Add hacky fix for rate-limiter segregation between pytests

backend/linkpulse/dependencies.py

@@ -1,3 +1,4 @@
+import os
 import structlog
 from fastapi import HTTPException, Request, Response, status
 from limits.strategies import MovingWindowRateLimiter
@@ -8,6 +9,7 @@ storage = MemoryStorage()
 strategy = MovingWindowRateLimiter(storage)
 
 logger = structlog.get_logger()
+is_pytest = os.environ.get("PYTEST_VERSION") is not None
 
 
 class RateLimiter:
@@ -16,15 +18,19 @@ class RateLimiter:
         self.retry_after = str(self.limit.get_expiry())
 
     async def __call__(self, request: Request, response: Response):
-        logger.debug("Rate limiting request", path=request.url.path)
-
         key = request.headers.get("X-Real-IP")
+
         if key is None:
             if request.client is None:
                 logger.warning("No client information available for request.")
                 return False
             key = request.client.host
 
+        if is_pytest:
+            # This is somewhat hacky, I'm not sure if there's a way it can break during pytesting, but look here if odd rate limiting errors occur during tests
+            # The reason for this is so tests don't compete with each other for rate limiting
+            key += "." + os.environ["PYTEST_CURRENT_TEST"]
+
         if not strategy.hit(self.limit, key):
             logger.warning("Rate limit exceeded", key=key)
             raise HTTPException(