ZarinT's picture
Update app.py
72f2043 verified
from huggingface_hub import HfApi
import tempfile
import streamlit as st
import pandas as pd
import io, os
import subprocess
import pdfplumber
from lxml import etree
from bs4 import BeautifulSoup
from PyPDF2 import PdfReader
from langchain_community.vectorstores import FAISS
from langchain.embeddings.base import Embeddings
from langchain_openai import ChatOpenAI
from langchain.agents import initialize_agent, AgentType
from langchain.agents import Tool
from langchain.memory import ConversationBufferMemory
from langchain.text_splitter import CharacterTextSplitter
from dotenv import load_dotenv
import google.generativeai as genai
from typing import List
from langchain_core.language_models import BaseLanguageModel
from langchain_core.runnables import Runnable
import google.generativeai as genai
from datetime import datetime
load_dotenv()
def load_environment():
# Ensure HF_TOKEN is available
if "HUGGINGFACEHUB_API_TOKEN" not in os.environ and "HF_TOKEN" in os.environ:
os.environ["HUGGINGFACEHUB_API_TOKEN"] = os.environ["HF_TOKEN"]
if "GOOGLE_API_KEY" not in os.environ:
raise ValueError("GOOGLE_API_KEY not found in environment variables.")
genai.configure(api_key=st.secrets["GOOGLE_API_KEY"])
from keybert import KeyBERT
from sentence_transformers import CrossEncoder
from sentence_transformers import SentenceTransformer
class GeminiLLM(Runnable):
def __init__(self, model_name="models/gemini-1.5-pro-latest", api_key=None):
self.api_key = api_key or os.environ["GOOGLE_API_KEY"]
if not self.api_key:
raise ValueError("GOOGLE_API_KEY not found.")
genai.configure(api_key=self.api_key)
self.model = genai.GenerativeModel(model_name)
def _call(self, prompt: str, stop=None) -> str:
response = self.model.generate_content(prompt)
return response.text
@property
def _llm_type(self) -> str:
return "custom_gemini"
def invoke(self, input, config=None):
response = self.model.generate_content(input)
return response.text.strip()
class GeminiEmbeddings(Embeddings):
def __init__(self, model_name="models/embedding-001", api_key=None):
api_key = "AIzaSyBIfGJRoet_wzzYXIiWXxStkIigEOzSR2o"
if not api_key:
raise ValueError("GOOGLE_API_KEY not found in environment variables.")
os.environ["GOOGLE_API_KEY"] = api_key
genai.configure(api_key=api_key)
self.model_name = model_name
def embed_documents(self, texts: List[str]) -> List[List[float]]:
return [
genai.embed_content(
model=self.model_name,
content=text,
task_type="retrieval_document"
)["embedding"]
for text in texts
]
def embed_query(self, text: str) -> List[float]:
return genai.embed_content(
model=self.model_name,
content=text,
task_type="retrieval_query"
)["embedding"]
vectorstore_global = None
if "feedback_log" not in st.session_state:
st.session_state["feedback_log"] = []
def preload_modtran_document():
global vectorstore_global
embeddings = GeminiEmbeddings()
st.session_state.vectorstore = FAISS.load_local("monte_vectorstore", embeddings, allow_dangerous_deserialization=True)
set_global_vectorstore(st.session_state.vectorstore)
st.session_state.chat_ready = True
def convert_pdf_to_xml(pdf_file, xml_path):
os.makedirs("temp", exist_ok=True)
pdf_path = os.path.join("temp", pdf_file.name)
with open(pdf_path, 'wb') as f:
f.write(pdf_file.getbuffer())
subprocess.run(["pdftohtml", "-xml", pdf_path, xml_path], check=True)
return xml_path
def extract_text_from_xml(xml_path, document_name):
tree = etree.parse(xml_path)
text_chunks = []
for page in tree.xpath("//page"):
page_num = int(page.get("number", 0))
texts = [text.text for text in page.xpath('.//text') if text.text]
combined_text = '\n'.join(texts)
text_chunks.append({"text": combined_text, "page": page_num, "document": document_name})
return text_chunks
def extract_text_from_pdf(pdf_file, document_name):
text_chunks = []
with pdfplumber.open(pdf_file) as pdf:
for i, page in enumerate(pdf.pages):
text = page.extract_text()
if text:
text_chunks.append({"text": text, "page": i + 1, "document": document_name})
return text_chunks
def get_uploaded_text(uploaded_files):
raw_text = []
for uploaded_file in uploaded_files:
document_name = uploaded_file.name
if document_name.endswith(".pdf"):
text_chunks = extract_text_from_pdf(uploaded_file, document_name)
raw_text.extend(text_chunks)
elif uploaded_file.name.endswith((".html", ".htm")):
soup = BeautifulSoup(uploaded_file.getvalue(), 'lxml')
raw_text.append({"text": soup.get_text(), "page": None, "document": document_name})
elif uploaded_file.name.endswith((".txt")):
content = uploaded_file.getvalue().decode("utf-8")
raw_text.append({"text": content, "page": None, "document": document_name})
return raw_text
def get_text_chunks(raw_text):
splitter = CharacterTextSplitter(separator='\n', chunk_size=500, chunk_overlap=100)
final_chunks = []
for chunk in raw_text:
for split_text in splitter.split_text(chunk["text"]):
final_chunks.append({"text": split_text, "page": chunk["page"], "document": chunk["document"]})
return final_chunks
def get_vectorstore(text_chunks):
if not text_chunks:
raise ValueError("text_chunks is empty. Cannot initialize FAISS vectorstore.")
embeddings = GeminiEmbeddings()
texts = [chunk["text"] for chunk in text_chunks]
metadatas = [{"page": chunk["page"], "document": chunk["document"]} for chunk in text_chunks]
return FAISS.from_texts(texts, embedding=embeddings, metadatas=metadatas)
def set_global_vectorstore(vectorstore):
global vectorstore_global
vectorstore_global = vectorstore
kw_model = None
reranker = None
def get_kw_model():
global kw_model
if kw_model is None:
# Load sentence transformer with HF token explicitly
model = SentenceTransformer(
'sentence-transformers/all-MiniLM-L6-v2',
use_auth_token=os.environ.get("HF_TOKEN")
)
kw_model = KeyBERT(model=model)
return kw_model
def self_reasoning(query, context):
print("πŸ§ͺ self_reasoning received context of length:", len(context))
llm = GeminiLLM()
reasoning_prompt = f"""
You are an AI assistant that analyzes the context provided to answer the user's query comprehensively and clearly.
Answer in a concise, factual way using the terminology from the context. Avoid extra explanation unless explicitly asked.
YOU MUST mention the document file name (e.g., tools.html, refguide.html) in your answer.
### Example 1:
**Question:** What is the purpose of the Monte GUI?
**Context:**
[From `tools.html`] The Monte GUI provides interfaces for setting up trajectory parameters and viewing output results.
**Answer:** The Monte GUI helps users configure trajectory parameters and visualize results. (From `tools.html`)
### Example 2:
**Question:** How do you perform covariance analysis in Monte?
**Context:**
[From `designEdition.html`] The Monte Design Edition includes support for statistical maneuver and covariance analysis during the design phase.
**Answer:** Monte supports covariance analysis through the Design Edition. (From `designEdition.html`)
### Now answer:
**Question:** {query}
**Context:**
{context}
**Answer:**
"""
try:
result = llm._call(reasoning_prompt)
print("βœ… Gemini returned a result.")
return result
except Exception as e:
print("❌ Error in self_reasoning:", e)
return f"⚠️ Gemini failed: {e}"
def faiss_search_with_keywords(query):
global vectorstore_global
if vectorstore_global is None:
raise ValueError("FAISS vectorstore is not initialized.")
kw_model = get_kw_model()
keywords = kw_model.extract_keywords(query, keyphrase_ngram_range=(1,2), stop_words='english', top_n=5)
refined_query = " ".join([keyword[0] for keyword in keywords])
retriever = vectorstore_global.as_retriever(search_kwargs={"k": 13})
docs = retriever.get_relevant_documents(refined_query)
context = '\n\n'.join([f"[From `{doc.metadata.get('document', 'unknown.html')}`] {doc.page_content}" for doc in docs])
return self_reasoning(query, context)
def get_reranker():
global reranker
if reranker is None:
reranker = CrossEncoder('cross-encoder/ms-marco-MiniLM-L-6-v2')
return reranker
def faiss_search_with_reasoning(query):
global vectorstore_global
if vectorstore_global is None:
raise ValueError("FAISS vectorstore is not initialized.")
reranker = get_reranker()
retriever = vectorstore_global.as_retriever(search_kwargs={"k": 13})
docs = retriever.get_relevant_documents(query)
pairs = [(query, doc.page_content) for doc in docs]
scores = reranker.predict(pairs)
reranked_docs = sorted(zip(scores, docs), key=lambda x: x[0], reverse=True)
top_docs = [doc for _, doc in reranked_docs[:5]]
context = '\n\n'.join([f"[From `{doc.metadata.get('document', 'unknown.html')}`] {doc.page_content.strip()}" for doc in top_docs])
return self_reasoning(query, context)
faiss_keyword_tool = Tool(
name="FAISS Keyword Search",
func=faiss_search_with_keywords,
description="Searches FAISS with a keyword-based approach to retrieve context."
)
faiss_reasoning_tool = Tool(
name="FAISS Reasoning Search",
func=faiss_search_with_reasoning,
description="Searches FAISS with detailed reasoning to retrieve context."
)
def initialize_chatbot_agent():
llm = GeminiLLM()
memory = ConversationBufferMemory(memory_key="chat_history", return_messages=True)
tools = [faiss_keyword_tool, faiss_reasoning_tool]
agent = initialize_agent(
tools=tools,
llm=llm,
agent=AgentType.ZERO_SHOT_REACT_DESCRIPTION,
memory=memory,
verbose=False,
handle_parsing_errors=True
)
return agent
def handle_user_query(query):
try:
global vectorstore_global
if vectorstore_global is None:
raise ValueError("Vectorstore is not initialized.")
print("πŸ” Starting handle_user_query with:", query)
if "how" in query.lower():
print("🧠 Routing to: faiss_search_with_reasoning")
context = faiss_search_with_reasoning(query)
else:
print("🧠 Routing to: faiss_search_with_keywords")
context = faiss_search_with_keywords(query)
print("πŸ“š Context length:", len(context))
print("✍️ Calling self_reasoning...")
answer = self_reasoning(query, context)
print("βœ… Answer generated.")
return answer
except Exception as e:
print("❌ Error in handle_user_query:", e)
return f"⚠️ Error: {e}"
def save_feedback_to_huggingface():
try:
if not st.session_state.feedback_log:
print("⚠️ No feedbacks collected yet.")
return
feedback_df = pd.DataFrame(st.session_state.feedback_log)
now = datetime.now().strftime("%Y%m%d_%H%M%S")
filename = f"feedback_{now}.csv"
with tempfile.TemporaryDirectory() as tmpdir:
filepath = os.path.join(tmpdir, filename)
feedback_df.to_csv(filepath, index=False)
token = os.environ.get("HF_TOKEN") or os.environ.get("HUGGINGFACEHUB_API_TOKEN")
if not token:
raise ValueError("❌ Hugging Face token not found!")
print(f"πŸ“€ Attempting upload to repo: ZarinT/chatbot-feedback as {filename}")
print("πŸ“ Feedback data:", feedback_df)
api = HfApi(token=token)
api.upload_file(
path_or_fileobj=filepath,
path_in_repo=filename,
repo_id="ZarinT/chatbot-feedback",
repo_type="dataset"
)
print("βœ… Feedback uploaded successfully.")
st.session_state.feedback_log.clear()
except Exception as e:
print("❌ Feedback upload failed:", e)
def clear_user_input():
st.session_state["user_input"] = ""
from datetime import datetime
def clear_user_input():
st.session_state.user_input = ""
from datetime import datetime
def clear_user_input():
st.session_state.user_input = ""
from datetime import datetime
def clear_user_input():
st.session_state.user_input = ""
def main():
load_environment()
# Initialize session state
if "chat_ready" not in st.session_state:
st.session_state.chat_ready = False
if "chat_history" not in st.session_state:
st.session_state.chat_history = []
if "vectorstore" not in st.session_state:
st.session_state.vectorstore = None
if "feedback_log" not in st.session_state:
st.session_state.feedback_log = []
if "feedback_submitted" not in st.session_state:
st.session_state.feedback_submitted = False
if "last_answered_question" not in st.session_state:
st.session_state.last_answered_question = ""
st.markdown("""
<h2>Chat with MONTE Documents πŸ“„</h2>
""", unsafe_allow_html=True)
# Inject custom CSS for chat bubbles
st.markdown("""
<style>
.chat-container {
display: flex;
flex-direction: column;
gap: 0.75rem;
}
.bot-bubble {
align-self: flex-start;
background-color: #e8f4fd;
color: #111;
padding: 1rem;
border-radius: 0.5rem 1rem 1rem 1rem;
max-width: 80%;
}
.user-bubble {
align-self: flex-end;
background-color: rgba(0, 0, 0, 0.85);
color: white;
padding: 1rem;
border-radius: 1rem 0.5rem 1rem 1rem;
max-width: 80%;
text-align: right;
}
.rating-line {
font-weight: bold;
color: #ffaa00;
font-size: 0.95rem;
max-width: 80%;
align-self: center;
text-align: center;
}
.feedback-counter {
font-style: italic;
color: #666;
font-size: 0.9rem;
margin: 1rem 0;
}
</style>
""", unsafe_allow_html=True)
# Load vectorstore and chatbot agent
if not st.session_state.chat_ready:
with st.spinner("Loading Monte documents..."):
preload_modtran_document()
st.session_state.agent = initialize_chatbot_agent()
st.session_state.chat_ready = True
st.success("Monte Docuemnts loaded successfully!")
# Render all previous Q&A in chat format
st.markdown('<div class="chat-container">', unsafe_allow_html=True)
for i, exchange in enumerate(st.session_state.chat_history):
# User's question (right side)
st.markdown(f'<div class="user-bubble"><strong>You:</strong> {exchange["user"]}</div>', unsafe_allow_html=True)
# MODTRAN Bot's answer (left side)
st.markdown(f'<div class="bot-bubble"><strong>MODTRAN Bot:</strong> {exchange["bot"]}</div>', unsafe_allow_html=True)
# If already rated
if "rating" in exchange:
st.markdown(f'<div class="rating-line">⭐️ You rated this: {exchange["rating"]}</div>', unsafe_allow_html=True)
# If not rated yet and it's the last message, show form
elif i == len(st.session_state.chat_history) - 1:
with st.form(key=f"feedback_form_{i}"):
rating = st.radio(
"Rate this response:",
options=["Not helpful", "Somewhat helpful", "Neutral", "Helpful", "Very helpful"],
key=f"rating_{i}",
horizontal=True
)
submitted = st.form_submit_button("Submit Rating")
if submitted:
st.session_state.chat_history[i]["rating"] = rating
st.session_state.feedback_log.append({
"question": exchange["user"],
"response": exchange["bot"],
"rating": rating,
"timestamp": datetime.now().isoformat()
})
if len(st.session_state.feedback_log) >= 2:
print("πŸ“¦ Upload threshold reached β€” saving feedback to Hugging Face.")
save_feedback_to_huggingface()
st.session_state.feedback_submitted = True
st.rerun()
st.markdown('</div>', unsafe_allow_html=True)
# Show current feedback progress
#st.markdown(f'<div class="feedback-counter">πŸ“„ Feedbacks collected: <strong>{len(st.session_state.feedback_log)} / 5</strong></div>', unsafe_allow_html=True)
# Input for next question
user_question = st.chat_input("Ask your next question:")
if user_question and user_question != st.session_state.last_answered_question:
with st.spinner("Generating answer..."):
try:
set_global_vectorstore(st.session_state.vectorstore)
response = handle_user_query(user_question)
except Exception as e:
response = f"⚠️ Something went wrong: {e}"
# Save new Q&A
st.session_state.chat_history.append({
"user": user_question,
"bot": response
})
st.session_state.last_answered_question = user_question
st.rerun()
if __name__ == "__main__":
load_environment()
main()