test(backend): enhance Drive file filtering and add unit tests for _drive_list_files

surfsense_backend/tests/e2e/fakes/composio_module.py

@@ -320,15 +320,54 @@ def _drive_list_files(args: dict[str, Any]) -> dict[str, Any]:
         except IndexError:
             folder_id = "root"
 
-    files = _DRIVE_FIXTURE.get(folder_id, [])
+    files = _filter_drive_files_for_query(q, _DRIVE_FIXTURE.get(folder_id, []))
     return {
         "data": {
-            "files": list(files),
+            "files": files,
             "nextPageToken": None,
         }
     }
 
 
+def _extract_quoted_value(q: str, anchor: str) -> str | None:
+    anchor_idx = q.find(anchor)
+    if anchor_idx == -1:
+        return None
+
+    after_anchor = q[anchor_idx + len(anchor) :]
+    first_quote_idx = after_anchor.find("'")
+    if first_quote_idx == -1:
+        return None
+
+    after_first_quote = after_anchor[first_quote_idx + 1 :]
+    second_quote_idx = after_first_quote.find("'")
+    if second_quote_idx == -1:
+        return None
+
+    return after_first_quote[:second_quote_idx]
+
+
+def _filter_drive_files_for_query(q: str, files: list[dict[str, Any]]) -> list[dict[str, Any]]:
+    filtered = list(files)
+
+    if "trashed = false" in q:
+        filtered = [entry for entry in filtered if entry.get("trashed") is not True]
+
+    excluded_mime_type = _extract_quoted_value(q, "mimeType !=")
+    if excluded_mime_type:
+        filtered = [
+            entry for entry in filtered if entry.get("mimeType") != excluded_mime_type
+        ]
+
+    included_mime_type = _extract_quoted_value(q, "mimeType =")
+    if included_mime_type:
+        filtered = [
+            entry for entry in filtered if entry.get("mimeType") == included_mime_type
+        ]
+
+    return filtered
+
+
 def _drive_download_file(args: dict[str, Any]) -> dict[str, Any]:
     """Mimic GOOGLEDRIVE_DOWNLOAD_FILE.
 

surfsense_backend/tests/e2e/fakes/fixtures/drive_files.json

@@ -34,6 +34,21 @@
       "mimeType": "text/csv",
       "modifiedTime": "2025-01-25T13:45:00.000Z",
       "createdTime": "2025-01-25T13:45:00.000Z"
+    },
+    {
+      "id": "fake-shortcut-canary",
+      "name": "Shortcut to Canary",
+      "mimeType": "application/vnd.google-apps.shortcut",
+      "modifiedTime": "2025-02-10T12:00:00.000Z",
+      "createdTime": "2025-02-10T12:00:00.000Z"
+    },
+    {
+      "id": "fake-file-trashed",
+      "name": "trashed-e2e-note.txt",
+      "mimeType": "text/plain",
+      "modifiedTime": "2025-02-11T09:00:00.000Z",
+      "createdTime": "2025-02-11T09:00:00.000Z",
+      "trashed": true
     }
   ],
   "fake-folder-projects": [

surfsense_backend/tests/unit/e2e_fakes/test_drive_list_files.py

@@ -0,0 +1,38 @@
+from tests.e2e.fakes.composio_module import _drive_list_files
+
+
+def _ids(result: dict) -> set[str]:
+    return {item["id"] for item in result["data"]["files"]}
+
+
+def test_drive_list_files_filters_shortcuts_and_trashed_items():
+    result = _drive_list_files(
+        {
+            "q": (
+                "'root' in parents and trashed = false and "
+                "mimeType != 'application/vnd.google-apps.shortcut'"
+            )
+        }
+    )
+
+    ids = _ids(result)
+
+    assert "fake-file-canary" in ids
+    assert "fake-shortcut-canary" not in ids
+    assert "fake-file-trashed" not in ids
+
+
+def test_drive_list_files_filters_to_exact_mime_type():
+    result = _drive_list_files(
+        {"q": "'root' in parents and trashed = false and mimeType = 'text/plain'"}
+    )
+
+    assert _ids(result) == {"fake-file-canary"}
+
+
+def test_drive_list_files_uses_requested_parent_folder():
+    result = _drive_list_files(
+        {"q": "'fake-folder-projects' in parents and trashed = false"}
+    )
+
+    assert _ids(result) == {"fake-file-roadmap"}