fix: Also load input from a file with a .json extension in file system storage

Author: janbuchar

src/apify/_configuration.py

@@ -3,6 +3,7 @@
 from datetime import datetime, timedelta
 from decimal import Decimal
 from logging import getLogger
+from pathlib import Path
 from typing import Annotated, Any
 
 from pydantic import AliasChoices, BeforeValidator, Field, model_validator
@@ -421,6 +422,14 @@ def disable_browser_sandbox_on_platform(self) -> Self:
             logger.warning('Actor is running on the Apify platform, `disable_browser_sandbox` was changed to True.')
         return self
 
+    @property
+    def canonical_input_key(self) -> str:
+        return Path(self.input_key).stem
+
+    @property
+    def input_key_candidates(self) -> set[str]:
+        return {self.input_key, self.canonical_input_key, Path(self.canonical_input_key).with_suffix('.json').name}
+
     @classmethod
     def get_global_configuration(cls) -> Configuration:
         """Retrieve the global instance of the configuration.

src/apify/storage_clients/_file_system/_key_value_store_client.py

@@ -2,10 +2,12 @@
 import json
 from pathlib import Path
 
+from more_itertools import flatten
 from typing_extensions import override
 
 from crawlee._consts import METADATA_FILENAME
 from crawlee.storage_clients._file_system import FileSystemKeyValueStoreClient
+from crawlee.storage_clients.models import KeyValueStoreRecord
 
 from apify._configuration import Configuration
 
@@ -24,16 +26,24 @@ async def purge(self) -> None:
         It deletes all files in the key-value store directory, except for the metadata file and
         the `INPUT.json` file. It also updates the metadata to reflect that the store has been purged.
         """
-        kvs_input_key = Configuration.get_global_configuration().input_key
+        configuration = Configuration.get_global_configuration()
 
         # First try to find the alternative format of the input file and process it if it exists.
         for file_path in self.path_to_kvs.glob('*'):
-            if file_path.name == f'{kvs_input_key}.json':
+            if (
+                file_path.name in configuration.input_key_candidates
+                and file_path.name != configuration.canonical_input_key
+            ):
                 await self._process_input_json(file_path)
 
         async with self._lock:
+            files_to_keep = set(
+                flatten([key, f'{key}.{METADATA_FILENAME}'] for key in configuration.input_key_candidates)
+            )
+            files_to_keep.add(METADATA_FILENAME)
+
             for file_path in self.path_to_kvs.glob('*'):
-                if file_path.name in {METADATA_FILENAME, kvs_input_key, f'{kvs_input_key}.{METADATA_FILENAME}'}:
+                if file_path.name in files_to_keep:
                     continue
                 if file_path.is_file():
                     await asyncio.to_thread(file_path.unlink, missing_ok=True)
@@ -55,3 +65,15 @@ async def _process_input_json(self, path: Path) -> None:
             f.close()
         await asyncio.to_thread(path.unlink, missing_ok=True)
         await self.set_value(key=path.stem, value=input_data)
+
+    @override
+    async def get_value(self, *, key: str) -> KeyValueStoreRecord | None:
+        configuration = Configuration.get_global_configuration()
+
+        if key in configuration.input_key_candidates:
+            for candidate in configuration.input_key_candidates:
+                value = await super().get_value(key=candidate)
+                if value is not None:
+                    return value
+
+        return await super().get_value(key=key)