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 internal response newlines

许文涛d84f4f344f14
diff --git a/app/services/llm.py b/app/services/llm.py
index 28a12da8..37fc53a7 100644
--- a/app/services/llm.py
+++ b/app/services/llm.py
@@ -64,7 +64,9 @@ def _normalize_text_response(content, llm_provider: str) -> str:
if not content:
raise ValueError(f"[{llm_provider}] returned empty text content")
- return content.replace("\n", "")
+ # 前面的 ``strip()`` 已经清理首尾空白。这里必须保留正文中的单换行和
+ # 双换行:脚本生成依赖双换行区分段落,字幕处理也会按行读取用户文案。
+ return content
def _sanitize_error_message(error: object) -> str:
diff --git a/test/services/test_llm.py b/test/services/test_llm.py
index cc1663c3..c766269c 100644
--- a/test/services/test_llm.py
+++ b/test/services/test_llm.py
@@ -32,6 +32,18 @@ RUN_INTEGRATION_TESTS = os.environ.get("MPT_RUN_INTEGRATION_TESTS", "").lower()
class TestScriptPromptOptions(unittest.TestCase):
+ def test_normalize_text_response_preserves_internal_newlines(self):
+ """
+ 归一化只清理首尾空白,不能删除正文内部的换行。双换行用于区分脚本
+ 段落,单换行也可能是模型按语义返回的字幕行。
+ """
+ result = llm._normalize_text_response(
+ "\n 第一行\n第二行\n\n第三段 \n",
+ "openai",
+ )
+
+ self.assertEqual(result, "第一行\n第二行\n\n第三段")
+
def test_normalize_text_response_removes_think_blocks(self):
"""
reasoning 模型可能返回 `<think>...</think>`。脚本生成链路必须只保留
@@ -665,7 +677,7 @@ class TestLiteLLMProvider(unittest.TestCase):
"messages": [{"role": "user", "content": "Say hello"}],
},
)
- self.assertEqual(result, "hellopollinations")
+ self.assertEqual(result, "hello\npollinations")
def test_gemini_uses_google_genai_client(self):
"""Gemini 适配器应通过新版 SDK 的统一 Client 发起内容生成请求。"""
@@ -698,7 +710,7 @@ class TestLiteLLMProvider(unittest.TestCase):
with patch("google.genai.Client", FakeClient):
result = llm._generate_response("Say hello")
- self.assertEqual(result, "hellogemini")
+ self.assertEqual(result, "hello\ngemini")
self.assertEqual(
captured["client_kwargs"],
{"api_key": "gemini-test-key", "http_options": None},
@@ -775,7 +787,7 @@ class TestLiteLLMProvider(unittest.TestCase):
"messages": [{"role": "user", "content": "Say hello"}],
},
)
- self.assertEqual(result, "gatewayresponse")
+ self.assertEqual(result, "gateway\nresponse")
def _use_litellm_provider(self, model_name="openai/gpt-4o-mini"):
config.app["llm_provider"] = "litellm"
@@ -808,7 +820,7 @@ class TestLiteLLMProvider(unittest.TestCase):
with patch.dict(sys.modules, {"litellm": fake_litellm}):
result = llm._generate_response("Say hello")
- self.assertEqual(result, "helloworld")
+ self.assertEqual(result, "hello\nworld")
def test_litellm_provider_uses_registry_default_model(self):
self._use_litellm_provider(model_name="")
@@ -971,7 +983,7 @@ class TestLiteLLMProvider(unittest.TestCase):
with self._patch_dashscope_generation(response):
result = llm._generate_response("Say hello")
- self.assertEqual(result, "你好世界")
+ self.assertEqual(result, "你好\n世界")
def test_qwen_provider_falls_back_to_output_text(self):
"""保留旧 DashScope completion 响应结构的兼容路径。"""
@@ -981,7 +993,7 @@ class TestLiteLLMProvider(unittest.TestCase):
with self._patch_dashscope_generation(response):
result = llm._generate_response("Say hello")
- self.assertEqual(result, "旧格式响应")
+ self.assertEqual(result, "旧格式\n响应")
def test_qwen_provider_reports_empty_text(self):
"""Qwen 空响应应返回可诊断错误,而不是底层 AttributeError。"""
@@ -1049,7 +1061,7 @@ class TestLiteLLMProvider(unittest.TestCase):
"messages": [{"role": "user", "content": "Say hello"}],
},
)
- self.assertEqual(result, "helloaihubmix")
+ self.assertEqual(result, "hello\naihubmix")
def test_aimlapi_provider_uses_openai_compatible_client(self):
config.app["llm_provider"] = "aimlapi"
@@ -1086,7 +1098,7 @@ class TestLiteLLMProvider(unittest.TestCase):
"messages": [{"role": "user", "content": "Say hello"}],
},
)
- self.assertEqual(result, "helloaimlapi")
+ self.assertEqual(result, "hello\naimlapi")
def test_evolink_provider_uses_openai_compatible_client(self):
"""
@@ -1128,7 +1140,7 @@ class TestLiteLLMProvider(unittest.TestCase):
"messages": [{"role": "user", "content": "Say hello"}],
},
)
- self.assertEqual(result, "helloevolink")
+ self.assertEqual(result, "hello\nevolink")
def test_volcengine_provider_uses_openai_compatible_client(self):
"""
@@ -1170,7 +1182,7 @@ class TestLiteLLMProvider(unittest.TestCase):
"messages": [{"role": "user", "content": "Say hello"}],
},
)
- self.assertEqual(result, "hellovolcengine")
+ self.assertEqual(result, "hello\nvolcengine")
def test_grok_provider_still_uses_existing_path(self):
config.app["llm_provider"] = "grok"
@@ -1225,7 +1237,7 @@ class TestLiteLLMProvider(unittest.TestCase):
api_key="groq-test-key",
base_url="https://api.groq.com/openai/v1",
)
- self.assertEqual(result, "hellogroq")
+ self.assertEqual(result, "hello\ngroq")
def _use_ollama_provider(self, base_url=""):
config.app["llm_provider"] = "ollama"
@@ -1263,7 +1275,7 @@ class TestLiteLLMProvider(unittest.TestCase):
"messages": [{"role": "user", "content": "Say hello"}],
},
)
- self.assertEqual(result, "helloollama")
+ self.assertEqual(result, "hello\nollama")
def test_ollama_default_base_url_uses_localhost_outside_container(self):
"""
@@ -1352,7 +1364,7 @@ class TestLiteLLMProvider(unittest.TestCase):
"messages": [{"role": "user", "content": "Say hello"}],
},
)
- self.assertEqual(result, "hellomimo")
+ self.assertEqual(result, "hello\nmimo")
def test_azure_provider_uses_azure_client_directly(self):
"""
@@ -1398,7 +1410,7 @@ class TestLiteLLMProvider(unittest.TestCase):
"messages": [{"role": "user", "content": "Say hello"}],
},
)
- self.assertEqual(result, "helloazure")
+ self.assertEqual(result, "hello\nazure")
def test_unsupported_provider_returns_clear_error(self):
config.app["llm_provider"] = "g" + "4f"