fix hf dataset

start_with_sync.sh

@@ -150,7 +150,7 @@ tar -xzf /tmp/caddy.tar.gz -C /tmp/caddy
 
 
 # Start the Node.js application
-git clone https://github.com/waxz/Gemini-CLI-2-API.git /tmp/Gemini-CLI-2-API
+git clone https://github.com/waxz/Gemini-CLI-2-API.git /tmp/Gemini-CLI-2-API -b waxz-patch-1
 git clone https://github.com/waxz/gemini-cli-openai /tmp/gemini-cli-openai
 
 git clone https://github.com/snailyp/gemini-balance.git /tmp/gemini-balance
@@ -303,7 +303,8 @@ ${HOME}/crypt.sh decrypt  ${HOME}/gemini/enc.dev.var.l -e GEMINI_AUTH_ENC_PASS
 
 cd /tmp/Gemini-CLI-2-API && npm install && node src/api-server.js &
 cd /tmp/gemini-balance && uvicorn app.main:app --host 0.0.0.0 --port 8000 --reload &
-cd /tmp/gemini-cli-openai && npm i && npx wrangler dev --port 8881 --host 0.0.0.0 --env x &
+cd /tmp/gemini-cli-openai && npm i
+cd /tmp/gemini-cli-openai && cp /tmp/gemini-cli-openai/.dev.var.x .dev.var  && npx wrangler dev --port 8881 --host 0.0.0.0 &
 # cd /tmp/gemini-cli-openai && npm i && npx wrangler dev --port 8882 --host 0.0.0.0 --env j &
 # cd /tmp/gemini-cli-openai && npm i && npx wrangler dev --port 8883 --host 0.0.0.0 --env w &
 # cd /tmp/gemini-cli-openai && npm i && npx wrangler dev --port 8884 --host 0.0.0.0 --env l &

sync_storage.py

@@ -375,6 +375,7 @@ This repository is automatically managed by the Open WebUI sync system.
             if not current_files:
                 logger.warning("Repository is empty or inaccessible. Nothing to prune.")
                 return
+            metadata = self._get_metadata()
 
             logger.info(f"Found {len(current_files)} files to preserve.")
 
@@ -382,6 +383,9 @@ This repository is automatically managed by the Open WebUI sync system.
             with tempfile.TemporaryDirectory() as tmpdir:
                 logger.info(f"Step 2/5: Downloading current files to a temporary location...")
                 for file_path in current_files:
+                    if file_path not in metadata["archives"]:
+                        logger.info(f"Skipping {file_path} (not in archive)")
+                        continue
                     self.api.hf_hub_download(
                         repo_id=self.repo_id,
                         filename=file_path,
@@ -390,7 +394,6 @@ This repository is automatically managed by the Open WebUI sync system.
                         local_dir=tmpdir,
                         local_dir_use_symlinks=False
                     )
-                
                 # 3. Delete the entire repository
                 logger.warning(f"Step 3/5: Deleting repository '{self.repo_id}'...")
                 delete_repo(self.repo_id, repo_type="dataset", token=self.token)