airbyte-source-github 1.8.35.dev202507171002__py3-none-any.whl → 1.8.36.dev202507211247__py3-none-any.whl

{airbyte_source_github-1.8.35.dev202507171002.dist-info → airbyte_source_github-1.8.36.dev202507211247.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.3
 Name: airbyte-source-github
-Version: 1.8.35.dev202507171002
+Version: 1.8.36.dev202507211247
 Summary: Source implementation for GitHub.
 License: MIT
 Author: Airbyte

{airbyte_source_github-1.8.35.dev202507171002.dist-info → airbyte_source_github-1.8.36.dev202507211247.dist-info}/RECORD

@@ -1,5 +1,5 @@
 source_github/__init__.py,sha256=punPc3v0mXEYOun7cbkfM5KUhgjv72B9DgDhI4VtzcQ,1134
-source_github/backoff_strategies.py,sha256=rzzSxuf9FEGOb7aXqPgl4A4dIKggW3uPnESwloVX1bM,3164
+source_github/backoff_strategies.py,sha256=0BWPR4_EtguVIIhN1-_fsPExiGs40FuXrtK_4qxHeKg,2936
 source_github/config_migrations.py,sha256=H58hHqAnuvb0B8IXHW4aEDZ3HotEg7HdA2rXDG9XW7A,3832
 source_github/constants.py,sha256=Hj3Q4y7OoU-Iff4m9gEC2CjwmWJYXhNbHVNjg8EBLmQ,238
 source_github/errors_handlers.py,sha256=POqpvbrNAoZztjwByOJxJbIaxhWC8KWLi3gK1WzbiK8,7259
@@ -57,8 +57,8 @@ source_github/schemas/workflows.json,sha256=gSNw8WZaVKbX4AL97PbjZHzvxcOltXqv9Ao1
 source_github/source.py,sha256=1o8eayigi4xSUeNHdCd-mhNswGUq_XQrVk2eihTjm1o,14246
 source_github/spec.json,sha256=7LOQm01fP_RvPF-HifhNPJ7i0AxT2LTNPaLAA3uOfNY,7443
 source_github/streams.py,sha256=h5YMPLIsLTv7WX_mcURVC2LmmWBLraZvKH8J_GzV1IE,77441
-source_github/utils.py,sha256=Ztd8VWwzTNUg_A96_8R9XSKorIcBa8wJ6aYUqygRkyk,5492
-airbyte_source_github-1.8.35.dev202507171002.dist-info/METADATA,sha256=CAKHjaasoVkCAFMszFj0iAOPjvJGsmzXz6JgCav5Lws,5218
-airbyte_source_github-1.8.35.dev202507171002.dist-info/WHEEL,sha256=b4K_helf-jlQoXBBETfwnf4B04YC67LOev0jo4fX5m8,88
-airbyte_source_github-1.8.35.dev202507171002.dist-info/entry_points.txt,sha256=gYhqVrTAZvMwuYByg0b_-o115yUFLLcfNxMrLZmiW9k,55
-airbyte_source_github-1.8.35.dev202507171002.dist-info/RECORD,,
+source_github/utils.py,sha256=SzjCYAOOtpzmpj7A3KOAMSlY-v2HSWOXOWUjxOk2Ges,5959
+airbyte_source_github-1.8.36.dev202507211247.dist-info/METADATA,sha256=NS3dCMe40d7sDPv3VEiGAp8__vLr6nGGKRb_hjhiEAk,5218
+airbyte_source_github-1.8.36.dev202507211247.dist-info/WHEEL,sha256=b4K_helf-jlQoXBBETfwnf4B04YC67LOev0jo4fX5m8,88
+airbyte_source_github-1.8.36.dev202507211247.dist-info/entry_points.txt,sha256=gYhqVrTAZvMwuYByg0b_-o115yUFLLcfNxMrLZmiW9k,55
+airbyte_source_github-1.8.36.dev202507211247.dist-info/RECORD,,

source_github/backoff_strategies.py

@@ -9,15 +9,12 @@ import requests
 
 from airbyte_cdk import BackoffStrategy
 from airbyte_cdk.sources.streams.http import HttpStream
-from airbyte_cdk.utils import AirbyteTracedException
-from airbyte_protocol.models import FailureType
 
 
 class GithubStreamABCBackoffStrategy(BackoffStrategy):
     min_backoff_time = 60.0
-    min_backoff_time_when_refreshing_auth_token = 60.0 * 10  # 10 minutes as default and as default time for access token live period
-    # 3600 - max seconds between messages from metadata
-    max_seconds_between_messages = 3600.0
+    # https://docs.github.com/en/rest/using-the-rest-api/troubleshooting-the-rest-api?apiVersion=2022-11-28
+    RATE_LIMITS_STATUS_CODES = [403, 429]
 
     def __init__(self, stream: HttpStream, **kwargs):  # type: ignore # noqa
         self.stream = stream
@@ -33,26 +30,28 @@ class GithubStreamABCBackoffStrategy(BackoffStrategy):
             retry_after = response_or_exception.headers.get("Retry-After")
             if retry_after is not None:
                 backoff_time_in_seconds = max(float(retry_after), self.min_backoff_time)
-                return self.get_waiting_time(backoff_time_in_seconds)
+                return self.get_waiting_time(backoff_time_in_seconds, response_or_exception)
 
             reset_time = response_or_exception.headers.get("X-RateLimit-Reset")
             if reset_time:
                 backoff_time_in_seconds = max(float(reset_time) - time.time(), self.min_backoff_time)
-                return self.get_waiting_time(backoff_time_in_seconds)
+                return self.get_waiting_time(backoff_time_in_seconds, response_or_exception)
         return None
 
-    def get_waiting_time(self, backoff_time_in_seconds: float) -> Optional[float]:
+    def get_waiting_time(self, backoff_time_in_seconds: float, response: requests.Response) -> Optional[float]:
         if backoff_time_in_seconds < 60 * 10:  # type: ignore[operator]
             return backoff_time_in_seconds
-        elif backoff_time_in_seconds > self.max_seconds_between_messages:
-            raise AirbyteTracedException(
-                internal_message="Waiting time from header is too long.",
-                message=f"The stream {self.stream.name} have faced rate limits, but waiting time is too long. The stream will sync data in the next sync when rate limits are refreshed.",
-                failure_type=FailureType.transient_error,
-            )
         else:
-            self.stream._http_client._session.auth.update_token()  # New token will be used in the next request
-            return self.stream._http_client._session.auth.max_time or self.min_backoff_time_when_refreshing_auth_token
+            # New token will be used in the next request or fail if all tokens are exhausted
+            if response.status_code in self.RATE_LIMITS_STATUS_CODES:
+                self.stream._http_client._session.auth.find_available_token(response.request)
+            else:
+                self.stream._http_client._session.auth.update_token()
+
+            # update headers in the request itself so the next request will use new token
+            response.request.headers.update(self.stream._http_client._session.auth.get_auth_header())
+
+            return 1
 
 
 class ContributorActivityBackoffStrategy(BackoffStrategy):

source_github/utils.py

@@ -76,8 +76,14 @@ class MultipleTokenAuthenticatorWithRateLimiter(AbstractHeaderAuthenticator):
             return {self.auth_header: self.token}
         return {}
 
-    def __call__(self, request):
-        """Attach the HTTP headers required to authenticate on the HTTP request"""
+    def find_available_token(self, request: requests.Request):
+        """
+        Iterates over each provided token and check their availability.
+        If no tokens are available or reset time is more than 10 minutes, an error is raised.
+        Method process_token() returns True is current token is available,
+        otherwise goes to the next token and sets it as current and checks for availability.
+        """
+
         while True:
             current_token = self._tokens[self.current_active_token]
             if "graphql" in request.path_url:
@@ -87,6 +93,11 @@ class MultipleTokenAuthenticatorWithRateLimiter(AbstractHeaderAuthenticator):
                 if self.process_token(current_token, "count_rest", "reset_at_rest"):
                     break
 
+    def __call__(self, request):
+        """Attach the HTTP headers required to authenticate on the HTTP request"""
+
+        self.find_available_token(request)
+
         request.headers.update(self.get_auth_header())
 
         return request