// ====================================
// Voice Bot Utilities
// Helper functions for text processing, metrics calculation,
// and pricing calculations for the voice bot system
// ====================================

const { encoding_for_model } = require('tiktoken');
const prices = require('./prices.json');
require('dotenv').config();

// ====================================
// Text Processing Functions
// ====================================

/**
 * Checks if a string ends with a sentence terminator (., !, ?, or :)
 * Used to determine when to send text for speech synthesis
 * @param {string} text - The text to check
 * @returns {boolean} - True if the text ends with a sentence terminator
 */
function matchesSentenceEnding(text) {
    return /([.!?:]([\s]|$|\n))/.test(text);
}

/**
 * Removes special characters and formatting markers from text
 * Cleans up text before processing or displaying
 * @param {string} text - The text to clean
 * @returns {string} - Cleaned text
 */
function removeSpecialCharacters(text) {
    return text
        .replace(/[*#\n]/g, '')           // Remove asterisks, hashtags, and newlines
        .replace(/【\d+:\d+†[^】]+】/g, '') // Remove timestamp-like markers
        .trim();
}

/**
 * Calculates the duration of an audio buffer
 * @param {Buffer} buffer - The audio buffer
 * @param {number} sampleRate - The sample rate of the audio
 * @param {number} channels - Number of audio channels
 * @returns {number} - Duration in seconds
 */
function getAudioDuration(buffer, sampleRate, channels) {
    const bytesPerSample = 2; // Assuming 16-bit audio
    const totalBytes = buffer.length;
    const duration = totalBytes / (sampleRate * channels * bytesPerSample);
    return Math.ceil(duration);
}

/**
 * Converts a hex string to a UUID format string
 * @param {string} hex - The hex string to convert to UUID
 * @returns {string} - Formatted UUID string
 */
function toUUID(hex) {
    return hex.replace(/(.{8})(.{4})(.{4})(.{4})(.{12})/, '$1-$2-$3-$4-$5');
}

// ====================================
// Metrics and Pricing Calculator
// ====================================

/**
 * Calculates various metrics and costs for processing text
 * Handles different roles (system, user, assistant) with different pricing models
 * 
 * @param {string} role - The role of the message (system, user, or assistant)
 * @param {string} text - The text content to analyze
 * @param {number} [wordsPerMinute=130] - Words per minute rate for duration calculation
 * @param {number} [costPerWord=0.00001] - Cost per word
 * @param {number} [costPerChar=0.0000075] - Cost per character
 * @param {number} [costPerSecond=0.00025] - Cost per second of audio
 * @returns {Object} - Object containing various metrics and costs
 */
function calculateMetricsAndPricing(
    role,
    text,
    wordsPerMinute = 130,
    costPerWord = 0.00001,
    costPerChar = 0.0000075,
    costPerSecond = 0.00025
) {
    // Determine which model to use, fallback to gpt-3.5-turbo if specified model not in prices
    const model = prices[process.env.OPENAI_MODEL || 'gpt-3.5-turbo']
        ? process.env.OPENAI_MODEL || 'gpt-3.5-turbo'
        : 'gpt-4';

    // Initialize tokenizer for the selected model
    const encoder = encoding_for_model(model);

    // Calculate basic metrics
    const charCount = text.length;
    const wordCount = text.trim().split(/\s+/).length;
    const durationInSeconds = Math.ceil((wordCount / wordsPerMinute) * 60);
    const tokenCount = encoder.encode(text).length;

    // Calculate costs with precision
    const costByWord = parseFloat((wordCount * costPerWord).toFixed(7));
    const costByCharacter = parseFloat((charCount * costPerChar).toFixed(7));
    const costBySecond = parseFloat((durationInSeconds * costPerSecond).toFixed(7));
    const costByToken = parseFloat(
        (tokenCount * prices[model][role === 'assistant' ? 'output' : 'input'] / 1000)
        .toFixed(7)
    );

    // Free up encoder resources
    encoder.free();

    // Return appropriate metrics based on role
    switch (role) {
        case 'system':
            return {
                tokenCount,
                costByToken,
                model
            };

        case 'user':
            return {
                durationInSeconds,
                tokenCount,
                costBySecond,
                costByToken,
                model
            };

        case 'assistant':
            return {
                charCount,
                tokenCount,
                costByCharacter,
                costByToken,
                model
            };

        default:
            // Return all metrics if role is not specified
            return {
                charCount,
                wordCount,
                tokenCount,
                durationInSeconds,
                costByWord,
                costByCharacter,
                costBySecond,
                costByToken,
                model
            };
    }
}

// Export utility functions
module.exports = {
    removeSpecialCharacters,
    matchesSentenceEnding,
    calculateMetricsAndPricing,
    getAudioDuration,
    toUUID
};