app.py

import numpy as np
import pandas as pd
import logging
from sentence_transformers import SentenceTransformer

from sklearn.metrics import DistanceMetric

import warnings
warnings.filterwarnings("ignore")
import gradio as gr
# from transformers.utils.hub import move_cache
#
# move_cache()
from telegram import Bot

logging.basicConfig(level=logging.INFO)
logger = logging.getLogger(__name__)

token = '7370905765:AAHvmlw68cW7RxWzsJE1yxTzgf3xQFiokDo'
chat_id = '7431171535'
pd.set_option('display.max_colwidth', None)
df_resume = pd.read_csv('QA.csv',encoding='latin-1')
# df_resume['role'][df_resume['role'].iloc[-1] == df_resume['role']] = "Other" # relabel random role as "other"
print(df_resume.head())
model = SentenceTransformer("all-MiniLM-L6-v2")
# file_name = 'questions.txt'

def send_telegram_message(token, chat_id, message):
    try:
        bot = Bot(token=token)
        bot.send_message(chat_id=chat_id, text=message)
        logger.info("Message sent successfully.")
    except TelegramError as e:
        logger.error(f"Failed to send message: {e}")


def savevectorstore():
    # store embed vectors
    embedding_arr = model.encode(df_resume['question'])
    print(embedding_arr.shape)
    np.save('embeddings.npy', embedding_arr)
# savevectorstore()
# load embed vectors
def rag_chain(question,name):
    embedding_arr = np.load('embeddings.npy')
    # print(embedding_arr.shape)

    # pca = PCA(n_components=2).fit(embedding_arr)
    # print(pca.explained_variance_ratio_)

    query = question

    query_embedding = model.encode(query)

    dist = DistanceMetric.get_metric('euclidean') # other distances: manhattan, chebyshev

    # compute pair wise distances between query embedding and all resume embeddings
    dist_arr = dist.pairwise(embedding_arr, query_embedding.reshape(1, -1)).flatten()
    # sort results
    idist_arr_sorted = np.argsort(dist_arr)

    # print(df_resume['name'].iloc[idist_arr_sorted[:1]].to_string(index=False))
    # print(df_resume['name'].iloc[idist_arr_sorted[:1]])

    que = f"**Most relevant question.**<br><br>{str(df_resume['question'].iloc[idist_arr_sorted[:1]].to_string(index=False)).replace('||','<br>')}"
    # que = df_resume['question'].iloc[idist_arr_sorted[:1]]
    ans = f"**Answer ideas.**<br><br>{str(df_resume['answer'].iloc[idist_arr_sorted[:1]].to_string(index=False)).replace('||','<br>')}"
    # ans = df_resume['answer'].iloc[idist_arr_sorted[:1]]
    # profit = df_resume['profit'].iloc[idist_arr_sorted[:1]].to_string(index=False)
    # product = df_resume['product'].iloc[idist_arr_sorted[:1]].to_string(index=False)
#     que1 = que
#     ans1 = ans
#     print(que1)
#     print(ans1)
#     que2= f'''Most relevant question
#
#
# {que1}'''
#     ans2= f'''Answer ideas
#
#
# {ans1}'''
#     print(que2)
#     print(ans2)
    # with open('questions.txt', 'a',encoding='utf-8') as file:
    #     file.write(f"\n\n{question}\n\n{country}\n\n{whatsapp}\n\n{name}||{sales}||{profit}||{product}")
    message = f"{name}\n\n{question}\n\n{ans}"
    if not token or not chat_id:
        logger.error("Please set the TELEGRAM_BOT_TOKEN and TELEGRAM_CHAT_ID environment variables.")
    else:
        send_telegram_message(token, chat_id, message)
    return que,ans
# rag_chain('I am very hungry.')

desc = "This is an awesome ML App.  I'm really excited to show you"
long_desc = "如果我没有回答你的问题，把问题发给Eric吧。"
search_interface = gr.Interface(
    fn=rag_chain,
    inputs=[gr.Textbox(label="Question"),gr.Textbox(label="Name")],
    outputs=[gr.Markdown(label="Most relevant question"),gr.Markdown(label="Answer ideas")],
    title="Ask Eric",
    description="Hi,我是数字分身，欢迎提问！",
    # theme=gr.themes.Glass
    article=long_desc
)

search_interface.launch(share=True,debug=True)