fix(file-system): Reclaim orphaned in-progress requests on RQ recovery (#1825)

## Summary

- When a process crashed after `fetch_next_request` but before
`mark_request_as_handled` or `reclaim_request`, the request remained in
persisted `in_progress_requests` and was silently skipped on restart
- Clear `in_progress_requests` during `_discover_existing_requests` so
orphaned requests are retried
- Add unit test simulating the crash-recovery scenario

---

Co-authored-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

Author: vdusek

src/crawlee/storage_clients/_file_system/_request_queue_client.py

@@ -798,10 +798,20 @@ async def _parse_request_file(cls, file_path: Path) -> Request | None:
             return None
 
     async def _discover_existing_requests(self) -> None:
-        """Discover and load existing requests into the state when opening an existing request queue."""
+        """Discover and load existing requests into the state when opening an existing request queue.
+
+        On recovery after a crash, any requests that were previously in-progress are reclaimed as pending,
+        since there is no active processing after a restart.
+        """
         request_files = await self._get_request_files(self.path_to_rq)
         state = self._state.current_value
 
+        if state.in_progress_requests:
+            logger.info(
+                f'Reclaiming {len(state.in_progress_requests)} in-progress request(s) from previous run.',
+            )
+            state.in_progress_requests.clear()
+
         for request_file in request_files:
             request = await self._parse_request_file(request_file)
             if request is None:

tests/unit/storage_clients/_file_system/test_fs_rq_client.py

@@ -175,6 +175,51 @@ async def test_data_persistence_across_reopens() -> None:
     await reopened_client.drop()
 
 
+async def test_in_progress_requests_recovered_after_crash() -> None:
+    """Test that requests left in-progress at crash time are recovered as pending on reopen.
+
+    Simulates a crash: requests are added, one is fetched (in-progress), state is persisted,
+    then the queue is reopened. The in-progress request should be available for fetching again.
+    """
+    storage_client = FileSystemStorageClient()
+
+    # Create RQ and add requests.
+    original_client = await storage_client.create_rq_client(name='crash-recovery-test')
+
+    test_requests = [
+        Request.from_url('https://example.com/1'),
+        Request.from_url('https://example.com/2'),
+        Request.from_url('https://example.com/3'),
+    ]
+    await original_client.add_batch_of_requests(test_requests)
+
+    # Fetch one request, putting it in-progress (simulating work before crash).
+    fetched = await original_client.fetch_next_request()
+    assert fetched is not None
+
+    # Persist state explicitly (simulating what happens periodically or at crash boundary).
+    await original_client._state.persist_state()
+
+    rq_id = (await original_client.get_metadata()).id
+
+    # Simulate crash: reopen the queue without calling mark_request_as_handled or reclaim_request.
+    reopened_client = await storage_client.create_rq_client(id=rq_id)
+
+    # All 3 requests should be fetchable (the in-progress one should have been reclaimed).
+    fetched_urls = set()
+    for _ in range(3):
+        req = await reopened_client.fetch_next_request()
+        assert req is not None, f'Expected 3 fetchable requests, only got {len(fetched_urls)}'
+        fetched_urls.add(req.url)
+
+    assert fetched_urls == {'https://example.com/1', 'https://example.com/2', 'https://example.com/3'}
+
+    # No more requests should be available.
+    assert await reopened_client.fetch_next_request() is None
+
+    await reopened_client.drop()
+
+
 async def test_get_request_does_not_mark_in_progress(rq_client: FileSystemRequestQueueClient) -> None:
     """Test that get_request does not block a request from being fetched."""
     request = Request.from_url('https://example.com/blocked')