import numpy as np

BASE_FREQ = 600      
FREQ_STEP = 25       
TONE_DURATION = 0.1  
SAMPLE_RATE = 44100

FREQ_QUERY = 1200
FREQ_ALERT = 1400
FREQ_RESP_YES = 1600
FREQ_RESP_NO = 1800

def generate_service_tone(frequency, duration=0.4):
    t = np.linspace(0, duration, int(SAMPLE_RATE * duration), False)
    tone = np.sin(frequency * t * 2 * np.pi) * 0.7
    envelope = np.ones_like(tone)
    fade_len = int(SAMPLE_RATE * 0.05)
    envelope[:fade_len] = np.linspace(0, 1, fade_len)
    envelope[-fade_len:] = np.linspace(1, 0, fade_len)
    return (tone * envelope).astype(np.float32).tobytes()

def text_to_audio(text):
    audio_signals = []
    t_start = np.linspace(0, 0.3, int(SAMPLE_RATE * 0.3), False)
    audio_signals.append(np.sin(1000 * t_start * 2 * np.pi))
    for char in text:
        freq = BASE_FREQ + ord(char) * FREQ_STEP
        t = np.linspace(0, TONE_DURATION, int(SAMPLE_RATE * TONE_DURATION), False)
        audio_signals.append(np.sin(freq * t * 2 * np.pi))
    full_audio = np.concatenate(audio_signals).astype(np.float32)
    if np.max(np.abs(full_audio)) > 0:
        full_audio = full_audio / np.max(np.abs(full_audio))
    return full_audio.tobytes()

def detect_service_tone(audio_data):
    try:
        window_data = np.hanning(len(audio_data))
        fft_data = np.abs(np.fft.rfft(audio_data * window_data))
        frequencies = np.fft.rfftfreq(len(audio_data), d=1/SAMPLE_RATE)
        detected_freq = frequencies[np.argmax(fft_data)]
        for target in [FREQ_QUERY, FREQ_ALERT, FREQ_RESP_YES, FREQ_RESP_NO]:
            if abs(detected_freq - target) <= 15:
                return target
        return None
    except:
        return None

def fast_decode(audio_data):
    try:
        samples_per_tone = int(SAMPLE_RATE * TONE_DURATION)
        current_sample = int(SAMPLE_RATE * 0.3)
        text = ""
        while current_sample < len(audio_data):
            chunk = audio_data[current_sample : current_sample + samples_per_tone]
            if len(chunk) < samples_per_tone: 
                break
            window_data = np.hanning(len(chunk))
            fft_data = np.abs(np.fft.rfft(chunk * window_data))
            frequencies = np.fft.rfftfreq(len(chunk), d=1/SAMPLE_RATE)
            detected_freq = frequencies[np.argmax(fft_data)]
            ascii_code = int(round((detected_freq - BASE_FREQ) / FREQ_STEP))
            if 0 <= ascii_code < 65535:
                text += chr(ascii_code)
            current_sample += samples_per_tone
        return text
    except:
        return ""