airbyte-cdk 6.37.0.dev0__tar.gz → 6.37.2.dev1__tar.gz

{airbyte_cdk-6.37.0.dev0 → airbyte_cdk-6.37.2.dev1}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: airbyte-cdk
-Version: 6.37.0.dev0
+Version: 6.37.2.dev1
 Summary: A framework for writing Airbyte Connectors.
 Home-page: https://airbyte.com
 License: MIT

{airbyte_cdk-6.37.0.dev0 → airbyte_cdk-6.37.2.dev1}/airbyte_cdk/entrypoint.py

@@ -37,8 +37,8 @@ from airbyte_cdk.sources import Source
 from airbyte_cdk.sources.connector_state_manager import HashableStreamDescriptor
 from airbyte_cdk.sources.utils.schema_helpers import check_config_against_spec_or_exit, split_config
 
-from airbyte_cdk.utils import PrintBuffer, is_cloud_environment, message_utils  # add PrintBuffer back once fixed
-# from airbyte_cdk.utils import is_cloud_environment, message_utils
+# from airbyte_cdk.utils import PrintBuffer, is_cloud_environment, message_utils  # add PrintBuffer back once fixed
+from airbyte_cdk.utils import is_cloud_environment, message_utils
 from airbyte_cdk.utils.airbyte_secrets_utils import get_secrets, update_secrets
 from airbyte_cdk.utils.constants import ENV_REQUEST_CACHE_PATH
 from airbyte_cdk.utils.traced_exception import AirbyteTracedException
@@ -337,11 +337,11 @@ def launch(source: Source, args: List[str]) -> None:
     parsed_args = source_entrypoint.parse_args(args)
     # temporarily removes the PrintBuffer because we're seeing weird print behavior for concurrent syncs
     # Refer to: https://github.com/airbytehq/oncall/issues/6235
-    with PrintBuffer():
-        for message in source_entrypoint.run(parsed_args):
-            # simply printing is creating issues for concurrent CDK as Python uses different two instructions to print: one for the message and
-            # the other for the break line. Adding `\n` to the message ensure that both are printed at the same time
-            print(f"{message}\n", end="", flush=True)
+    # with PrintBuffer():
+    for message in source_entrypoint.run(parsed_args):
+        # simply printing is creating issues for concurrent CDK as Python uses different two instructions to print: one for the message and
+        # the other for the break line. Adding `\n` to the message ensure that both are printed at the same time
+        print(f"{message}\n", end="", flush=True)
 
 
 def _init_internal_request_filter() -> None:

{airbyte_cdk-6.37.0.dev0 → airbyte_cdk-6.37.2.dev1}/airbyte_cdk/sources/declarative/decoders/composite_raw_decoder.py

@@ -5,7 +5,7 @@ import json
 import logging
 from abc import ABC, abstractmethod
 from dataclasses import dataclass
-from io import BufferedIOBase, TextIOWrapper
+from io import BufferedIOBase, BytesIO, TextIOWrapper
 from typing import Any, Generator, MutableMapping, Optional
 
 import orjson
@@ -124,7 +124,8 @@ class CsvParser(Parser):
         """
         Parse CSV data from decompressed bytes.
         """
-        text_data = TextIOWrapper(data, encoding=self.encoding)  # type: ignore
+        bytes_data = BytesIO(data.read())
+        text_data = TextIOWrapper(bytes_data, encoding=self.encoding)  # type: ignore
         reader = csv.DictReader(text_data, delimiter=self._get_delimiter() or ",")
         for row in reader:
             yield row

{airbyte_cdk-6.37.0.dev0 → airbyte_cdk-6.37.2.dev1}/airbyte_cdk/sources/declarative/extractors/response_to_file_extractor.py

@@ -136,6 +136,7 @@ class ResponseToFileExtractor(RecordExtractor):
         """
 
         try:
+            # TODO: Add support for other file types, like `json`, with `pd.read_json()`
             with open(path, "r", encoding=file_encoding) as data:
                 chunks = pd.read_csv(
                     data, chunksize=chunk_size, iterator=True, dialect="unix", dtype=object

{airbyte_cdk-6.37.0.dev0 → airbyte_cdk-6.37.2.dev1}/airbyte_cdk/sources/declarative/requesters/http_job_repository.py

@@ -23,6 +23,7 @@ from airbyte_cdk.sources.declarative.extractors.response_to_file_extractor impor
 )
 from airbyte_cdk.sources.declarative.requesters.requester import Requester
 from airbyte_cdk.sources.declarative.retrievers.simple_retriever import SimpleRetriever
+from airbyte_cdk.sources.http_logger import format_http_message
 from airbyte_cdk.sources.types import Record, StreamSlice
 from airbyte_cdk.utils import AirbyteTracedException
 
@@ -71,7 +72,15 @@ class AsyncHttpJobRepository(AsyncJobRepository):
         """
 
         polling_response: Optional[requests.Response] = self.polling_requester.send_request(
-            stream_slice=stream_slice
+            stream_slice=stream_slice,
+            log_formatter=lambda polling_response: format_http_message(
+                response=polling_response,
+                title="Async Job -- Polling",
+                description="Poll the status of the server-side async job.",
+                stream_name=None,
+                is_auxiliary=True,
+                type="ASYNC_POLL",
+            ),
         )
         if polling_response is None:
             raise AirbyteTracedException(
@@ -118,8 +127,17 @@ class AsyncHttpJobRepository(AsyncJobRepository):
         """
 
         response: Optional[requests.Response] = self.creation_requester.send_request(
-            stream_slice=stream_slice
+            stream_slice=stream_slice,
+            log_formatter=lambda response: format_http_message(
+                response=response,
+                title="Async Job -- Create",
+                description="Create the server-side async job.",
+                stream_name=None,
+                is_auxiliary=True,
+                type="ASYNC_CREATE",
+            ),
         )
+
         if not response:
             raise AirbyteTracedException(
                 internal_message="Always expect a response or an exception from creation_requester",
@@ -217,13 +235,33 @@ class AsyncHttpJobRepository(AsyncJobRepository):
         if not self.abort_requester:
             return
 
-        self.abort_requester.send_request(stream_slice=self._get_create_job_stream_slice(job))
+        abort_response = self.abort_requester.send_request(
+            stream_slice=self._get_create_job_stream_slice(job),
+            log_formatter=lambda abort_response: format_http_message(
+                response=abort_response,
+                title="Async Job -- Abort",
+                description="Abort the running server-side async job.",
+                stream_name=None,
+                is_auxiliary=True,
+                type="ASYNC_ABORT",
+            ),
+        )
 
     def delete(self, job: AsyncJob) -> None:
         if not self.delete_requester:
             return
 
-        self.delete_requester.send_request(stream_slice=self._get_create_job_stream_slice(job))
+        delete_job_reponse = self.delete_requester.send_request(
+            stream_slice=self._get_create_job_stream_slice(job),
+            log_formatter=lambda delete_job_reponse: format_http_message(
+                response=delete_job_reponse,
+                title="Async Job -- Delete",
+                description="Delete the specified job from the list of Jobs.",
+                stream_name=None,
+                is_auxiliary=True,
+                type="ASYNC_DELETE",
+            ),
+        )
         self._clean_up_job(job.api_job_id())
 
     def _clean_up_job(self, job_id: str) -> None:

{airbyte_cdk-6.37.0.dev0 → airbyte_cdk-6.37.2.dev1}/airbyte_cdk/sources/declarative/retrievers/async_retriever.py

@@ -1,13 +1,12 @@
 # Copyright (c) 2024 Airbyte, Inc., all rights reserved.
 
 
-from dataclasses import InitVar, dataclass
+from dataclasses import InitVar, dataclass, field
 from typing import Any, Iterable, Mapping, Optional
 
 from typing_extensions import deprecated
 
 from airbyte_cdk.sources.declarative.async_job.job import AsyncJob
-from airbyte_cdk.sources.declarative.async_job.job_orchestrator import AsyncPartition
 from airbyte_cdk.sources.declarative.extractors.record_selector import RecordSelector
 from airbyte_cdk.sources.declarative.partition_routers.async_job_partition_router import (
     AsyncJobPartitionRouter,
@@ -16,6 +15,7 @@ from airbyte_cdk.sources.declarative.retrievers.retriever import Retriever
 from airbyte_cdk.sources.source import ExperimentalClassWarning
 from airbyte_cdk.sources.streams.core import StreamData
 from airbyte_cdk.sources.types import Config, StreamSlice, StreamState
+from airbyte_cdk.sources.utils.slice_logger import AlwaysLogSliceLogger
 
 
 @deprecated(
@@ -28,6 +28,10 @@ class AsyncRetriever(Retriever):
     parameters: InitVar[Mapping[str, Any]]
     record_selector: RecordSelector
     stream_slicer: AsyncJobPartitionRouter
+    slice_logger: AlwaysLogSliceLogger = field(
+        init=False,
+        default_factory=lambda: AlwaysLogSliceLogger(),
+    )
 
     def __post_init__(self, parameters: Mapping[str, Any]) -> None:
         self._parameters = parameters
@@ -75,13 +79,16 @@ class AsyncRetriever(Retriever):
         return stream_slice.extra_fields.get("jobs", []) if stream_slice else []
 
     def stream_slices(self) -> Iterable[Optional[StreamSlice]]:
-        return self.stream_slicer.stream_slices()
+        yield from self.stream_slicer.stream_slices()
 
     def read_records(
         self,
         records_schema: Mapping[str, Any],
         stream_slice: Optional[StreamSlice] = None,
     ) -> Iterable[StreamData]:
+        # emit the slice_descriptor log message, for connector builder TestRead
+        yield self.slice_logger.create_slice_log_message(stream_slice.cursor_slice)  # type: ignore
+
         stream_state: StreamState = self._get_stream_state()
         jobs: Iterable[AsyncJob] = self._validate_and_get_stream_slice_jobs(stream_slice)
         records: Iterable[Mapping[str, Any]] = self.stream_slicer.fetch_records(jobs)

{airbyte_cdk-6.37.0.dev0 → airbyte_cdk-6.37.2.dev1}/pyproject.toml

@@ -23,7 +23,7 @@ classifiers = [
 keywords = ["airbyte", "connector-development-kit", "cdk"]
 
 # Python CDK uses dynamic versioning: https://github.com/mtkennerly/poetry-dynamic-versioning
-version = "6.37.0dev0"  # Version will be calculated dynamically.
+version = "6.37.2dev1"  # Version will be calculated dynamically.
 
 [tool.poetry-dynamic-versioning]
 enable = false