better handling of 503 errors

pyproject.toml

@@ -1,6 +1,6 @@
 [tool.poetry]
 name = "tap-salesforce-connect"
-version = "0.1.1"
+version = "0.1.2"
 description = "`tap-salesforce-connect` is a Singer tap for SalesforceConnect, built with the Meltano Singer SDK."
 readme = "README.md"
 authors = ["Josh Lloyd"]

tap_salesforce_connect/client.py

@@ -7,7 +7,6 @@
 from pathlib import Path
 from typing import Any, Callable, Generator, Iterable
 
-import backoff
 import requests
 from singer_sdk.helpers.jsonpath import extract_jsonpath
 from singer_sdk.streams import RESTStream
@@ -110,24 +109,15 @@ def post_process(self, row: dict, context: dict | None = None) -> dict | None:
                 stringified_row[k] = v
         return stringified_row
 
-    # there is a one-hour rate limit on the SalesforceConnect API
-    def backoff_wait_generator(self) -> Generator[float, None, None]:
-        """Use wait generator for the backoff decorator on request failure.
-
-        See for options:
-        https://github.com/litl/backoff/blob/master/backoff/_wait_gen.py
+    def get_wait_time_based_on_response(self, exception):
+        """Return the number of seconds to wait before retrying.
 
-        And see for examples: `Code Samples <../code_samples.html#custom-backoff>`_
-
-        Returns:
-            The wait generator
+        Salesforce Connect API has a rate limit scoped to an hour
         """
-        return backoff.expo(factor=3)  # type: ignore # ignore 'Returning Any'
+        if exception.response.status_code == 503:
+            return 60 * 60
+        return exception.response.headers.get("Retry-After", 0)
 
-    def backoff_max_tries(self) -> int:
-        """Provide the number of attempts before giving up when retrying requests.
-
-        Returns:
-            Number of max retries.
-        """
-        return 20
+    def backoff_wait_generator(self) -> Generator[float, None, None]:
+        """Return a generator of wait times between retries."""
+        return self.backoff_runtime(value=self.get_wait_time_based_on_response)