|
import requests |
|
import json |
|
import gradio as gr |
|
import fitz |
|
import logging |
|
import base64 |
|
from flask import Flask, request, jsonify |
|
import io |
|
import os |
|
import base64 |
|
import io |
|
from PyPDF2 import PdfReader |
|
|
|
|
|
logging.basicConfig(level=logging.ERROR) |
|
logger = logging.getLogger(__name__) |
|
|
|
app = Flask(__name__) |
|
|
|
|
|
OPENROUTER_API_KEY = "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" |
|
YOUR_SITE_URL = "votre-site.com" |
|
YOUR_APP_NAME = "MonChatbot" |
|
|
|
def extract_text_from_pdf(pdf_file): |
|
doc = fitz.open(pdf_file) |
|
text = "" |
|
for page in doc: |
|
text += page.get_text() |
|
return text |
|
|
|
def encode_image(image_path): |
|
with open(image_path, "rb") as image_file: |
|
encoded_string = base64.b64encode(image_file.read()).decode('utf-8') |
|
return encoded_string |
|
|
|
def chatbot_response(message, history, pdf_text=None, image_path=None): |
|
messages = [{"role": "system", "content": "Vous êtes un assistant IA utile et amical, capable d'analyser des images et du texte."}] |
|
|
|
if pdf_text: |
|
messages.append({"role": "system", "content": f"Le contenu du PDF est : {pdf_text}"}) |
|
|
|
for human, assistant in history: |
|
messages.append({"role": "user", "content": human}) |
|
if assistant is not None: |
|
messages.append({"role": "assistant", "content": assistant}) |
|
|
|
message_content = message |
|
if image_path: |
|
encoded_image = encode_image(image_path) |
|
message_content = [ |
|
{"type": "text", "text": message}, |
|
{"type": "image_url", "image_url": {"url": f"data:image/jpeg;base64,{encoded_image}"}} |
|
] |
|
|
|
messages.append({"role": "user", "content": message_content}) |
|
|
|
try: |
|
response = requests.post( |
|
url="https://openrouter.ai/api/v1/chat/completions", |
|
headers={ |
|
"Authorization": f"Bearer {OPENROUTER_API_KEY}", |
|
"HTTP-Referer": f"{YOUR_SITE_URL}", |
|
"X-Title": f"{YOUR_APP_NAME}", |
|
"Content-Type": "application/json" |
|
}, |
|
data=json.dumps({ |
|
"model": "mistralai/pixtral-12b:free", |
|
"messages": messages |
|
}) |
|
) |
|
if response.status_code == 200: |
|
data = response.json() |
|
return data['choices'][0]['message']['content'] |
|
else: |
|
return f"Erreur {response.status_code}: {response.text}" |
|
except Exception as e: |
|
logger.error(f"Erreur lors de l'appel API: {str(e)}") |
|
return f"Erreur: {str(e)}" |
|
|
|
@app.route('/api/chatbot', methods=['POST']) |
|
def api_chatbot(): |
|
try: |
|
|
|
message = request.json.get('message') |
|
pdf_base64 = request.json.get('pdf_content') |
|
|
|
if not pdf_base64: |
|
return jsonify({'error': 'Aucun contenu PDF reçu.'}), 400 |
|
|
|
|
|
pdf_data = base64.b64decode(pdf_base64) |
|
pdf_file = io.BytesIO(pdf_data) |
|
|
|
|
|
pdf_reader = PdfReader(pdf_file) |
|
pdf_text = "" |
|
for page in pdf_reader.pages: |
|
pdf_text += page.extract_text() |
|
|
|
if not pdf_text: |
|
return jsonify({'error': 'Impossible d\'extraire le texte du PDF.'}), 500 |
|
|
|
|
|
response = chatbot_response(message, history=[], pdf_text=pdf_text) |
|
|
|
return jsonify({'response': response}) |
|
except Exception as e: |
|
return jsonify({'error': str(e)}), 500 |
|
|
|
|
|
|
|
def user(user_message, history, pdf_text, image): |
|
|
|
return "", history + [[user_message, None]], pdf_text, image |
|
|
|
def bot(history, pdf_text, image): |
|
if history: |
|
|
|
bot_message = chatbot_response(history[-1][0], history[:-1], pdf_text, image) |
|
history[-1][1] = bot_message |
|
return history |
|
return [] |
|
|
|
def clear_chat(): |
|
return [], None, None |
|
|
|
|
|
with gr.Blocks(theme=gr.themes.Soft()) as demo: |
|
chatbot = gr.Chatbot(label="Historique de la conversation",type="messages") |
|
msg = gr.Textbox(label="Votre message", placeholder="Tapez votre message ici...") |
|
pdf_upload = gr.File(label="Téléchargez un fichier PDF", file_types=[".pdf"]) |
|
image_upload = gr.Image(type="filepath", label="Téléchargez une image") |
|
clear = gr.Button("Effacer la conversation") |
|
pdf_text = gr.State() |
|
|
|
|
|
pdf_upload.change(lambda file: extract_text_from_pdf(file), pdf_upload, pdf_text) |
|
|
|
|
|
msg.submit(user, [msg, chatbot, pdf_text, image_upload], [msg, chatbot, pdf_text, image_upload], queue=False).then( |
|
bot, [chatbot, pdf_text, image_upload], chatbot |
|
) |
|
|
|
|
|
clear.click(clear_chat, None, [chatbot, pdf_text, image_upload], queue=False) |
|
|
|
demo.launch(server_name="0.0.0.0", server_port=int(os.environ.get("PORT", 5000))) |
|
|
|
|
|
|
|
if __name__ == "__main__": |
|
port = int(os.environ.get("PORT", 5000)) |
|
app.run(host="0.0.0.0", port=port) |