flowllm 0.1.2__tar.gz → 0.1.3__tar.gz

{flowllm-0.1.2 → flowllm-0.1.3}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: flowllm
-Version: 0.1.2
+Version: 0.1.3
 Summary: A flexible framework for building LLM-powered flows and mcp services
 Author-email: FlowLLM Team <flowllm@example.com>
 Maintainer-email: FlowLLM Team <flowllm@example.com>

{flowllm-0.1.2 → flowllm-0.1.3}/flowllm/__init__.py

@@ -1,5 +1,9 @@
 import os
 
+from flowllm.utils.logger_utils import init_logger
+
+init_logger()
+
 from flowllm.utils.common_utils import load_env
 
 load_env()
@@ -17,5 +21,5 @@ from flowllm import service
 from flowllm.context.service_context import C
 from flowllm.op import BaseOp, BaseRayOp, BaseLLMOp
 
-__version__ = "0.1.2"
+__version__ = "0.1.3"
 

{flowllm-0.1.2 → flowllm-0.1.3}/flowllm/context/service_context.py

@@ -39,10 +39,12 @@ class ServiceContext(BaseContext):
 
         self.tool_flow_dict: dict = {}
 
-    def set_default_service_config(self):
-        from flowllm.config.pydantic_config_parser import PydanticConfigParser
+    def set_default_service_config(self, parser=None):
+        if parser is None:
+            from flowllm.config.pydantic_config_parser import PydanticConfigParser
+            parser = PydanticConfigParser
 
-        config_parser = PydanticConfigParser(ServiceConfig)
+        config_parser = parser(ServiceConfig)
         self.service_config = config_parser.parse_args("config=default")
         return self
 

{flowllm-0.1.2 → flowllm-0.1.3}/flowllm/flow/base_flow.py

@@ -17,7 +17,7 @@ class BaseFlow(ABC):
         self.name: str = name or camel_to_snake(self.__class__.__name__)
         self.flow_params: dict = kwargs
 
-        self.flow_op: Optional[BaseOp] = self.build_flow()
+        self.flow_op: BaseOp = self.build_flow()
         self.print_flow()
 
     @abstractmethod
@@ -55,6 +55,7 @@ class BaseFlow(ABC):
             logger.info(f"{prefix}Operation: {op.name}")
 
     def return_callback(self, context: FlowContext):
+        logger.info(f"context.response={context.response.model_dump_json()}")
         return context.response
 
     def __call__(self, **kwargs) -> FlowResponse:
@@ -62,7 +63,8 @@ class BaseFlow(ABC):
         logger.info(f"request.params={kwargs}")
 
         try:
-            self.flow_op(context=context)
+            flow_op = self.build_flow()
+            flow_op(context=context)
 
         except Exception as e:
             logger.exception(f"flow_name={self.name} encounter error={e.args}")

{flowllm-0.1.2 → flowllm-0.1.3}/flowllm/flow/gallery/mock_tool_flow.py

@@ -25,11 +25,6 @@ class MockToolFlow(BaseToolFlow):
             "type": "function",
             "name": "mock_data_processor",
             "description": "A mock tool that processes data through multiple operations and returns structured results",
-            "arguments": {
-                "input_data": "sample_data",
-                "processing_mode": "advanced",
-                "output_format": "json"
-            },
             "input_schema": {
                 "input_data": ParamAttrs(
                     type="string",

{flowllm-0.1.2 → flowllm-0.1.3}/flowllm/flow/gallery/terminate_tool_flow.py

@@ -18,7 +18,7 @@ class TerminateToolFlow(BaseToolFlow):
             "input_schema": {
                 "status": {
                     "type": "str",
-                    "description": "Please determine whether the user's question has been completed. (success / failure)",
+                    "description": "If the user's question can be answered, return success, otherwise return failure.",
                     "required": True,
                     "enum": ["success", "failure"],
                 }

{flowllm-0.1.2 → flowllm-0.1.3}/flowllm/llm/litellm_llm.py

@@ -2,7 +2,6 @@ import asyncio
 import os
 from typing import List, Dict
 
-from litellm import completion, acompletion
 from loguru import logger
 from pydantic import Field, PrivateAttr, model_validator
 
@@ -98,6 +97,7 @@ class LiteLLMBaseLLM(BaseLLM):
         Yields:
             Tuple of (chunk_content, ChunkEnum) for each streaming piece
         """
+        from litellm import completion
         for i in range(self.max_retries):
             try:
                 # Prepare parameters for LiteLLM
@@ -200,6 +200,7 @@ class LiteLLMBaseLLM(BaseLLM):
         Yields:
             Tuple of (chunk_content, ChunkEnum) for each streaming piece
         """
+        from litellm import acompletion
         for i in range(self.max_retries):
             try:
                 # Prepare parameters for LiteLLM

{flowllm-0.1.2 → flowllm-0.1.3}/flowllm/op/__init__.py

@@ -9,3 +9,4 @@ from . import akshare
 from . import code
 from . import gallery
 from . import search
+from . import agent

flowllm-0.1.3/flowllm/op/agent/__init__.py

@@ -0,0 +1 @@
+from .react_v1_op import ReactV1Op

flowllm-0.1.2/flowllm/op/agent/react_op.py → flowllm-0.1.3/flowllm/op/agent/react_v1_op.py

@@ -1,32 +1,45 @@
 import datetime
+import json
 import time
 from typing import List, Dict
 
 from loguru import logger
 
-from flowllm import C, BaseLLMOp
-from flowllm.flow.base_tool_flow import BaseToolFlow
-from flowllm.flow.gallery import DashscopeSearchToolFlow, CodeToolFlow, TerminateToolFlow
+from flowllm.context.flow_context import FlowContext
+from flowllm.context.service_context import C
+from flowllm.op.base_llm_op import BaseLLMOp
+from flowllm.schema.flow_response import FlowResponse
 from flowllm.schema.message import Message, Role
 
 
 @C.register_op()
-class ReactOp(BaseLLMOp):
-    # TODO: test react op
+class ReactV1Op(BaseLLMOp):
     file_path: str = __file__
 
     def execute(self):
         query: str = self.context.query
 
         max_steps: int = int(self.op_params.get("max_steps", 10))
+        from flowllm.flow.base_tool_flow import BaseToolFlow
+        from flowllm.flow.gallery import DashscopeSearchToolFlow, CodeToolFlow, TerminateToolFlow
+
         tools: List[BaseToolFlow] = [DashscopeSearchToolFlow(), CodeToolFlow(), TerminateToolFlow()]
-        tool_dict: Dict[str, BaseToolFlow] = {x.name: x for x in tools}
+
+        """
+        NOTE : x.tool_call.name != x.name
+        `x.tool_call.name` is tool's namex.name is flow's name(unique service name)
+        """
+        tool_dict: Dict[str, BaseToolFlow] = {x.tool_call.name: x for x in tools}
+        for name, tool_call in tool_dict.items():
+            logger.info(f"name={name} "
+                        f"tool_call={json.dumps(tool_call.tool_call.simple_input_dump(), ensure_ascii=False)}")
+
         now_time = datetime.datetime.now().strftime('%Y-%m-%d %H:%M:%S')
         has_terminate_tool = False
 
         user_prompt = self.prompt_format(prompt_name="role_prompt",
                                          time=now_time,
-                                         tools=",".join([x.name for x in tools]),
+                                         tools=",".join(list(tool_dict.keys())),
                                          query=query)
         messages: List[Message] = [Message(role=Role.USER, content=user_prompt)]
         logger.info(f"step.0 user_prompt={user_prompt}")
@@ -59,6 +72,7 @@ class ReactOp(BaseLLMOp):
                 logger.info(f"submit step={i} tool_calls.name={tool_call.name} argument_dict={tool_call.argument_dict}")
 
                 if tool_call.name not in tool_dict:
+                    logger.warning(f"step={i} no tool_call.name={tool_call.name}")
                     continue
 
                 self.submit_task(tool_dict[tool_call.name].__call__, **tool_call.argument_dict)
@@ -68,7 +82,10 @@ class ReactOp(BaseLLMOp):
                 user_content_list = []
                 for tool_result, tool_call in zip(self.join_task(), assistant_message.tool_calls):
                     logger.info(f"submit step={i} tool_calls.name={tool_call.name} tool_result={tool_result}")
-                    assert isinstance(tool_result, str)
+                    if isinstance(tool_result, FlowResponse):
+                        tool_result = tool_result.answer
+                    else:
+                        tool_result = str(tool_result)
                     user_content_list.append(f"<tool_response>\n{tool_result}\n</tool_response>")
                 user_content_list.append(self.prompt_format(prompt_name="next_prompt"))
                 assistant_message.tool_calls.clear()
@@ -76,8 +93,17 @@ class ReactOp(BaseLLMOp):
 
             else:
                 assistant_message.tool_calls.clear()
-                messages.append(Message(role=Role.USER, content=self.prompt_format(prompt_name="final_prompt")))
+                query = self.prompt_format(prompt_name="final_prompt", query=query)
+                messages.append(Message(role=Role.USER, content=query))
 
         # Store results in context instead of response
-        self.context.messages = messages
-        self.context.answer = messages[-1].content
+        self.context.response.messages = messages
+        self.context.response.answer = messages[-1].content
+
+
+if __name__ == "__main__":
+    C.set_default_service_config().init_by_service_config()
+    context = FlowContext(query="茅台和五粮现在股价多少？")
+
+    op = ReactV1Op()
+    op(context=context)

flowllm-0.1.2/flowllm/op/agent/react_prompt.yaml → flowllm-0.1.3/flowllm/op/agent/react_v1_prompt.yaml

@@ -26,3 +26,29 @@ final_prompt: |
   # User's Question
   {query}
 
+
+
+
+role_prompt_zh: |
+  你是一个有用的助手。
+  当前时间是 {time}。
+  请根据用户的问题，主动选择最合适的工具或工具组合，包括 {tools} 等。
+  请先思考如何将问题分解为子任务，每个子任务应使用哪些工具和参数，最后提供工具调用名称和参数。
+  尝试多次使用相同的工具，但使用不同的参数，从多个角度获取信息。
+  请根据用户问题的语言来确定回复的语言。
+  
+  {query}
+
+next_prompt_zh: |
+  根据当前内容和用户的问题进行思考：当前上下文是否足以回答用户的问题？
+  - 如果当前上下文不足以回答用户的问题，请考虑缺少哪些信息。
+    重新规划并思考如何将缺失的信息分解为子任务。
+    对于每个子任务，确定应使用哪些工具和参数进行查询。
+    请先提供推理过程，然后给出工具调用名称和参数。
+  - 如果当前上下文足以回答用户的问题，请使用 **terminate** 工具。
+
+final_prompt_zh: |
+  请整合上下文，为用户的问题提供一个完整的答案。
+  
+  # 用户的问题
+  {query}

flowllm-0.1.3/flowllm/op/agent/react_v2_op.py

@@ -0,0 +1,86 @@
+import datetime
+import json
+import time
+from typing import List, Dict
+
+from loguru import logger
+
+from flowllm.context.flow_context import FlowContext
+from flowllm.context.service_context import C
+from flowllm.op.base_llm_op import BaseLLMOp
+from flowllm.schema.flow_response import FlowResponse
+from flowllm.schema.message import Message, Role
+
+
+@C.register_op()
+class ReactV2Op(BaseLLMOp):
+    file_path: str = __file__
+
+    def execute(self):
+        query: str = self.context.query
+
+        max_steps: int = int(self.op_params.get("max_steps", 10))
+        from flowllm.flow.base_tool_flow import BaseToolFlow
+        from flowllm.flow.gallery import DashscopeSearchToolFlow, CodeToolFlow
+
+        tools: List[BaseToolFlow] = [DashscopeSearchToolFlow(), CodeToolFlow()]
+
+        """
+        NOTE : x.tool_call.name != x.name
+        `x.tool_call.name` is tool's namex.name is flow's name(unique service name)
+        """
+        tool_dict: Dict[str, BaseToolFlow] = {x.tool_call.name: x for x in tools}
+        for name, tool_call in tool_dict.items():
+            logger.info(f"name={name} "
+                        f"tool_call={json.dumps(tool_call.tool_call.simple_input_dump(), ensure_ascii=False)}")
+
+        now_time = datetime.datetime.now().strftime('%Y-%m-%d %H:%M:%S')
+        has_terminate_tool = False
+
+        user_prompt = self.prompt_format(prompt_name="role_prompt",
+                                         time=now_time,
+                                         tools=",".join(list(tool_dict.keys())),
+                                         query=query)
+        messages: List[Message] = [Message(role=Role.USER, content=user_prompt)]
+        logger.info(f"step.0 user_prompt={user_prompt}")
+
+        for i in range(max_steps):
+            assistant_message: Message = self.llm.chat(messages, tools=[x.tool_call for x in tools])
+            messages.append(assistant_message)
+            logger.info(f"assistant.round{i}.reasoning_content={assistant_message.reasoning_content}\n"
+                        f"content={assistant_message.content}\n"
+                        f"tool.size={len(assistant_message.tool_calls)}")
+
+            if not assistant_message.tool_calls:
+                break
+
+            for j, tool_call in enumerate(assistant_message.tool_calls):
+                logger.info(f"submit step={i} tool_calls.name={tool_call.name} argument_dict={tool_call.argument_dict}")
+
+                if tool_call.name not in tool_dict:
+                    logger.warning(f"step={i} no tool_call.name={tool_call.name}")
+                    continue
+
+                self.submit_task(tool_dict[tool_call.name].__call__, **tool_call.argument_dict)
+                time.sleep(1)
+
+            for i, (tool_result, tool_call) in enumerate(zip(self.join_task(), assistant_message.tool_calls)):
+                logger.info(f"submit step={i} tool_calls.name={tool_call.name} tool_result={tool_result}")
+                if isinstance(tool_result, FlowResponse):
+                    tool_result = tool_result.answer
+                else:
+                    tool_result = str(tool_result)
+                tool_message = Message(role=Role.TOOL, content=tool_result, tool_call_id=tool_call.id)
+                messages.append(tool_message)
+
+        # Store results in context instead of response
+        self.context.response.messages = messages
+        self.context.response.answer = messages[-1].content
+
+
+if __name__ == "__main__":
+    C.set_default_service_config().init_by_service_config()
+    context = FlowContext(query="茅台和五粮现在股价多少？")
+
+    op = ReactV2Op()
+    op(context=context)

flowllm-0.1.3/flowllm/op/agent/react_v2_prompt.yaml

@@ -0,0 +1,35 @@
+role_prompt: |
+  You are a helpful assistant.
+  The current time is {time}.
+  Please proactively choose the most suitable tool or combination of tools based on the user's question, including {tools} etc.
+  Please first think about how to break down the problem into subtasks, what tools and parameters should be used for each subtask, and finally provide the tool call name and parameters.
+  Try calling the same tool multiple times with different parameters to obtain information from various perspectives.
+  Please determine the response language based on the language of the user's question.
+  
+  {query}
+
+next_prompt: |
+  Think based on the current content and the user's question: Is the current context sufficient to answer the user's question?    
+  - If the current context is not sufficient to answer the user's question, consider what information is missing. 
+    Re-plan and think about how to break down the missing information into subtasks. 
+    For each subtask, determine what tools and parameters should be used for the query.
+    Please first provide the reasoning process, then give the tool call name and parameters.
+  - If the current context is sufficient to answer the user's question, please integrate the context and provide a complete answer to the user's question.
+
+role_prompt_zh: |
+  你是一个有用的助手。
+  当前时间是 {time}。
+  请根据用户的问题，主动选择最合适的工具或工具组合，包括 {tools} 等。
+  请先思考如何将问题分解为子任务，每个子任务应使用哪些工具和参数，最后提供工具调用名称和参数。
+  尝试多次使用相同的工具，但使用不同的参数，从多个角度获取信息。
+  请根据用户问题的语言来确定回复的语言。
+  
+  {query}
+
+next_prompt_zh: |
+  根据当前内容和用户的问题进行思考：当前上下文是否足以回答用户的问题？
+  - 如果当前上下文不足以回答用户的问题，请考虑缺少哪些信息。
+    重新规划并思考如何将缺失的信息分解为子任务。
+    对于每个子任务，确定应使用哪些工具和参数进行查询。
+    请先提供推理过程，然后给出工具调用名称和参数。
+  - 如果当前上下文足以回答用户的问题，请整合上下文，为用户的问题提供一个完整的答案。

{flowllm-0.1.2 → flowllm-0.1.3}/flowllm/op/base_llm_op.py

@@ -23,8 +23,9 @@ class BaseLLMOp(BaseOp, ABC):
         super().__init__(**kwargs)
 
         self.language: str = language or C.language
-        self.prompt_path: Path = Path(prompt_path) if prompt_path else \
-            Path(self.file_path).parent / self.name.replace("_op", "_prompt.yaml")
+        default_prompt_path = self.file_path.replace("op.py", "prompt.yaml")
+        self.prompt_path: Path = Path(prompt_path) if prompt_path else default_prompt_path
+
         self._llm: BaseLLM | str = llm
         self._embedding_model: BaseEmbeddingModel | str = embedding_model
         self._vector_store: BaseVectorStore | str = vector_store

{flowllm-0.1.2 → flowllm-0.1.3}/flowllm/op/base_op.py

@@ -16,11 +16,13 @@ class BaseOp(ABC):
     def __init__(self,
                  name: str = "",
                  raise_exception: bool = True,
+                 enable_multithread: bool = True,
                  **kwargs):
         super().__init__()
 
         self.name: str = name or camel_to_snake(self.__class__.__name__)
         self.raise_exception: bool = raise_exception
+        self.enable_multithread: bool = enable_multithread
         self.op_params: dict = kwargs
 
         self.task_list: List[Future] = []
@@ -48,19 +50,34 @@ class BaseOp(ABC):
         return self.context.response if self.context else None
 
     def submit_task(self, fn, *args, **kwargs):
-        task = C.thread_pool.submit(fn, *args, **kwargs)
-        self.task_list.append(task)
+        if self.enable_multithread:
+            task = C.thread_pool.submit(fn, *args, **kwargs)
+            self.task_list.append(task)
+
+        else:
+            result = fn(*args, **kwargs)
+            if result:
+                if isinstance(result, list):
+                    result.extend(result)
+                else:
+                    result.append(result)
+
         return self
 
     def join_task(self, task_desc: str = None) -> list:
         result = []
-        for task in tqdm(self.task_list, desc=task_desc or self.name):
-            t_result = task.result()
-            if t_result:
-                if isinstance(t_result, list):
-                    result.extend(t_result)
-                else:
-                    result.append(t_result)
+        if self.enable_multithread:
+            for task in tqdm(self.task_list, desc=task_desc or self.name):
+                t_result = task.result()
+                if t_result:
+                    if isinstance(t_result, list):
+                        result.extend(t_result)
+                    else:
+                        result.append(t_result)
+
+        else:
+            result.extend(self.task_list)
+
         self.task_list.clear()
         return result
 

{flowllm-0.1.2 → flowllm-0.1.3}/flowllm/op/search/dashscope_deep_research_op.py

@@ -40,8 +40,15 @@ class DashscopeDeepResearchOp(BaseLLMOp):
         self.return_only_content = return_only_content
 
         # Ensure API key is available
-        self.api_key = os.getenv("FLOW_DASHSCOPE_API_KEY")
-        self.cache = DataCache(cache_path) if self.enable_cache else None
+        self.api_key = os.environ["FLOW_DASHSCOPE_API_KEY"]
+        self.cache_path: str = cache_path
+        self._cache: DataCache | None = None
+
+    @property
+    def cache(self):
+        if self.enable_cache and self._cache is None:
+            self._cache = DataCache(self.cache_path)
+        return self._cache
 
     def process_responses(self, responses, step_name):
         """Process streaming responses from the deep research model"""

{flowllm-0.1.2 → flowllm-0.1.3}/flowllm/op/search/dashscope_search_op.py

@@ -45,8 +45,15 @@ class DashscopeSearchOp(BaseLLMOp):
         self.enable_role_prompt = enable_role_prompt
 
         # Ensure API key is available
-        self.api_key = os.getenv("FLOW_DASHSCOPE_API_KEY")
-        self.cache = DataCache(cache_path) if self.enable_cache else None
+        self.api_key = os.environ["FLOW_DASHSCOPE_API_KEY"]
+        self.cache_path: str = cache_path
+        self._cache: DataCache | None = None
+
+    @property
+    def cache(self):
+        if self.enable_cache and self._cache is None:
+            self._cache = DataCache(self.cache_path)
+        return self._cache
 
     @staticmethod
     def format_search_results(search_results: List[Dict[str, Any]]) -> str:

{flowllm-0.1.2 → flowllm-0.1.3}/flowllm/op/search/tavily_search_op.py

@@ -16,8 +16,8 @@ from flowllm.storage.cache.data_cache import DataCache
 class TavilySearchOp(BaseOp):
     def __init__(self,
                  enable_print: bool = True,
-                 enable_cache: bool = False,
-                 cache_path: str = "./web_search_cache",
+                 enable_cache: bool = True,
+                 cache_path: str = "./tavily_search_cache",
                  cache_expire_hours: float = 0.1,
                  topic: Literal["general", "news", "finance"] = "general",
                  max_retries: int = 3,
@@ -33,8 +33,15 @@ class TavilySearchOp(BaseOp):
         self.return_only_content = return_only_content
 
         # Initialize DataCache if caching is enabled
-        self.cache = DataCache(cache_path) if self.enable_cache else None
         self._client = TavilyClient(api_key=os.getenv("FLOW_TAVILY_API_KEY", ""))
+        self.cache_path: str = cache_path
+        self._cache: DataCache | None = None
+
+    @property
+    def cache(self):
+        if self.enable_cache and self._cache is None:
+            self._cache = DataCache(self.cache_path)
+        return self._cache
 
     def post_process(self, response):
         if self.enable_print:

{flowllm-0.1.2 → flowllm-0.1.3}/flowllm/schema/message.py

@@ -1,3 +1,4 @@
+import datetime
 from typing import List
 
 from pydantic import BaseModel, Field
@@ -12,6 +13,7 @@ class Message(BaseModel):
     reasoning_content: str = Field(default="")
     tool_calls: List[ToolCall] = Field(default_factory=list)
     tool_call_id: str = Field(default="")
+    time_created: str = Field(default_factory=lambda: datetime.datetime.now().strftime("%Y-%m-%d %H:%M:%S"))
     metadata: dict = Field(default_factory=dict)
 
     def simple_dump(self, add_reason_content: bool = True) -> dict:

{flowllm-0.1.2 → flowllm-0.1.3}/flowllm/schema/tool_call.py

@@ -58,12 +58,16 @@ class ToolCall(BaseModel):
     type: str = Field(default="function")
     name: str = Field(default="")
 
-    arguments: dict = Field(default_factory=dict, description="tool execution arguments")
+    arguments: str = Field(default="", description="tool execution arguments")
 
     description: str = Field(default="")
     input_schema: Dict[str, ParamAttrs] = Field(default_factory=dict)
     output_schema: Dict[str, ParamAttrs] = Field(default_factory=dict)
 
+    @property
+    def argument_dict(self) -> dict:
+        return json.loads(self.arguments)
+
     def simple_input_dump(self, version: str = "v1") -> dict:
         if version == "v1":
             required_list = [name for name, tool_param in self.input_schema.items() if tool_param.required]
@@ -91,7 +95,7 @@ class ToolCall(BaseModel):
                 "index": self.index,
                 "id": self.id,
                 self.type: {
-                    "arguments": json.dumps(self.arguments, ensure_ascii=False),
+                    "arguments": self.arguments,
                     "name": self.name
                 },
                 "type": self.type,
@@ -111,7 +115,7 @@ class ToolCall(BaseModel):
                 if name:
                     self.name = name
                 if arguments:
-                    self.arguments = json.loads(arguments)
+                    self.arguments = arguments
         else:
             raise NotImplementedError(f"version {version} not supported")
 

{flowllm-0.1.2 → flowllm-0.1.3}/flowllm/service/http_service.py

@@ -65,8 +65,8 @@ class HttpService(BaseService):
 
             return await loop.run_in_executor(executor=C.thread_pool, func=list_tool_flows)  # noqa
 
-        endpoint_path = "/list"
-        self.app.get(endpoint_path, response_model=list)(execute_endpoint)
+        endpoint_path = "list"
+        self.app.get("/" + endpoint_path, response_model=list)(execute_endpoint)
         logger.info(f"integrate endpoint={endpoint_path}")
 
     def __call__(self):

{flowllm-0.1.2 → flowllm-0.1.3}/flowllm/storage/vector_store/base_vector_store.py

@@ -24,6 +24,9 @@ class BaseVectorStore(BaseModel, ABC):
     def _iter_workspace_nodes(self, workspace_id: str, **kwargs) -> Iterable[VectorNode]:
         raise NotImplementedError
 
+    def iter_workspace_nodes(self, workspace_id: str, **kwargs) -> Iterable[VectorNode]:
+        return self._iter_workspace_nodes(workspace_id, **kwargs)
+
     def dump_workspace(self, workspace_id: str, path: str | Path = "", callback_fn=None, **kwargs):
         raise NotImplementedError
 

{flowllm-0.1.2 → flowllm-0.1.3}/flowllm/storage/vector_store/es_vector_store.py

@@ -62,7 +62,7 @@ class EsVectorStore(LocalVectorStore):
         node.workspace_id = workspace_id
         node.unique_id = doc["_id"]
         if "_score" in doc:
-            node.metadata["_score"] = doc["_score"] - 1
+            node.metadata["score"] = doc["_score"] - 1
         return node
 
     def add_term_filter(self, key: str, value):
@@ -111,7 +111,7 @@ class EsVectorStore(LocalVectorStore):
         self.retrieve_filters.clear()
         return nodes
 
-    def insert(self, nodes: VectorNode | List[VectorNode], workspace_id: str, refresh: bool = False, **kwargs):
+    def insert(self, nodes: VectorNode | List[VectorNode], workspace_id: str, refresh: bool = True, **kwargs):
         if not self.exist_workspace(workspace_id=workspace_id):
             self.create_workspace(workspace_id=workspace_id)
 
@@ -140,7 +140,7 @@ class EsVectorStore(LocalVectorStore):
         if refresh:
             self.refresh(workspace_id=workspace_id)
 
-    def delete(self, node_ids: str | List[str], workspace_id: str, refresh: bool = False, **kwargs):
+    def delete(self, node_ids: str | List[str], workspace_id: str, refresh: bool = True, **kwargs):
         if not self.exist_workspace(workspace_id=workspace_id):
             logger.warning(f"workspace_id={workspace_id} is not exists!")
             return

flowllm-0.1.3/flowllm/utils/logger_utils.py

@@ -0,0 +1,28 @@
+import os
+import sys
+from datetime import datetime
+
+
+def init_logger():
+    from loguru import logger
+    logger.remove()
+
+    log_dir = "logs"
+    os.makedirs(log_dir, exist_ok=True)
+
+    current_ts = datetime.now().strftime("%Y-%m-%d_%H-%M-%S")
+
+    log_filename = f"{current_ts}.log"
+    log_filepath = os.path.join(log_dir, log_filename)
+
+    logger.add(log_filepath,
+               level="DEBUG",
+               rotation="00:00",
+               retention="7 days",
+               compression="zip",
+               encoding="utf-8")
+
+    logger.add(sink=sys.stdout,
+               level="INFO",
+               format="{time:YYYY-MM-DD HH:mm:ss} | {level} | {message}",
+               colorize=True)

{flowllm-0.1.2 → flowllm-0.1.3}/flowllm.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: flowllm
-Version: 0.1.2
+Version: 0.1.3
 Summary: A flexible framework for building LLM-powered flows and mcp services
 Author-email: FlowLLM Team <flowllm@example.com>
 Maintainer-email: FlowLLM Team <flowllm@example.com>

{flowllm-0.1.2 → flowllm-0.1.3}/flowllm.egg-info/SOURCES.txt

@@ -56,8 +56,10 @@ flowllm/op/base_ray_op.py
 flowllm/op/parallel_op.py
 flowllm/op/sequential_op.py
 flowllm/op/agent/__init__.py
-flowllm/op/agent/react_op.py
-flowllm/op/agent/react_prompt.yaml
+flowllm/op/agent/react_v1_op.py
+flowllm/op/agent/react_v1_prompt.yaml
+flowllm/op/agent/react_v2_op.py
+flowllm/op/agent/react_v2_prompt.yaml
 flowllm/op/akshare/__init__.py
 flowllm/op/akshare/get_ak_a_code_op.py
 flowllm/op/akshare/get_ak_a_code_prompt.yaml
@@ -97,6 +99,7 @@ flowllm/utils/__init__.py
 flowllm/utils/common_utils.py
 flowllm/utils/fetch_url.py
 flowllm/utils/llm_utils.py
+flowllm/utils/logger_utils.py
 flowllm/utils/ridge_v2.py
 flowllm/utils/singleton.py
 flowllm/utils/timer.py

{flowllm-0.1.2 → flowllm-0.1.3}/pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "setuptools.build_meta"
 
 [project]
 name = "flowllm"
-version = "0.1.2"
+version = "0.1.3"
 description = "A flexible framework for building LLM-powered flows and mcp services"
 authors = [
     {name = "FlowLLM Team", email = "flowllm@example.com"}