Always refresh metadata

Author: janbuchar

src/apify/storage_clients/_file_system/_key_value_store_client.py

@@ -55,12 +55,11 @@ async def _sanitize_input_json_files(self) -> None:
         alternative_keys = configuration.input_key_candidates - {configuration.canonical_input_key}
 
         if (self.path_to_kvs / configuration.canonical_input_key).exists():
-            # Handle missing metadata
-            if not await self.record_exists(key=configuration.canonical_input_key):
-                input_data = await asyncio.to_thread(
-                    lambda: json.loads((self.path_to_kvs / configuration.canonical_input_key).read_text())
-                )
-                await self.set_value(key=configuration.canonical_input_key, value=input_data)
+            # Refresh metadata to prevent inconsistencies
+            input_data = await asyncio.to_thread(
+                lambda: json.loads((self.path_to_kvs / configuration.canonical_input_key).read_text())
+            )
+            await self.set_value(key=configuration.canonical_input_key, value=input_data)
 
             for alternative_key in alternative_keys:
                 if (alternative_input_file := self.path_to_kvs / alternative_key).exists():
@@ -69,11 +68,10 @@ async def _sanitize_input_json_files(self) -> None:
             for alternative_key in alternative_keys:
                 alternative_input_file = self.path_to_kvs / alternative_key
 
-                # Handle missing metadata
-                if alternative_input_file.exists() and not await self.record_exists(key=alternative_key):
-                    with alternative_input_file.open() as f:
-                        input_data = await asyncio.to_thread(lambda: json.load(f))
-                    await self.set_value(key=configuration.canonical_input_key, value=input_data)
+                # Refresh metadata to prevent inconsistencies
+                with alternative_input_file.open() as f:
+                    input_data = await asyncio.to_thread(lambda: json.load(f))
+                await self.set_value(key=configuration.canonical_input_key, value=input_data)
 
     @override
     async def get_value(self, *, key: str) -> KeyValueStoreRecord | None:

tests/unit/storage_clients/test_file_system.py

@@ -2,7 +2,7 @@
 
 import asyncio
 import json
-from typing import TYPE_CHECKING
+from pathlib import Path
 
 import pytest
 
@@ -11,9 +11,6 @@
 from apify import Actor, Configuration
 from apify.storage_clients._file_system import ApifyFileSystemKeyValueStoreClient
 
-if TYPE_CHECKING:
-    from pathlib import Path
-
 
 async def test_purge_preserves_input_file_and_metadata() -> None:
     """Test that purge() preserves INPUT.json and metadata files but removes other files."""
@@ -67,13 +64,15 @@ async def test_purge_preserves_input_file_and_metadata() -> None:
 
 
 @pytest.mark.parametrize('input_file_name', ['INPUT', 'INPUT.json'])
-async def test_pre_existing_input_used_by_actor(tmp_path: Path, input_file_name: str) -> None:
+async def test_pre_existing_input_used_by_actor(input_file_name: str) -> None:
+    configuration = Configuration.get_global_configuration()
+
     pre_existing_input = {
         'foo': 'bar',
     }
 
     # Create pre-existing INPUT.json file
-    path_to_input = tmp_path / 'key_value_stores' / 'default'
+    path_to_input = Path(configuration.storage_dir) / 'key_value_stores' / 'default'
     path_to_input.mkdir(parents=True)
     (path_to_input / input_file_name).write_text(json.dumps(pre_existing_input))