modules/register.py

from models.whisper import model
import modules.audio as audio
import os
from modules.r2 import upload_to_s3,get_url
import json
import os
import redis

redis_host = os.getenv('REDIS_HOST', 'localhost') 
redis_pass = os.getenv('REDIS_PASS', 'localhost') 
redis_port = os.getenv('REDIS_PORT', 6379)

cache = redis.Redis(host=redis_host, port=redis_port, password=redis_pass)

isOnline = os.environ.get("IS_ONLINE")
isOnline = "True" if isOnline == "True" else "False"

def prepare_audio(audio_path, key):
    """Prepara o áudio"""
    audio_duration = audio.get_audio_duration(audio_path)
    audio_duration = round(audio_duration, 2)
    cuted = False
    
    if audio_duration > 15:
        audio_path = audio.cut_audio(audio_path, start_time="0", end_time="12", output_path=f"{key}_cut.wav")
        cuted = True
    
    if audio_duration < 8:
        raise ValueError("audio_too_short")
    
    audio_path = audio.add_silence_to_audio(audio_path, start_silence=0.5, end_silence=0.8, output_path=f"{key}.wav")
    if cuted:
        os.remove(f"{key}_cut.wav")
    return audio_path

def get_audio_transcription(key):
    """Obtém a transcrição do áudio usando o Redis"""
    if(isOnline == "True"):
        data = cache.get(f"mimic:audio:translation:{key}")
        return data
    
    with open(f"{key}.txt", "r") as f:
        transcription = f.read()
    return transcription

def get_audio_path(key):
    """Obtém o caminho do áudio usando o Redis"""
    if(isOnline == "True"):
        data = cache.get(f"mimic:audio:url:{key}")
        if(data):
            return data
        
        new_url = get_url(f"{key}.wav")
                         
        cache.set(f"mimic:audio:url:{key}", new_url)
        cache.expire(f"mimic:audio:url:{key}", 600000)
        
        return new_url

    return f"{key}.wav"

def get_audio(key):
    """Obtém o áudio usando o Redis"""
    transcription = get_audio_transcription(key)
    audio_path = get_audio_path(key)
    return {
        "transcription": transcription,
        "audio_path": audio_path,
        "isOnline": isOnline
    }
    
def process_audio(audio_path, key):
    """Processa o áudio usando o Whisper"""
    if(isOnline == "True"):
        audio_exists = cache.exists(f"mimic:audio_exists:{key}")
        
        if audio_exists:
            return get_audio(key)
        
        audio_path = prepare_audio(audio_path, key)
        transcription, info = model.transcribe(audio_path, beam_size=5)
        content = ""
        for segment in transcription:
            content = f"{content} {segment.text}"
        url = upload_to_s3(audio_path, f"{key}.wav", "wav")
        
        cache.set(f"mimic:audio:translation:{key}", content)

        cache.set(f"mimic:audio:url:{key}", url)
        cache.expire(f"mimic:audio:url:{key}", 600000)
        
        cache.set(f"mimic:audio_exists:{key}", "true")
        return {"transcription": content, "audio_path": url, "isOnline": isOnline}

    audio_exists = os.path.exists(f"{key}.txt")
    if audio_exists:
        return get_audio(key)
    
    if os.path.exists(f"{key}.wav"):
        os.remove(f"{key}.wav")
    
    audio_path = prepare_audio(audio_path, key)
    transcription, info = model.transcribe(audio_path, beam_size=5)
    content = ""
    for segment in transcription:
        content = f"{content} {segment.text}"
        
    with open(f"{key}.txt", "w") as f:
        f.write(content)
        
    return {"transcription": content, "audio_path": audio_path, "isOnline": isOnline}