apify 3.0.2b6__tar.gz → 3.0.2b7__tar.gz

{apify-3.0.2b6 → apify-3.0.2b7}/CHANGELOG.md

@@ -9,6 +9,7 @@ All notable changes to this project will be documented in this file.
 
 - Handle None result in single consumer request queue client ([#623](https://github.com/apify/apify-sdk-python/pull/623)) ([451284a](https://github.com/apify/apify-sdk-python/commit/451284a5c633bc5613bd1e9060df286a1c20b259)) by [@janbuchar](https://github.com/janbuchar), closes [#1472](https://github.com/apify/apify-sdk-python/issues/1472)
 - Unify Actor context manager with init & exit methods ([#600](https://github.com/apify/apify-sdk-python/pull/600)) ([6b0d084](https://github.com/apify/apify-sdk-python/commit/6b0d0842ae66a3a206bbb682a3e5f81ad552f029)) by [@vdusek](https://github.com/vdusek), closes [#598](https://github.com/apify/apify-sdk-python/issues/598)
+- Handle truncated `unique_key` in `list_head` by fetching full request data ([#631](https://github.com/apify/apify-sdk-python/pull/631)) ([4238086](https://github.com/apify/apify-sdk-python/commit/423808678d9155a84a266bf50bb09f1a56466174)) by [@vdusek](https://github.com/vdusek), closes [#627](https://github.com/apify/apify-sdk-python/issues/627)
 
 
 <!-- git-cliff-unreleased-end -->

{apify-3.0.2b6 → apify-3.0.2b7}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: apify
-Version: 3.0.2b6
+Version: 3.0.2b7
 Summary: Apify SDK for Python
 Project-URL: Apify Homepage, https://apify.com
 Project-URL: Changelog, https://docs.apify.com/sdk/python/docs/changelog

{apify-3.0.2b6 → apify-3.0.2b7}/pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "hatchling.build"
 
 [project]
 name = "apify"
-version = "3.0.2b6"
+version = "3.0.2b7"
 description = "Apify SDK for Python"
 authors = [{ name = "Apify Technologies s.r.o.", email = "support@apify.com" }]
 license = { file = "LICENSE" }

{apify-3.0.2b6 → apify-3.0.2b7}/src/apify/storage_clients/_apify/_request_queue_client.py

@@ -12,8 +12,8 @@ from crawlee.storage_clients.models import AddRequestsResponse, ProcessedRequest
 from crawlee.storages import RequestQueue
 
 from ._models import ApifyRequestQueueMetadata, RequestQueueStats
-from ._request_queue_shared_client import _ApifyRequestQueueSharedClient
-from ._request_queue_single_client import _ApifyRequestQueueSingleClient
+from ._request_queue_shared_client import ApifyRequestQueueSharedClient
+from ._request_queue_single_client import ApifyRequestQueueSingleClient
 from ._utils import AliasResolver
 
 if TYPE_CHECKING:
@@ -47,14 +47,14 @@ class ApifyRequestQueueClient(RequestQueueClient):
         self._api_client = api_client
         """The Apify request queue client for API operations."""
 
-        self._implementation: _ApifyRequestQueueSingleClient | _ApifyRequestQueueSharedClient
+        self._implementation: ApifyRequestQueueSingleClient | ApifyRequestQueueSharedClient
         """Internal implementation used to communicate with the Apify platform based Request Queue."""
         if access == 'single':
-            self._implementation = _ApifyRequestQueueSingleClient(
+            self._implementation = ApifyRequestQueueSingleClient(
                 api_client=self._api_client, metadata=metadata, cache_size=self._MAX_CACHED_REQUESTS
             )
         elif access == 'shared':
-            self._implementation = _ApifyRequestQueueSharedClient(
+            self._implementation = ApifyRequestQueueSharedClient(
                 api_client=self._api_client,
                 metadata=metadata,
                 cache_size=self._MAX_CACHED_REQUESTS,

{apify-3.0.2b6 → apify-3.0.2b7}/src/apify/storage_clients/_apify/_request_queue_shared_client.py

@@ -23,7 +23,7 @@ if TYPE_CHECKING:
 logger = getLogger(__name__)
 
 
-class _ApifyRequestQueueSharedClient:
+class ApifyRequestQueueSharedClient:
     """An Apify platform implementation of the request queue client.
 
     This implementation supports multiple producers and multiple consumers scenario.
@@ -106,23 +106,19 @@ class _ApifyRequestQueueSharedClient:
                 # It could have been handled by another client in the meantime, so cached information about
                 # `request.was_already_handled` is not reliable.
                 already_present_requests.append(
-                    ProcessedRequest.model_validate(
-                        {
-                            'uniqueKey': request.unique_key,
-                            'wasAlreadyPresent': True,
-                            'wasAlreadyHandled': request.was_already_handled,
-                        }
+                    ProcessedRequest(
+                        unique_key=request.unique_key,
+                        was_already_present=True,
+                        was_already_handled=request.was_already_handled,
                     )
                 )
 
             else:
                 # Add new request to the cache.
-                processed_request = ProcessedRequest.model_validate(
-                    {
-                        'uniqueKey': request.unique_key,
-                        'wasAlreadyPresent': True,
-                        'wasAlreadyHandled': request.was_already_handled,
-                    }
+                processed_request = ProcessedRequest(
+                    unique_key=request.unique_key,
+                    was_already_present=True,
+                    was_already_handled=request.was_already_handled,
                 )
                 self._cache_request(
                     request.unique_key,

{apify-3.0.2b6 → apify-3.0.2b7}/src/apify/storage_clients/_apify/_request_queue_single_client.py

@@ -21,7 +21,7 @@ if TYPE_CHECKING:
 logger = getLogger(__name__)
 
 
-class _ApifyRequestQueueSingleClient:
+class ApifyRequestQueueSingleClient:
     """An Apify platform implementation of the request queue client with limited capability.
 
     This client is designed to use as little resources as possible, but has to be used in constrained context.
@@ -108,23 +108,19 @@ class _ApifyRequestQueueSingleClient:
             # Check if request is known to be already handled (it has to be present as well.)
             if request.unique_key in self._requests_already_handled:
                 already_present_requests.append(
-                    ProcessedRequest.model_validate(
-                        {
-                            'uniqueKey': request.unique_key,
-                            'wasAlreadyPresent': True,
-                            'wasAlreadyHandled': True,
-                        }
+                    ProcessedRequest(
+                        unique_key=request.unique_key,
+                        was_already_present=True,
+                        was_already_handled=True,
                     )
                 )
             # Check if request is known to be already present, but unhandled
             elif self._requests_cache.get(request.unique_key):
                 already_present_requests.append(
-                    ProcessedRequest.model_validate(
-                        {
-                            'uniqueKey': request.unique_key,
-                            'wasAlreadyPresent': True,
-                            'wasAlreadyHandled': request.was_already_handled,
-                        }
+                    ProcessedRequest(
+                        unique_key=request.unique_key,
+                        was_already_present=True,
+                        was_already_handled=request.was_already_handled,
                     )
                 )
             else:
@@ -158,8 +154,9 @@ class _ApifyRequestQueueSingleClient:
                 self._requests_cache.pop(unprocessed_request.unique_key, None)
 
         else:
-            api_response = AddRequestsResponse.model_validate(
-                {'unprocessedRequests': [], 'processedRequests': already_present_requests}
+            api_response = AddRequestsResponse(
+                unprocessed_requests=[],
+                processed_requests=already_present_requests,
             )
 
         # Update assumed total count for newly added requests.
@@ -236,11 +233,20 @@ class _ApifyRequestQueueSingleClient:
 
         # Update the cached data
         for request_data in response.get('items', []):
+            # Due to https://github.com/apify/apify-core/blob/v0.1377.0/src/api/src/lib/request_queues/request_queue.ts#L53,
+            # the list_head endpoint may return truncated fields for long requests (e.g., long URLs or unique keys).
+            # If truncation is detected, fetch the full request data by its ID from the API.
+            # This is a temporary workaround - the caching will be refactored to use request IDs instead of unique keys.
+            # See https://github.com/apify/apify-sdk-python/issues/630 for details.
+            if '[truncated]' in request_data['uniqueKey'] or '[truncated]' in request_data['url']:
+                request_data = await self._api_client.get_request(request_id=request_data['id'])  # noqa: PLW2901
+
             request = Request.model_validate(request_data)
 
             if request.unique_key in self._requests_in_progress:
                 # Ignore requests that are already in progress, we will not process them again.
                 continue
+
             if request.was_already_handled:
                 # Do not cache fully handled requests, we do not need them. Just cache their unique_key.
                 self._requests_already_handled.add(request.unique_key)
@@ -248,16 +254,20 @@ class _ApifyRequestQueueSingleClient:
                 # Only fetch the request if we do not know it yet.
                 if request.unique_key not in self._requests_cache:
                     request_id = unique_key_to_request_id(request.unique_key)
-                    complete_request_data = await self._api_client.get_request(request_id)
 
-                    if complete_request_data is not None:
-                        request = Request.model_validate(complete_request_data)
-                        self._requests_cache[request.unique_key] = request
-                    else:
+                    if request_data is not None and request_id != request_data['id']:
                         logger.warning(
-                            f'Could not fetch request data for unique_key=`{request.unique_key}` (id=`{request_id}`)'
+                            f'Request ID mismatch: {request_id} != {request_data["id"]}, '
+                            'this may cause unexpected behavior.'
                         )
 
+                    # See https://github.com/apify/apify-sdk-python/issues/630 for details.
+                    if '[truncated]' not in request.unique_key:
+                        request_data = await self._api_client.get_request(request_id=request_id)  # noqa: PLW2901
+                        request = Request.model_validate(request_data)
+
+                    self._requests_cache[request.unique_key] = request
+
                 # Add new requests to the end of the head, unless already present in head
                 if request.unique_key not in self._head_requests:
                     self._head_requests.appendleft(request.unique_key)

{apify-3.0.2b6 → apify-3.0.2b7}/tests/integration/test_request_queue.py

@@ -7,12 +7,14 @@ from typing import TYPE_CHECKING, Literal, cast
 import pytest
 
 from apify_shared.consts import ApifyEnvVars
-from crawlee import Request, service_locator
+from crawlee import service_locator
 from crawlee.crawlers import BasicCrawler
 
 from ._utils import generate_unique_resource_name
-from apify import Actor
+from apify import Actor, Request
 from apify.storage_clients import ApifyStorageClient
+from apify.storage_clients._apify._request_queue_shared_client import ApifyRequestQueueSharedClient
+from apify.storage_clients._apify._utils import unique_key_to_request_id
 from apify.storages import RequestQueue
 
 if TYPE_CHECKING:
@@ -1189,3 +1191,30 @@ async def test_request_queue_has_stats(request_queue_apify: RequestQueue) -> Non
     assert hasattr(metadata, 'stats')
     apify_metadata = cast('ApifyRequestQueueMetadata', metadata)
     assert apify_metadata.stats.write_count == add_request_count
+
+
+async def test_rq_long_url(request_queue_apify: RequestQueue) -> None:
+    # TODO: Remove the skip when issue #630 is resolved.
+    if isinstance(request_queue_apify._client._implementation, ApifyRequestQueueSharedClient):  # type: ignore[attr-defined]
+        pytest.skip('Skipping for the "shared" request queue - unskip after issue #630 is resolved.')
+
+    url = 'https://portal.isoss.gov.cz/irj/portal/anonymous/mvrest?path=/eosm-public-offer&officeLabels=%7B%7D&page=1&pageSize=100000&sortColumn=zdatzvsm&sortOrder=-1'
+
+    req = Request.from_url(
+        url=url,
+        use_extended_unique_key=True,
+        always_enqueue=True,
+    )
+
+    request_id = unique_key_to_request_id(req.unique_key)
+
+    processed_request = await request_queue_apify.add_request(req)
+    assert processed_request.id == request_id
+
+    request_obtained = await request_queue_apify.fetch_next_request()
+    assert request_obtained is not None
+
+    await request_queue_apify.mark_request_as_handled(request_obtained)
+
+    is_finished = await request_queue_apify.is_finished()
+    assert is_finished

{apify-3.0.2b6 → apify-3.0.2b7}/uv.lock

@@ -28,7 +28,7 @@ wheels = [
 
 [[package]]
 name = "apify"
-version = "3.0.2b6"
+version = "3.0.2b7"
 source = { editable = "." }
 dependencies = [
     { name = "apify-client" },