from fastapi import FastAPI, Query, HTTPException
from fastapi.responses import StreamingResponse
from TTS.api import TTS
import os
from io import BytesIO
from typing import Generator

app = FastAPI()
import os

# By using XTTS you agree to CPML license https://coqui.ai/cpml
os.environ["COQUI_TOS_AGREED"] = "1"

# Initialize the TTS model
tts = TTS("tts_models/multilingual/multi-dataset/xtts_v2", gpu=False)  # Set gpu=True if you have GPU support

# Predefined path to the sample voice clone
FIXED_SPEAKER_WAV = "Bible Verses About Community.wav"

# Function to split text into chunks
def split_text(text: str, words_per_chunk: int = 20):
    words = text.split()
    return [' '.join(words[i:i + words_per_chunk]) for i in range(0, len(words), words_per_chunk)]



@app.post("/generate-audio/")
async def generate_audio(
    text: str = Query(..., description="The input text to convert to speech."),
    language: str = Query("en", description="Language code for TTS (e.g., 'en' for English).")
):
    if not os.path.exists(FIXED_SPEAKER_WAV):
        raise HTTPException(status_code=400, detail="Fixed speaker WAV file not found.")

    # StreamingResponse to stream audio chunks
    def audio_stream():
            if tts.is_multi_lingual and not language:
                raise ValueError("Language must be specified for multi-lingual models.")
        
            text_chunks = split_text(text, 20)
            
            for idx, chunk in enumerate(text_chunks):
                # Generate audio for each chunk and yield as bytes
                output_file = f"out_{idx}.wav"
                tts.tts_to_file(
                    text=chunk,
                    file_path=output_file,
                    speaker_wav=FIXED_SPEAKER_WAV,
                    language=language
                )
                print(output_file)
                # Read the file content and yield as binary
                with open(output_file, "rb") as audio_file:
                    yield audio_file.read()
                # Optionally delete the file after streaming
                os.remove(output_file)
        
    return StreamingResponse(audio_stream(), media_type="audio/wav")