequestOption",
310
- "inject_into": "request_parameter",
311
- "field_name": "page_size",
312
- "name": _stream_name,
313
- "primary_key": _stream_primary_key,
314
- "url_base": _stream_url_base,
315
- "$parameters": _stream_options,
316
- },
317
- "page_token_option": {
318
- "type": "RequestPath",
319
- "inject_into": "path",
320
- "name": _stream_name,
321
- "primary_key": _stream_primary_key,
322
- "url_base": _stream_url_base,
323
- "$parameters": _stream_options,
324
- },
325
- "pagination_strategy": {
326
- "type": "CursorPagination",
327
- "cursor_value": "{{ response._metadata.next }}",
328
- "name": _stream_name,
329
- "primary_key": _stream_primary_key,
330
- "url_base": _stream_url_base,
331
- "$parameters": _stream_options,
332
- "page_size": _page_size,
333
- },
334
- "name": _stream_name,
335
- "primary_key": _stream_primary_key,
336
- "url_base": _stream_url_base,
337
- "$parameters": _stream_options,
338
- },
339
- "requester": {
340
- "type": "HttpRequester",
341
- "path": "/v3/marketing/lists",
342
- "authenticator": {
343
- "type": "BearerAuthenticator",
344
- "api_token": "{{ config.apikey }}",
345
- "name": _stream_name,
346
- "primary_key": _stream_primary_key,
347
- "url_base": _stream_url_base,
348
- "$parameters": _stream_options,
349
- },
350
- "request_parameters": {"a_param": "10"},
351
- "name": _stream_name,
352
- "primary_key": _stream_primary_key,
353
- "url_base": _stream_url_base,
354
- "$parameters": _stream_options,
355
- },
356
- "partition_router": {
357
- "type": "ListPartitionRouter",
358
- "values": ["0", "1", "2", "3", "4", "5", "6", "7"],
359
- "cursor_field": "item_id",
360
- "name": _stream_name,
361
- "primary_key": _stream_primary_key,
362
- "url_base": _stream_url_base,
363
- "$parameters": _stream_options,
364
- },
365
- "record_selector": {
366
- "type": "RecordSelector",
367
- "extractor": {
368
- "type": "DpathExtractor",
369
- "field_path": ["result"],
370
- "name": _stream_name,
371
- "primary_key": _stream_primary_key,
372
- "url_base": _stream_url_base,
373
- "$parameters": _stream_options,
374
- },
375
- "name": _stream_name,
376
- "primary_key": _stream_primary_key,
377
- "url_base": _stream_url_base,
378
- "$parameters": _stream_options,
379
- },
380
- "name": _stream_name,
381
- "primary_key": _stream_primary_key,
382
- "url_base": _stream_url_base,
383
- "$parameters": _stream_options,
384
- },
385
- "name": _stream_name,
386
- "primary_key": _stream_primary_key,
387
- "url_base": _stream_url_base,
388
- "$parameters": _stream_options,
389
- },
390
- ],
391
- "check": {"type": "CheckStream", "stream_names": ["lists"]},
392
- "spec": {
393
- "connection_specification": {
394
- "$schema": "http://json-schema.org/draft-07/schema#",
395
- "type": "object",
396
- "required": [],
397
- "properties": {},
398
- "additionalProperties": True,
399
- },
400
- "type": "Spec",
401
- },
402
- }
403
- assert resolved_manifest.record.data["manifest"] == expected_resolved_manifest
404
- assert resolved_manifest.record.stream == "resolve_manifest"
405
-
406
-
407
- def test_resolve_manifest_error_returns_error_response():
408
- class MockManifestDeclarativeSource:
409
- @property
410
- def resolved_manifest(self):
411
- raise ValueError
412
-
413
- source = MockManifestDeclarativeSource()
414
- response = resolve_manifest(source)
415
- assert "Error resolving manifest" in response.trace.error.message
416
-
417
-
418
- def test_read():
419
- config = TEST_READ_CONFIG
420
- source = ManifestDeclarativeSource(MANIFEST)
421
-
422
- real_record = AirbyteRecordMessage(data={"id": "1234", "key": "value"}, emitted_at=1, stream=_stream_name)
423
- stream_read = StreamRead(
424
- logs=[{"message": "here be a log message"}],
425
- slices=[
426
- StreamReadSlices(
427
- pages=[StreamReadPages(records=[real_record], request=None, response=None)],
428
- slice_descriptor=None,
429
- state=None,
430
- )
431
- ],
432
- auxiliary_requests=[],
433
- test_read_limit_reached=False,
434
- inferred_schema=None,
435
- inferred_datetime_formats=None,
436
- latest_config_update={},
437
- )
438
-
439
- expected_airbyte_message = AirbyteMessage(
440
- type=MessageType.RECORD,
441
- record=AirbyteRecordMessage(
442
- stream=_stream_name,
443
- data={
444
- "logs": [{"message": "here be a log message"}],
445
- "slices": [
446
- {"pages": [{"records": [real_record], "request": None, "response": None}], "slice_descriptor": None, "state": None}
447
- ],
448
- "test_read_limit_reached": False,
449
- "auxiliary_requests": [],
450
- "inferred_schema": None,
451
- "inferred_datetime_formats": None,
452
- "latest_config_update": {},
453
- },
454
- emitted_at=1,
455
- ),
456
- )
457
- limits = TestReadLimits()
458
- with patch("airbyte_cdk.connector_builder.message_grouper.MessageGrouper.get_message_groups", return_value=stream_read):
459
- output_record = handle_connector_builder_request(
460
- source, "test_read", config, ConfiguredAirbyteCatalog.parse_obj(CONFIGURED_CATALOG), limits
461
- )
462
- output_record.record.emitted_at = 1
463
- assert output_record == expected_airbyte_message
464
-
465
-
466
- def test_config_update():
467
- manifest = copy.deepcopy(MANIFEST)
468
- manifest["definitions"]["retriever"]["requester"]["authenticator"] = {
469
- "type": "OAuthAuthenticator",
470
- "token_refresh_endpoint": "https://oauth.endpoint.com/tokens/bearer",
471
- "client_id": "{{ config['credentials']['client_id'] }}",
472
- "client_secret": "{{ config['credentials']['client_secret'] }}",
473
- "refresh_token": "{{ config['credentials']['refresh_token'] }}",
474
- "refresh_token_updater": {},
475
- }
476
- config = copy.deepcopy(TEST_READ_CONFIG)
477
- config["__injected_declarative_manifest"] = manifest
478
- config["credentials"] = {
479
- "client_id": "a client id",
480
- "client_secret": "a client secret",
481
- "refresh_token": "a refresh token",
482
- }
483
- source = ManifestDeclarativeSource(manifest)
484
-
485
- refresh_request_response = {
486
- "access_token": "an updated access token",
487
- "refresh_token": "an updated refresh token",
488
- "expires_in": 3600,
489
- }
490
- with patch(
491
- "airbyte_cdk.sources.streams.http.requests_native_auth.SingleUseRefreshTokenOauth2Authenticator._get_refresh_access_token_response",
492
- return_value=refresh_request_response,
493
- ):
494
- output = handle_connector_builder_request(
495
- source, "test_read", config, ConfiguredAirbyteCatalog.parse_obj(CONFIGURED_CATALOG), TestReadLimits()
496
- )
497
- assert output.record.data["latest_config_update"]
498
-
499
-
500
- @patch("traceback.TracebackException.from_exception")
501
- def test_read_returns_error_response(mock_from_exception):
502
- class MockManifestDeclarativeSource:
503
- def read(self, logger, config, catalog, state):
504
- raise ValueError("error_message")
505
-
506
- def spec(self, logger: logging.Logger) -> ConnectorSpecification:
507
- connector_specification = mock.Mock()
508
- connector_specification.connectionSpecification = {}
509
- return connector_specification
510
-
511
- @property
512
- def check_config_against_spec(self):
513
- return False
514
-
515
- stack_trace = "a stack trace"
516
- mock_from_exception.return_value = stack_trace
517
-
518
- source = MockManifestDeclarativeSource()
519
- limits = TestReadLimits()
520
- response = read_stream(source, TEST_READ_CONFIG, ConfiguredAirbyteCatalog.parse_obj(CONFIGURED_CATALOG), limits)
521
-
522
- expected_stream_read = StreamRead(
523
- logs=[LogMessage("error_message - a stack trace", "ERROR")],
524
- slices=[],
525
- test_read_limit_reached=False,
526
- auxiliary_requests=[],
527
- inferred_schema=None,
528
- inferred_datetime_formats={},
529
- latest_config_update=None,
530
- )
531
-
532
- expected_message = AirbyteMessage(
533
- type=MessageType.RECORD,
534
- record=AirbyteRecordMessage(stream=_stream_name, data=dataclasses.asdict(expected_stream_read), emitted_at=1),
535
- )
536
- response.record.emitted_at = 1
537
- assert response == expected_message
538
-
539
-
540
- @pytest.mark.parametrize(
541
- "command",
542
- [
543
- pytest.param("check", id="test_check_command_error"),
544
- pytest.param("spec", id="test_spec_command_error"),
545
- pytest.param("discover", id="test_discover_command_error"),
546
- pytest.param(None, id="test_command_is_none_error"),
547
- pytest.param("", id="test_command_is_empty_error"),
548
- ],
549
- )
550
- def test_invalid_protocol_command(command, valid_resolve_manifest_config_file):
551
- config = copy.deepcopy(RESOLVE_MANIFEST_CONFIG)
552
- config["__command"] = "resolve_manifest"
553
- with pytest.raises(SystemExit):
554
- handle_request([command, "--config", str(valid_resolve_manifest_config_file), "--catalog", ""])
555
-
556
-
557
- def test_missing_command(valid_resolve_manifest_config_file):
558
- with pytest.raises(SystemExit):
559
- handle_request(["--config", str(valid_resolve_manifest_config_file), "--catalog", ""])
560
-
561
-
562
- def test_missing_catalog(valid_resolve_manifest_config_file):
563
- with pytest.raises(SystemExit):
564
- handle_request(["read", "--config", str(valid_resolve_manifest_config_file)])
565
-
566
-
567
- def test_missing_config(valid_resolve_manifest_config_file):
568
- with pytest.raises(SystemExit):
569
- handle_request(["read", "--catalog", str(valid_resolve_manifest_config_file)])
570
-
571
-
572
- def test_invalid_config_command(invalid_config_file, dummy_catalog):
573
- with pytest.raises(ValueError):
574
- handle_request(["read", "--config", str(invalid_config_file), "--catalog", str(dummy_catalog)])
575
-
576
-
577
- @pytest.fixture
578
- def manifest_declarative_source():
579
- return mock.Mock(spec=ManifestDeclarativeSource, autospec=True)
580
-
581
-
582
- def create_mock_retriever(name, url_base, path):
583
- http_stream = mock.Mock(spec=SimpleRetriever, autospec=True)
584
- http_stream.name = name
585
- http_stream.requester = MagicMock()
586
- http_stream.requester.get_url_base.return_value = url_base
587
- http_stream.requester.get_path.return_value = path
588
- http_stream._paginator_path.return_value = None
589
- return http_stream
590
-
591
-
592
- def create_mock_declarative_stream(http_stream):
593
- declarative_stream = mock.Mock(spec=DeclarativeStream, autospec=True)
594
- declarative_stream.retriever = http_stream
595
- return declarative_stream
596
-
597
-
598
- @pytest.mark.parametrize(
599
- "test_name, config, expected_max_records, expected_max_slices, expected_max_pages_per_slice",
600
- [
601
- (
602
- "test_no_test_read_config",
603
- {},
604
- DEFAULT_MAXIMUM_RECORDS,
605
- DEFAULT_MAXIMUM_NUMBER_OF_SLICES,
606
- DEFAULT_MAXIMUM_NUMBER_OF_PAGES_PER_SLICE,
607
- ),
608
- (
609
- "test_no_values_set",
610
- {"__test_read_config": {}},
611
- DEFAULT_MAXIMUM_RECORDS,
612
- DEFAULT_MAXIMUM_NUMBER_OF_SLICES,
613
- DEFAULT_MAXIMUM_NUMBER_OF_PAGES_PER_SLICE,
614
- ),
615
- ("test_values_are_set", {"__test_read_config": {"max_slices": 1, "max_pages_per_slice": 2, "max_records": 3}}, 3, 1, 2),
616
- ],
617
- )
618
- def test_get_limits(test_name, config, expected_max_records, expected_max_slices, expected_max_pages_per_slice):
619
- limits = get_limits(config)
620
- assert limits.max_records == expected_max_records
621
- assert limits.max_pages_per_slice == expected_max_pages_per_slice
622
- assert limits.max_slices == expected_max_slices
623
-
624
-
625
- def test_create_source():
626
- max_records = 3
627
- max_pages_per_slice = 2
628
- max_slices = 1
629
- limits = TestReadLimits(max_records, max_pages_per_slice, max_slices)
630
-
631
- config = {"__injected_declarative_manifest": MANIFEST}
632
-
633
- source = create_source(config, limits)
634
-
635
- assert isinstance(source, ManifestDeclarativeSource)
636
- assert source._constructor._limit_pages_fetched_per_slice == limits.max_pages_per_slice
637
- assert source._constructor._limit_slices_fetched == limits.max_slices
638
- assert source.streams(config={})[0].retriever.requester.max_retries == 0
639
-
640
-
641
- def request_log_message(request: dict) -> AirbyteMessage:
642
- return AirbyteMessage(type=Type.LOG, log=AirbyteLogMessage(level=Level.INFO, message=f"request:{json.dumps(request)}"))
643
-
644
-
645
- def response_log_message(response: dict) -> AirbyteMessage:
646
- return AirbyteMessage(type=Type.LOG, log=AirbyteLogMessage(level=Level.INFO, message=f"response:{json.dumps(response)}"))
647
-
648
-
649
- def _create_request():
650
- url = "https://example.com/api"
651
- headers = {"Content-Type": "application/json"}
652
- return requests.Request("POST", url, headers=headers, json={"key": "value"}).prepare()
653
-
654
-
655
- def _create_response(body, request):
656
- response = requests.Response()
657
- response.status_code = 200
658
- response._content = bytes(json.dumps(body), "utf-8")
659
- response.headers["Content-Type"] = "application/json"
660
- response.request = request
661
- return response
662
-
663
-
664
- def _create_page_response(response_body):
665
- request = _create_request()
666
- return _create_response(response_body, request)
667
-
668
-
669
- @patch.object(
670
- requests.Session,
671
- "send",
672
- side_effect=(
673
- _create_page_response({"result": [{"id": 0}, {"id": 1}], "_metadata": {"next": "next"}}),
674
- _create_page_response({"result": [{"id": 2}], "_metadata": {"next": "next"}}),
675
- )
676
- * 10,
677
- )
678
- def test_read_source(mock_http_stream):
679
- """
680
- This test sort of acts as an integration test for the connector builder.
681
-
682
- Each slice has two pages
683
- The first page has two records
684
- The second page one record
685
-
686
- The response._metadata.next field in the first page tells the paginator to fetch the next page.
687
- """
688
- max_records = 100
689
- max_pages_per_slice = 2
690
- max_slices = 3
691
- limits = TestReadLimits(max_records, max_pages_per_slice, max_slices)
692
-
693
- catalog = ConfiguredAirbyteCatalog(
694
- streams=[
695
- ConfiguredAirbyteStream(
696
- stream=AirbyteStream(name=_stream_name, json_schema={}, supported_sync_modes=[SyncMode.full_refresh]),
697
- sync_mode=SyncMode.full_refresh,
698
- destination_sync_mode=DestinationSyncMode.append,
699
- )
700
- ]
701
- )
702
-
703
- config = {"__injected_declarative_manifest": MANIFEST}
704
-
705
- source = create_source(config, limits)
706
-
707
- output_data = read_stream(source, config, catalog, limits).record.data
708
- slices = output_data["slices"]
709
-
710
- assert len(slices) == max_slices
711
- for s in slices:
712
- pages = s["pages"]
713
- assert len(pages) == max_pages_per_slice
714
-
715
- first_page, second_page = pages[0], pages[1]
716
- assert len(first_page["records"]) == _page_size
717
- assert len(second_page["records"]) == 1
718
-
719
- streams = source.streams(config)
720
- for s in streams:
721
- assert isinstance(s.retriever, SimpleRetrieverTestReadDecorator)
722
-
723
-
724
- @patch.object(
725
- requests.Session,
726
- "send",
727
- side_effect=(
728
- _create_page_response({"result": [{"id": 0}, {"id": 1}], "_metadata": {"next": "next"}}),
729
- _create_page_response({"result": [{"id": 2}], "_metadata": {"next": "next"}}),
730
- ),
731
- )
732
- def test_read_source_single_page_single_slice(mock_http_stream):
733
- max_records = 100
734
- max_pages_per_slice = 1
735
- max_slices = 1
736
- limits = TestReadLimits(max_records, max_pages_per_slice, max_slices)
737
-
738
- catalog = ConfiguredAirbyteCatalog(
739
- streams=[
740
- ConfiguredAirbyteStream(
741
- stream=AirbyteStream(name=_stream_name, json_schema={}, supported_sync_modes=[SyncMode.full_refresh]),
742
- sync_mode=SyncMode.full_refresh,
743
- destination_sync_mode=DestinationSyncMode.append,
744
- )
745
- ]
746
- )
747
-
748
- config = {"__injected_declarative_manifest": MANIFEST}
749
-
750
- source = create_source(config, limits)
751
-
752
- output_data = read_stream(source, config, catalog, limits).record.data
753
- slices = output_data["slices"]
754
-
755
- assert len(slices) == max_slices
756
- for s in slices:
757
- pages = s["pages"]
758
- assert len(pages) == max_pages_per_slice
759
-
760
- first_page = pages[0]
761
- assert len(first_page["records"]) == _page_size
762
-
763
- streams = source.streams(config)
764
- for s in streams:
765
- assert isinstance(s.retriever, SimpleRetrieverTestReadDecorator)
766
-
767
-
768
- @pytest.mark.parametrize(
769
- "deployment_mode, url_base, expected_error",
770
- [
771
- pytest.param("CLOUD", "https://airbyte.com/api/v1/characters", None, id="test_cloud_read_with_public_endpoint"),
772
- pytest.param("CLOUD", "https://10.0.27.27", "AirbyteTracedException", id="test_cloud_read_with_private_endpoint"),
773
- pytest.param("CLOUD", "https://localhost:80/api/v1/cast", "AirbyteTracedException", id="test_cloud_read_with_localhost"),
774
- pytest.param("CLOUD", "http://unsecured.protocol/api/v1", "InvalidSchema", id="test_cloud_read_with_unsecured_endpoint"),
775
- pytest.param("CLOUD", "https://domainwithoutextension", "Invalid URL", id="test_cloud_read_with_invalid_url_endpoint"),
776
- pytest.param("OSS", "https://airbyte.com/api/v1/", None, id="test_oss_read_with_public_endpoint"),
777
- pytest.param("OSS", "https://10.0.27.27/api/v1/", None, id="test_oss_read_with_private_endpoint"),
778
- ],
779
- )
780
- @patch.object(requests.Session, "send", _mocked_send)
781
- def test_handle_read_external_requests(deployment_mode, url_base, expected_error):
782
- """
783
- This test acts like an integration test for the connector builder when it receives Test Read requests.
784
-
785
- The scenario being tested is whether requests should be denied if they are done on an unsecure channel or are made to internal
786
- endpoints when running on Cloud or OSS deployments
787
- """
788
-
789
- limits = TestReadLimits(max_records=100, max_pages_per_slice=1, max_slices=1)
790
-
791
- catalog = ConfiguredAirbyteCatalog(
792
- streams=[
793
- ConfiguredAirbyteStream(
794
- stream=AirbyteStream(name=_stream_name, json_schema={}, supported_sync_modes=[SyncMode.full_refresh]),
795
- sync_mode=SyncMode.full_refresh,
796
- destination_sync_mode=DestinationSyncMode.append,
797
- )
798
- ]
799
- )
800
-
801
- test_manifest = MANIFEST
802
- test_manifest["streams"][0]["$parameters"]["url_base"] = url_base
803
- config = {"__injected_declarative_manifest": test_manifest}
804
-
805
- source = create_source(config, limits)
806
-
807
- with mock.patch.dict(os.environ, {"DEPLOYMENT_MODE": deployment_mode}, clear=False):
808
- output_data = read_stream(source, config, catalog, limits).record.data
809
- if expected_error:
810
- assert len(output_data["logs"]) > 0, "Expected at least one log message with the expected error"
811
- error_message = output_data["logs"][0]
812
- assert error_message["level"] == "ERROR"
813
- assert expected_error in error_message["message"]
814
- else:
815
- page_records = output_data["slices"][0]["pages"][0]
816
- assert len(page_records) == len(MOCK_RESPONSE["result"])
817
-
818
-
819
- @pytest.mark.parametrize(
820
- "deployment_mode, token_url, expected_error",
821
- [
822
- pytest.param("CLOUD", "https://airbyte.com/tokens/bearer", None, id="test_cloud_read_with_public_endpoint"),
823
- pytest.param("CLOUD", "https://10.0.27.27/tokens/bearer", "AirbyteTracedException", id="test_cloud_read_with_private_endpoint"),
824
- pytest.param("CLOUD", "http://unsecured.protocol/tokens/bearer", "InvalidSchema", id="test_cloud_read_with_unsecured_endpoint"),
825
- pytest.param("CLOUD", "https://domainwithoutextension", "Invalid URL", id="test_cloud_read_with_invalid_url_endpoint"),
826
- pytest.param("OSS", "https://airbyte.com/tokens/bearer", None, id="test_oss_read_with_public_endpoint"),
827
- pytest.param("OSS", "https://10.0.27.27/tokens/bearer", None, id="test_oss_read_with_private_endpoint"),
828
- ],
829
- )
830
- @patch.object(requests.Session, "send", _mocked_send)
831
- def test_handle_read_external_oauth_request(deployment_mode, token_url, expected_error):
832
- """
833
- This test acts like an integration test for the connector builder when it receives Test Read requests.
834
-
835
- The scenario being tested is whether requests should be denied if they are done on an unsecure channel or are made to internal
836
- endpoints when running on Cloud or OSS deployments
837
- """
838
-
839
- limits = TestReadLimits(max_records=100, max_pages_per_slice=1, max_slices=1)
840
-
841
- catalog = ConfiguredAirbyteCatalog(
842
- streams=[
843
- ConfiguredAirbyteStream(
844
- stream=AirbyteStream(name=_stream_name, json_schema={}, supported_sync_modes=[SyncMode.full_refresh]),
845
- sync_mode=SyncMode.full_refresh,
846
- destination_sync_mode=DestinationSyncMode.append,
847
- )
848
- ]
849
- )
850
-
851
- oauth_authenticator_config: dict[str, str] = {
852
- "type": "OAuthAuthenticator",
853
- "token_refresh_endpoint": token_url,
854
- "client_id": "greta",
855
- "client_secret": "teo",
856
- "refresh_token": "john",
857
- }
858
-
859
- test_manifest = MANIFEST
860
- test_manifest["definitions"]["retriever"]["requester"]["authenticator"] = oauth_authenticator_config
861
- config = {"__injected_declarative_manifest": test_manifest}
862
-
863
- source = create_source(config, limits)
864
-
865
- with mock.patch.dict(os.environ, {"DEPLOYMENT_MODE": deployment_mode}, clear=False):
866
- output_data = read_stream(source, config, catalog, limits).record.data
867
- if expected_error:
868
- assert len(output_data["logs"]) > 0, "Expected at least one log message with the expected error"
869
- error_message = output_data["logs"][0]
870
- assert error_message["level"] == "ERROR"
871
- assert expected_error in error_message["message"]