From 0174f935c314452b7b9c8755cde321e62ce1ad49 Mon Sep 17 00:00:00 2001
From: Your Name <119736744+aborayan2022@users.noreply.github.com>
Date: Fri, 20 Mar 2026 10:07:48 +0200
Subject: [PATCH] fix: resolve download 404 caused by file UUID / Celery task
 ID mismatch\n\nThe download route checked access using the file UUID from the
 URL,\nbut the session and usage_events only stored the Celery task ID.\nThese
 are different UUIDs, causing all downloads to return 404.\n\nFixes:\n- Add
 has_download_access() to check file_history table as fallback\n- Update
 assert_web/api_task_access to use file_history lookup\n- Remember file UUID
 in session when task status returns SUCCESS"

---
 backend/app/routes/tasks.py             | 12 ++++++++++++
 backend/app/services/account_service.py | 17 +++++++++++++++++
 backend/app/services/policy_service.py  | 11 ++++++++++-
 3 files changed, 39 insertions(+), 1 deletion(-)

diff --git a/backend/app/routes/tasks.py b/backend/app/routes/tasks.py
index b2bddc0..4d28e33 100644
--- a/backend/app/routes/tasks.py
+++ b/backend/app/routes/tasks.py
@@ -11,6 +11,7 @@ from app.services.policy_service import (
     resolve_api_actor,
     resolve_web_actor,
 )
+from app.utils.auth import remember_task_access
 
 tasks_bp = Blueprint("tasks", __name__)
 
@@ -52,6 +53,17 @@ def get_task_status(task_id: str):
         task_result = result.result or {}
         response["result"] = task_result
 
+        # Remember the file UUID in the session so the download route can verify access.
+        # The download URL contains a different UUID than the Celery task ID.
+        download_url = task_result.get("download_url", "")
+        if download_url:
+            parts = download_url.split("/")
+            # URL format: /api/download/<file_uuid>/<filename>
+            if len(parts) >= 4:
+                file_uuid = parts[3]
+                if file_uuid != task_id:
+                    remember_task_access(file_uuid)
+
     elif result.state == "FAILURE":
         response["error"] = str(result.info) if result.info else "Task failed."
 
diff --git a/backend/app/services/account_service.py b/backend/app/services/account_service.py
index 88af715..5177439 100644
--- a/backend/app/services/account_service.py
+++ b/backend/app/services/account_service.py
@@ -678,6 +678,23 @@ def has_task_access(user_id: int, source: str, task_id: str) -> bool:
     return row is not None
 
 
+def has_download_access(user_id: int, file_task_id: str) -> bool:
+    """Return whether one user owns a file_history entry whose download_url contains the given file task id."""
+    pattern = f"/api/download/{file_task_id}/"
+    with _connect() as conn:
+        row = conn.execute(
+            """
+            SELECT 1
+            FROM file_history
+            WHERE user_id = ? AND download_url LIKE ?
+            LIMIT 1
+            """,
+            (user_id, f"%{pattern}%"),
+        ).fetchone()
+
+    return row is not None
+
+
 # ---------------------------------------------------------------------------
 # Password reset tokens
 # ---------------------------------------------------------------------------
diff --git a/backend/app/services/policy_service.py b/backend/app/services/policy_service.py
index 6de8406..3651c10 100644
--- a/backend/app/services/policy_service.py
+++ b/backend/app/services/policy_service.py
@@ -8,6 +8,7 @@ from app.services.account_service import (
     get_api_key_actor,
     get_user_by_id,
     get_current_period_month,
+    has_download_access,
     has_task_access,
     normalize_plan,
     record_usage_event,
@@ -227,8 +228,13 @@ def build_task_tracking_kwargs(actor: ActorContext) -> dict:
 
 def assert_api_task_access(actor: ActorContext, task_id: str):
     """Ensure one API actor can poll one task id."""
-    if actor.user_id is None or not has_task_access(actor.user_id, "api", task_id):
+    if actor.user_id is None:
         raise PolicyError("Task not found.", 404)
+    if has_task_access(actor.user_id, "api", task_id):
+        return
+    if has_download_access(actor.user_id, task_id):
+        return
+    raise PolicyError("Task not found.", 404)
 
 
 def assert_web_task_access(actor: ActorContext, task_id: str):
@@ -236,6 +242,9 @@ def assert_web_task_access(actor: ActorContext, task_id: str):
     if actor.user_id is not None and has_task_access(actor.user_id, "web", task_id):
         return
 
+    if actor.user_id is not None and has_download_access(actor.user_id, task_id):
+        return
+
     if has_session_task_access(task_id):
         return