ws-bom-robot-app 0.0.76__py3-none-any.whl → 0.0.78__py3-none-any.whl

ws_bom_robot_app/llm/agent_handler.py

@@ -5,10 +5,9 @@ from langchain.callbacks.base import AsyncCallbackHandler
 from ws_bom_robot_app.llm.utils.print import print_json, print_string
 from typing import Any, Dict, List, Optional, Union
 from uuid import UUID
-import ws_bom_robot_app.llm.settings as settings
 from langchain_core.callbacks.base import AsyncCallbackHandler
 from langchain_core.outputs import ChatGenerationChunk, GenerationChunk
-from langchain_core.messages import BaseMessage, HumanMessage, AIMessage
+from langchain_core.messages import BaseMessage, AIMessage
 import json, logging, re
 
 # Here is a custom handler that will print the tokens to stdout.
@@ -118,11 +117,6 @@ class AgentHandler(AsyncCallbackHandler):
         tags: List[str] = None,
         **kwargs: Any,
     ) -> None:
-        settings.chat_history.extend(
-            [
-                AIMessage(content=_parse_token(self.llm,finish.return_values["output"])),
-            ]
-        )
         # end-of-stream: flush any remaining text
         if self.in_json_block:
             try:
@@ -169,9 +163,4 @@ class RawAgentHandler(AsyncCallbackHandler):
         tags: List[str] = None,
         **kwargs: Any,
     ) -> None:
-        settings.chat_history.extend(
-            [
-                AIMessage(content=_parse_token(self.llm,finish.return_values["output"]))
-            ]
-        )
         await self.queue.put(None)

ws_bom_robot_app/llm/api.py

@@ -21,9 +21,9 @@ async def root():
 async def _invoke(rq: InvokeRequest):
     return await invoke(rq)
 
-def _stream_headers(rq: StreamRequest) -> Mapping[str, str]:
+def _rs_stream_headers(rq: StreamRequest) -> Mapping[str, str]:
     return {
-        "X-thread-id": rq.msg_id or str(uuid4()),
+        "X-thread-id": rq.thread_id or str(uuid4()),
         "X-msg-id": rq.msg_id or str(uuid4()),
         }
 
@@ -40,11 +40,11 @@ async def cms_app_by_id(id: str):
 
 @router.post("/stream")
 async def _stream(rq: StreamRequest, ctx: Request) -> StreamingResponse:
-    return StreamingResponse(stream(rq, ctx), media_type="application/json", headers=_stream_headers(rq))
+    return StreamingResponse(stream(rq, ctx), media_type="application/json", headers=_rs_stream_headers(rq))
 
 @router.post("/stream/raw")
 async def _stream_raw(rq: StreamRequest, ctx: Request) -> StreamingResponse:
-    return StreamingResponse(stream(rq, ctx, formatted=False), media_type="application/json", headers=_stream_headers(rq))
+    return StreamingResponse(stream(rq, ctx, formatted=False), media_type="application/json", headers=_rs_stream_headers(rq))
 
 @router.post("/kb")
 async def _kb(rq: KbRequest) -> VectorDbResponse:

ws_bom_robot_app/llm/main.py

@@ -3,7 +3,7 @@ import asyncio, json, logging, os, traceback, re
 from fastapi import Request
 from langchain.callbacks.tracers import LangChainTracer
 from langchain_core.callbacks.base import AsyncCallbackHandler
-from langchain_core.messages import AIMessage, HumanMessage
+from langchain_core.messages import BaseMessage, AIMessage, HumanMessage
 from langsmith import Client as LangSmithClient
 from typing import AsyncGenerator, List
 from ws_bom_robot_app.config import config
@@ -14,7 +14,6 @@ from ws_bom_robot_app.llm.models.api import InvokeRequest, StreamRequest
 from ws_bom_robot_app.llm.providers.llm_manager import LlmInterface
 from ws_bom_robot_app.llm.tools.tool_builder import get_structured_tools
 from ws_bom_robot_app.llm.nebuly_handler import NebulyHandler
-import ws_bom_robot_app.llm.settings as settings
 
 async def invoke(rq: InvokeRequest) -> str:
   await rq.initialize()
@@ -40,21 +39,19 @@ def _parse_formatted_message(message: str) -> str:
   except:
     result = message
   return result
-async def __stream(rq: StreamRequest, ctx: Request, queue: Queue,formatted: bool = True) -> None:
-  await rq.initialize()
+async def __stream(rq: StreamRequest, ctx: Request, queue: Queue, formatted: bool = True) -> None:
   #os.environ["KMP_DUPLICATE_LIB_OK"] = "TRUE"
-  if formatted:
-    agent_handler = AgentHandler(queue,rq.provider,rq.thread_id)
-  else:
-    agent_handler = RawAgentHandler(queue,rq.provider)
-  os.environ["AGENT_HANDLER_FORMATTED"] = str(formatted)
-  callbacks: List[AsyncCallbackHandler] = [agent_handler]
-  settings.init()
 
-  #CREATION OF CHAT HISTORY FOR AGENT
+  # rq initialization
+  await rq.initialize()
+  for tool in rq.app_tools:
+    tool.thread_id = rq.thread_id
+
+  #chat history
+  chat_history: list[BaseMessage] = []
   for message in rq.messages:
     if message.role in ["human","user"]:
-      settings.chat_history.append(HumanMessage(content=message.content))
+      chat_history.append(HumanMessage(content=message.content))
     elif message.role in ["ai","assistant"]:
       message_content = ""
       if formatted:
@@ -79,39 +76,53 @@ async def __stream(rq: StreamRequest, ctx: Request, queue: Queue,formatted: bool
       else:
         message_content = message.content
       if message_content:
-        settings.chat_history.append(AIMessage(content=message_content))
+        chat_history.append(AIMessage(content=message_content))
+
+  #llm
+  __llm: LlmInterface = rq.get_llm()
+
+  #agent handler
+  if formatted:
+    agent_handler = AgentHandler(queue, rq.provider, rq.thread_id)
+  else:
+    agent_handler = RawAgentHandler(queue, rq.provider)
+  #TODO: move from os.environ to rq
+  os.environ["AGENT_HANDLER_FORMATTED"] = str(formatted)
 
+  #callbacks
+  ## agent
+  callbacks: List[AsyncCallbackHandler] = [agent_handler]
+  ## langchain tracing
   if rq.lang_chain_tracing:
     client = LangSmithClient(
       api_key= rq.secrets.get("langChainApiKey", "")
     )
     trace = LangChainTracer(project_name=rq.lang_chain_project,client=client,tags=[str(ctx.base_url) if ctx else ''])
     callbacks.append(trace)
-
-  __llm: LlmInterface =rq.get_llm()
-  for tool in rq.app_tools:
-    tool.thread_id = rq.thread_id
-  processor = AgentLcel(
-      llm=__llm,
-      sys_message=rq.system_message,
-      sys_context=rq.system_context,
-      tools=get_structured_tools(__llm, tools=rq.app_tools, callbacks=[callbacks], queue=queue),
-      rules=rq.rules
-  )
+  ## nebuly tracing
   if rq.secrets.get("nebulyApiKey","") != "":
     user_id = rq.system_context.user.id if rq.system_context and rq.system_context.user and rq.system_context.user.id else None
     nebuly_callback = NebulyHandler(
         llm_model=__llm.config.model,
         threadId=rq.thread_id,
+        chat_history=chat_history,
         url=config.NEBULY_API_URL,
         api_key=rq.secrets.get("nebulyApiKey", None),
         user_id=user_id
       )
     callbacks.append(nebuly_callback)
 
+  # chain
+  processor = AgentLcel(
+      llm=__llm,
+      sys_message=rq.system_message,
+      sys_context=rq.system_context,
+      tools=get_structured_tools(__llm, tools=rq.app_tools, callbacks=[callbacks], queue=queue),
+      rules=rq.rules
+  )
   try:
     await processor.executor.ainvoke(
-        {"chat_history": settings.chat_history},
+        {"chat_history": chat_history},
         {"callbacks": callbacks},
     )
   except Exception as e:
@@ -122,7 +133,7 @@ async def __stream(rq: StreamRequest, ctx: Request, queue: Queue,formatted: bool
       await queue.put(_error)
     await queue.put(None)
 
-  # Signal the end of streaming
+  # signal the end of streaming
   await queue.put(None)
 
 async def stream(rq: StreamRequest, ctx: Request, formatted: bool = True) -> AsyncGenerator[str, None]:

ws_bom_robot_app/llm/models/api.py

@@ -169,8 +169,8 @@ class InvokeRequest(LlmApp):
   mode: str
 
 class StreamRequest(LlmApp):
-  thread_id: Optional[str] = Field(None, validation_alias=AliasChoices("threadId","thread_id"))
-  msg_id: Optional[str] = Field(None, validation_alias=AliasChoices("msgId","msg_id"))
+  thread_id: Optional[str] = Field(default=str(uuid.uuid4()), validation_alias=AliasChoices("threadId","thread_id"))
+  msg_id: Optional[str] = Field(default=str(uuid.uuid4()), validation_alias=AliasChoices("msgId","msg_id"))
 #endregion
 
 #region vector_db

ws_bom_robot_app/llm/nebuly_handler.py

@@ -2,16 +2,16 @@ from typing import Union
 from ws_bom_robot_app.llm.models.api import NebulyInteraction, NebulyLLMTrace, NebulyRetrievalTrace
 from datetime import datetime, timezone
 from langchain_core.callbacks.base import AsyncCallbackHandler
-import ws_bom_robot_app.llm.settings as settings
 from langchain_core.messages import BaseMessage, HumanMessage, AIMessage
 from langchain_core.outputs import ChatGenerationChunk, GenerationChunk
 
 class NebulyHandler(AsyncCallbackHandler):
-    def __init__(self, llm_model: str | None, threadId: str = None, url: str = None, api_key: str = None, user_id: str | None = None):
+    def __init__(self, llm_model: str | None, threadId: str = None, chat_history: list[BaseMessage] = [],  url: str = None, api_key: str = None, user_id: str | None = None):
         super().__init__()
         self.__started: bool = False
         self.__url: str = url
         self.__api_key: str = api_key
+        self.chat_history = chat_history
         self.interaction = NebulyInteraction(
             conversation_id=threadId,
             input="",
@@ -76,7 +76,7 @@ class NebulyHandler(AsyncCallbackHandler):
         self.interaction.output = finish.return_values["output"]
         # Trace
         self.llm_trace.output = finish.return_values["output"]
-        message_history =  self._convert_to_json_format(settings.chat_history)[:-1]
+        message_history =  self._convert_to_json_format(self.chat_history)
         self.llm_trace.messages = self.__parse_multimodal_history(message_history)
         await self.__send_interaction()
 
@@ -153,7 +153,7 @@ class NebulyHandler(AsyncCallbackHandler):
                 parsed_input += item.get("text", "")
             elif item.get("type") == "image_url":
                 parsed_input += " <image>"
-        print(parsed_input)
+        # print(parsed_input)
         return parsed_input
 
     def __parse_multimodal_history(self, messages: list[dict]) -> list[dict]:

{ws_bom_robot_app-0.0.76.dist-info → ws_bom_robot_app-0.0.78.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: ws_bom_robot_app
-Version: 0.0.76
+Version: 0.0.78
 Summary: A FastAPI application serving ws bom/robot/llm platform ai.
 Home-page: https://github.com/websolutespa/bom
 Author: Websolute Spa
@@ -25,7 +25,7 @@ Requires-Dist: langchain-anthropic==0.3.6
 Requires-Dist: langchain-ibm==0.3.14
 Requires-Dist: langchain-google-genai==2.0.7
 Requires-Dist: langchain-google-vertexai==2.0.27
-Requires-Dist: langchain-groq==0.3.5
+Requires-Dist: langchain-groq==0.3.6
 Requires-Dist: langchain-ollama==0.3.3
 Requires-Dist: faiss-cpu==1.11.0
 Requires-Dist: chromadb==1.0.15

{ws_bom_robot_app-0.0.76.dist-info → ws_bom_robot_app-0.0.78.dist-info}/RECORD

@@ -8,17 +8,16 @@ ws_bom_robot_app/util.py,sha256=RjVD6B9sHje788Lndqq5DHy6TJM0KLs9qx3JYt81Wyk,4834
 ws_bom_robot_app/llm/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 ws_bom_robot_app/llm/agent_context.py,sha256=uatHJ8wcRly6h0S762BgfzDMpmcwCHwNzwo37aWjeE0,1305
 ws_bom_robot_app/llm/agent_description.py,sha256=yK4aVU3RNk1oP4bEneV3QPAi-208JwWk4R6qHlzqYIg,4656
-ws_bom_robot_app/llm/agent_handler.py,sha256=BQ-f--Z5QCJDp-7tzSG_CKrANUCqG65S09psgWVNxa4,7597
+ws_bom_robot_app/llm/agent_handler.py,sha256=TnpfChHLWVQ-gCEHNQPW3UXiuS8AmiP8JYwRz9pqbCg,7203
 ws_bom_robot_app/llm/agent_lcel.py,sha256=tVa1JJOuL1CG0tXS5AwOB4gli0E2rGqSBD5oEehHvOY,2480
-ws_bom_robot_app/llm/api.py,sha256=2bF-UFczY9LuBqPxKObM0TOWYbZgVztX1RiIz5MSorU,5042
+ws_bom_robot_app/llm/api.py,sha256=EUllZaJvtm1CQPxJ6QfufpBhZG_-ew8gSK-vxVg5r58,5054
 ws_bom_robot_app/llm/defaut_prompt.py,sha256=LlCd_nSMkMmHESfiiiQYfnJyB6Pp-LSs4CEKdYW4vFk,1106
-ws_bom_robot_app/llm/main.py,sha256=vzUfaLCRk2SYujD00hnrTiHEVLYgZcbSw6LUea43siU,5235
-ws_bom_robot_app/llm/nebuly_handler.py,sha256=d4TI5XbvIYJLCxHcCUa6QUxsgwKW_4ItCYe4ocn7IJo,7900
-ws_bom_robot_app/llm/settings.py,sha256=EkFGCppORenStH9W4e6_dYvQ-5p6xiEMpmUHBqNqG9M,117
+ws_bom_robot_app/llm/main.py,sha256=BXTIfVc9Ck7izZ893qry7C_uz1A8ZupbcHivrZrjpxY,5372
+ws_bom_robot_app/llm/nebuly_handler.py,sha256=hbkiTc0Jl4EzwXltpICiUXM5i5wOsmEX_Chyr1NhvSc,7924
 ws_bom_robot_app/llm/feedbacks/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 ws_bom_robot_app/llm/feedbacks/feedback_manager.py,sha256=WcKgzlOb8VFG7yqHoIOO_R6LAzdzE4YIRFCVOGBSgfM,2856
 ws_bom_robot_app/llm/models/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
-ws_bom_robot_app/llm/models/api.py,sha256=8hllPUnPM6Rp6TgcvTQtpUh5Nk36r1qcvIEc2O7LZJE,10968
+ws_bom_robot_app/llm/models/api.py,sha256=qr9BO3NqQ4juYcJGARPiKas5Mj2gIg6gUH9MpmDkMLE,11010
 ws_bom_robot_app/llm/models/base.py,sha256=1TqxuTK3rjJEALn7lvgoen_1ba3R2brAgGx6EDTtDZo,152
 ws_bom_robot_app/llm/models/feedback.py,sha256=zh1jLqPRLzNlxInkCMoiJbfSu0-tiOEYHM7FhC46PkM,1692
 ws_bom_robot_app/llm/models/kb.py,sha256=oVSw6_dmNxikAHrPqcfxDXz9M0ezLIYuxpgvzfs_Now,9514
@@ -67,7 +66,7 @@ ws_bom_robot_app/llm/vector_store/loader/__init__.py,sha256=47DEQpj8HBSa-_TImW-5
 ws_bom_robot_app/llm/vector_store/loader/base.py,sha256=L_ugekNuAq0N9O-24wtlHSNHkqSeD-KsJrfGt_FX9Oc,5340
 ws_bom_robot_app/llm/vector_store/loader/docling.py,sha256=yP0zgXLeFAlByaYuj-6cYariuknckrFds0dxdRcnVz8,3456
 ws_bom_robot_app/llm/vector_store/loader/json_loader.py,sha256=qo9ejRZyKv_k6jnGgXnu1W5uqsMMtgqK_uvPpZQ0p74,833
-ws_bom_robot_app-0.0.76.dist-info/METADATA,sha256=e6QyzVw9spgo-qPmhDmX-O5oYtDvGN_zJLnhy2gHTuk,8609
-ws_bom_robot_app-0.0.76.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
-ws_bom_robot_app-0.0.76.dist-info/top_level.txt,sha256=Yl0akyHVbynsBX_N7wx3H3ZTkcMLjYyLJs5zBMDAKcM,17
-ws_bom_robot_app-0.0.76.dist-info/RECORD,,
+ws_bom_robot_app-0.0.78.dist-info/METADATA,sha256=96-Ate6TbTUzRqNCLEe6gJEblJOA4r9BoVY6Ajbb2_4,8609
+ws_bom_robot_app-0.0.78.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
+ws_bom_robot_app-0.0.78.dist-info/top_level.txt,sha256=Yl0akyHVbynsBX_N7wx3H3ZTkcMLjYyLJs5zBMDAKcM,17
+ws_bom_robot_app-0.0.78.dist-info/RECORD,,

ws_bom_robot_app/llm/settings.py

@@ -1,4 +0,0 @@
-def init():
-    """Initialize the chat history list as a global var"""
-    global chat_history
-    chat_history = []