cleaned code, updated requirmenets

hf_mixtral_agent.py

@@ -1,15 +1,9 @@
 # HF libraries
 from langchain_community.llms import HuggingFaceEndpoint
-from langchain_core.prompts import ChatPromptTemplate
-from langchain import hub
-import gradio as gr
 from langchain.agents import AgentExecutor
 from langchain.agents.format_scratchpad import format_log_to_str
-from langchain.agents.output_parsers import (
-    ReActJsonSingleInputOutputParser,
-)
+from langchain.agents.output_parsers import ReActJsonSingleInputOutputParser
 # Import things that are needed generically
-from typing import List, Dict
 from langchain.tools.render import render_text_description
 import os
 from dotenv import load_dotenv
@@ -17,12 +11,11 @@ from innovation_pathfinder_ai.structured_tools.structured_tools import (
     arxiv_search, get_arxiv_paper, google_search, wikipedia_search
 )
 
-# hacky and should be replaced with a database
-from innovation_pathfinder_ai.source_container.container import (
-    all_sources
-)
 from langchain import PromptTemplate
 from innovation_pathfinder_ai.templates.react_json_with_memory import template_system
+from innovation_pathfinder_ai.utils import logger
+
+logger = logger.get_console_logger("hf_mixtral_agent")
 
 config = load_dotenv(".env")
 HUGGINGFACEHUB_API_TOKEN = os.getenv('HUGGINGFACEHUB_API_TOKEN')
@@ -49,13 +42,6 @@ tools = [
 #    get_arxiv_paper,
     ]
 
-tools_papers = [
-    arxiv_search,
-    get_arxiv_paper,
-
-]
-
-
 prompt = PromptTemplate.from_template(
     template=template_system
 )
@@ -87,15 +73,4 @@ agent_executor = AgentExecutor(
     #max_execution_time=60,  # timout at 60 sec
     return_intermediate_steps=True,
     handle_parsing_errors=True,
-    )
-
-# instantiate AgentExecutor
-agent_executor_noweb = AgentExecutor(
-    agent=agent, 
-    tools=tools_papers, 
-    verbose=True,
-    max_iterations=6,       # cap number of iterations
-    #max_execution_time=60,  # timout at 60 sec
-    return_intermediate_steps=True,
-    handle_parsing_errors=True,
     )

innovation_pathfinder_ai/utils.py

@@ -1,42 +0,0 @@
-def create_wikipedia_urls_from_text(text):
-    """
-    Extracts page titles from a given text and constructs Wikipedia URLs for each title.
-    
-    Args:
-    - text (str): A string containing multiple sections, each starting with "Page:" followed by the title.
-    
-    Returns:
-    - list: A list of Wikipedia URLs constructed from the extracted titles.
-    """
-    # Split the text into sections based on "Page:" prefix
-    sections = text.split("Page: ")
-    # Remove the first item if it's empty (in case the text starts with "Page:")
-    if sections[0].strip() == "":
-        sections = sections[1:]
-    
-    urls = []  # Initialize an empty list to store the URLs
-    for section in sections:
-        # Extract the title, which is the string up to the first newline
-        title = section.split("\n", 1)[0]
-        # Replace spaces with underscores for the URL
-        url_title = title.replace(" ", "_")
-        # Construct the URL and add it to the list
-        url = f"https://en.wikipedia.org/wiki/{url_title}"
-        urls.append(url)
-    
-    return urls
-
-def collect_urls(data_list):
-    urls = []
-    for item in data_list:
-        # Check if item is a string and contains 'link:'
-        if isinstance(item, str) and 'link:' in item:
-            start = item.find('link:') + len('link: ')
-            end = item.find(',', start)
-            url = item[start:end if end != -1 else None].strip()
-            urls.append(url)
-        # Check if item is a dictionary and has 'Entry ID'
-        elif isinstance(item, dict) and 'Entry ID' in item:
-            urls.append(item['Entry ID'])
-    last_sources = urls[-3:]
-    return last_sources

requirements.txt

@@ -8,4 +8,6 @@ wikipedia
 gradio==3.48.0
 chromadb
 google_api_python_client
-pypdf2
+pypdf2
+sqlmodel
+rich