revert the asyncio to thread lambda changes

Author: vdusek

docs/guides/code_examples/http_crawlers/selectolax_parser.py

@@ -22,7 +22,7 @@ async def parse(self, response: HttpResponse) -> LexborHTMLParser:
         """Parse HTTP response body into a document object."""
         response_body = await response.read()
         # Run parsing in a thread to avoid blocking the event loop.
-        return await asyncio.to_thread(lambda: LexborHTMLParser(response_body))
+        return await asyncio.to_thread(LexborHTMLParser, response_body)
 
     @override
     async def parse_text(self, text: str) -> LexborHTMLParser:

src/crawlee/browsers/_playwright_browser.py

@@ -79,7 +79,7 @@ async def new_context(self, **context_options: Any) -> BrowserContext:
     async def _delete_temp_dir(self, _: BrowserContext | None) -> None:
         if self._temp_dir and self._temp_dir.exists():
             temp_dir = self._temp_dir
-            await asyncio.to_thread(lambda: shutil.rmtree(temp_dir, ignore_errors=True))
+            await asyncio.to_thread(shutil.rmtree, temp_dir, ignore_errors=True)
 
     @override
     async def close(self, **kwargs: Any) -> None:

src/crawlee/crawlers/_parsel/_parsel_parser.py

@@ -22,7 +22,7 @@ class ParselParser(AbstractHttpParser[Selector, Selector]):
     @override
     async def parse(self, response: HttpResponse) -> Selector:
         response_body = await response.read()
-        return await asyncio.to_thread(lambda: Selector(body=response_body))
+        return await asyncio.to_thread(Selector, body=response_body)
 
     @override
     async def parse_text(self, text: str) -> Selector:

src/crawlee/storage_clients/_file_system/_dataset_client.py

@@ -120,7 +120,7 @@ async def open(
         dataset_base_path = Path(configuration.storage_dir) / cls._STORAGE_SUBDIR
 
         if not dataset_base_path.exists():
-            await asyncio.to_thread(lambda: dataset_base_path.mkdir(parents=True, exist_ok=True))
+            await asyncio.to_thread(dataset_base_path.mkdir, parents=True, exist_ok=True)
 
         # Get a new instance by ID.
         if id:
@@ -134,7 +134,7 @@ async def open(
                     continue
 
                 try:
-                    file = await asyncio.to_thread(lambda p=path_to_metadata: p.open(mode='r', encoding='utf-8'))
+                    file = await asyncio.to_thread(path_to_metadata.open, mode='r', encoding='utf-8')
                     try:
                         file_content = json.load(file)
                         metadata = DatasetMetadata(**file_content)
@@ -163,7 +163,7 @@ async def open(
 
             # If the dataset directory exists, reconstruct the client from the metadata file.
             if path_to_dataset.exists() and path_to_metadata.exists():
-                file = await asyncio.to_thread(lambda: path_to_metadata.open(mode='r', encoding='utf-8'))
+                file = await asyncio.to_thread(path_to_metadata.open, mode='r', encoding='utf-8')
                 try:
                     file_content = json.load(file)
                 finally:
@@ -211,7 +211,7 @@ async def drop(self) -> None:
     async def purge(self) -> None:
         async with self._lock:
             for file_path in await self._get_sorted_data_files():
-                await asyncio.to_thread(lambda f=file_path: f.unlink(missing_ok=True))
+                await asyncio.to_thread(file_path.unlink, missing_ok=True)
 
             await self._update_metadata(
                 update_accessed_at=True,
@@ -435,7 +435,7 @@ async def _update_metadata(
             self._metadata.item_count = new_item_count
 
         # Ensure the parent directory for the metadata file exists.
-        await asyncio.to_thread(lambda: self.path_to_metadata.parent.mkdir(parents=True, exist_ok=True))
+        await asyncio.to_thread(self.path_to_metadata.parent.mkdir, parents=True, exist_ok=True)
 
         # Dump the serialized metadata to the file.
         data = await json_dumps(self._metadata.model_dump())
@@ -456,7 +456,7 @@ async def _push_item(self, item: dict[str, Any], item_id: int) -> None:
         file_path = self.path_to_dataset / filename
 
         # Ensure the dataset directory exists.
-        await asyncio.to_thread(lambda: self.path_to_dataset.mkdir(parents=True, exist_ok=True))
+        await asyncio.to_thread(self.path_to_dataset.mkdir, parents=True, exist_ok=True)
 
         # Dump the serialized item to the file.
         data = await json_dumps(item)

src/crawlee/storage_clients/_file_system/_key_value_store_client.py

@@ -1,6 +1,7 @@
 from __future__ import annotations
 
 import asyncio
+import functools
 import json
 import shutil
 import urllib.parse
@@ -119,7 +120,7 @@ async def open(
         kvs_base_path = Path(configuration.storage_dir) / cls._STORAGE_SUBDIR
 
         if not kvs_base_path.exists():
-            await asyncio.to_thread(lambda: kvs_base_path.mkdir(parents=True, exist_ok=True))
+            await asyncio.to_thread(kvs_base_path.mkdir, parents=True, exist_ok=True)
 
         # Get a new instance by ID.
         if id:
@@ -133,7 +134,7 @@ async def open(
                     continue
 
                 try:
-                    file = await asyncio.to_thread(lambda p=path_to_metadata: p.open(mode='r', encoding='utf-8'))
+                    file = await asyncio.to_thread(path_to_metadata.open, mode='r', encoding='utf-8')
                     try:
                         file_content = json.load(file)
                         metadata = KeyValueStoreMetadata(**file_content)
@@ -162,7 +163,7 @@ async def open(
 
             # If the key-value store directory exists, reconstruct the client from the metadata file.
             if path_to_kvs.exists() and path_to_metadata.exists():
-                file = await asyncio.to_thread(lambda: path_to_metadata.open(mode='r', encoding='utf-8'))
+                file = await asyncio.to_thread(path_to_metadata.open, mode='r', encoding='utf-8')
                 try:
                     file_content = json.load(file)
                 finally:
@@ -212,7 +213,7 @@ async def purge(self) -> None:
             for file_path in self.path_to_kvs.glob('*'):
                 if file_path.name == METADATA_FILENAME:
                     continue
-                await asyncio.to_thread(lambda f=file_path: f.unlink(missing_ok=True))
+                await asyncio.to_thread(file_path.unlink, missing_ok=True)
 
             await self._update_metadata(
                 update_accessed_at=True,
@@ -239,7 +240,9 @@ async def get_value(self, *, key: str) -> KeyValueStoreRecord | None:
         # Read the metadata file
         async with self._lock:
             try:
-                file = await asyncio.to_thread(lambda: record_metadata_filepath.open(mode='r', encoding='utf-8'))
+                file = await asyncio.to_thread(
+                    functools.partial(record_metadata_filepath.open, mode='r', encoding='utf-8'),
+                )
             except FileNotFoundError:
                 logger.warning(f'Metadata file disappeared for key "{key}", aborting get_value')
                 return None
@@ -346,11 +349,11 @@ async def delete_value(self, *, key: str) -> None:
         async with self._lock:
             # Delete the value file and its metadata if found
             if record_path.exists():
-                await asyncio.to_thread(lambda: record_path.unlink(missing_ok=True))
+                await asyncio.to_thread(record_path.unlink, missing_ok=True)
 
                 # Delete the metadata file if it exists
                 if metadata_path.exists():
-                    await asyncio.to_thread(lambda: metadata_path.unlink(missing_ok=True))
+                    await asyncio.to_thread(metadata_path.unlink, missing_ok=True)
                 else:
                     logger.warning(f'Found value file for key "{key}" but no metadata file when trying to delete it.')
 
@@ -395,7 +398,7 @@ async def iterate_keys(
 
             # Try to read and parse the metadata file
             try:
-                metadata_content = await asyncio.to_thread(lambda f=file_path: f.read_text(encoding='utf-8'))
+                metadata_content = await asyncio.to_thread(file_path.read_text, encoding='utf-8')
             except FileNotFoundError:
                 logger.warning(f'Metadata file disappeared for key "{key_name}", skipping it.')
                 continue
@@ -475,7 +478,7 @@ async def _update_metadata(
             self._metadata.modified_at = now
 
         # Ensure the parent directory for the metadata file exists.
-        await asyncio.to_thread(lambda: self.path_to_metadata.parent.mkdir(parents=True, exist_ok=True))
+        await asyncio.to_thread(self.path_to_metadata.parent.mkdir, parents=True, exist_ok=True)
 
         # Dump the serialized metadata to the file.
         data = await json_dumps(self._metadata.model_dump())

src/crawlee/storage_clients/_file_system/_request_queue_client.py

@@ -1,6 +1,7 @@
 from __future__ import annotations
 
 import asyncio
+import functools
 import json
 import shutil
 from collections import deque
@@ -183,7 +184,7 @@ async def open(
         rq_base_path = Path(configuration.storage_dir) / cls._STORAGE_SUBDIR
 
         if not rq_base_path.exists():
-            await asyncio.to_thread(lambda: rq_base_path.mkdir(parents=True, exist_ok=True))
+            await asyncio.to_thread(rq_base_path.mkdir, parents=True, exist_ok=True)
 
         # Open an existing RQ by its ID, raise an error if not found.
         if id:
@@ -197,7 +198,7 @@ async def open(
                     continue
 
                 try:
-                    file = await asyncio.to_thread(lambda p=path_to_metadata: p.open(mode='r', encoding='utf-8'))
+                    file = await asyncio.to_thread(path_to_metadata.open, mode='r', encoding='utf-8')
                     try:
                         file_content = json.load(file)
                         metadata = RequestQueueMetadata(**file_content)
@@ -232,7 +233,7 @@ async def open(
 
             # If the RQ directory exists, reconstruct the client from the metadata file.
             if path_to_rq.exists() and path_to_metadata.exists():
-                file = await asyncio.to_thread(lambda: path_to_metadata.open(encoding='utf-8'))
+                file = await asyncio.to_thread(path_to_metadata.open, encoding='utf-8')
                 try:
                     file_content = json.load(file)
                 finally:
@@ -300,7 +301,7 @@ async def purge(self) -> None:
             request_files = await self._get_request_files(self.path_to_rq)
 
             for file_path in request_files:
-                await asyncio.to_thread(lambda f=file_path: f.unlink(missing_ok=True))
+                await asyncio.to_thread(file_path.unlink, missing_ok=True)
 
             # Clear recoverable state
             await self._state.reset()
@@ -675,7 +676,7 @@ async def _update_metadata(
             self._metadata.had_multiple_clients = True
 
         # Ensure the parent directory for the metadata file exists.
-        await asyncio.to_thread(lambda: self.path_to_metadata.parent.mkdir(parents=True, exist_ok=True))
+        await asyncio.to_thread(self.path_to_metadata.parent.mkdir, parents=True, exist_ok=True)
 
         # Dump the serialized metadata to the file.
         data = await json_dumps(self._metadata.model_dump())
@@ -753,7 +754,7 @@ async def _get_request_files(cls, path_to_rq: Path) -> list[Path]:
             A list of paths to all request files.
         """
         # Create the requests directory if it doesn't exist.
-        await asyncio.to_thread(lambda: path_to_rq.mkdir(parents=True, exist_ok=True))
+        await asyncio.to_thread(path_to_rq.mkdir, parents=True, exist_ok=True)
 
         # List all the json files.
         files = await asyncio.to_thread(lambda: list(path_to_rq.glob('*.json')))
@@ -775,7 +776,7 @@ async def _parse_request_file(cls, file_path: Path) -> Request | None:
         """
         # Open the request file.
         try:
-            file = await asyncio.to_thread(lambda f=file_path: f.open(mode='r', encoding='utf-8'))
+            file = await asyncio.to_thread(functools.partial(file_path.open, mode='r', encoding='utf-8'))
         except FileNotFoundError:
             logger.warning(f'Request file "{file_path}" not found.')
             return None