sycommon-python-lib 0.1.56b12__tar.gz → 0.1.56b14__tar.gz

{sycommon_python_lib-0.1.56b12 → sycommon_python_lib-0.1.56b14}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: sycommon-python-lib
-Version: 0.1.56b12
+Version: 0.1.56b14
 Summary: Add your description here
 Requires-Python: >=3.11
 Description-Content-Type: text/markdown
@@ -13,6 +13,7 @@ Requires-Dist: kafka-python>=2.3.0
 Requires-Dist: langchain>=1.2.0
 Requires-Dist: langchain-core>=1.2.6
 Requires-Dist: langchain-openai>=1.1.6
+Requires-Dist: langfuse>=3.11.2
 Requires-Dist: langgraph>=1.0.5
 Requires-Dist: loguru>=0.7.3
 Requires-Dist: mysql-connector-python>=9.5.0

{sycommon_python_lib-0.1.56b12 → sycommon_python_lib-0.1.56b14}/pyproject.toml

@@ -1,6 +1,6 @@
 [project]
 name = "sycommon-python-lib"
-version = "0.1.56-beta12"
+version = "0.1.56-beta14"
 description = "Add your description here"
 readme = "README.md"
 requires-python = ">=3.11"
@@ -14,6 +14,7 @@ dependencies = [
     "langchain>=1.2.0",
     "langchain-core>=1.2.6",
     "langchain-openai>=1.1.6",
+    "langfuse>=3.11.2",
     "langgraph>=1.0.5",
     "loguru>=0.7.3",
     "mysql-connector-python>=9.5.0",

{sycommon_python_lib-0.1.56b12 → sycommon_python_lib-0.1.56b14}/src/sycommon/config/Config.py

@@ -21,6 +21,7 @@ class Config(metaclass=SingletonMeta):
         self.embedding_configs = []
         self.reranker_configs = []
         self.sentry_configs = []
+        self.langfuse_configs = []
         self._process_config()
 
     def get_llm_config(self, model_name):
@@ -47,6 +48,12 @@ class Config(metaclass=SingletonMeta):
                 return sentry
         raise ValueError(f"No configuration found for server: {name}")
 
+    def get_langfuse_config(self, name):
+        for langfuse in self.langfuse_configs:
+            if langfuse.get('name') == name:
+                return langfuse
+        raise ValueError(f"No configuration found for server: {name}")
+
     def _process_config(self):
         llm_config_list = self.config.get('LLMConfig', [])
         for llm_config in llm_config_list:

sycommon_python_lib-0.1.56b14/src/sycommon/config/LangfuseConfig.py

@@ -0,0 +1,15 @@
+from pydantic import BaseModel
+
+
+class LangfuseConfig(BaseModel):
+    name: str
+    secretKey: str
+    publicKey: str
+    baseUrl: str
+    enable: bool
+
+    @classmethod
+    def from_config(cls, server_name: str):
+        from sycommon.config.Config import Config
+        langfuse_config = Config().get_langfuse_config(server_name)
+        return cls(**langfuse_config)

{sycommon_python_lib-0.1.56b12 → sycommon_python_lib-0.1.56b14}/src/sycommon/llm/embedding.py

@@ -25,15 +25,21 @@ class Embedding(metaclass=SingletonMeta):
 
         # 并发信号量
         self.semaphore = asyncio.Semaphore(self.max_concurrency)
+        # 全局默认超时：永不超时（None）
+        self.default_timeout = aiohttp.ClientTimeout(total=None)
 
     async def _get_embeddings_http_async(
         self,
         input: Union[str, List[str]],
         encoding_format: str = None,
         model: str = None,
+        timeout: aiohttp.ClientTimeout = None,
         **kwargs
     ):
         async with self.semaphore:
+            # 优先使用传入的超时，无则用全局默认
+            request_timeout = timeout or self.default_timeout
+
             # 优先使用传入的模型名，无则用默认值
             target_model = model or self.default_embedding_model
             target_base_url = EmbeddingConfig.from_config(target_model).baseUrl
@@ -46,14 +52,23 @@ class Embedding(metaclass=SingletonMeta):
             }
             request_body.update(kwargs)
 
-            async with aiohttp.ClientSession() as session:
-                async with session.post(url, json=request_body) as response:
-                    if response.status != 200:
-                        error_detail = await response.text()
-                        SYLogger.error(
-                            f"Embedding request failed (model: {target_model}): {error_detail}")
-                        return None
-                    return await response.json()
+            try:
+                async with aiohttp.ClientSession(timeout=request_timeout) as session:
+                    async with session.post(url, json=request_body) as response:
+                        if response.status != 200:
+                            error_detail = await response.text()
+                            SYLogger.error(
+                                f"Embedding request failed (model: {target_model}): {error_detail}")
+                            return None
+                        return await response.json()
+            except asyncio.TimeoutError:
+                SYLogger.error(
+                    f"Embedding request timeout (model: {target_model})")
+                return None
+            except Exception as e:
+                SYLogger.error(
+                    f"Embedding request unexpected error (model: {target_model}): {str(e)}")
+                return None
 
     async def _get_reranker_http_async(
         self,
@@ -64,9 +79,13 @@ class Embedding(metaclass=SingletonMeta):
         max_chunks_per_doc: Optional[int] = None,
         return_documents: Optional[bool] = True,
         return_len: Optional[bool] = True,
+        timeout: aiohttp.ClientTimeout = None,
         **kwargs
     ):
         async with self.semaphore:
+            # 优先使用传入的超时，无则用全局默认
+            request_timeout = timeout or self.default_timeout
+
             # 优先使用传入的模型名，无则用默认值
             target_model = model or self.default_reranker_model
             target_base_url = RerankerConfig.from_config(target_model).baseUrl
@@ -84,19 +103,29 @@ class Embedding(metaclass=SingletonMeta):
             }
             request_body.update(kwargs)
 
-            async with aiohttp.ClientSession() as session:
-                async with session.post(url, json=request_body) as response:
-                    if response.status != 200:
-                        error_detail = await response.text()
-                        SYLogger.error(
-                            f"Rerank request failed (model: {target_model}): {error_detail}")
-                        return None
-                    return await response.json()
+            try:
+                async with aiohttp.ClientSession(timeout=request_timeout) as session:
+                    async with session.post(url, json=request_body) as response:
+                        if response.status != 200:
+                            error_detail = await response.text()
+                            SYLogger.error(
+                                f"Rerank request failed (model: {target_model}): {error_detail}")
+                            return None
+                        return await response.json()
+            except asyncio.TimeoutError:
+                SYLogger.error(
+                    f"Rerank request timeout (model: {target_model})")
+                return None
+            except Exception as e:
+                SYLogger.error(
+                    f"Rerank request unexpected error (model: {target_model}): {str(e)}")
+                return None
 
     async def get_embeddings(
         self,
         corpus: List[str],
-        model: str = None
+        model: str = None,
+        timeout: Optional[Union[int, float]] = None
     ):
         """
         获取语料库的嵌入向量，结果顺序与输入语料库顺序一致
@@ -104,12 +133,24 @@ class Embedding(metaclass=SingletonMeta):
         Args:
             corpus: 待生成嵌入向量的文本列表
             model: 可选，指定使用的embedding模型名称，默认使用bge-large-zh-v1.5
+            timeout: 可选，超时时间（秒）：
+                     - 传int/float：表示总超时时间（秒）
+                     - 不传/None：使用默认永不超时配置
         """
+        request_timeout = None
+        if timeout is not None:
+            if isinstance(timeout, (int, float)):
+                request_timeout = aiohttp.ClientTimeout(total=timeout)
+            else:
+                SYLogger.warning(
+                    f"Invalid timeout type: {type(timeout)}, must be int/float, use default timeout")
+
         SYLogger.info(
-            f"Requesting embeddings for corpus: {corpus} (model: {model or self.default_embedding_model}, max_concurrency: {self.max_concurrency})")
-        # 给每个异步任务传入模型名称
+            f"Requesting embeddings for corpus: {corpus} (model: {model or self.default_embedding_model}, max_concurrency: {self.max_concurrency}, timeout: {timeout or 'None'})")
+
+        # 给每个异步任务传入模型名称和超时配置
         tasks = [self._get_embeddings_http_async(
-            text, model=model) for text in corpus]
+            text, model=model, timeout=request_timeout) for text in corpus]
         results = await asyncio.gather(*tasks)
 
         vectors = []
@@ -131,7 +172,8 @@ class Embedding(metaclass=SingletonMeta):
         self,
         top_results: List[str],
         query: str,
-        model: str = None
+        model: str = None,
+        timeout: Optional[Union[int, float]] = None
     ):
         """
         对搜索结果进行重排序
@@ -140,10 +182,23 @@ class Embedding(metaclass=SingletonMeta):
             top_results: 待重排序的文本列表
             query: 排序参考的查询语句
             model: 可选，指定使用的reranker模型名称，默认使用bge-reranker-large
+            timeout: 可选，超时时间（秒）：
+                     - 传int/float：表示总超时时间（秒）
+                     - 不传/None：使用默认永不超时配置
         """
+        request_timeout = None
+        if timeout is not None:
+            if isinstance(timeout, (int, float)):
+                request_timeout = aiohttp.ClientTimeout(total=timeout)
+            else:
+                SYLogger.warning(
+                    f"Invalid timeout type: {type(timeout)}, must be int/float, use default timeout")
+
         SYLogger.info(
-            f"Requesting reranker for top_results: {top_results} (model: {model or self.default_reranker_model}, max_concurrency: {self.max_concurrency})")
-        data = await self._get_reranker_http_async(top_results, query, model=model)
+            f"Requesting reranker for top_results: {top_results} (model: {model or self.default_reranker_model}, max_concurrency: {self.max_concurrency}, timeout: {timeout or 'None'})")
+
+        data = await self._get_reranker_http_async(
+            top_results, query, model=model, timeout=request_timeout)
         SYLogger.info(
             f"Reranker for top_results: {top_results} completed (model: {model or self.default_reranker_model})")
         return data

{sycommon_python_lib-0.1.56b12 → sycommon_python_lib-0.1.56b14}/src/sycommon/llm/get_llm.py

@@ -1,4 +1,6 @@
+import os
 from typing import Dict, Type, List, Optional, Callable, Any
+from sycommon.config.Config import Config
 from sycommon.llm.llm_logger import LLMLogger
 from langchain_core.language_models import BaseChatModel
 from langchain_core.runnables import Runnable, RunnableLambda, RunnableConfig
@@ -10,6 +12,7 @@ from pydantic import BaseModel, ValidationError, Field
 from sycommon.config.LLMConfig import LLMConfig
 from sycommon.llm.llm_tokens import TokensCallbackHandler
 from sycommon.logging.kafka_log import SYLogger
+from langfuse.langchain import CallbackHandler
 
 
 class StructuredRunnableWithToken(Runnable):
@@ -230,6 +233,21 @@ def get_llm(
     if not llmConfig:
         raise Exception(f"无效的模型配置：{model}")
 
+    callbacks = [LLMLogger()]
+
+    config = Config().config
+    server_name = config.get('Name', '')
+    langfuse_configs = config.get('LangfuseConfig', [])
+    target_config = next(
+        (item for item in langfuse_configs if item.get('name') == server_name), None)
+    if target_config and target_config.get('enable', False):
+        os.environ["LANGFUSE_SECRET_KEY"] = target_config.get('secretKey', '')
+        os.environ["LANGFUSE_PUBLIC_KEY"] = target_config.get('publicKey', '')
+        os.environ["LANGFUSE_BASE_URL"] = target_config.get('baseUrl', '')
+
+        langfuse_handler = CallbackHandler()
+        callbacks += [langfuse_handler]
+
     llm = init_chat_model(
         model_provider=llmConfig.provider,
         model=llmConfig.model,
@@ -237,7 +255,7 @@ def get_llm(
         api_key="-",
         temperature=0.1,
         streaming=streaming,
-        callbacks=[LLMLogger()]
+        callbacks=callbacks
     )
 
     if llm is None:

{sycommon_python_lib-0.1.56b12 → sycommon_python_lib-0.1.56b14}/src/sycommon_python_lib.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: sycommon-python-lib
-Version: 0.1.56b12
+Version: 0.1.56b14
 Summary: Add your description here
 Requires-Python: >=3.11
 Description-Content-Type: text/markdown
@@ -13,6 +13,7 @@ Requires-Dist: kafka-python>=2.3.0
 Requires-Dist: langchain>=1.2.0
 Requires-Dist: langchain-core>=1.2.6
 Requires-Dist: langchain-openai>=1.1.6
+Requires-Dist: langfuse>=3.11.2
 Requires-Dist: langgraph>=1.0.5
 Requires-Dist: loguru>=0.7.3
 Requires-Dist: mysql-connector-python>=9.5.0

{sycommon_python_lib-0.1.56b12 → sycommon_python_lib-0.1.56b14}/src/sycommon_python_lib.egg-info/SOURCES.txt

@@ -7,6 +7,7 @@ src/sycommon/config/Config.py
 src/sycommon/config/DatabaseConfig.py
 src/sycommon/config/EmbeddingConfig.py
 src/sycommon/config/LLMConfig.py
+src/sycommon/config/LangfuseConfig.py
 src/sycommon/config/MQConfig.py
 src/sycommon/config/RerankerConfig.py
 src/sycommon/config/SentryConfig.py

{sycommon_python_lib-0.1.56b12 → sycommon_python_lib-0.1.56b14}/src/sycommon_python_lib.egg-info/requires.txt

@@ -7,6 +7,7 @@ kafka-python>=2.3.0
 langchain>=1.2.0
 langchain-core>=1.2.6
 langchain-openai>=1.1.6
+langfuse>=3.11.2
 langgraph>=1.0.5
 loguru>=0.7.3
 mysql-connector-python>=9.5.0