import os
import subprocess
import multiprocessing
import atexit

audio_device = os.getenv("WHISPER_AUDIO_DEVICE", "-1")
whisper_model = os.getenv("WHISPER_MODEL_PATH", "models/ggml-tiny.bin")
command = ["whisper-cpp-stream", "-kc", "-m", whisper_model, "-c", audio_device, "-t", "4"]

filter_strings = ["", "*", "\r", "\n","\t", "(inaudible)", "[BLANK_AUDIO]", "[Start speaking]", "(gunshot)", "(wind howling)", "[Music]", "(footsteps)"]  # Example strings to filter out

class SharedString:
    def __init__(self):
        manager = multiprocessing.Manager()
        self.namespace = manager.Namespace()
        self.namespace.value = ""

    def get_value(self):
        with multiprocessing.Lock():
            return self.namespace.value

    def set_value(self, new_value):
        with multiprocessing.Lock():
            self.namespace.value = new_value

    def append(self, append_value):
        with multiprocessing.Lock():
            self.namespace.value += append_value

buffer = SharedString()
process = None
process_thread = None

def read_output(proc, buffer):
    while True:
        output = proc.stdout.readline()
        if output == b"" and proc.poll() is not None:
            break
        if output:
            # print(output.decode("utf-8"))
            buffer.append(output.decode("utf-8"))

def start():
    global process, process_thread
    process = subprocess.Popen(command, stdout=subprocess.PIPE, stderr=subprocess.PIPE)
    process_thread = multiprocessing.Process(target=read_output, args=(process, buffer))
    process_thread.start()

    # Register cleanup function to be called when script exits
    atexit.register(stop)

def stop():
    global process, process_thread
    if process:
        process.terminate()
        process_thread.join()
        process = None
        process_thread = None

def filter_buffer(data):
    for f_str in filter_strings:
        data = data.replace(f_str, "")
    return data.strip()

def get_buffer():
    data = buffer.get_value()
    buffer.set_value("")
    return filter_buffer(data)
    #return data

def main():
    start()

    try:
        while process.poll() is None:
            data = get_buffer()
            if data:
                print(data)
    except KeyboardInterrupt:
        stop()

if __name__ == "__main__":
    main()