Auto re-upload expired Google files on node run

Add _check_google_file_active() to verify cached Google file URIs
before reuse. If expired (48h limit), clears cache and transparently
re-uploads via ensure_google_file_upload().

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

1 files changed, 38 insertions, 9 deletions

diff --git a/backend/app/main.py b/backend/app/main.py
index 75491bb..a6f4f10 100644
--- a/backend/app/main.py
+++ b/backend/app/main.py
@@ -808,6 +808,29 @@ def save_files_index(user: str, items: List[FileMeta]):
         json.dump([item.model_dump() for item in items], f, ensure_ascii=False, indent=2)
 
 
+async def _check_google_file_active(uri_or_name: str, api_key: str = None) -> bool:
+    """Check if a Google file reference is still ACTIVE (not expired)."""
+    key = api_key or os.getenv("GOOGLE_API_KEY")
+    if not key:
+        return False
+    try:
+        client = genai.Client(api_key=key)
+        # Google file names look like "files/abc123", URIs like "https://..."
+        # files.get() needs the name, but we can extract it or just try
+        name = uri_or_name
+        if "://" in uri_or_name:
+            # Try to get by URI — extract the name from the path
+            # URI format: https://generativelanguage.googleapis.com/v1beta/files/abc123
+            parts = uri_or_name.rstrip("/").split("/")
+            name = f"files/{parts[-1]}" if parts else uri_or_name
+        info = await asyncio.to_thread(client.files.get, name=name)
+        state = str(getattr(info, "state", "")).upper()
+        return state.endswith("ACTIVE") or state == "ACTIVE"
+    except Exception as e:
+        logger.debug("Google file check failed for %s: %s", uri_or_name, e)
+        return False
+
+
 async def prepare_attachments(
     user: str,
     target_provider: str,
@@ -882,20 +905,26 @@ async def prepare_attachments(
                 raise HTTPException(status_code=500, detail=f"OpenAI upload failed: {str(e)}")
 
         elif provider_norm == ModelProvider.GOOGLE or provider_norm == "google":
-            # Reuse cached google_file_uri
+            # Reuse cached google_file_uri — but verify it's still ACTIVE
             cached_uri = meta.google_file_uri or (
                 meta.provider_file_id if meta.provider == "google" and meta.provider_file_id and "://" in meta.provider_file_id else None
             )
             if cached_uri:
-                attachments.append({
-                    "provider": "google",
-                    "uri": cached_uri,
-                    "name": meta.name,
-                    "mime": meta.mime,
-                })
-                continue
+                still_valid = await _check_google_file_active(cached_uri, llm_config.api_key)
+                if still_valid:
+                    attachments.append({
+                        "provider": "google",
+                        "uri": cached_uri,
+                        "name": meta.name,
+                        "mime": meta.mime,
+                    })
+                    continue
+                # Expired — clear cache, fall through to re-upload
+                logger.info("Google file expired, re-uploading: %s (%s)", meta.name, cached_uri)
+                meta.google_file_uri = None
+                changed = True
 
-            # On-demand upload via shared helper
+            # On-demand upload (or re-upload after expiry)
             uri = await ensure_google_file_upload(meta, path, llm_config.api_key)
             if not uri:
                 raise HTTPException(status_code=500, detail=f"Google upload failed for {meta.name}")