|
from flask import Flask, request |
|
import os |
|
from langchain.vectorstores import Chroma |
|
from langchain.document_loaders import PyPDFLoader |
|
from langchain.embeddings import HuggingFaceEmbeddings |
|
from langchain.text_splitter import RecursiveCharacterTextSplitter |
|
import requests |
|
from twilio.rest import Client |
|
|
|
|
|
|
|
app = Flask(__name__) |
|
|
|
|
|
CHROMA_PATH = '/code/chroma_db' |
|
if not os.path.exists(CHROMA_PATH): |
|
os.makedirs(CHROMA_PATH) |
|
|
|
from ai71 import AI71 |
|
|
|
def generate_response(query, chat_history): |
|
response = '' |
|
try: |
|
ai71_client = AI71(api_key=AI71_API_KEY) |
|
chat_completion = ai71_client.chat.completions.create( |
|
model="tiiuae/falcon-180b-chat", |
|
messages=[ |
|
{"role": "system", "content": "You are the best agricultural assistant. Remember to give a response in not more than 2 sentences."}, |
|
{"role": "user", "content": f"Answer the query based on history {chat_history}: {query}"} |
|
], |
|
stream=True |
|
) |
|
|
|
for chunk in chat_completion: |
|
if chunk.choices[0].delta.content: |
|
response += chunk.choices[0].delta.content |
|
|
|
|
|
response = response.replace("###", '').replace('\nUser:', '') |
|
|
|
except Exception as e: |
|
print(f"Error generating response: {e}") |
|
response = "An error occurred while generating the response." |
|
|
|
return response |
|
|
|
|
|
|
|
def initialize_chroma(): |
|
try: |
|
embedding_function = HuggingFaceEmbeddings() |
|
db = Chroma(persist_directory=CHROMA_PATH, embedding_function=embedding_function) |
|
|
|
|
|
db.similarity_search_with_score("test query", k=1) |
|
print("Chroma initialized successfully.") |
|
except Exception as e: |
|
print(f"Error initializing Chroma: {e}") |
|
|
|
initialize_chroma() |
|
|
|
|
|
AI71_API_KEY = os.environ.get('AI71_API_KEY') |
|
account_sid = os.environ.get('TWILIO_ACCOUNT_SID') |
|
auth_token = os.environ.get('TWILIO_AUTH_TOKEN') |
|
client = Client(account_sid, auth_token) |
|
from_whatsapp_number = 'whatsapp:+14155238886' |
|
|
|
|
|
def download_file(url, ext): |
|
local_filename = f'/code/uploaded_file{ext}' |
|
with requests.get(url, stream=True) as r: |
|
with open(local_filename, 'wb') as f: |
|
for chunk in r.iter_content(chunk_size=8192): |
|
f.write(chunk) |
|
return local_filename |
|
|
|
|
|
|
|
|
|
import fitz |
|
|
|
def extract_text_from_pdf(pdf_filepath): |
|
text = '' |
|
try: |
|
pdf_document = fitz.open(pdf_filepath) |
|
for page_num in range(len(pdf_document)): |
|
page = pdf_document.load_page(page_num) |
|
text += page.get_text() |
|
pdf_document.close() |
|
except Exception as e: |
|
print(f"Error extracting text from PDF: {e}") |
|
return None |
|
return text |
|
|
|
def query_rag(query_text: str, chat_history): |
|
try: |
|
embedding_function = HuggingFaceEmbeddings() |
|
db = Chroma(persist_directory=CHROMA_PATH, embedding_function=embedding_function) |
|
|
|
results = db.similarity_search_with_score(query_text, k=5) |
|
|
|
if not results: |
|
return "Sorry, I couldn't find any relevant information." |
|
|
|
context_text = "\n\n---\n\n".join([doc.page_content for doc, _score in results]) |
|
|
|
prompt = f"Context:\n{context_text}\n\nQuestion:\n{query_text}" |
|
response = generate_response(prompt, chat_history) |
|
|
|
return response |
|
except Exception as e: |
|
print(f"Error querying RAG system: {e}") |
|
return "An error occurred while querying the RAG system." |
|
|
|
|
|
def save_pdf_and_update_database(pdf_filepath): |
|
try: |
|
text = extract_text_from_pdf(pdf_filepath) |
|
if not text: |
|
print("Error extracting text from PDF.") |
|
return |
|
|
|
text_splitter = RecursiveCharacterTextSplitter( |
|
chunk_size=800, |
|
chunk_overlap=80, |
|
length_function=len, |
|
is_separator_regex=False, |
|
) |
|
chunks = text_splitter.split_text(text) |
|
|
|
embedding_function = HuggingFaceEmbeddings() |
|
db = Chroma(persist_directory=CHROMA_PATH, embedding_function=embedding_function) |
|
|
|
db.add_documents(chunks) |
|
db.persist() |
|
print("PDF processed and data updated in Chroma.") |
|
except Exception as e: |
|
print(f"Error processing PDF: {e}") |
|
|
|
|
|
@app.route('/whatsapp', methods=['POST']) |
|
def whatsapp_webhook(): |
|
incoming_msg = request.values.get('Body', '').lower() |
|
sender = request.values.get('From') |
|
num_media = int(request.values.get('NumMedia', 0)) |
|
|
|
chat_history = [] |
|
|
|
if num_media > 0: |
|
media_url = request.values.get('MediaUrl0') |
|
content_type = request.values.get('MediaContentType0') |
|
|
|
if content_type == 'application/pdf': |
|
filepath = download_file(media_url, ".pdf") |
|
save_pdf_and_update_database(filepath) |
|
response_text = "PDF has been processed. You can now ask questions related to its content." |
|
else: |
|
response_text = "Unsupported file type. Please upload a PDF document." |
|
else: |
|
|
|
response_text = query_rag(incoming_msg, chat_history) |
|
|
|
|
|
send_message(sender, response_text) |
|
return '', 204 |
|
|
|
|
|
def send_message(to, body): |
|
try: |
|
message = client.messages.create( |
|
from_=from_whatsapp_number, |
|
body=body, |
|
to=to |
|
) |
|
print(f"Message sent with SID: {message.sid}") |
|
except Exception as e: |
|
print(f"Error sending message: {e}") |
|
def send_initial_message(to_number): |
|
send_message( |
|
f'whatsapp:{to_number}', |
|
'Welcome to the Agri AI Chatbot! How can I assist you today? You can send an image with "pest" or "disease" to classify it.' |
|
) |
|
|
|
if __name__ == "__main__": |
|
send_initial_message('919080522395') |
|
send_initial_message('916382792828') |
|
app.run(host='0.0.0.0', port=7860) |