From 79206f391912b917135037b4877da52cfbff92c8 Mon Sep 17 00:00:00 2001
From: Alex <a@tushynski.me>
Date: Fri, 3 Apr 2026 17:57:49 +0100
Subject: [PATCH] fix: harden faiss

---
 application/vectorstore/faiss.py | 32 +++++++++++++++++++++++++++-----
 tests/vectorstore/test_faiss.py  | 22 ++++++++++++++++++++++
 2 files changed, 49 insertions(+), 5 deletions(-)

diff --git a/application/vectorstore/faiss.py b/application/vectorstore/faiss.py
index 3e86203b..5c926d4d 100644
--- a/application/vectorstore/faiss.py
+++ b/application/vectorstore/faiss.py
@@ -11,11 +11,33 @@ from application.storage.storage_creator import StorageCreator
 
 
 def get_vectorstore(path: str) -> str:
-    if path:
-        vectorstore = f"indexes/{path}"
-    else:
-        vectorstore = "indexes"
-    return vectorstore
+    """Build a safe local path for a FAISS index.
+
+    Args:
+        path: Source identifier provided by the caller.
+
+    Returns:
+        The validated vectorstore path rooted under ``indexes``.
+
+    Raises:
+        ValueError: If ``path`` escapes the ``indexes`` directory.
+    """
+    base_dir = "indexes"
+    if not path:
+        return base_dir
+
+    normalized = str(path).strip()
+    if "\\" in normalized:
+        raise ValueError("Invalid source_id path")
+
+    candidate = os.path.normpath(os.path.join(base_dir, normalized))
+    base_abs = os.path.abspath(base_dir)
+    candidate_abs = os.path.abspath(candidate)
+
+    if not candidate_abs.startswith(base_abs + os.sep) and candidate_abs != base_abs:
+        raise ValueError("Invalid source_id path")
+
+    return candidate
 
 
 class FaissStore(BaseVectorStore):
diff --git a/tests/vectorstore/test_faiss.py b/tests/vectorstore/test_faiss.py
index c9fd40e3..4529f856 100644
--- a/tests/vectorstore/test_faiss.py
+++ b/tests/vectorstore/test_faiss.py
@@ -363,6 +363,28 @@ class TestGetVectorstore:
 
         assert get_vectorstore("user/source123") == "indexes/user/source123"
 
+    @pytest.mark.parametrize(
+        "malicious_path",
+        [
+            "../outside",
+            "../../etc/passwd",
+            "nested/../../../outside",
+            "/tmp/evil",
+            "..\\outside",
+            "valid/../../escape",
+        ],
+    )
+    def test_rejects_path_traversal(self, malicious_path):
+        from application.vectorstore.faiss import get_vectorstore
+
+        with pytest.raises(ValueError, match="Invalid source_id path"):
+            get_vectorstore(malicious_path)
+
+    def test_allows_mongodb_style_ids(self):
+        from application.vectorstore.faiss import get_vectorstore
+
+        assert get_vectorstore("65e8f6a8a7a96b1bdad4154f") == "indexes/65e8f6a8a7a96b1bdad4154f"
+
 
 @pytest.mark.unit
 class TestFaissStoreAddChunk: