apps/api/src/utils/metrics.ts

import levenshtein from 'fast-levenshtein';

/**
 * Word Error Rate (WER) calculation using Levenshtein distance at word level.
 */
export const calculateWER = (reference: string, hypothesis: string): number => {
    const clean = (str: string) => str.toLowerCase().replace(/[.,/#!$%^&*;:{}=\-_`~()]/g, "").split(/\s+/).filter(w => w);
    
    const refWords = clean(reference);
    const hypWords = clean(hypothesis);
    
    if (refWords.length === 0) return hypWords.length > 0 ? 1 : 0;

    const wordMap = new Map<string, string>();
    let charCode = 0xE000;
    
    const getChar = (word: string) => {
        if (!wordMap.has(word)) wordMap.set(word, String.fromCharCode(charCode++));
        return wordMap.get(word)!;
    };
    
    const refChars = refWords.map(getChar).join('');
    const hypChars = hypWords.map(getChar).join('');
    
    return levenshtein.get(refChars, hypChars) / refWords.length;
};

/**
 * Standard error formatter for catch blocks.
 */
export const formatError = (err: unknown): string => {
    if (err instanceof Error) return err.message;
    return String(err);
};