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.
Python ai-video-generatorcontent-creationffmpeginstagram-reelsllmpythonshort-videosubtitlestext-to-speechtiktokvideo-automationvideo-workflowworkflow-automationyoutube-shorts
fix(twelvelabs): keep invalid embeddings out of the session cache
Rudy Celeklic8bace5bb4c0
diff --git a/app/services/twelvelabs.py b/app/services/twelvelabs.pyindex f6f3946e..aa32b629 100644--- a/app/services/twelvelabs.py+++ b/app/services/twelvelabs.py@@ -102,7 +102,12 @@ def _embed_text_cached(text: str, model: str) -> List[float]:with _managed_client() as client:resp = client.embed.create(model_name=model, text=text)# SDK aliases the raw JSON 'float' vector key to `float_`.- return list(resp.text_embedding.segments[0].float_)+ vector = list(resp.text_embedding.segments[0].float_)+ if not vector or not all(math.isfinite(value) for value in vector):+ raise ValueError("embedding must contain finite components")+ if not any(vector):+ raise ValueError("embedding must have a nonzero norm")+ return vectordef rerank_terms_by_subject(diff --git a/test/services/test_twelvelabs_invalid_embedding_native.py b/test/services/test_twelvelabs_invalid_embedding_native.pynew file mode 100644index 00000000..66515e6b--- /dev/null+++ b/test/services/test_twelvelabs_invalid_embedding_native.py@@ -0,0 +1,91 @@+import http.server+import json+import threading++import pytest++from app.services import twelvelabs as service++[email protected]("invalid", [[], [0.0] * 512])+def test_native_sdk_does_not_cache_invalid_embeddings(tmp_path, monkeypatch, invalid):+ TwelveLabs = pytest.importorskip("twelvelabs").TwelveLabs++ requests = []+ good = [1.0] + [0.0] * 511++ class Handler(http.server.BaseHTTPRequestHandler):+ def do_POST(self):+ requests.append(self.rfile.read(int(self.headers["Content-Length"])))+ vector = invalid if len(requests) == 1 else good+ body = json.dumps(+ {+ "model_name": "marengo3.0",+ "text_embedding": {"segments": [{"float": vector}]},+ }+ ).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)++ def log_message(self, *args):+ pass++ server = http.server.ThreadingHTTPServer(("127.0.0.1", 0), Handler)+ worker = threading.Thread(target=server.serve_forever, daemon=True)+ worker.start()+ monkeypatch.setitem(service.config.app, "twelvelabs_api_keys", ["synthetic"])+ monkeypatch.setattr(+ service,+ "_client",+ lambda httpx_client=None: TwelveLabs(+ api_key="synthetic",+ base_url=f"http://127.0.0.1:{server.server_port}",+ httpx_client=httpx_client,+ ),+ )+ service._embed_text_cached.cache_clear()+ try:+ assert service.embed_text("topic") is None+ assert service.embed_text("topic") == good+ assert len(requests) == 2+ assert service.embed_text("topic") == good+ assert len(requests) == 2+ finally:+ service._embed_text_cached.cache_clear()+ server.shutdown()+ server.server_close()+ worker.join(2)+++ "invalid", [[], [0.0, 0.0], [float("inf"), 1.0], [float("nan"), 1.0]]+)+def test_invalid_embedding_retry_and_cache_without_optional_sdk(monkeypatch, invalid):+ from contextlib import contextmanager+ from types import SimpleNamespace++ calls = []+ good = [1.0, 0.0]++ def create(**kwargs):+ calls.append(kwargs)+ segment = SimpleNamespace(float_=invalid if len(calls) == 1 else good)+ return SimpleNamespace(text_embedding=SimpleNamespace(segments=[segment]))++ @contextmanager+ def managed_client():+ yield SimpleNamespace(embed=SimpleNamespace(create=create))++ monkeypatch.setitem(service.config.app, "twelvelabs_api_keys", ["synthetic"])+ monkeypatch.setattr(service, "_managed_client", managed_client)+ service._embed_text_cached.cache_clear()+ try:+ assert service.embed_text("topic") is None+ assert service.embed_text("topic") == good+ assert service.embed_text("topic") == good+ assert len(calls) == 2+ finally:+ service._embed_text_cached.cache_clear()