Merge remote-tracking branch 'origin/master' into test-new-storage-services-creation-2

Author: Pijukatel

CHANGELOG.md

@@ -16,6 +16,7 @@ All notable changes to this project will be documented in this file.
 - Expose `APIFY_USER_IS_PAYING` env var to the configuration ([#507](https://github.com/apify/apify-sdk-python/pull/507)) ([0801e54](https://github.com/apify/apify-sdk-python/commit/0801e54887317c1280cc6828ecd3f2cc53287e76)) by [@stepskop](https://github.com/stepskop)
 - Resolve DeprecationWarning in ApifyEventManager ([#555](https://github.com/apify/apify-sdk-python/pull/555)) ([0c5111d](https://github.com/apify/apify-sdk-python/commit/0c5111dafe19796ec1fb9652a44c031bed9758df)) by [@vdusek](https://github.com/vdusek), closes [#343](https://github.com/apify/apify-sdk-python/issues/343)
 - Use same `client_key` for `Actor` created `request_queue` and improve its metadata estimation ([#552](https://github.com/apify/apify-sdk-python/pull/552)) ([7e4e5da](https://github.com/apify/apify-sdk-python/commit/7e4e5da81dd87e84ebeef2bd336c6c1d422cb9a7)) by [@Pijukatel](https://github.com/Pijukatel), closes [#536](https://github.com/apify/apify-sdk-python/issues/536)
+- Properly process pre-existing Actor input file ([#591](https://github.com/apify/apify-sdk-python/pull/591)) ([cc5075f](https://github.com/apify/apify-sdk-python/commit/cc5075fab8c72ca5711cfd97932037b34e6997cd)) by [@Pijukatel](https://github.com/Pijukatel), closes [#590](https://github.com/apify/apify-sdk-python/issues/590)
 
 ### Chore
 
@@ -101,6 +102,18 @@ All notable changes to this project will be documented in this file.
 - Tagline overlap ([#501](https://github.com/apify/apify-sdk-python/pull/501)) ([bae8340](https://github.com/apify/apify-sdk-python/commit/bae8340c46fea756ea35ea4d591da84c09d478e2)) by [@katzino](https://github.com/katzino)
 
 
+## [2.7.0](https://github.com/apify/apify-sdk-python/releases/tag/v2.7.0) (2025-07-14)
+
+### 🚀 Features
+
+- **crypto:** Decrypt secret objects ([#482](https://github.com/apify/apify-sdk-python/pull/482)) ([ce9daf7](https://github.com/apify/apify-sdk-python/commit/ce9daf7381212b8dc194e8a643e5ca0dedbc0078)) by [@MFori](https://github.com/MFori)
+
+### 🐛 Bug Fixes
+
+- Sync `@docusaurus` theme version [internal] ([#500](https://github.com/apify/apify-sdk-python/pull/500)) ([a7485e7](https://github.com/apify/apify-sdk-python/commit/a7485e7d2276fde464ce862573d5b95e7d4d836a)) by [@katzino](https://github.com/katzino)
+- Tagline overlap ([#501](https://github.com/apify/apify-sdk-python/pull/501)) ([bae8340](https://github.com/apify/apify-sdk-python/commit/bae8340c46fea756ea35ea4d591da84c09d478e2)) by [@katzino](https://github.com/katzino)
+
+
 
 ## [2.7.3](https://github.com/apify/apify-sdk-python/releases/tag/v2.7.3) (2025-08-11)
 

src/apify/storage_clients/_file_system/_key_value_store_client.py

@@ -1,4 +1,6 @@
 import asyncio
+import json
+from pathlib import Path
 
 from typing_extensions import override
 
@@ -23,9 +25,15 @@ async def purge(self) -> None:
         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
+
+        # 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':
+                await self._process_input_json(file_path)
+
         async with self._lock:
             for file_path in self.path_to_kvs.glob('*'):
-                if file_path.name in {METADATA_FILENAME, f'{kvs_input_key}.json'}:
+                if file_path.name in {METADATA_FILENAME, kvs_input_key, f'{kvs_input_key}.{METADATA_FILENAME}'}:
                     continue
                 if file_path.is_file():
                     await asyncio.to_thread(file_path.unlink, missing_ok=True)
@@ -34,3 +42,16 @@ async def purge(self) -> None:
                 update_accessed_at=True,
                 update_modified_at=True,
             )
+
+    async def _process_input_json(self, path: Path) -> None:
+        """Process simple input json file to format expected by the FileSystemKeyValueStoreClient.
+
+        For example: INPUT.json -> INPUT, INPUT.json.metadata
+        """
+        try:
+            f = await asyncio.to_thread(path.open)
+            input_data = json.load(f)
+        finally:
+            f.close()
+        await asyncio.to_thread(path.unlink, missing_ok=True)
+        await self.set_value(key=path.stem, value=input_data)

tests/unit/storage_clients/test_file_system.py

@@ -1,11 +1,17 @@
 from __future__ import annotations
 
 import asyncio
+import json
+from typing import TYPE_CHECKING
 
+from crawlee import service_locator
 from crawlee._consts import METADATA_FILENAME
 
-from apify import Configuration
-from apify.storage_clients._file_system import ApifyFileSystemKeyValueStoreClient
+from apify import Actor, Configuration
+from apify.storage_clients._file_system import ApifyFileSystemKeyValueStoreClient, ApifyFileSystemStorageClient
+
+if TYPE_CHECKING:
+    from pathlib import Path
 
 
 async def test_purge_preserves_input_file_and_metadata() -> None:
@@ -23,18 +29,21 @@ async def test_purge_preserves_input_file_and_metadata() -> None:
     kvs_path = kvs_storage_client.path_to_kvs
 
     # Create various files
-    input_file = kvs_path / f'{configuration.input_key}.json'
+    input_file = kvs_path / f'{configuration.input_key}'
+    input_metadata_file = kvs_path / f'{configuration.input_key}.{METADATA_FILENAME}.json'
     metadata_file = kvs_path / METADATA_FILENAME
     regular_file1 = kvs_path / 'regular_file1.json'
     regular_file2 = kvs_path / 'another_file.txt'
 
     # Write content to files
     await asyncio.to_thread(input_file.write_text, '{"test": "input"}')
+    await asyncio.to_thread(input_metadata_file.write_text, 'some text content')
     await asyncio.to_thread(regular_file1.write_text, '{"test": "data1"}')
     await asyncio.to_thread(regular_file2.write_text, 'some text content')
 
     # Verify all files exist before purge
     assert input_file.exists()
+    assert input_metadata_file.exists()
     assert metadata_file.exists()  # Should exist from client creation
     assert regular_file1.exists()
     assert regular_file2.exists()
@@ -53,3 +62,21 @@ async def test_purge_preserves_input_file_and_metadata() -> None:
     # Verify INPUT.json content is unchanged
     input_content = await asyncio.to_thread(input_file.read_text)
     assert input_content == '{"test": "input"}'
+
+
+async def test_pre_existing_input_used_by_actor(tmp_path: Path) -> None:
+    pre_existing_input = {
+        'foo': 'bar',
+    }
+
+    configuration = Configuration.get_global_configuration()
+    # Create pre-existing INPUT.json file
+    path_to_input = tmp_path / 'key_value_stores' / 'default'
+    path_to_input.mkdir(parents=True)
+    (path_to_input / f'{configuration.input_key}.json').write_text(json.dumps(pre_existing_input))
+
+    # Remove this line after https://github.com/apify/apify-sdk-python/pull/576
+    service_locator.set_storage_client(ApifyFileSystemStorageClient())
+
+    async with Actor():
+        assert pre_existing_input == await Actor.get_input()