Skip to content

利用 AI 大模型和自动化工作流,根据主题或关键词一键生成高清短视频。Generate HD short videos from a topic or keyword with an automated AI workflow.

fix(llm): preserve captions containing error-message text

Rudy Celeklie1a9ffae9727
diff --git a/app/services/llm.py b/app/services/llm.py
index 7868d337..ac82b766 100644
--- a/app/services/llm.py
+++ b/app/services/llm.py
@@ -1212,7 +1212,7 @@ def generate_social_metadata(
for i in range(_max_retries):
try:
response = _generate_response(prompt)
- if isinstance(response, str) and "Error: " in response:
+ if isinstance(response, str) and response.startswith("Error: "):
logger.error(f"failed to generate social metadata: {response}")
break
metadata = _parse_social_metadata(response, platform)
diff --git a/test/services/test_social_error_text.py b/test/services/test_social_error_text.py
new file mode 100644
index 00000000..5233e8c8
--- /dev/null
+++ b/test/services/test_social_error_text.py
@@ -0,0 +1,47 @@
+import json
+import threading
+from http.server import BaseHTTPRequestHandler, ThreadingHTTPServer
+from unittest.mock import patch
+
+from app.config import config
+from app.services import llm
+
+
+def test_social_metadata_accepts_error_text_inside_successful_provider_json():
+ metadata = {"title": "Disk recovery tutorial", "caption": "Fix Error: disk full without losing files", "hashtags": ["#tech", "#storage", "#repair"]}
+ received = []
+
+ class Handler(BaseHTTPRequestHandler):
+ def log_message(self, *args):
+ pass
+
+ def do_POST(self):
+ received.append((self.path, json.loads(self.rfile.read(int(self.headers["Content-Length"])))))
+ body = json.dumps({"id": "fixture", "object": "chat.completion", "created": 1, "model": "fixture", "choices": [{"index": 0, "message": {"role": "assistant", "content": json.dumps(metadata)}, "finish_reason": "stop"}]}).encode()
+ self.send_response(200)
+ self.send_header("Content-Type", "application/json")
+ self.send_header("Content-Length", str(len(body)))
+ self.end_headers()
+ self.wfile.write(body)
+
+ server = ThreadingHTTPServer(("127.0.0.1", 0), Handler)
+ thread = threading.Thread(target=server.serve_forever, daemon=True)
+ thread.start()
+ try:
+ settings = {"llm_provider": "openai", "openai_api_key": "fixture-key", "openai_model_name": "fixture", "openai_base_url": f"http://127.0.0.1:{server.server_port}/v1"}
+ with patch.object(config, "app", settings):
+ actual = llm.generate_social_metadata(video_subject="Storage tips", platform="youtube_shorts")
+ assert actual == metadata
+ assert len(received) == 1
+ assert received[0][0] == "/v1/chat/completions"
+ finally:
+ server.shutdown()
+ server.server_close()
+ thread.join(timeout=5)
+
+
+def test_provider_error_prefix_still_returns_the_fallback():
+ with patch.object(llm, "_generate_response", return_value="Error: provider unavailable") as response:
+ actual = llm.generate_social_metadata(video_subject="Storage tips", platform="youtube_shorts")
+ assert actual["title"] == "Storage tips"
+ assert response.call_count == 1