main.py

from flask import Flask, request
from werkzeug.utils import secure_filename
import os
import json
from dotenv import load_dotenv
from openai import OpenAI

from transcribe import transcribe
from process_transcribe import process_transcribe

load_dotenv()

app = Flask(__name__,
            static_url_path='', 
            static_folder='./webapp',
            template_folder='./webapp')

if (os.getenv("OPENAI_API_KEY") is None):
  print("OPENAI_API_KEY is not set")
  exit(1) 

client = OpenAI()

# only one file at a time
file_processing = False

@app.route('/')
def index():
  return app.send_static_file('index.html')

@app.route('/api/transcribe', methods=['POST'])
def process():
  global file_processing
  if file_processing:
    return "File already processing", 400
  print(request.files)
  if 'audio' in request.files:
    file_processing = True

    file = request.files['audio']
    filename = secure_filename(file.filename)

    # save file to uploads folder
    num_files = len(os.listdir("uploads"))
    if not os.path.exists(str(num_files)):
      os.makedirs(os.path.join("uploads", str(num_files)))
    extension = filename.split(".")[-1]
    filename = os.path.join("uploads", str(num_files), "audio." + extension)
    file.save(filename)
    
    # convert to wav if needed
    if extension != "wav":
      original_filename = filename
      filename = filename.replace(extension, "wav")
      os.system("ffmpeg -i " + original_filename + " -acodec pcm_s16le -ac 1 -ar 48000 " + filename)

    # transcribe
    print("Transcribing", filename)
    data = transcribe(filename)
    data_json = json.dumps(data, ensure_ascii=False)

    # save result to json file
    f = open(os.path.join("uploads", str(num_files), "result.json"), "w")
    f.write(data_json)
    f.close()

    file_processing = False

    return process_transcribe(data)
  else:
    return "No file uploaded", 400 

@app.route('/api/summarize')
def summarize():
  # get transcription from last file in uploads folder
  num_files = len(os.listdir("uploads"))
  f = open(os.path.join("uploads", str(num_files - 1), "result.json"), "r")
  data = json.loads(f.read())
  f.close()

  transcription = process_transcribe(data)

  prompt = "Tu es un assistant virtuel qui aide les gens à résumer des conversations. Tu dois résumer la conversation. N'utilise pas le terme \"Personne X\" pour désigner les interlocuteurs. Au début du résumé, tu dois indiquer de quel type de conversation il s'agit (interview, conversation entre amis, etc.) et tu dois aussi indiquer le nombre de personnes qui participent à la conversation. Marque ces informations là de cette façon : \"Résumé de l'interview/conversation/etc. (X personnes) :\n\n\""
  conversation = "\n".join([("Personne " + str(x["speaker"]) + " : " + x["text"]) for x in transcription])

  chat_completion = client.chat.completions.create(
    messages=[
      {
        "role": "system",
        "content": prompt
      },
      {
        "role": "user",
        "content": conversation
      }
    ],
    model="gpt-3.5-turbo-1106",
  )

  summary = chat_completion.choices[0].message.content

  return { "summary": summary }

@app.route('/api/chat', methods=['POST'])
def chat():
  text = request.json["text"] # user input

  # get transcription from last file in uploads folder
  num_files = len(os.listdir("uploads"))
  f = open(os.path.join("uploads", str(num_files - 1), "result.json"), "r")
  data = json.loads(f.read())
  f.close()

  transcription = process_transcribe(data)

  prompt = "Tu es un assistant virtuel qui aide les gens à répondre à des questions sur une conversation. N'utilise pas le terme \"Personne X\" pour désigner les interlocuteurs."

  conversation = "\n".join([("Personne " + str(x["speaker"]) + " : " + x["text"]) for x in transcription])

  chat_completion = client.chat.completions.create(
    messages=[
      {
        "role": "system",
        "content": prompt
      },
      {
        "role": "user",
        "content": "QUESTION : " + text + "\nCONVERSATION :\n" + conversation
      }
    ],
    model="gpt-3.5-turbo-1106",
  )

  return { "response": chat_completion.choices[0].message.content }


if __name__ == '__main__':
  app.run(debug=True, host='0.0.0.0', port=5000)