Merge pull request #1523 from frappe/add-tar-safety

fix: add safety filter for untarring

bench/app.py

@@ -24,6 +24,7 @@ from bench.exceptions import NotInBenchDirectoryError
 from bench.utils import (
 	UNSET_ARG,
 	fetch_details_from_tag,
+	get_app_cache_extract_filter,
 	get_available_folder_name,
 	get_bench_cache_path,
 	is_bench_directory,
@@ -343,7 +344,12 @@ class App(AppMeta):
 		
 		click.secho(f"Getting {self.app_name} from cache", fg="yellow")
 		with tarfile.open(cache_path, mode) as tar:
-			tar.extractall(app_path.parent)
+			try:
+				tar.extractall(app_path.parent, filter=get_app_cache_extract_filter())
+			except Exception:
+				logger.exception(f"Cache extraction failed for {self.app_name}")
+				shutil.rmtree(app_path)
+				return False
 
 		return True
 	

bench/utils/__init__.py

@@ -9,6 +9,7 @@ from functools import lru_cache
 from glob import glob
 from pathlib import Path
 from shlex import split
+from tarfile import TarInfo
 from typing import List, Optional, Tuple
 
 # imports - third party imports
@@ -569,3 +570,35 @@ def get_cmd_from_sysargv():
 		break
 
 	return cmd_from_ctx
+
+
+def get_app_cache_extract_filter(
+	count_threshold: int = 10_000,
+	size_threshold: int = 1_000_000_000,
+): # -> Callable[[TarInfo, str], TarInfo | None]
+	state = dict(count=0, size=0)
+
+	AbsoluteLinkError = Exception
+	def data_filter(m: TarInfo, _:str) -> TarInfo:
+		return m
+
+	if (sys.version_info.major == 3 and sys.version_info.minor > 7) or sys.version_info.major > 3:
+		from tarfile import data_filter, AbsoluteLinkError
+
+	def filter_function(member: TarInfo, dest_path: str) -> Optional[TarInfo]:
+		state["count"] += 1
+		state["size"] += member.size
+
+		if state["count"] > count_threshold:
+			raise RuntimeError(f"Number of entries exceeds threshold ({state['count']})")
+
+		if state["size"] > size_threshold:
+			raise RuntimeError(f"Extracted size exceeds threshold ({state['size']})")
+
+		try:
+			return data_filter(member, dest_path)
+		except AbsoluteLinkError:
+			# Links created by `frappe` after extraction
+			return None
+
+	return filter_function