apps/api/src/services/training-service.ts

import { prisma } from './prisma';
import { calculateWER } from '../utils/metrics';
import { normalizeWolof } from '../scripts/normalizeWolof';

export class TrainingService {
    static async getPendingAudios(page: number, limit: number) {
        const [pending, total] = await Promise.all([
            prisma.trainingData.findMany({
                where: { status: 'PENDING' },
                orderBy: { createdAt: 'desc' },
                skip: (page - 1) * limit,
                take: limit,
            }),
            prisma.trainingData.count({ where: { status: 'PENDING' } })
        ]);
        return { pending, total, page, limit };
    }

    static async submitCorrection(id: string | undefined, audioUrl: string, transcription: string, manualCorrection: string) {
        const normResult = normalizeWolof(transcription);
        const rawWER = calculateWER(manualCorrection, transcription);
        const normalizedWER = calculateWER(manualCorrection, normResult.normalizedText);

        // Analyze missing words
        const manualWords = manualCorrection.toLowerCase().replace(/[.,/#!$%^&*;:{}=\-_`~()]/g, "").split(/\s+/).filter(w => w);
        const normWords = normResult.normalizedText.toLowerCase().replace(/[.,/#!$%^&*;:{}=\-_`~()]/g, "").split(/\s+/).filter((w: string) => w);
        const missingWords = Array.from(new Set(manualWords.filter((w: string) => !normWords.includes(w))));

        const data = await prisma.trainingData.upsert({
            where: { id: id || '00000000-0000-0000-0000-000000000000' },
            update: { manualCorrection, rawWER, normalizedWER, status: 'REVIEWED' },
            create: { audioUrl, transcription, manualCorrection, rawWER, normalizedWER, status: 'REVIEWED' }
        });

        return { data, missingWords, rawWER, normalizedWER };
    }

    static async createTrainingRecord(audioUrl: string, transcription: string) {
        return prisma.trainingData.create({
            data: {
                audioUrl,
                transcription: transcription || '',
                status: 'PENDING'
            }
        });
    }
}