airbyte-cdk 6.36.2__py3-none-any.whl → 6.36.3__py3-none-any.whl

airbyte_cdk/sources/declarative/declarative_component_schema.yaml

@@ -1779,6 +1779,9 @@ definitions:
           - stream_interval
           - stream_partition
           - stream_slice
+          - creation_response
+          - polling_response
+          - download_target
         examples:
           - "/products"
           - "/quotes/{{ stream_partition['id'] }}/quote_line_groups"
@@ -3223,7 +3226,7 @@ definitions:
       - polling_requester
       - download_requester
       - status_extractor
-      - urls_extractor
+      - download_target_extractor
     properties:
       type:
         type: string
@@ -3240,7 +3243,7 @@ definitions:
         anyOf:
           - "$ref": "#/definitions/CustomRecordExtractor"
           - "$ref": "#/definitions/DpathExtractor"
-      urls_extractor:
+      download_target_extractor:
         description: Responsible for fetching the final result `urls` provided by the completed / finished / ready async job.
         anyOf:
           - "$ref": "#/definitions/CustomRecordExtractor"
@@ -3261,7 +3264,7 @@ definitions:
         anyOf:
           - "$ref": "#/definitions/CustomRequester"
           - "$ref": "#/definitions/HttpRequester"
-      url_requester:
+      download_target_requester:
         description: Requester component that describes how to prepare HTTP requests to send to the source API to extract the url from polling response by the completed async job.
         anyOf:
           - "$ref": "#/definitions/CustomRequester"
@@ -3667,6 +3670,21 @@ interpolation:
             self: https://api.sendgrid.com/v3/marketing/lists?page_size=1&page_token=
             next: https://api.sendgrid.com/v3/marketing/lists?page_size=1&page_token=0236d6d2
             count: 82
+    - title: creation_response
+      description: The response received from the creation_requester in the AsyncRetriever component.
+      type: object
+      examples:
+        - id: "1234"
+    - title: polling_response
+      description: The response received from the polling_requester in the AsyncRetriever component.
+      type: object
+      examples:
+        - id: "1234"
+    - title: download_target
+      description: The `URL` received from the polling_requester in the AsyncRetriever with jobStatus as `COMPLETED`.
+      type: string
+      examples:
+        - "https://api.sendgrid.com/v3/marketing/lists?page_size=1&page_token=0236d6d2&filename=xxx_yyy_zzz.csv"
     - title: stream_interval
       description: The current stream interval being processed. The keys are defined by the incremental sync component. Default keys are `start_time` and `end_time`.
       type: object

airbyte_cdk/sources/declarative/models/declarative_component_schema.py

@@ -2263,7 +2263,7 @@ class AsyncRetriever(BaseModel):
     status_extractor: Union[CustomRecordExtractor, DpathExtractor] = Field(
         ..., description="Responsible for fetching the actual status of the async job."
     )
-    urls_extractor: Union[CustomRecordExtractor, DpathExtractor] = Field(
+    download_target_extractor: Union[CustomRecordExtractor, DpathExtractor] = Field(
         ...,
         description="Responsible for fetching the final result `urls` provided by the completed / finished / ready async job.",
     )
@@ -2278,7 +2278,7 @@ class AsyncRetriever(BaseModel):
         ...,
         description="Requester component that describes how to prepare HTTP requests to send to the source API to fetch the status of the running async job.",
     )
-    url_requester: Optional[Union[CustomRequester, HttpRequester]] = Field(
+    download_target_requester: Optional[Union[CustomRequester, HttpRequester]] = Field(
         None,
         description="Requester component that describes how to prepare HTTP requests to send to the source API to extract the url from polling response by the completed async job.",
     )

airbyte_cdk/sources/declarative/parsers/model_to_component_factory.py

@@ -2744,32 +2744,32 @@ class ModelToComponentFactory:
             if model.delete_requester
             else None
         )
-        url_requester = (
+        download_target_requester = (
             self._create_component_from_model(
-                model=model.url_requester,
+                model=model.download_target_requester,
                 decoder=decoder,
                 config=config,
                 name=f"job extract_url - {name}",
             )
-            if model.url_requester
+            if model.download_target_requester
             else None
         )
         status_extractor = self._create_component_from_model(
             model=model.status_extractor, decoder=decoder, config=config, name=name
         )
-        urls_extractor = self._create_component_from_model(
-            model=model.urls_extractor, decoder=decoder, config=config, name=name
+        download_target_extractor = self._create_component_from_model(
+            model=model.download_target_extractor, decoder=decoder, config=config, name=name
         )
         job_repository: AsyncJobRepository = AsyncHttpJobRepository(
             creation_requester=creation_requester,
             polling_requester=polling_requester,
             download_retriever=download_retriever,
-            url_requester=url_requester,
+            download_target_requester=download_target_requester,
             abort_requester=abort_requester,
             delete_requester=delete_requester,
             status_extractor=status_extractor,
             status_mapping=self._create_async_job_status_mapping(model.status_mapping, config),
-            urls_extractor=urls_extractor,
+            download_target_extractor=download_target_extractor,
         )
 
         async_job_partition_router = AsyncJobPartitionRouter(

airbyte_cdk/sources/declarative/requesters/README.md

@@ -1,8 +1,8 @@
 # AsyncHttpJobRepository sequence diagram
 
 - Components marked as optional are not required and can be ignored.
-- if `url_requester` is not provided, `urls_extractor` will get urls from the `polling_job_response`
-- interpolation_context, e.g. `create_job_response` or `polling_job_response` can be obtained from stream_slice
+- if `download_target_requester` is not provided, `download_target_extractor` will get urls from the `polling_response`
+- interpolation_context, e.g. `creation_response` or `polling_response` can be obtained from stream_slice
 
 ```mermaid
 ---
@@ -12,7 +12,7 @@ sequenceDiagram
     participant AsyncHttpJobRepository as AsyncOrchestrator
     participant CreationRequester as creation_requester
     participant PollingRequester as polling_requester
-    participant UrlRequester as url_requester (Optional)
+    participant UrlRequester as download_target_requester (Optional)
     participant DownloadRetriever as download_retriever
     participant AbortRequester as abort_requester (Optional)
     participant DeleteRequester as delete_requester (Optional)
@@ -25,14 +25,14 @@ sequenceDiagram
 
     loop Poll for job status
         AsyncHttpJobRepository ->> PollingRequester: Check job status
-        PollingRequester ->> Reporting Server: Status request (interpolation_context: `create_job_response`)
+        PollingRequester ->> Reporting Server: Status request (interpolation_context: `creation_response`)
         Reporting Server -->> PollingRequester: Status response
         PollingRequester -->> AsyncHttpJobRepository: Job status
     end
 
     alt Status: Ready
         AsyncHttpJobRepository ->> UrlRequester: Request download URLs (if applicable)
-        UrlRequester ->> Reporting Server: URL request (interpolation_context: `polling_job_response`)
+        UrlRequester ->> Reporting Server: URL request (interpolation_context: `polling_response`)
         Reporting Server -->> UrlRequester: Download URLs
         UrlRequester -->> AsyncHttpJobRepository: Download URLs
 

airbyte_cdk/sources/declarative/requesters/http_job_repository.py

@@ -43,13 +43,13 @@ class AsyncHttpJobRepository(AsyncJobRepository):
     delete_requester: Optional[Requester]
     status_extractor: DpathExtractor
     status_mapping: Mapping[str, AsyncJobStatus]
-    urls_extractor: DpathExtractor
+    download_target_extractor: DpathExtractor
 
     job_timeout: Optional[timedelta] = None
     record_extractor: RecordExtractor = field(
         init=False, repr=False, default_factory=lambda: ResponseToFileExtractor({})
     )
-    url_requester: Optional[Requester] = (
+    download_target_requester: Optional[Requester] = (
         None  # use it in case polling_requester provides some <id> and extra request is needed to obtain list of urls to download from
     )
 
@@ -211,12 +211,15 @@ class AsyncHttpJobRepository(AsyncJobRepository):
 
         """
 
-        for url in self._get_download_url(job):
+        for target_url in self._get_download_targets(job):
             job_slice = job.job_parameters()
             stream_slice = StreamSlice(
                 partition=job_slice.partition,
                 cursor_slice=job_slice.cursor_slice,
-                extra_fields={**job_slice.extra_fields, "url": url},
+                extra_fields={
+                    **job_slice.extra_fields,
+                    "download_target": target_url,
+                },
             )
             for message in self.download_retriever.read_records({}, stream_slice):
                 if isinstance(message, Record):
@@ -269,27 +272,29 @@ class AsyncHttpJobRepository(AsyncJobRepository):
         del self._polling_job_response_by_id[job_id]
 
     def _get_create_job_stream_slice(self, job: AsyncJob) -> StreamSlice:
+        creation_response = self._create_job_response_by_id[job.api_job_id()].json()
         stream_slice = StreamSlice(
-            partition={"create_job_response": self._create_job_response_by_id[job.api_job_id()]},
+            partition={},
             cursor_slice={},
+            extra_fields={"creation_response": creation_response},
         )
         return stream_slice
 
-    def _get_download_url(self, job: AsyncJob) -> Iterable[str]:
-        if not self.url_requester:
+    def _get_download_targets(self, job: AsyncJob) -> Iterable[str]:
+        if not self.download_target_requester:
             url_response = self._polling_job_response_by_id[job.api_job_id()]
         else:
+            polling_response = self._polling_job_response_by_id[job.api_job_id()].json()
             stream_slice: StreamSlice = StreamSlice(
-                partition={
-                    "polling_job_response": self._polling_job_response_by_id[job.api_job_id()]
-                },
+                partition={},
                 cursor_slice={},
+                extra_fields={"polling_response": polling_response},
             )
-            url_response = self.url_requester.send_request(stream_slice=stream_slice)  # type: ignore # we expect url_requester to always be presented, otherwise raise an exception as we cannot proceed with the report
+            url_response = self.download_target_requester.send_request(stream_slice=stream_slice)  # type: ignore # we expect download_target_requester to always be presented, otherwise raise an exception as we cannot proceed with the report
             if not url_response:
                 raise AirbyteTracedException(
-                    internal_message="Always expect a response or an exception from url_requester",
+                    internal_message="Always expect a response or an exception from download_target_requester",
                     failure_type=FailureType.system_error,
                 )
 
-        yield from self.urls_extractor.extract_records(url_response)  # type: ignore # we expect urls_extractor to always return list of strings
+        yield from self.download_target_extractor.extract_records(url_response)  # type: ignore # we expect download_target_extractor to always return list of strings

airbyte_cdk/sources/declarative/requesters/http_requester.py

@@ -85,7 +85,7 @@ class HttpRequester(Requester):
         self._parameters = parameters
 
         if self.error_handler is not None and hasattr(self.error_handler, "backoff_strategies"):
-            backoff_strategies = self.error_handler.backoff_strategies
+            backoff_strategies = self.error_handler.backoff_strategies  # type: ignore
         else:
             backoff_strategies = None
 
@@ -125,6 +125,12 @@ class HttpRequester(Requester):
         kwargs = {
             "stream_slice": stream_slice,
             "next_page_token": next_page_token,
+            # update the interpolation context with extra fields, if passed.
+            **(
+                stream_slice.extra_fields
+                if stream_slice is not None and hasattr(stream_slice, "extra_fields")
+                else {}
+            ),
         }
         path = str(self._path.eval(self.config, **kwargs))
         return path.lstrip("/")

{airbyte_cdk-6.36.2.dist-info → airbyte_cdk-6.36.3.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: airbyte-cdk
-Version: 6.36.2
+Version: 6.36.3
 Summary: A framework for writing Airbyte Connectors.
 Home-page: https://airbyte.com
 License: MIT

{airbyte_cdk-6.36.2.dist-info → airbyte_cdk-6.36.3.dist-info}/RECORD

@@ -71,7 +71,7 @@ airbyte_cdk/sources/declarative/concurrent_declarative_source.py,sha256=KBF9wdPC
 airbyte_cdk/sources/declarative/datetime/__init__.py,sha256=l9LG7Qm6e5r_qgqfVKnx3mXYtg1I9MmMjomVIPfU4XA,177
 airbyte_cdk/sources/declarative/datetime/datetime_parser.py,sha256=SX9JjdesN1edN2WVUVMzU_ptqp2QB1OnsnjZ4mwcX7w,2579
 airbyte_cdk/sources/declarative/datetime/min_max_datetime.py,sha256=0BHBtDNQZfvwM45-tY5pNlTcKAFSGGNxemoi0Jic-0E,5785
-airbyte_cdk/sources/declarative/declarative_component_schema.yaml,sha256=5o5GsltzbVL2jyXvjWzUoV_r5xpwG_YdLSVUuG_d_34,144548
+airbyte_cdk/sources/declarative/declarative_component_schema.yaml,sha256=ezTm6hN-IvM_0FSgcxc-Axv_9lJBqmJc_vAR_RgsrO4,145340
 airbyte_cdk/sources/declarative/declarative_source.py,sha256=nF7wBqFd3AQmEKAm4CnIo29CJoQL562cJGSCeL8U8bA,1531
 airbyte_cdk/sources/declarative/declarative_stream.py,sha256=venZjfpvtqr3oFSuvMBWtn4h9ayLhD4L65ACuXCDZ64,10445
 airbyte_cdk/sources/declarative/decoders/__init__.py,sha256=JHb_0d3SE6kNY10mxA5YBEKPeSbsWYjByq1gUQxepoE,953
@@ -113,13 +113,13 @@ airbyte_cdk/sources/declarative/migrations/__init__.py,sha256=47DEQpj8HBSa-_TImW
 airbyte_cdk/sources/declarative/migrations/legacy_to_per_partition_state_migration.py,sha256=iemy3fKLczcU0-Aor7tx5jcT6DRedKMqyK7kCOp01hg,3924
 airbyte_cdk/sources/declarative/migrations/state_migration.py,sha256=KWPjealMLKSMtajXgkdGgKg7EmTLR-CqqD7UIh0-eDU,794
 airbyte_cdk/sources/declarative/models/__init__.py,sha256=nUFxNCiKeYRVXuZEKA7GD-lTHxsiKcQ8FitZjKhPIvE,100
-airbyte_cdk/sources/declarative/models/declarative_component_schema.py,sha256=gNL9DqajD2A8UBnKAz7F7YQuYH7frQyHiPQPIMGq2xo,101958
+airbyte_cdk/sources/declarative/models/declarative_component_schema.py,sha256=N9zrGClmsg0st7YKWL6KHE0-haYVVeUF_MZc3NJ6j5Q,101981
 airbyte_cdk/sources/declarative/parsers/__init__.py,sha256=ZnqYNxHsKCgO38IwB34RQyRMXTs4GTvlRi3ImKnIioo,61
 airbyte_cdk/sources/declarative/parsers/custom_code_compiler.py,sha256=958MMX6_ZOJUlDDdNr9Krosgi2bCKGx2Z765M2Woz18,5505
 airbyte_cdk/sources/declarative/parsers/custom_exceptions.py,sha256=Rir9_z3Kcd5Es0-LChrzk-0qubAsiK_RSEnLmK2OXm8,553
 airbyte_cdk/sources/declarative/parsers/manifest_component_transformer.py,sha256=CXwTfD3wSQq3okcqwigpprbHhSURUokh4GK2OmOyKC8,9132
 airbyte_cdk/sources/declarative/parsers/manifest_reference_resolver.py,sha256=IWUOdF03o-aQn0Occo1BJCxU0Pz-QILk5L67nzw2thw,6803
-airbyte_cdk/sources/declarative/parsers/model_to_component_factory.py,sha256=dj7b3s8jaDV7Tb7EXVSzkvC8QY-mxyOf48rxkSMws6A,134851
+airbyte_cdk/sources/declarative/parsers/model_to_component_factory.py,sha256=8gjOsCS8VdkH2-lImilDq8M8-dmGKtWpWyzd5ajzNok,134955
 airbyte_cdk/sources/declarative/partition_routers/__init__.py,sha256=HJ-Syp3p7RpyR_OK0X_a2kSyISfu3W-PKrRI16iY0a8,957
 airbyte_cdk/sources/declarative/partition_routers/async_job_partition_router.py,sha256=VelO7zKqKtzMJ35jyFeg0ypJLQC0plqqIBNXoBW1G2E,3001
 airbyte_cdk/sources/declarative/partition_routers/cartesian_product_stream_slicer.py,sha256=c5cuVFM6NFkuQqG8Z5IwkBuwDrvXZN1CunUOM_L0ezg,6892
@@ -127,7 +127,7 @@ airbyte_cdk/sources/declarative/partition_routers/list_partition_router.py,sha25
 airbyte_cdk/sources/declarative/partition_routers/partition_router.py,sha256=YyEIzdmLd1FjbVP3QbQ2VFCLW_P-OGbVh6VpZShp54k,2218
 airbyte_cdk/sources/declarative/partition_routers/single_partition_router.py,sha256=SKzKjSyfccq4dxGIh-J6ejrgkCHzaiTIazmbmeQiRD4,1942
 airbyte_cdk/sources/declarative/partition_routers/substream_partition_router.py,sha256=LlWj-Ofs-xfjlqmDzH8OYpyblP2Pb8bPDdR9g1UZyt0,17693
-airbyte_cdk/sources/declarative/requesters/README.md,sha256=eL1I4iLkxaw7hJi9S9d18_XcRl-R8lUSjqBVJJzvXmg,2656
+airbyte_cdk/sources/declarative/requesters/README.md,sha256=DQll2qsIzzTiiP35kJp16ONpr7cFeUQNgPfhl5krB24,2675
 airbyte_cdk/sources/declarative/requesters/__init__.py,sha256=d7a3OoHbqaJDyyPli3nqqJ2yAW_SLX6XDaBAKOwvpxw,364
 airbyte_cdk/sources/declarative/requesters/error_handlers/__init__.py,sha256=SkEDcJxlT1683rNx93K9whoS0OyUukkuOfToGtgpF58,776
 airbyte_cdk/sources/declarative/requesters/error_handlers/backoff_strategies/__init__.py,sha256=1WZdpFmWL6W_Dko0qjflTaKIWeqt8jHT-D6HcujIp3s,884
@@ -142,8 +142,8 @@ airbyte_cdk/sources/declarative/requesters/error_handlers/default_error_handler.
 airbyte_cdk/sources/declarative/requesters/error_handlers/default_http_response_filter.py,sha256=q0YkeYUUWO6iErUy0vjqiOkhg8_9d5YcCmtlpXAJJ9E,1314
 airbyte_cdk/sources/declarative/requesters/error_handlers/error_handler.py,sha256=Tan66odx8VHzfdyyXMQkXz2pJYksllGqvxmpoajgcK4,669
 airbyte_cdk/sources/declarative/requesters/error_handlers/http_response_filter.py,sha256=E-fQbt4ShfxZVoqfnmOx69C6FUPWZz8BIqI3DN9Kcjs,7935
-airbyte_cdk/sources/declarative/requesters/http_job_repository.py,sha256=Wjh5ARW1kcuRfihfFEloQ_l1N6yY1NN_4qFZJiqlH4o,11612
-airbyte_cdk/sources/declarative/requesters/http_requester.py,sha256=pR2uR5b9eGyvYIOYwus3mz3OaqRu1ozwja_ys1SE7hc,14952
+airbyte_cdk/sources/declarative/requesters/http_job_repository.py,sha256=4wpP0ZNTMLugi-Rc1OFdFaxWfRZSl45nzhHqMFCE8SQ,11924
+airbyte_cdk/sources/declarative/requesters/http_requester.py,sha256=Sie8IyntFu66UoJASwpWV0WrRDBr9lpHWSOws7vZfM0,15228
 airbyte_cdk/sources/declarative/requesters/paginators/__init__.py,sha256=uArbKs9JKNCt7t9tZoeWwjDpyI1HoPp29FNW0JzvaEM,644
 airbyte_cdk/sources/declarative/requesters/paginators/default_paginator.py,sha256=ZW4lwWNAzb4zL0jKc-HjowP5-y0Zg9xi0YlK6tkx_XY,12057
 airbyte_cdk/sources/declarative/requesters/paginators/no_pagination.py,sha256=j6j9QRPaTbKQ2N661RFVKthhkWiodEp6ut0tKeEd0Ng,2019
@@ -360,9 +360,9 @@ airbyte_cdk/utils/slice_hasher.py,sha256=EDxgROHDbfG-QKQb59m7h_7crN1tRiawdf5uU7G
 airbyte_cdk/utils/spec_schema_transformations.py,sha256=-5HTuNsnDBAhj-oLeQXwpTGA0HdcjFOf2zTEMUTTg_Y,816
 airbyte_cdk/utils/stream_status_utils.py,sha256=ZmBoiy5HVbUEHAMrUONxZvxnvfV9CesmQJLDTAIWnWw,1171
 airbyte_cdk/utils/traced_exception.py,sha256=C8uIBuCL_E4WnBAOPSxBicD06JAldoN9fGsQDp463OY,6292
-airbyte_cdk-6.36.2.dist-info/LICENSE.txt,sha256=Wfe61S4BaGPj404v8lrAbvhjYR68SHlkzeYrg3_bbuM,1051
-airbyte_cdk-6.36.2.dist-info/LICENSE_SHORT,sha256=aqF6D1NcESmpn-cqsxBtszTEnHKnlsp8L4x9wAh3Nxg,55
-airbyte_cdk-6.36.2.dist-info/METADATA,sha256=un3DK_o5RZ4nnhj5U_QBHoYqSfUYbAHBb4cfAEModjk,6010
-airbyte_cdk-6.36.2.dist-info/WHEEL,sha256=Nq82e9rUAnEjt98J6MlVmMCZb-t9cYE2Ir1kpBmnWfs,88
-airbyte_cdk-6.36.2.dist-info/entry_points.txt,sha256=fj-e3PAQvsxsQzyyq8UkG1k8spunWnD4BAH2AwlR6NM,95
-airbyte_cdk-6.36.2.dist-info/RECORD,,
+airbyte_cdk-6.36.3.dist-info/LICENSE.txt,sha256=Wfe61S4BaGPj404v8lrAbvhjYR68SHlkzeYrg3_bbuM,1051
+airbyte_cdk-6.36.3.dist-info/LICENSE_SHORT,sha256=aqF6D1NcESmpn-cqsxBtszTEnHKnlsp8L4x9wAh3Nxg,55
+airbyte_cdk-6.36.3.dist-info/METADATA,sha256=hOSYqconZWZ42hn8s2vyw_43Ou2dMvbj6M6XoIFMPmo,6010
+airbyte_cdk-6.36.3.dist-info/WHEEL,sha256=Nq82e9rUAnEjt98J6MlVmMCZb-t9cYE2Ir1kpBmnWfs,88
+airbyte_cdk-6.36.3.dist-info/entry_points.txt,sha256=fj-e3PAQvsxsQzyyq8UkG1k8spunWnD4BAH2AwlR6NM,95
+airbyte_cdk-6.36.3.dist-info/RECORD,,