LCORE-579 Anonymize user ID in transcripts

src/utils/transcripts.py

@@ -9,6 +9,7 @@
 import logging
 import os
 from pathlib import Path
+import hashlib
 
 from configuration import configuration
 from models.requests import Attachment, QueryRequest
@@ -18,11 +19,17 @@
 logger = logging.getLogger("utils.transcripts")
 
 
+def _hash_user_id(user_id: str) -> str:
+    """Hash the user ID using SHA-256."""
+    return hashlib.sha256(user_id.encode("utf-8")).hexdigest()
+
+
 def construct_transcripts_path(user_id: str, conversation_id: str) -> Path:
     """Construct path to transcripts."""
     # these two normalizations are required by Snyk as it detects
     # this Path sanitization pattern
-    uid = os.path.normpath("/" + user_id).lstrip("/")
+    hashed_user_id = _hash_user_id(user_id)
+    uid = os.path.normpath("/" + hashed_user_id).lstrip("/")
     cid = os.path.normpath("/" + conversation_id).lstrip("/")
     file_path = (
         configuration.user_data_collection_configuration.transcripts_storage or ""
@@ -59,13 +66,15 @@ def store_transcript(  # pylint: disable=too-many-arguments,too-many-positional-
     transcripts_path = construct_transcripts_path(user_id, conversation_id)
     transcripts_path.mkdir(parents=True, exist_ok=True)
 
+    hashed_user_id = _hash_user_id(user_id)
+
     data_to_store = {
         "metadata": {
             "provider": provider_id,
             "model": model_id,
             "query_provider": query_request.provider,
             "query_model": query_request.model,
-            "user_id": user_id,
+            "user_id": hashed_user_id,
             "conversation_id": conversation_id,
             "timestamp": datetime.now(UTC).isoformat(),
         },

tests/unit/utils/test_transcripts.py

@@ -1,5 +1,6 @@
 """Unit tests for functions defined in utils.transcripts module."""
 
+import hashlib
 from configuration import AppConfig
 from models.requests import QueryRequest
 
@@ -39,11 +40,13 @@ def test_construct_transcripts_path(mocker):
 
     user_id = "user123"
     conversation_id = "123e4567-e89b-12d3-a456-426614174000"
+    hashed_user_id = hashlib.sha256(user_id.encode("utf-8")).hexdigest()
 
     path = construct_transcripts_path(user_id, conversation_id)
 
     assert (
-        str(path) == "/tmp/transcripts/user123/123e4567-e89b-12d3-a456-426614174000"
+        str(path)
+        == f"/tmp/transcripts/{hashed_user_id}/123e4567-e89b-12d3-a456-426614174000"
     ), "Path should be constructed correctly"
 
 
@@ -97,14 +100,15 @@ def test_store_transcript(mocker):
     )
 
     # Assert that the transcript was stored correctly
+    hashed_user_id = hashlib.sha256(user_id.encode("utf-8")).hexdigest()
     mock_json.dump.assert_called_once_with(
         {
             "metadata": {
                 "provider": "fake-provider",
                 "model": "fake-model",
                 "query_provider": query_request.provider,
                 "query_model": query_request.model,
-                "user_id": user_id,
+                "user_id": hashed_user_id,
                 "conversation_id": conversation_id,
                 "timestamp": mocker.ANY,
             },