airbyte-cdk 0.72.1__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 +1185 -85
  61. airbyte_cdk/sources/declarative/declarative_source.py +5 -2
  62. airbyte_cdk/sources/declarative/declarative_stream.py +95 -9
  63. airbyte_cdk/sources/declarative/decoders/__init__.py +23 -2
  64. airbyte_cdk/sources/declarative/decoders/composite_raw_decoder.py +97 -0
  65. airbyte_cdk/sources/declarative/decoders/decoder.py +11 -4
  66. airbyte_cdk/sources/declarative/decoders/json_decoder.py +92 -5
  67. airbyte_cdk/sources/declarative/decoders/noop_decoder.py +21 -0
  68. airbyte_cdk/sources/declarative/decoders/pagination_decoder_decorator.py +39 -0
  69. airbyte_cdk/sources/declarative/decoders/xml_decoder.py +98 -0
  70. airbyte_cdk/sources/declarative/extractors/__init__.py +12 -1
  71. airbyte_cdk/sources/declarative/extractors/dpath_extractor.py +29 -24
  72. airbyte_cdk/sources/declarative/extractors/http_selector.py +4 -5
  73. airbyte_cdk/sources/declarative/extractors/record_extractor.py +2 -3
  74. airbyte_cdk/sources/declarative/extractors/record_filter.py +65 -8
  75. airbyte_cdk/sources/declarative/extractors/record_selector.py +85 -26
  76. airbyte_cdk/sources/declarative/extractors/response_to_file_extractor.py +177 -0
  77. airbyte_cdk/sources/declarative/extractors/type_transformer.py +55 -0
  78. airbyte_cdk/sources/declarative/incremental/__init__.py +25 -3
  79. airbyte_cdk/sources/declarative/incremental/datetime_based_cursor.py +156 -48
  80. airbyte_cdk/sources/declarative/incremental/declarative_cursor.py +13 -0
  81. airbyte_cdk/sources/declarative/incremental/global_substream_cursor.py +350 -0
  82. airbyte_cdk/sources/declarative/incremental/per_partition_cursor.py +159 -74
  83. airbyte_cdk/sources/declarative/incremental/per_partition_with_global.py +200 -0
  84. airbyte_cdk/sources/declarative/incremental/resumable_full_refresh_cursor.py +122 -0
  85. airbyte_cdk/sources/declarative/interpolation/filters.py +27 -1
  86. airbyte_cdk/sources/declarative/interpolation/interpolated_boolean.py +23 -5
  87. airbyte_cdk/sources/declarative/interpolation/interpolated_mapping.py +12 -8
  88. airbyte_cdk/sources/declarative/interpolation/interpolated_nested_mapping.py +13 -6
  89. airbyte_cdk/sources/declarative/interpolation/interpolated_string.py +21 -6
  90. airbyte_cdk/sources/declarative/interpolation/interpolation.py +9 -3
  91. airbyte_cdk/sources/declarative/interpolation/jinja.py +72 -37
  92. airbyte_cdk/sources/declarative/interpolation/macros.py +72 -17
  93. airbyte_cdk/sources/declarative/manifest_declarative_source.py +193 -52
  94. airbyte_cdk/sources/declarative/migrations/legacy_to_per_partition_state_migration.py +98 -0
  95. airbyte_cdk/sources/declarative/migrations/state_migration.py +24 -0
  96. airbyte_cdk/sources/declarative/models/__init__.py +1 -1
  97. airbyte_cdk/sources/declarative/models/declarative_component_schema.py +1319 -603
  98. airbyte_cdk/sources/declarative/parsers/custom_exceptions.py +2 -2
  99. airbyte_cdk/sources/declarative/parsers/manifest_component_transformer.py +26 -4
  100. airbyte_cdk/sources/declarative/parsers/manifest_reference_resolver.py +26 -15
  101. airbyte_cdk/sources/declarative/parsers/model_to_component_factory.py +1695 -225
  102. airbyte_cdk/sources/declarative/partition_routers/__init__.py +24 -4
  103. airbyte_cdk/sources/declarative/partition_routers/async_job_partition_router.py +65 -0
  104. airbyte_cdk/sources/declarative/partition_routers/cartesian_product_stream_slicer.py +176 -0
  105. airbyte_cdk/sources/declarative/partition_routers/list_partition_router.py +39 -9
  106. airbyte_cdk/sources/declarative/partition_routers/partition_router.py +62 -0
  107. airbyte_cdk/sources/declarative/partition_routers/single_partition_router.py +15 -3
  108. airbyte_cdk/sources/declarative/partition_routers/substream_partition_router.py +222 -39
  109. airbyte_cdk/sources/declarative/requesters/error_handlers/__init__.py +19 -5
  110. airbyte_cdk/sources/declarative/requesters/error_handlers/backoff_strategies/__init__.py +3 -1
  111. airbyte_cdk/sources/declarative/requesters/error_handlers/backoff_strategies/constant_backoff_strategy.py +19 -7
  112. airbyte_cdk/sources/declarative/requesters/error_handlers/backoff_strategies/exponential_backoff_strategy.py +19 -7
  113. airbyte_cdk/sources/declarative/requesters/error_handlers/backoff_strategies/header_helper.py +4 -2
  114. airbyte_cdk/sources/declarative/requesters/error_handlers/backoff_strategies/wait_time_from_header_backoff_strategy.py +41 -9
  115. airbyte_cdk/sources/declarative/requesters/error_handlers/backoff_strategies/wait_until_time_from_header_backoff_strategy.py +29 -14
  116. airbyte_cdk/sources/declarative/requesters/error_handlers/backoff_strategy.py +5 -13
  117. airbyte_cdk/sources/declarative/requesters/error_handlers/composite_error_handler.py +32 -16
  118. airbyte_cdk/sources/declarative/requesters/error_handlers/default_error_handler.py +46 -56
  119. airbyte_cdk/sources/declarative/requesters/error_handlers/default_http_response_filter.py +40 -0
  120. airbyte_cdk/sources/declarative/requesters/error_handlers/error_handler.py +6 -32
  121. airbyte_cdk/sources/declarative/requesters/error_handlers/http_response_filter.py +119 -41
  122. airbyte_cdk/sources/declarative/requesters/http_job_repository.py +228 -0
  123. airbyte_cdk/sources/declarative/requesters/http_requester.py +98 -344
  124. airbyte_cdk/sources/declarative/requesters/paginators/__init__.py +14 -3
  125. airbyte_cdk/sources/declarative/requesters/paginators/default_paginator.py +105 -46
  126. airbyte_cdk/sources/declarative/requesters/paginators/no_pagination.py +14 -8
  127. airbyte_cdk/sources/declarative/requesters/paginators/paginator.py +19 -8
  128. airbyte_cdk/sources/declarative/requesters/paginators/strategies/__init__.py +9 -3
  129. airbyte_cdk/sources/declarative/requesters/paginators/strategies/cursor_pagination_strategy.py +53 -21
  130. airbyte_cdk/sources/declarative/requesters/paginators/strategies/offset_increment.py +42 -19
  131. airbyte_cdk/sources/declarative/requesters/paginators/strategies/page_increment.py +25 -12
  132. airbyte_cdk/sources/declarative/requesters/paginators/strategies/pagination_strategy.py +13 -10
  133. airbyte_cdk/sources/declarative/requesters/paginators/strategies/stop_condition.py +26 -13
  134. airbyte_cdk/sources/declarative/requesters/request_options/__init__.py +15 -2
  135. airbyte_cdk/sources/declarative/requesters/request_options/datetime_based_request_options_provider.py +91 -0
  136. airbyte_cdk/sources/declarative/requesters/request_options/default_request_options_provider.py +60 -0
  137. airbyte_cdk/sources/declarative/requesters/request_options/interpolated_nested_request_input_provider.py +31 -14
  138. airbyte_cdk/sources/declarative/requesters/request_options/interpolated_request_input_provider.py +27 -15
  139. airbyte_cdk/sources/declarative/requesters/request_options/interpolated_request_options_provider.py +63 -10
  140. airbyte_cdk/sources/declarative/requesters/request_options/request_options_provider.py +1 -1
  141. airbyte_cdk/sources/declarative/requesters/requester.py +9 -17
  142. airbyte_cdk/sources/declarative/resolvers/__init__.py +41 -0
  143. airbyte_cdk/sources/declarative/resolvers/components_resolver.py +55 -0
  144. airbyte_cdk/sources/declarative/resolvers/config_components_resolver.py +136 -0
  145. airbyte_cdk/sources/declarative/resolvers/http_components_resolver.py +112 -0
  146. airbyte_cdk/sources/declarative/retrievers/__init__.py +6 -2
  147. airbyte_cdk/sources/declarative/retrievers/async_retriever.py +100 -0
  148. airbyte_cdk/sources/declarative/retrievers/retriever.py +1 -3
  149. airbyte_cdk/sources/declarative/retrievers/simple_retriever.py +228 -72
  150. airbyte_cdk/sources/declarative/schema/__init__.py +14 -1
  151. airbyte_cdk/sources/declarative/schema/default_schema_loader.py +5 -3
  152. airbyte_cdk/sources/declarative/schema/dynamic_schema_loader.py +236 -0
  153. airbyte_cdk/sources/declarative/schema/json_file_schema_loader.py +8 -8
  154. airbyte_cdk/sources/declarative/spec/spec.py +12 -5
  155. airbyte_cdk/sources/declarative/stream_slicers/__init__.py +1 -2
  156. airbyte_cdk/sources/declarative/stream_slicers/declarative_partition_generator.py +88 -0
  157. airbyte_cdk/sources/declarative/stream_slicers/stream_slicer.py +9 -14
  158. airbyte_cdk/sources/declarative/transformations/add_fields.py +19 -11
  159. airbyte_cdk/sources/declarative/transformations/flatten_fields.py +52 -0
  160. airbyte_cdk/sources/declarative/transformations/keys_replace_transformation.py +61 -0
  161. airbyte_cdk/sources/declarative/transformations/keys_to_lower_transformation.py +22 -0
  162. airbyte_cdk/sources/declarative/transformations/keys_to_snake_transformation.py +68 -0
  163. airbyte_cdk/sources/declarative/transformations/remove_fields.py +13 -10
  164. airbyte_cdk/sources/declarative/transformations/transformation.py +5 -5
  165. airbyte_cdk/sources/declarative/types.py +19 -110
  166. airbyte_cdk/sources/declarative/yaml_declarative_source.py +31 -10
  167. airbyte_cdk/sources/embedded/base_integration.py +16 -5
  168. airbyte_cdk/sources/embedded/catalog.py +16 -4
  169. airbyte_cdk/sources/embedded/runner.py +19 -3
  170. airbyte_cdk/sources/embedded/tools.py +5 -2
  171. airbyte_cdk/sources/file_based/README.md +152 -0
  172. airbyte_cdk/sources/file_based/__init__.py +24 -0
  173. airbyte_cdk/sources/file_based/availability_strategy/__init__.py +9 -2
  174. airbyte_cdk/sources/file_based/availability_strategy/abstract_file_based_availability_strategy.py +22 -6
  175. airbyte_cdk/sources/file_based/availability_strategy/default_file_based_availability_strategy.py +46 -10
  176. airbyte_cdk/sources/file_based/config/abstract_file_based_spec.py +58 -10
  177. airbyte_cdk/sources/file_based/config/avro_format.py +2 -1
  178. airbyte_cdk/sources/file_based/config/csv_format.py +29 -10
  179. airbyte_cdk/sources/file_based/config/excel_format.py +18 -0
  180. airbyte_cdk/sources/file_based/config/file_based_stream_config.py +16 -4
  181. airbyte_cdk/sources/file_based/config/jsonl_format.py +2 -1
  182. airbyte_cdk/sources/file_based/config/parquet_format.py +2 -1
  183. airbyte_cdk/sources/file_based/config/unstructured_format.py +13 -5
  184. airbyte_cdk/sources/file_based/discovery_policy/__init__.py +6 -2
  185. airbyte_cdk/sources/file_based/discovery_policy/abstract_discovery_policy.py +2 -4
  186. airbyte_cdk/sources/file_based/discovery_policy/default_discovery_policy.py +7 -2
  187. airbyte_cdk/sources/file_based/exceptions.py +52 -15
  188. airbyte_cdk/sources/file_based/file_based_source.py +163 -33
  189. airbyte_cdk/sources/file_based/file_based_stream_reader.py +83 -5
  190. airbyte_cdk/sources/file_based/file_types/__init__.py +14 -1
  191. airbyte_cdk/sources/file_based/file_types/avro_parser.py +75 -24
  192. airbyte_cdk/sources/file_based/file_types/csv_parser.py +116 -34
  193. airbyte_cdk/sources/file_based/file_types/excel_parser.py +196 -0
  194. airbyte_cdk/sources/file_based/file_types/file_transfer.py +37 -0
  195. airbyte_cdk/sources/file_based/file_types/file_type_parser.py +4 -1
  196. airbyte_cdk/sources/file_based/file_types/jsonl_parser.py +24 -8
  197. airbyte_cdk/sources/file_based/file_types/parquet_parser.py +60 -18
  198. airbyte_cdk/sources/file_based/file_types/unstructured_parser.py +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-0.72.1.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-0.72.1.dist-info/METADATA +0 -243
  326. airbyte_cdk-0.72.1.dist-info/RECORD +0 -466
  327. airbyte_cdk-0.72.1.dist-info/top_level.txt +0 -3
  328. source_declarative_manifest/main.py +0 -29
  329. unit_tests/connector_builder/__init__.py +0 -3
  330. unit_tests/connector_builder/test_connector_builder_handler.py +0 -871
  331. unit_tests/connector_builder/test_message_grouper.py +0 -713
  332. unit_tests/connector_builder/utils.py +0 -27
  333. unit_tests/destinations/test_destination.py +0 -243
  334. unit_tests/singer/test_singer_helpers.py +0 -56
  335. unit_tests/singer/test_singer_source.py +0 -112
  336. unit_tests/sources/__init__.py +0 -0
  337. unit_tests/sources/concurrent_source/__init__.py +0 -3
  338. unit_tests/sources/concurrent_source/test_concurrent_source_adapter.py +0 -106
  339. unit_tests/sources/declarative/__init__.py +0 -3
  340. unit_tests/sources/declarative/auth/__init__.py +0 -3
  341. unit_tests/sources/declarative/auth/test_oauth.py +0 -331
  342. unit_tests/sources/declarative/auth/test_selective_authenticator.py +0 -39
  343. unit_tests/sources/declarative/auth/test_session_token_auth.py +0 -182
  344. unit_tests/sources/declarative/auth/test_token_auth.py +0 -200
  345. unit_tests/sources/declarative/auth/test_token_provider.py +0 -73
  346. unit_tests/sources/declarative/checks/__init__.py +0 -3
  347. unit_tests/sources/declarative/checks/test_check_stream.py +0 -146
  348. unit_tests/sources/declarative/decoders/__init__.py +0 -0
  349. unit_tests/sources/declarative/decoders/test_json_decoder.py +0 -16
  350. unit_tests/sources/declarative/external_component.py +0 -13
  351. unit_tests/sources/declarative/extractors/__init__.py +0 -3
  352. unit_tests/sources/declarative/extractors/test_dpath_extractor.py +0 -55
  353. unit_tests/sources/declarative/extractors/test_record_filter.py +0 -55
  354. unit_tests/sources/declarative/extractors/test_record_selector.py +0 -179
  355. unit_tests/sources/declarative/incremental/__init__.py +0 -0
  356. unit_tests/sources/declarative/incremental/test_datetime_based_cursor.py +0 -860
  357. unit_tests/sources/declarative/incremental/test_per_partition_cursor.py +0 -406
  358. unit_tests/sources/declarative/incremental/test_per_partition_cursor_integration.py +0 -332
  359. unit_tests/sources/declarative/interpolation/__init__.py +0 -3
  360. unit_tests/sources/declarative/interpolation/test_filters.py +0 -80
  361. unit_tests/sources/declarative/interpolation/test_interpolated_boolean.py +0 -40
  362. unit_tests/sources/declarative/interpolation/test_interpolated_mapping.py +0 -35
  363. unit_tests/sources/declarative/interpolation/test_interpolated_nested_mapping.py +0 -45
  364. unit_tests/sources/declarative/interpolation/test_interpolated_string.py +0 -25
  365. unit_tests/sources/declarative/interpolation/test_jinja.py +0 -240
  366. unit_tests/sources/declarative/interpolation/test_macros.py +0 -73
  367. unit_tests/sources/declarative/parsers/__init__.py +0 -3
  368. unit_tests/sources/declarative/parsers/test_manifest_component_transformer.py +0 -406
  369. unit_tests/sources/declarative/parsers/test_manifest_reference_resolver.py +0 -139
  370. unit_tests/sources/declarative/parsers/test_model_to_component_factory.py +0 -1847
  371. unit_tests/sources/declarative/parsers/testing_components.py +0 -36
  372. unit_tests/sources/declarative/partition_routers/__init__.py +0 -3
  373. unit_tests/sources/declarative/partition_routers/test_list_partition_router.py +0 -155
  374. unit_tests/sources/declarative/partition_routers/test_single_partition_router.py +0 -14
  375. unit_tests/sources/declarative/partition_routers/test_substream_partition_router.py +0 -404
  376. unit_tests/sources/declarative/requesters/__init__.py +0 -3
  377. unit_tests/sources/declarative/requesters/error_handlers/__init__.py +0 -3
  378. unit_tests/sources/declarative/requesters/error_handlers/backoff_strategies/__init__.py +0 -3
  379. unit_tests/sources/declarative/requesters/error_handlers/backoff_strategies/test_constant_backoff.py +0 -34
  380. unit_tests/sources/declarative/requesters/error_handlers/backoff_strategies/test_exponential_backoff.py +0 -36
  381. unit_tests/sources/declarative/requesters/error_handlers/backoff_strategies/test_header_helper.py +0 -38
  382. unit_tests/sources/declarative/requesters/error_handlers/backoff_strategies/test_wait_time_from_header.py +0 -35
  383. unit_tests/sources/declarative/requesters/error_handlers/backoff_strategies/test_wait_until_time_from_header.py +0 -64
  384. unit_tests/sources/declarative/requesters/error_handlers/test_composite_error_handler.py +0 -213
  385. unit_tests/sources/declarative/requesters/error_handlers/test_default_error_handler.py +0 -178
  386. unit_tests/sources/declarative/requesters/error_handlers/test_http_response_filter.py +0 -121
  387. unit_tests/sources/declarative/requesters/error_handlers/test_response_status.py +0 -44
  388. unit_tests/sources/declarative/requesters/paginators/__init__.py +0 -3
  389. unit_tests/sources/declarative/requesters/paginators/test_cursor_pagination_strategy.py +0 -64
  390. unit_tests/sources/declarative/requesters/paginators/test_default_paginator.py +0 -313
  391. unit_tests/sources/declarative/requesters/paginators/test_no_paginator.py +0 -12
  392. unit_tests/sources/declarative/requesters/paginators/test_offset_increment.py +0 -58
  393. unit_tests/sources/declarative/requesters/paginators/test_page_increment.py +0 -70
  394. unit_tests/sources/declarative/requesters/paginators/test_request_option.py +0 -43
  395. unit_tests/sources/declarative/requesters/paginators/test_stop_condition.py +0 -105
  396. unit_tests/sources/declarative/requesters/request_options/__init__.py +0 -3
  397. unit_tests/sources/declarative/requesters/request_options/test_interpolated_request_options_provider.py +0 -101
  398. unit_tests/sources/declarative/requesters/test_http_requester.py +0 -974
  399. unit_tests/sources/declarative/requesters/test_interpolated_request_input_provider.py +0 -32
  400. unit_tests/sources/declarative/retrievers/__init__.py +0 -3
  401. unit_tests/sources/declarative/retrievers/test_simple_retriever.py +0 -542
  402. unit_tests/sources/declarative/schema/__init__.py +0 -6
  403. unit_tests/sources/declarative/schema/source_test/SourceTest.py +0 -8
  404. unit_tests/sources/declarative/schema/source_test/__init__.py +0 -3
  405. unit_tests/sources/declarative/schema/test_default_schema_loader.py +0 -32
  406. unit_tests/sources/declarative/schema/test_inline_schema_loader.py +0 -19
  407. unit_tests/sources/declarative/schema/test_json_file_schema_loader.py +0 -26
  408. unit_tests/sources/declarative/states/__init__.py +0 -3
  409. unit_tests/sources/declarative/stream_slicers/__init__.py +0 -3
  410. unit_tests/sources/declarative/stream_slicers/test_cartesian_product_stream_slicer.py +0 -225
  411. unit_tests/sources/declarative/test_create_partial.py +0 -83
  412. unit_tests/sources/declarative/test_declarative_stream.py +0 -103
  413. unit_tests/sources/declarative/test_manifest_declarative_source.py +0 -1260
  414. unit_tests/sources/declarative/test_types.py +0 -39
  415. unit_tests/sources/declarative/test_yaml_declarative_source.py +0 -148
  416. unit_tests/sources/file_based/__init__.py +0 -0
  417. unit_tests/sources/file_based/availability_strategy/__init__.py +0 -0
  418. unit_tests/sources/file_based/availability_strategy/test_default_file_based_availability_strategy.py +0 -100
  419. unit_tests/sources/file_based/config/__init__.py +0 -0
  420. unit_tests/sources/file_based/config/test_abstract_file_based_spec.py +0 -28
  421. unit_tests/sources/file_based/config/test_csv_format.py +0 -34
  422. unit_tests/sources/file_based/config/test_file_based_stream_config.py +0 -84
  423. unit_tests/sources/file_based/discovery_policy/__init__.py +0 -0
  424. unit_tests/sources/file_based/discovery_policy/test_default_discovery_policy.py +0 -31
  425. unit_tests/sources/file_based/file_types/__init__.py +0 -0
  426. unit_tests/sources/file_based/file_types/test_avro_parser.py +0 -243
  427. unit_tests/sources/file_based/file_types/test_csv_parser.py +0 -546
  428. unit_tests/sources/file_based/file_types/test_jsonl_parser.py +0 -158
  429. unit_tests/sources/file_based/file_types/test_parquet_parser.py +0 -274
  430. unit_tests/sources/file_based/file_types/test_unstructured_parser.py +0 -593
  431. unit_tests/sources/file_based/helpers.py +0 -70
  432. unit_tests/sources/file_based/in_memory_files_source.py +0 -211
  433. unit_tests/sources/file_based/scenarios/__init__.py +0 -0
  434. unit_tests/sources/file_based/scenarios/avro_scenarios.py +0 -744
  435. unit_tests/sources/file_based/scenarios/check_scenarios.py +0 -220
  436. unit_tests/sources/file_based/scenarios/concurrent_incremental_scenarios.py +0 -2844
  437. unit_tests/sources/file_based/scenarios/csv_scenarios.py +0 -3105
  438. unit_tests/sources/file_based/scenarios/file_based_source_builder.py +0 -91
  439. unit_tests/sources/file_based/scenarios/incremental_scenarios.py +0 -1926
  440. unit_tests/sources/file_based/scenarios/jsonl_scenarios.py +0 -930
  441. unit_tests/sources/file_based/scenarios/parquet_scenarios.py +0 -754
  442. unit_tests/sources/file_based/scenarios/scenario_builder.py +0 -234
  443. unit_tests/sources/file_based/scenarios/unstructured_scenarios.py +0 -608
  444. unit_tests/sources/file_based/scenarios/user_input_schema_scenarios.py +0 -746
  445. unit_tests/sources/file_based/scenarios/validation_policy_scenarios.py +0 -726
  446. unit_tests/sources/file_based/stream/__init__.py +0 -0
  447. unit_tests/sources/file_based/stream/concurrent/__init__.py +0 -0
  448. unit_tests/sources/file_based/stream/concurrent/test_adapters.py +0 -362
  449. unit_tests/sources/file_based/stream/concurrent/test_file_based_concurrent_cursor.py +0 -458
  450. unit_tests/sources/file_based/stream/test_default_file_based_cursor.py +0 -310
  451. unit_tests/sources/file_based/stream/test_default_file_based_stream.py +0 -244
  452. unit_tests/sources/file_based/test_file_based_scenarios.py +0 -320
  453. unit_tests/sources/file_based/test_file_based_stream_reader.py +0 -272
  454. unit_tests/sources/file_based/test_scenarios.py +0 -253
  455. unit_tests/sources/file_based/test_schema_helpers.py +0 -346
  456. unit_tests/sources/fixtures/__init__.py +0 -3
  457. unit_tests/sources/fixtures/source_test_fixture.py +0 -153
  458. unit_tests/sources/message/__init__.py +0 -0
  459. unit_tests/sources/message/test_repository.py +0 -153
  460. unit_tests/sources/streams/__init__.py +0 -0
  461. unit_tests/sources/streams/concurrent/__init__.py +0 -3
  462. unit_tests/sources/streams/concurrent/scenarios/__init__.py +0 -3
  463. unit_tests/sources/streams/concurrent/scenarios/incremental_scenarios.py +0 -250
  464. unit_tests/sources/streams/concurrent/scenarios/stream_facade_builder.py +0 -140
  465. unit_tests/sources/streams/concurrent/scenarios/stream_facade_scenarios.py +0 -452
  466. unit_tests/sources/streams/concurrent/scenarios/test_concurrent_scenarios.py +0 -76
  467. unit_tests/sources/streams/concurrent/scenarios/thread_based_concurrent_stream_scenarios.py +0 -418
  468. unit_tests/sources/streams/concurrent/scenarios/thread_based_concurrent_stream_source_builder.py +0 -142
  469. unit_tests/sources/streams/concurrent/scenarios/utils.py +0 -55
  470. unit_tests/sources/streams/concurrent/test_adapters.py +0 -380
  471. unit_tests/sources/streams/concurrent/test_concurrent_read_processor.py +0 -684
  472. unit_tests/sources/streams/concurrent/test_cursor.py +0 -139
  473. unit_tests/sources/streams/concurrent/test_datetime_state_converter.py +0 -369
  474. unit_tests/sources/streams/concurrent/test_default_stream.py +0 -197
  475. unit_tests/sources/streams/concurrent/test_partition_enqueuer.py +0 -90
  476. unit_tests/sources/streams/concurrent/test_partition_reader.py +0 -67
  477. unit_tests/sources/streams/concurrent/test_thread_pool_manager.py +0 -106
  478. unit_tests/sources/streams/http/__init__.py +0 -0
  479. unit_tests/sources/streams/http/auth/__init__.py +0 -0
  480. unit_tests/sources/streams/http/auth/test_auth.py +0 -173
  481. unit_tests/sources/streams/http/requests_native_auth/__init__.py +0 -0
  482. unit_tests/sources/streams/http/requests_native_auth/test_requests_native_auth.py +0 -423
  483. unit_tests/sources/streams/http/test_availability_strategy.py +0 -180
  484. unit_tests/sources/streams/http/test_http.py +0 -635
  485. unit_tests/sources/streams/test_availability_strategy.py +0 -70
  486. unit_tests/sources/streams/test_call_rate.py +0 -300
  487. unit_tests/sources/streams/test_stream_read.py +0 -405
  488. unit_tests/sources/streams/test_streams_core.py +0 -184
  489. unit_tests/sources/test_abstract_source.py +0 -1442
  490. unit_tests/sources/test_concurrent_source.py +0 -112
  491. unit_tests/sources/test_config.py +0 -92
  492. unit_tests/sources/test_connector_state_manager.py +0 -482
  493. unit_tests/sources/test_http_logger.py +0 -252
  494. unit_tests/sources/test_integration_source.py +0 -86
  495. unit_tests/sources/test_source.py +0 -684
  496. unit_tests/sources/test_source_read.py +0 -460
  497. unit_tests/test/__init__.py +0 -0
  498. unit_tests/test/mock_http/__init__.py +0 -0
  499. unit_tests/test/mock_http/test_matcher.py +0 -53
  500. unit_tests/test/mock_http/test_mocker.py +0 -214
  501. unit_tests/test/mock_http/test_request.py +0 -117
  502. unit_tests/test/mock_http/test_response_builder.py +0 -177
  503. unit_tests/test/test_entrypoint_wrapper.py +0 -240
  504. unit_tests/utils/__init__.py +0 -0
  505. unit_tests/utils/test_datetime_format_inferrer.py +0 -60
  506. unit_tests/utils/test_mapping_helpers.py +0 -54
  507. unit_tests/utils/test_message_utils.py +0 -91
  508. unit_tests/utils/test_rate_limiting.py +0 -26
  509. unit_tests/utils/test_schema_inferrer.py +0 -202
  510. unit_tests/utils/test_secret_utils.py +0 -135
  511. unit_tests/utils/test_stream_status_utils.py +0 -61
  512. unit_tests/utils/test_traced_exception.py +0 -107
  513. /airbyte_cdk/sources/{deprecated → declarative/async_job}/__init__.py +0 -0
  514. {source_declarative_manifest → airbyte_cdk/sources/declarative/migrations}/__init__.py +0 -0
  515. {unit_tests/destinations → airbyte_cdk/sql}/__init__.py +0 -0
  516. {unit_tests/singer → airbyte_cdk/sql/_util}/__init__.py +0 -0
  517. {airbyte_cdk-0.72.1.dist-info → airbyte_cdk-6.13.1.dev4106.dist-info}/LICENSE.txt +0 -0
@@ -1,744 +0,0 @@
1
- #
2
- # Copyright (c) 2023 Airbyte, Inc., all rights reserved.
3
- #
4
-
5
- import datetime
6
- import decimal
7
-
8
- from unit_tests.sources.file_based.in_memory_files_source import TemporaryAvroFilesStreamReader
9
- from unit_tests.sources.file_based.scenarios.file_based_source_builder import FileBasedSourceBuilder
10
- from unit_tests.sources.file_based.scenarios.scenario_builder import TestScenarioBuilder
11
-
12
- _single_avro_file = {
13
- "a.avro": {
14
- "schema": {
15
- "type": "record",
16
- "name": "sampleAvro",
17
- "fields": [
18
- {"name": "col1", "type": "string"},
19
- {"name": "col2", "type": "int"},
20
- ],
21
- },
22
- "contents": [
23
- ("val11", 12),
24
- ("val21", 22),
25
- ],
26
- "last_modified": "2023-06-05T03:54:07.000Z",
27
- }
28
- }
29
-
30
- _multiple_avro_combine_schema_file = {
31
- "a.avro": {
32
- "schema": {
33
- "type": "record",
34
- "name": "sampleAvro",
35
- "fields": [
36
- {"name": "col_double", "type": "double"},
37
- {"name": "col_string", "type": "string"},
38
- {"name": "col_album", "type": {"type": "record", "name": "Album", "fields": [{"name": "album", "type": "string"}]}},
39
- ],
40
- },
41
- "contents": [
42
- (20.02, "Robbers", {"album": "The 1975"}),
43
- (20.23, "Somebody Else", {"album": "I Like It When You Sleep, for You Are So Beautiful yet So Unaware of It"}),
44
- ],
45
- "last_modified": "2023-06-05T03:54:07.000Z",
46
- },
47
- "b.avro": {
48
- "schema": {
49
- "type": "record",
50
- "name": "sampleAvro",
51
- "fields": [
52
- {"name": "col_double", "type": "double"},
53
- {"name": "col_string", "type": "string"},
54
- {"name": "col_song", "type": {"type": "record", "name": "Song", "fields": [{"name": "title", "type": "string"}]}},
55
- ],
56
- },
57
- "contents": [
58
- (1975.1975, "It's Not Living (If It's Not with You)", {"title": "Love It If We Made It"}),
59
- (5791.5791, "The 1975", {"title": "About You"}),
60
- ],
61
- "last_modified": "2023-06-06T03:54:07.000Z",
62
- },
63
- }
64
-
65
- _avro_all_types_file = {
66
- "a.avro": {
67
- "schema": {
68
- "type": "record",
69
- "name": "sampleAvro",
70
- "fields": [
71
- # Primitive Types
72
- {"name": "col_bool", "type": "boolean"},
73
- {"name": "col_int", "type": "int"},
74
- {"name": "col_long", "type": "long"},
75
- {"name": "col_float", "type": "float"},
76
- {"name": "col_double", "type": "double"},
77
- {"name": "col_bytes", "type": "bytes"},
78
- {"name": "col_string", "type": "string"},
79
- # Complex Types
80
- {
81
- "name": "col_record",
82
- "type": {
83
- "type": "record",
84
- "name": "SongRecord",
85
- "fields": [
86
- {"name": "artist", "type": "string"},
87
- {"name": "song", "type": "string"},
88
- {"name": "year", "type": "int"},
89
- ],
90
- },
91
- },
92
- {"name": "col_enum", "type": {"type": "enum", "name": "Genre", "symbols": ["POP_ROCK", "INDIE_ROCK", "ALTERNATIVE_ROCK"]}},
93
- {"name": "col_array", "type": {"type": "array", "items": "string"}},
94
- {"name": "col_map", "type": {"type": "map", "values": "string"}},
95
- {"name": "col_fixed", "type": {"type": "fixed", "name": "MyFixed", "size": 4}},
96
- # Logical Types
97
- {"name": "col_decimal", "type": {"type": "bytes", "logicalType": "decimal", "precision": 10, "scale": 5}},
98
- {"name": "col_uuid", "type": {"type": "string", "logicalType": "uuid"}},
99
- {"name": "col_date", "type": {"type": "int", "logicalType": "date"}},
100
- {"name": "col_time_millis", "type": {"type": "int", "logicalType": "time-millis"}},
101
- {"name": "col_time_micros", "type": {"type": "long", "logicalType": "time-micros"}},
102
- {"name": "col_timestamp_millis", "type": {"type": "long", "logicalType": "timestamp-millis"}},
103
- {"name": "col_timestamp_micros", "type": {"type": "long", "logicalType": "timestamp-micros"}},
104
- ],
105
- },
106
- "contents": [
107
- (
108
- True,
109
- 27,
110
- 1992,
111
- 999.09723456,
112
- 9123456.12394,
113
- b"\x00\x01\x02\x03",
114
- "Love It If We Made It",
115
- {"artist": "The 1975", "song": "About You", "year": 2022},
116
- "POP_ROCK",
117
- [
118
- "The 1975",
119
- "I Like It When You Sleep, for You Are So Beautiful yet So Unaware of It",
120
- "The 1975 A Brief Inquiry into Online Relationships",
121
- "Notes on a Conditional Form",
122
- "Being Funny in a Foreign Language",
123
- ],
124
- {"lead_singer": "Matty Healy", "lead_guitar": "Adam Hann", "bass_guitar": "Ross MacDonald", "drummer": "George Daniel"},
125
- b"\x12\x34\x56\x78",
126
- decimal.Decimal("1234.56789"),
127
- "123e4567-e89b-12d3-a456-426655440000",
128
- datetime.date(2022, 5, 29),
129
- datetime.time(6, 0, 0, 456000),
130
- datetime.time(12, 0, 0, 456789),
131
- datetime.datetime(2022, 5, 29, 0, 0, 0, 456000, tzinfo=datetime.timezone.utc),
132
- datetime.datetime(2022, 5, 30, 0, 0, 0, 456789, tzinfo=datetime.timezone.utc),
133
- ),
134
- ],
135
- "last_modified": "2023-06-05T03:54:07.000Z",
136
- }
137
- }
138
-
139
- _multiple_avro_stream_file = {
140
- "odesza_songs.avro": {
141
- "schema": {
142
- "type": "record",
143
- "name": "sampleAvro",
144
- "fields": [
145
- {"name": "col_title", "type": "string"},
146
- {
147
- "name": "col_album",
148
- "type": {
149
- "type": "enum",
150
- "name": "Album",
151
- "symbols": ["SUMMERS_GONE", "IN_RETURN", "A_MOMENT_APART", "THE_LAST_GOODBYE"],
152
- },
153
- },
154
- {"name": "col_year", "type": "int"},
155
- {"name": "col_vocals", "type": "boolean"},
156
- ],
157
- },
158
- "contents": [
159
- ("Late Night", "A_MOMENT_APART", 2017, False),
160
- ("White Lies", "IN_RETURN", 2014, True),
161
- ("Wide Awake", "THE_LAST_GOODBYE", 2022, True),
162
- ("Sun Models", "SUMMERS_GONE", 2012, True),
163
- ("All We Need", "IN_RETURN", 2014, True),
164
- ],
165
- "last_modified": "2023-06-05T03:54:07.000Z",
166
- },
167
- "california_festivals.avro": {
168
- "schema": {
169
- "type": "record",
170
- "name": "sampleAvro",
171
- "fields": [
172
- {"name": "col_name", "type": "string"},
173
- {
174
- "name": "col_location",
175
- "type": {
176
- "type": "record",
177
- "name": "LocationRecord",
178
- "fields": [
179
- {"name": "country", "type": "string"},
180
- {"name": "state", "type": "string"},
181
- {"name": "city", "type": "string"},
182
- ],
183
- },
184
- },
185
- {"name": "col_attendance", "type": "long"},
186
- ],
187
- },
188
- "contents": [
189
- ("Coachella", {"country": "USA", "state": "California", "city": "Indio"}, 250000),
190
- ("CRSSD", {"country": "USA", "state": "California", "city": "San Diego"}, 30000),
191
- ("Lightning in a Bottle", {"country": "USA", "state": "California", "city": "Buena Vista Lake"}, 18000),
192
- ("Outside Lands", {"country": "USA", "state": "California", "city": "San Francisco"}, 220000),
193
- ],
194
- "last_modified": "2023-06-06T03:54:07.000Z",
195
- },
196
- }
197
-
198
- single_avro_scenario = (
199
- TestScenarioBuilder()
200
- .set_name("single_avro_stream")
201
- .set_config(
202
- {
203
- "streams": [
204
- {
205
- "name": "stream1",
206
- "format": {"filetype": "avro"},
207
- "globs": ["*"],
208
- "validation_policy": "Emit Record",
209
- }
210
- ]
211
- }
212
- )
213
- .set_source_builder(
214
- FileBasedSourceBuilder()
215
- .set_stream_reader(TemporaryAvroFilesStreamReader(files=_single_avro_file, file_type="avro"))
216
- .set_file_type("avro")
217
- )
218
- .set_expected_check_status("SUCCEEDED")
219
- .set_expected_records(
220
- [
221
- {
222
- "data": {
223
- "col1": "val11",
224
- "col2": 12,
225
- "_ab_source_file_last_modified": "2023-06-05T03:54:07.000000Z",
226
- "_ab_source_file_url": "a.avro",
227
- },
228
- "stream": "stream1",
229
- },
230
- {
231
- "data": {
232
- "col1": "val21",
233
- "col2": 22,
234
- "_ab_source_file_last_modified": "2023-06-05T03:54:07.000000Z",
235
- "_ab_source_file_url": "a.avro",
236
- },
237
- "stream": "stream1",
238
- },
239
- ]
240
- )
241
- .set_expected_catalog(
242
- {
243
- "streams": [
244
- {
245
- "default_cursor_field": ["_ab_source_file_last_modified"],
246
- "json_schema": {
247
- "type": "object",
248
- "properties": {
249
- "col1": {"type": ["null", "string"]},
250
- "col2": {"type": ["null", "integer"]},
251
- "_ab_source_file_last_modified": {"type": "string"},
252
- "_ab_source_file_url": {"type": "string"},
253
- },
254
- },
255
- "name": "stream1",
256
- "source_defined_cursor": True,
257
- "supported_sync_modes": ["full_refresh", "incremental"],
258
- }
259
- ]
260
- }
261
- )
262
- ).build()
263
-
264
- multiple_avro_combine_schema_scenario = (
265
- TestScenarioBuilder()
266
- .set_name("multiple_avro_combine_schema_stream")
267
- .set_config(
268
- {
269
- "streams": [
270
- {
271
- "name": "stream1",
272
- "format": {"filetype": "avro"},
273
- "globs": ["*"],
274
- "validation_policy": "Emit Record",
275
- }
276
- ]
277
- }
278
- )
279
- .set_source_builder(
280
- FileBasedSourceBuilder()
281
- .set_stream_reader(TemporaryAvroFilesStreamReader(files=_multiple_avro_combine_schema_file, file_type="avro"))
282
- .set_file_type("avro")
283
- )
284
- .set_expected_records(
285
- [
286
- {
287
- "data": {
288
- "col_double": 20.02,
289
- "col_string": "Robbers",
290
- "col_album": {"album": "The 1975"},
291
- "_ab_source_file_last_modified": "2023-06-05T03:54:07.000000Z",
292
- "_ab_source_file_url": "a.avro",
293
- },
294
- "stream": "stream1",
295
- },
296
- {
297
- "data": {
298
- "col_double": 20.23,
299
- "col_string": "Somebody Else",
300
- "col_album": {"album": "I Like It When You Sleep, for You Are So Beautiful yet So Unaware of It"},
301
- "_ab_source_file_last_modified": "2023-06-05T03:54:07.000000Z",
302
- "_ab_source_file_url": "a.avro",
303
- },
304
- "stream": "stream1",
305
- },
306
- {
307
- "data": {
308
- "col_double": 1975.1975,
309
- "col_string": "It's Not Living (If It's Not with You)",
310
- "col_song": {"title": "Love It If We Made It"},
311
- "_ab_source_file_last_modified": "2023-06-06T03:54:07.000000Z",
312
- "_ab_source_file_url": "b.avro",
313
- },
314
- "stream": "stream1",
315
- },
316
- {
317
- "data": {
318
- "col_double": 5791.5791,
319
- "col_string": "The 1975",
320
- "col_song": {"title": "About You"},
321
- "_ab_source_file_last_modified": "2023-06-06T03:54:07.000000Z",
322
- "_ab_source_file_url": "b.avro",
323
- },
324
- "stream": "stream1",
325
- },
326
- ]
327
- )
328
- .set_expected_catalog(
329
- {
330
- "streams": [
331
- {
332
- "default_cursor_field": ["_ab_source_file_last_modified"],
333
- "json_schema": {
334
- "type": "object",
335
- "properties": {
336
- "col_double": {"type": ["null", "number"]},
337
- "col_string": {"type": ["null", "string"]},
338
- "col_album": {
339
- "properties": {
340
- "album": {"type": ["null", "string"]},
341
- },
342
- "type": ["null", "object"],
343
- },
344
- "col_song": {
345
- "properties": {
346
- "title": {"type": ["null", "string"]},
347
- },
348
- "type": ["null", "object"],
349
- },
350
- "_ab_source_file_last_modified": {"type": "string"},
351
- "_ab_source_file_url": {"type": "string"},
352
- },
353
- },
354
- "name": "stream1",
355
- "source_defined_cursor": True,
356
- "supported_sync_modes": ["full_refresh", "incremental"],
357
- }
358
- ]
359
- }
360
- )
361
- ).build()
362
-
363
- avro_all_types_scenario = (
364
- TestScenarioBuilder()
365
- .set_name("avro_all_types_stream")
366
- .set_config(
367
- {
368
- "streams": [
369
- {
370
- "name": "stream1",
371
- "format": {"filetype": "avro"},
372
- "globs": ["*"],
373
- "validation_policy": "Emit Record",
374
- }
375
- ]
376
- }
377
- )
378
- .set_source_builder(
379
- FileBasedSourceBuilder()
380
- .set_stream_reader(TemporaryAvroFilesStreamReader(files=_avro_all_types_file, file_type="avro"))
381
- .set_file_type("avro")
382
- )
383
- .set_expected_records(
384
- [
385
- {
386
- "data": {
387
- "col_bool": True,
388
- "col_int": 27,
389
- "col_long": 1992,
390
- "col_float": 999.09723456,
391
- "col_double": 9123456.12394,
392
- "col_bytes": "\x00\x01\x02\x03",
393
- "col_string": "Love It If We Made It",
394
- "col_record": {"artist": "The 1975", "song": "About You", "year": 2022},
395
- "col_enum": "POP_ROCK",
396
- "col_array": [
397
- "The 1975",
398
- "I Like It When You Sleep, for You Are So Beautiful yet So Unaware of It",
399
- "The 1975 A Brief Inquiry into Online Relationships",
400
- "Notes on a Conditional Form",
401
- "Being Funny in a Foreign Language",
402
- ],
403
- "col_map": {
404
- "lead_singer": "Matty Healy",
405
- "lead_guitar": "Adam Hann",
406
- "bass_guitar": "Ross MacDonald",
407
- "drummer": "George Daniel",
408
- },
409
- "col_fixed": "\x12\x34\x56\x78",
410
- "col_decimal": "1234.56789",
411
- "col_uuid": "123e4567-e89b-12d3-a456-426655440000",
412
- "col_date": "2022-05-29",
413
- "col_time_millis": "06:00:00.456000",
414
- "col_time_micros": "12:00:00.456789",
415
- "col_timestamp_millis": "2022-05-29T00:00:00.456000+00:00",
416
- "col_timestamp_micros": "2022-05-30T00:00:00.456789+00:00",
417
- "_ab_source_file_last_modified": "2023-06-05T03:54:07.000000Z",
418
- "_ab_source_file_url": "a.avro",
419
- },
420
- "stream": "stream1",
421
- },
422
- ]
423
- )
424
- .set_expected_catalog(
425
- {
426
- "streams": [
427
- {
428
- "default_cursor_field": ["_ab_source_file_last_modified"],
429
- "json_schema": {
430
- "type": "object",
431
- "properties": {
432
- "col_array": {"items": {"type": ["null", "string"]}, "type": ["null", "array"]},
433
- "col_bool": {"type": ["null", "boolean"]},
434
- "col_bytes": {"type": ["null", "string"]},
435
- "col_double": {"type": ["null", "number"]},
436
- "col_enum": {"enum": ["POP_ROCK", "INDIE_ROCK", "ALTERNATIVE_ROCK"], "type": ["null", "string"]},
437
- "col_fixed": {"pattern": "^[0-9A-Fa-f]{8}$", "type": ["null", "string"]},
438
- "col_float": {"type": ["null", "number"]},
439
- "col_int": {"type": ["null", "integer"]},
440
- "col_long": {"type": ["null", "integer"]},
441
- "col_map": {"additionalProperties": {"type": ["null", "string"]}, "type": ["null", "object"]},
442
- "col_record": {
443
- "properties": {
444
- "artist": {"type": ["null", "string"]},
445
- "song": {"type": ["null", "string"]},
446
- "year": {"type": ["null", "integer"]},
447
- },
448
- "type": ["null", "object"],
449
- },
450
- "col_string": {"type": ["null", "string"]},
451
- "col_decimal": {"pattern": "^-?\\d{(1, 5)}(?:\\.\\d(1, 5))?$", "type": ["null", "string"]},
452
- "col_uuid": {"type": ["null", "string"]},
453
- "col_date": {"format": "date", "type": ["null", "string"]},
454
- "col_time_millis": {"type": ["null", "integer"]},
455
- "col_time_micros": {"type": ["null", "integer"]},
456
- "col_timestamp_millis": {"format": "date-time", "type": ["null", "string"]},
457
- "col_timestamp_micros": {"type": ["null", "string"]},
458
- "_ab_source_file_last_modified": {"type": "string"},
459
- "_ab_source_file_url": {"type": "string"},
460
- },
461
- },
462
- "name": "stream1",
463
- "source_defined_cursor": True,
464
- "supported_sync_modes": ["full_refresh", "incremental"],
465
- }
466
- ]
467
- }
468
- )
469
- ).build()
470
-
471
- multiple_streams_avro_scenario = (
472
- TestScenarioBuilder()
473
- .set_name("multiple_streams_avro_stream")
474
- .set_config(
475
- {
476
- "streams": [
477
- {
478
- "name": "songs_stream",
479
- "format": {"filetype": "avro"},
480
- "globs": ["*_songs.avro"],
481
- "validation_policy": "Emit Record",
482
- },
483
- {
484
- "name": "festivals_stream",
485
- "format": {"filetype": "avro"},
486
- "globs": ["*_festivals.avro"],
487
- "validation_policy": "Emit Record",
488
- },
489
- ]
490
- }
491
- )
492
- .set_source_builder(
493
- FileBasedSourceBuilder()
494
- .set_stream_reader(TemporaryAvroFilesStreamReader(files=_multiple_avro_stream_file, file_type="avro"))
495
- .set_file_type("avro")
496
- )
497
- .set_expected_records(
498
- [
499
- {
500
- "data": {
501
- "col_title": "Late Night",
502
- "col_album": "A_MOMENT_APART",
503
- "col_year": 2017,
504
- "col_vocals": False,
505
- "_ab_source_file_last_modified": "2023-06-05T03:54:07.000000Z",
506
- "_ab_source_file_url": "odesza_songs.avro",
507
- },
508
- "stream": "songs_stream",
509
- },
510
- {
511
- "data": {
512
- "col_title": "White Lies",
513
- "col_album": "IN_RETURN",
514
- "col_year": 2014,
515
- "col_vocals": True,
516
- "_ab_source_file_last_modified": "2023-06-05T03:54:07.000000Z",
517
- "_ab_source_file_url": "odesza_songs.avro",
518
- },
519
- "stream": "songs_stream",
520
- },
521
- {
522
- "data": {
523
- "col_title": "Wide Awake",
524
- "col_album": "THE_LAST_GOODBYE",
525
- "col_year": 2022,
526
- "col_vocals": True,
527
- "_ab_source_file_last_modified": "2023-06-05T03:54:07.000000Z",
528
- "_ab_source_file_url": "odesza_songs.avro",
529
- },
530
- "stream": "songs_stream",
531
- },
532
- {
533
- "data": {
534
- "col_title": "Sun Models",
535
- "col_album": "SUMMERS_GONE",
536
- "col_year": 2012,
537
- "col_vocals": True,
538
- "_ab_source_file_last_modified": "2023-06-05T03:54:07.000000Z",
539
- "_ab_source_file_url": "odesza_songs.avro",
540
- },
541
- "stream": "songs_stream",
542
- },
543
- {
544
- "data": {
545
- "col_title": "All We Need",
546
- "col_album": "IN_RETURN",
547
- "col_year": 2014,
548
- "col_vocals": True,
549
- "_ab_source_file_last_modified": "2023-06-05T03:54:07.000000Z",
550
- "_ab_source_file_url": "odesza_songs.avro",
551
- },
552
- "stream": "songs_stream",
553
- },
554
- {
555
- "data": {
556
- "col_name": "Coachella",
557
- "col_location": {"country": "USA", "state": "California", "city": "Indio"},
558
- "col_attendance": 250000,
559
- "_ab_source_file_last_modified": "2023-06-06T03:54:07.000000Z",
560
- "_ab_source_file_url": "california_festivals.avro",
561
- },
562
- "stream": "festivals_stream",
563
- },
564
- {
565
- "data": {
566
- "col_name": "CRSSD",
567
- "col_location": {"country": "USA", "state": "California", "city": "San Diego"},
568
- "col_attendance": 30000,
569
- "_ab_source_file_last_modified": "2023-06-06T03:54:07.000000Z",
570
- "_ab_source_file_url": "california_festivals.avro",
571
- },
572
- "stream": "festivals_stream",
573
- },
574
- {
575
- "data": {
576
- "col_name": "Lightning in a Bottle",
577
- "col_location": {"country": "USA", "state": "California", "city": "Buena Vista Lake"},
578
- "col_attendance": 18000,
579
- "_ab_source_file_last_modified": "2023-06-06T03:54:07.000000Z",
580
- "_ab_source_file_url": "california_festivals.avro",
581
- },
582
- "stream": "festivals_stream",
583
- },
584
- {
585
- "data": {
586
- "col_name": "Outside Lands",
587
- "col_location": {"country": "USA", "state": "California", "city": "San Francisco"},
588
- "col_attendance": 220000,
589
- "_ab_source_file_last_modified": "2023-06-06T03:54:07.000000Z",
590
- "_ab_source_file_url": "california_festivals.avro",
591
- },
592
- "stream": "festivals_stream",
593
- },
594
- ]
595
- )
596
- .set_expected_catalog(
597
- {
598
- "streams": [
599
- {
600
- "default_cursor_field": ["_ab_source_file_last_modified"],
601
- "json_schema": {
602
- "type": "object",
603
- "properties": {
604
- "col_title": {"type": ["null", "string"]},
605
- "col_album": {
606
- "type": ["null", "string"],
607
- "enum": ["SUMMERS_GONE", "IN_RETURN", "A_MOMENT_APART", "THE_LAST_GOODBYE"],
608
- },
609
- "col_year": {"type": ["null", "integer"]},
610
- "col_vocals": {"type": ["null", "boolean"]},
611
- "_ab_source_file_last_modified": {"type": "string"},
612
- "_ab_source_file_url": {"type": "string"},
613
- },
614
- },
615
- "name": "songs_stream",
616
- "source_defined_cursor": True,
617
- "supported_sync_modes": ["full_refresh", "incremental"],
618
- },
619
- {
620
- "default_cursor_field": ["_ab_source_file_last_modified"],
621
- "json_schema": {
622
- "type": "object",
623
- "properties": {
624
- "col_name": {"type": ["null", "string"]},
625
- "col_location": {
626
- "properties": {
627
- "country": {"type": ["null", "string"]},
628
- "state": {"type": ["null", "string"]},
629
- "city": {"type": ["null", "string"]},
630
- },
631
- "type": ["null", "object"],
632
- },
633
- "col_attendance": {"type": ["null", "integer"]},
634
- "_ab_source_file_last_modified": {"type": "string"},
635
- "_ab_source_file_url": {"type": "string"},
636
- },
637
- },
638
- "name": "festivals_stream",
639
- "source_defined_cursor": True,
640
- "supported_sync_modes": ["full_refresh", "incremental"],
641
- },
642
- ]
643
- }
644
- )
645
- ).build()
646
-
647
- avro_file_with_double_as_number_scenario = (
648
- TestScenarioBuilder()
649
- .set_name("avro_file_with_double_as_number_stream")
650
- .set_config(
651
- {
652
- "streams": [
653
- {
654
- "name": "stream1",
655
- "globs": ["*"],
656
- "validation_policy": "Emit Record",
657
- "format": {"filetype": "avro", "double_as_string": False},
658
- }
659
- ]
660
- }
661
- )
662
- .set_source_builder(
663
- FileBasedSourceBuilder()
664
- .set_stream_reader(TemporaryAvroFilesStreamReader(files=_multiple_avro_combine_schema_file, file_type="avro"))
665
- .set_file_type("avro")
666
- )
667
- .set_expected_records(
668
- [
669
- {
670
- "data": {
671
- "col_double": 20.02,
672
- "col_string": "Robbers",
673
- "col_album": {"album": "The 1975"},
674
- "_ab_source_file_last_modified": "2023-06-05T03:54:07.000000Z",
675
- "_ab_source_file_url": "a.avro",
676
- },
677
- "stream": "stream1",
678
- },
679
- {
680
- "data": {
681
- "col_double": 20.23,
682
- "col_string": "Somebody Else",
683
- "col_album": {"album": "I Like It When You Sleep, for You Are So Beautiful yet So Unaware of It"},
684
- "_ab_source_file_last_modified": "2023-06-05T03:54:07.000000Z",
685
- "_ab_source_file_url": "a.avro",
686
- },
687
- "stream": "stream1",
688
- },
689
- {
690
- "data": {
691
- "col_double": 1975.1975,
692
- "col_string": "It's Not Living (If It's Not with You)",
693
- "col_song": {"title": "Love It If We Made It"},
694
- "_ab_source_file_last_modified": "2023-06-06T03:54:07.000000Z",
695
- "_ab_source_file_url": "b.avro",
696
- },
697
- "stream": "stream1",
698
- },
699
- {
700
- "data": {
701
- "col_double": 5791.5791,
702
- "col_string": "The 1975",
703
- "col_song": {"title": "About You"},
704
- "_ab_source_file_last_modified": "2023-06-06T03:54:07.000000Z",
705
- "_ab_source_file_url": "b.avro",
706
- },
707
- "stream": "stream1",
708
- },
709
- ]
710
- )
711
- .set_expected_catalog(
712
- {
713
- "streams": [
714
- {
715
- "default_cursor_field": ["_ab_source_file_last_modified"],
716
- "json_schema": {
717
- "type": "object",
718
- "properties": {
719
- "col_double": {"type": ["null", "number"]},
720
- "col_string": {"type": ["null", "string"]},
721
- "col_album": {
722
- "properties": {
723
- "album": {"type": ["null", "string"]},
724
- },
725
- "type": ["null", "object"],
726
- },
727
- "col_song": {
728
- "properties": {
729
- "title": {"type": ["null", "string"]},
730
- },
731
- "type": ["null", "object"],
732
- },
733
- "_ab_source_file_last_modified": {"type": "string"},
734
- "_ab_source_file_url": {"type": "string"},
735
- },
736
- },
737
- "name": "stream1",
738
- "source_defined_cursor": True,
739
- "supported_sync_modes": ["full_refresh", "incremental"],
740
- }
741
- ]
742
- }
743
- )
744
- ).build()