security patch

Author: Chenglong-MS

py-src/data_formulator/datalake/azure_blob_workspace.py

@@ -720,7 +720,12 @@ def import_session_zip(self, zip_data: io.BytesIO) -> dict:
                 self.cleanup()
                 for entry in workspace_entries:
                     rel = entry[len("workspace/"):]
-                    self._upload_bytes(rel, zf.read(entry))
+                    # Guard against zip-slip: secure_filename strips
+                    # path separators and ".." components.
+                    safe_rel = secure_filename(rel)
+                    if not safe_rel:
+                        continue  # skip entries that sanitise to empty
+                    self._upload_bytes(safe_rel, zf.read(entry))
                 if not self._blob_exists(METADATA_FILENAME):
                     self._init_metadata()
                 # Invalidate caches since workspace blobs were replaced

py-src/data_formulator/datalake/cached_azure_blob_workspace.py

@@ -221,8 +221,18 @@ def __init__(
     # ------------------------------------------------------------------
 
     def _cache_path(self, filename: str) -> Path:
-        """Return the local cache path for *filename*."""
-        return self._cache_dir / filename
+        """Return the local cache path for *filename*.
+
+        Raises ``ValueError`` if the resolved path escapes the cache
+        directory (defence-in-depth against path-traversal attacks).
+        """
+        resolved = (self._cache_dir / filename).resolve()
+        if not resolved.is_relative_to(self._cache_dir.resolve()):
+            raise ValueError(
+                f"Path traversal detected: {filename!r} resolves outside "
+                f"the cache directory"
+            )
+        return resolved
 
     # ------------------------------------------------------------------
     # Low-level blob overrides (write-through cache)