created a FastAPI backend

innovation_pathfinder_ai/agents/hf_mixtral_agent.py

@@ -1,81 +0,0 @@
-# HF libraries
-from langchain_community.llms import HuggingFaceEndpoint
-from langchain.agents import AgentExecutor
-from langchain.agents.format_scratchpad import format_log_to_str
-from langchain.agents.output_parsers import ReActJsonSingleInputOutputParser
-# Import things that are needed generically
-from langchain.tools.render import render_text_description
-import os
-from dotenv import load_dotenv
-from innovation_pathfinder_ai.structured_tools.structured_tools import (
-    arxiv_search, get_arxiv_paper, google_search, wikipedia_search, knowledgeBase_search, memory_search
-)
-
-from langchain.prompts import PromptTemplate
-from innovation_pathfinder_ai.templates.react_json_with_memory import template_system
-from innovation_pathfinder_ai.utils import logger
-from langchain.globals import set_llm_cache
-from langchain.cache import SQLiteCache
-
-set_llm_cache(SQLiteCache(database_path=".cache.db"))
-logger = logger.get_console_logger("hf_mixtral_agent")
-
-config = load_dotenv(".env")
-HUGGINGFACEHUB_API_TOKEN = os.getenv('HUGGINGFACEHUB_API_TOKEN')
-GOOGLE_CSE_ID = os.getenv('GOOGLE_CSE_ID')
-GOOGLE_API_KEY = os.getenv('GOOGLE_API_KEY')
-LANGCHAIN_TRACING_V2 = "true"
-LANGCHAIN_ENDPOINT = "https://api.smith.langchain.com"
-LANGCHAIN_API_KEY = os.getenv('LANGCHAIN_API_KEY')
-LANGCHAIN_PROJECT = os.getenv('LANGCHAIN_PROJECT')
-
-# Load the model from the Hugging Face Hub
-llm = HuggingFaceEndpoint(repo_id="mistralai/Mixtral-8x7B-Instruct-v0.1", 
-                          temperature=0.1, 
-                          max_new_tokens=1024,
-                          repetition_penalty=1.2,
-                          return_full_text=False
-    )
-
-
-tools = [
-    memory_search,
-    knowledgeBase_search,
-    arxiv_search,
-    wikipedia_search,
-    google_search,
-#    get_arxiv_paper,
-    ]
-
-prompt = PromptTemplate.from_template(
-    template=template_system
-)
-prompt = prompt.partial(
-    tools=render_text_description(tools),
-    tool_names=", ".join([t.name for t in tools]),
-)
-
-
-# define the agent
-chat_model_with_stop = llm.bind(stop=["\nObservation"])
-agent = (
-    {
-        "input": lambda x: x["input"],
-        "agent_scratchpad": lambda x: format_log_to_str(x["intermediate_steps"]),
-        "chat_history": lambda x: x["chat_history"],
-    }
-    | prompt
-    | chat_model_with_stop
-    | ReActJsonSingleInputOutputParser()
-)
-
-# instantiate AgentExecutor
-agent_executor = AgentExecutor(
-    agent=agent, 
-    tools=tools, 
-    verbose=True,
-    max_iterations=10,       # cap number of iterations
-    #max_execution_time=60,  # timout at 60 sec
-    return_intermediate_steps=True,
-    handle_parsing_errors=True,
-    )

innovation_pathfinder_ai/backend/app/api/v1/agents/hf_mixtral_agent.py

@@ -0,0 +1,112 @@
+# HF libraries
+from langchain_community.llms import HuggingFaceEndpoint
+from langchain.agents import AgentExecutor
+from langchain.agents.format_scratchpad import format_log_to_str
+from langchain.agents.output_parsers import ReActJsonSingleInputOutputParser
+# Import things that are needed generically
+from langchain.tools.render import render_text_description
+import os
+from dotenv import load_dotenv
+from innovation_pathfinder_ai.backend.app.structured_tools.structured_tools import (
+    arxiv_search, get_arxiv_paper, google_search, wikipedia_search, knowledgeBase_search, memory_search
+)
+from fastapi import APIRouter, WebSocket, WebSocketDisconnect
+from langchain.prompts import PromptTemplate
+from innovation_pathfinder_ai.backend.app.templates.react_json_with_memory import template_system
+from innovation_pathfinder_ai.backend.app.utils import logger
+from innovation_pathfinder_ai.backend.app.utils import generate_uuid
+from langchain.globals import set_llm_cache
+from langchain.cache import SQLiteCache
+
+set_llm_cache(SQLiteCache(database_path=".cache.db"))
+logger = logger.get_console_logger("hf_mixtral_agent")
+
+config = load_dotenv(".env")
+HUGGINGFACEHUB_API_TOKEN = os.getenv('HUGGINGFACEHUB_API_TOKEN')
+GOOGLE_CSE_ID = os.getenv('GOOGLE_CSE_ID')
+GOOGLE_API_KEY = os.getenv('GOOGLE_API_KEY')
+LANGCHAIN_TRACING_V2 = "true"
+LANGCHAIN_ENDPOINT = "https://api.smith.langchain.com"
+LANGCHAIN_API_KEY = os.getenv('LANGCHAIN_API_KEY')
+LANGCHAIN_PROJECT = os.getenv('LANGCHAIN_PROJECT')
+
+router = APIRouter()
+
+@router.websocket("")
+async def websocket_endpoint(websocket: WebSocket):
+    await websocket.accept()
+
+    while True:
+        try:
+            data = await websocket.receive_json()
+            user_message = data["message"]
+
+            resp = IChatResponse(
+                sender="you",
+                message=user_message_card.to_dict(),
+                type="start",
+                message_id=generate_uuid(),
+                id=generate_uuid(),
+            )
+
+            await websocket.send_json(resp.dict())
+            message_id: str = generate_uuid()
+            # custom_handler = CustomFinalStreamingStdOutCallbackHandler(
+            #     websocket, message_id=message_id
+            # )
+
+            # Load the model from the Hugging Face Hub
+            llm = HuggingFaceEndpoint(repo_id="mistralai/Mixtral-8x7B-Instruct-v0.1", 
+                                    temperature=0.1, 
+                                    max_new_tokens=1024,
+                                    repetition_penalty=1.2,
+                                    return_full_text=False
+                )
+
+
+            tools = [
+                memory_search,
+                knowledgeBase_search,
+                arxiv_search,
+                wikipedia_search,
+                google_search,
+            #    get_arxiv_paper,
+                ]
+
+            prompt = PromptTemplate.from_template(
+                template=template_system
+            )
+            prompt = prompt.partial(
+                tools=render_text_description(tools),
+                tool_names=", ".join([t.name for t in tools]),
+            )
+
+
+            # define the agent
+            chat_model_with_stop = llm.bind(stop=["\nObservation"])
+            agent = (
+                {
+                    "input": lambda x: x["input"],
+                    "agent_scratchpad": lambda x: format_log_to_str(x["intermediate_steps"]),
+                    "chat_history": lambda x: x["chat_history"],
+                }
+                | prompt
+                | chat_model_with_stop
+                | ReActJsonSingleInputOutputParser()
+            )
+
+            # instantiate AgentExecutor
+            agent_executor = AgentExecutor(
+                agent=agent, 
+                tools=tools, 
+                verbose=True,
+                max_iterations=10,       # cap number of iterations
+                #max_execution_time=60,  # timout at 60 sec
+                return_intermediate_steps=True,
+                handle_parsing_errors=True,
+                )
+            
+            await agent_executor.arun(input=user_message) #, callbacks=[custom_handler]
+        except WebSocketDisconnect:
+            logger.info("websocket disconnect")
+            break

innovation_pathfinder_ai/backend/app/api/v1/api.py

@@ -0,0 +1,12 @@
+from fastapi import APIRouter
+from app.api.v1.agents import (
+    hf_mixtral_agent,
+)
+from app.api.v1.endpoints import (
+    add_to_kb
+)
+
+api_router = APIRouter()
+
+api_router.include_router(hf_mixtral_agent.router, prefix="/chat", tags=["chat"])
+api_router.include_router(add_to_kb.router, prefix="/add-documents", tags=["documents"])

innovation_pathfinder_ai/backend/app/api/v1/endpoints/add_to_kb.py

@@ -0,0 +1,33 @@
+from fastapi import APIRouter
+from innovation_pathfinder_ai.backend.app.utils.utils import extract_urls
+from innovation_pathfinder_ai.backend.app.utils import logger
+from innovation_pathfinder_ai.backend.app.vector_store import initialize_chroma_db 
+from innovation_pathfinder_ai.backend.app.utils.utils import (
+    generate_uuid    
+)
+from langchain_community.vectorstores import Chroma
+
+router = APIRouter()
+
+@router.post("/add-document", status_code=201)
+async def add_document(conversation):
+
+    db = initialize_chroma_db()
+    all_sources = conversation.sources
+    documents = conversation.documents
+    message = conversation.query
+
+    sources = extract_urls(all_sources)
+    src_list = '\n'.join(sources)
+    current_id = generate_uuid()
+    db.add(
+        ids=[current_id],
+        documents=documents,
+        metadatas=[
+            {
+                "human_message":message,
+                "sources": 'Internal Knowledge Base From: \n\n' + src_list
+            }
+        ]
+    )
+    return {"message": "Document added successfully", "ids": current_id}

innovation_pathfinder_ai/backend/app/core/config.py

@@ -0,0 +1,34 @@
+import os
+from pydantic import AnyHttpUrl, BaseSettings
+from enum import Enum
+
+
+class ModeEnum(str, Enum):
+    development = "development"
+    production = "production"
+    testing = "testing"
+
+
+class Settings(BaseSettings):
+    PROJECT_NAME: str = "app"
+    BACKEND_CORS_ORIGINS: list[str] | list[AnyHttpUrl]
+    MODE: ModeEnum = ModeEnum.development
+    API_VERSION: str = "v1"
+    API_V1_STR: str = f"/api/{API_VERSION}"
+    OPENAI_API_KEY: str
+    UNSPLASH_API_KEY: str
+    SERP_API_KEY: str
+    HUGGINGFACEHUB_API_TOKEN: str
+    GOOGLE_CSE_ID: str
+    GOOGLE_API_KEY: str
+    VECTOR_DATABASE_LOCATION: str
+    CONVERSATION_COLLECTION_NAME: str
+    EMBEDDING_MODEL: str
+    SOURCES_CACHE: str
+
+    class Config:
+        case_sensitive = True
+        env_file = os.path.expanduser("~/.env")
+
+
+settings = Settings()

innovation_pathfinder_ai/{database → backend/app/crud}/db_handler.py

@@ -1,6 +1,6 @@
 from sqlmodel import SQLModel, create_engine, Session, select
-from innovation_pathfinder_ai.database.schema import Sources
-from innovation_pathfinder_ai.utils.logger import get_console_logger
+from innovation_pathfinder_ai.backend.app.database.db_schema import Sources
+from innovation_pathfinder_ai.backend.app.utils.logger import get_console_logger
 import os
 from dotenv import load_dotenv
 

innovation_pathfinder_ai/backend/app/main.py

@@ -0,0 +1,27 @@
+from fastapi import FastAPI
+from innovation_pathfinder_ai.backend.app.api.v1.api import api_router as api_router_v1
+#from app.core.config import settings
+from fastapi.middleware.cors import CORSMiddleware
+
+app = FastAPI()
+BACKEND_CORS_ORIGINS = ["*"]
+# CORS Middleware setup for allowing frontend requests
+# ToDO: replace with settings.BACKEND_CORS_ORIGINS once core/config.py is implemented
+if BACKEND_CORS_ORIGINS:
+    app.add_middleware(
+        CORSMiddleware,
+        allow_origins=[str(origin) for origin in BACKEND_CORS_ORIGINS],
+        allow_credentials=True,
+        allow_methods=["*"],
+        allow_headers=["*"],
+    )
+
+@app.get("/", tags=["Root"])
+async def root():
+    """
+    Simple root endpoint to verify the API is running.
+    """
+    return {"message": "API is running"}
+
+# Include the versioned API router from api.py
+app.include_router(api_router_v1)

innovation_pathfinder_ai/backend/app/schemas/message_schema.py

@@ -0,0 +1,10 @@
+from pydantic import BaseModel
+from typing import List, Tuple, Optional
+
+class InferRequest(BaseModel):
+    question: str
+    history: List[Tuple[str, str]]  
+
+class BotRequest(BaseModel):
+    history: List[Tuple[str, str]]
+

innovation_pathfinder_ai/backend/app/schemas/response_schema.py

@@ -0,0 +1,18 @@
+from pydantic import BaseModel
+from typing import List, Tuple, Optional
+
+class SourceData(BaseModel):
+    human_message: str
+    sources: str
+
+class DocumentAddResponse(BaseModel):
+    success: bool
+    message: Optional[str] = None
+
+class InferResponse(BaseModel):
+    output: str
+    sources: Optional[List[str]] = []
+
+class BotResponse(BaseModel):
+    history: List[Tuple[str, str]]
+    response_with_sources: str  # Consolidated response with sources appended

innovation_pathfinder_ai/{structured_tools → backend/app/structured_tools}/structured_tools.py

@@ -21,7 +21,7 @@ from innovation_pathfinder_ai.source_container.container import (
 from innovation_pathfinder_ai.utils.utils import (
     parse_list_to_dicts, format_wiki_summaries, format_arxiv_documents, format_search_results
 )
-from innovation_pathfinder_ai.database.db_handler import (
+from backend.app.crud.db_handler import (
     add_many
 )
 

innovation_pathfinder_ai/{utils → backend/app/utils}/utils.py

@@ -3,7 +3,7 @@ import datetime
 import os
 import uuid
 
-from innovation_pathfinder_ai.utils import logger
+from innovation_pathfinder_ai.backend.app.utils import logger
 
 logger = logger.get_console_logger("utils")
 

innovation_pathfinder_ai/backend/app/vector_store/initialize_chroma_db.py

@@ -0,0 +1,20 @@
+from langchain_community.vectorstores import Chroma
+import chromadb
+import dotenv
+import os
+
+dotenv.load_dotenv()
+persist_directory = os.getenv('VECTOR_DATABASE_LOCATION')
+
+def initialize_chroma_db() -> Chroma:
+    collection_name = os.getenv('CONVERSATION_COLLECTION_NAME')
+    
+    client = chromadb.PersistentClient(
+        path=persist_directory
+        )
+    
+    collection = client.get_or_create_collection(
+    name=collection_name,
+    )
+    
+    return collection

app.py → innovation_pathfinder_ai/frontend/app.py

@@ -1,14 +1,14 @@
 from fastapi import FastAPI
 import gradio as gr
 from gradio.themes.base import Base
-from innovation_pathfinder_ai.agents.hf_mixtral_agent import agent_executor
+from innovation_pathfinder_ai.backend.app.api.v1.agents.hf_mixtral_agent import agent_executor
 from innovation_pathfinder_ai.source_container.container import (
     all_sources
 )
-from innovation_pathfinder_ai.utils.utils import extract_urls
-from innovation_pathfinder_ai.utils import logger
+from innovation_pathfinder_ai.backend.app.utils.utils import extract_urls
+from innovation_pathfinder_ai.backend.app.utils import logger
 
-from innovation_pathfinder_ai.utils.utils import (
+from innovation_pathfinder_ai.backend.app.utils.utils import (
     generate_uuid    
 )
 from langchain_community.vectorstores import Chroma
@@ -49,6 +49,10 @@ if __name__ == "__main__":
 
     def bot(history):
         response = infer(history[-1][0], history)
+        # Existing logic remains the same up to the point where you need to call backend operations
+        # Example for calling generate_uuid from the backend
+        # response = requests.post("http://localhost:8000/add-document")
+        #current_id = response.text
         sources = extract_urls(all_sources)
         src_list = '\n'.join(sources)
         current_id = generate_uuid()

requirements.txt

@@ -12,4 +12,5 @@ pypdf2
 sqlmodel
 rich
 fastapi
+uvicorn
 sentence-transformers