Momento/memento-note/lib/integrations/gmail-scanner.service.ts

/**
 * Gmail Scanner — extrait vols, colis, abonnements → notes + rappels
 */

import prisma from '@/lib/prisma'
import { getChatProvider } from '@/lib/ai/factory'
import { getSystemConfig } from '@/lib/config'
import { getGoogleTokens, refreshGoogleAccessToken } from '@/lib/integrations/google-integration-tokens'

const GMAIL_API = 'https://gmail.googleapis.com/gmail/v1/users/me'

const SCAN_QUERIES: { category: string; q: string }[] = [
  { category: 'flight', q: 'subject:(flight OR boarding OR vol OR "boarding pass") newer_than:7d' },
  { category: 'package', q: 'from:(amazon OR ups OR fedex OR dhl OR laposte OR colissimo) newer_than:7d' },
  { category: 'subscription', q: 'subject:(renewal OR reconduction OR renouvellement OR abonnement) newer_than:30d' },
]

function decodeBase64Url(data: string): string {
  const normalized = data.replace(/-/g, '+').replace(/_/g, '/')
  return Buffer.from(normalized, 'base64').toString('utf-8')
}

function extractBody(payload: any): string {
  if (!payload) return ''
  if (payload.body?.data) {
    const raw = decodeBase64Url(payload.body.data)
    return raw.replace(/<[^>]+>/g, ' ').replace(/\s+/g, ' ').trim().slice(0, 4000)
  }
  if (Array.isArray(payload.parts)) {
    for (const part of payload.parts) {
      if (part.mimeType === 'text/plain' && part.body?.data) {
        return decodeBase64Url(part.body.data).slice(0, 4000)
      }
    }
    for (const part of payload.parts) {
      const nested = extractBody(part)
      if (nested) return nested
    }
  }
  return ''
}

async function gmailFetch(userId: string, path: string, accessToken: string, refreshToken: string) {
  let token = accessToken
  let res = await fetch(`${GMAIL_API}${path}`, {
    headers: { Authorization: `Bearer ${token}` },
  })
  if (res.status === 401 && refreshToken) {
    const refreshed = await refreshGoogleAccessToken(userId, 'gmail', refreshToken)
    if (refreshed) {
      token = refreshed
      res = await fetch(`${GMAIL_API}${path}`, {
        headers: { Authorization: `Bearer ${token}` },
      })
    }
  }
  return res
}

async function extractStructured(
  category: string,
  subject: string,
  body: string,
): Promise<Record<string, unknown> | null> {
  const config = await getSystemConfig()
  const provider = getChatProvider(config)
  if (!provider) return null

  const schemas: Record<string, string> = {
    flight: '{"airline":"","flightNumber":"","departureDate":"ISO8601","departureAirport":"","arrivalAirport":"","confirmationCode":""}',
    package: '{"carrier":"","trackingNumber":"","expectedDelivery":"ISO8601","status":""}',
    subscription: '{"serviceName":"","renewalDate":"ISO8601","amount":0,"currency":"EUR"}',
  }

  const prompt = `Extrais les données structurées de cet email (${category}). Retourne UNIQUEMENT du JSON valide selon ce schéma: ${schemas[category] || '{}'}
Sujet: ${subject}
Corps: ${body.slice(0, 2500)}`

  try {
    const raw = await provider.generateText(prompt)
    const match = raw.match(/\{[\s\S]*\}/)
    if (!match) return null
    return JSON.parse(match[0])
  } catch {
    return null
  }
}

function buildNoteFromExtraction(category: string, subject: string, data: Record<string, unknown>) {
  if (category === 'flight') {
    const title = `Vol ${data.flightNumber || data.airline || ''}`.trim() || subject.slice(0, 80)
    const dep = data.departureDate ? new Date(String(data.departureDate)) : null
    const reminder = dep && !Number.isNaN(dep.getTime())
      ? new Date(dep.getTime() - 3 * 60 * 60 * 1000)
      : null
    const html = `<p><strong>${title}</strong></p><p>${data.departureAirport || ''} → ${data.arrivalAirport || ''}</p><p>Confirmation: ${data.confirmationCode || '—'}</p>`
    return { title, content: html, reminder }
  }
  if (category === 'package') {
    const title = `Colis ${data.trackingNumber || data.carrier || ''}`.trim() || subject.slice(0, 80)
    const del = data.expectedDelivery ? new Date(String(data.expectedDelivery)) : null
    const reminder = del && !Number.isNaN(del.getTime()) ? del : null
    const html = `<p><strong>${title}</strong></p><p>Transporteur: ${data.carrier || '—'}</p><p>Suivi: ${data.trackingNumber || '—'}</p><p>Statut: ${data.status || '—'}</p>`
    return { title, content: html, reminder }
  }
  const title = `Abonnement ${data.serviceName || ''}`.trim() || subject.slice(0, 80)
  const ren = data.renewalDate ? new Date(String(data.renewalDate)) : null
  const reminder = ren && !Number.isNaN(ren.getTime())
    ? new Date(ren.getTime() - 7 * 24 * 60 * 60 * 1000)
    : null
  const html = `<p><strong>${title}</strong></p><p>Renouvellement: ${data.renewalDate || '—'}</p><p>Montant: ${data.amount ?? '—'} ${data.currency || ''}</p>`
  return { title, content: html, reminder }
}

export class GmailScannerService {
  async scanUser(userId: string): Promise<{ processed: number; created: number }> {
    const tokens = await getGoogleTokens(userId, 'gmail')
    if (!tokens) return { processed: 0, created: 0 }

    let processed = 0
    let created = 0

    for (const { category, q } of SCAN_QUERIES) {
      const listRes = await gmailFetch(
        userId,
        `/messages?maxResults=5&q=${encodeURIComponent(q)}`,
        tokens.accessToken,
        tokens.refreshToken,
      )
      if (!listRes.ok) continue
      const listJson = await listRes.json()
      const ids: string[] = (listJson.messages || []).map((m: { id: string }) => m.id)

      for (const messageId of ids) {
        const exists = await prisma.gmailScanHistory.findUnique({
          where: { userId_messageId: { userId, messageId } },
        })
        if (exists) continue

        const msgRes = await gmailFetch(
          userId,
          `/messages/${messageId}?format=full`,
          tokens.accessToken,
          tokens.refreshToken,
        )
        if (!msgRes.ok) continue
        processed++

        const msg = await msgRes.json()
        const headers: { name: string; value: string }[] = msg.payload?.headers || []
        const subject = headers.find(h => h.name.toLowerCase() === 'subject')?.value || '(Sans objet)'
        const body = extractBody(msg.payload)

        const extracted = await extractStructured(category, subject, body)
        if (!extracted) {
          await prisma.gmailScanHistory.create({
            data: { userId, messageId, category, data: { subject, skipped: true } },
          })
          continue
        }

        const built = buildNoteFromExtraction(category, subject, extracted)
        const note = await prisma.note.create({
          data: {
            userId,
            title: built.title,
            content: built.content,
            type: 'richtext',
            labels: JSON.stringify(['gmail', category]),
            reminder: built.reminder,
            autoGenerated: true,
          },
        })

        await prisma.gmailScanHistory.create({
          data: {
            userId,
            messageId,
            category,
            data: extracted as object,
            noteId: note.id,
          },
        })
        created++
      }
    }

    return { processed, created }
  }

  async getStatus(userId: string) {
    const tokens = await getGoogleTokens(userId, 'gmail')
    const recent = await prisma.gmailScanHistory.count({
      where: {
        userId,
        scannedAt: { gte: new Date(Date.now() - 7 * 24 * 60 * 60 * 1000) },
        noteId: { not: null },
      },
    })
    return { connected: !!tokens, recentCaptures: recent }
  }
}

export const gmailScannerService = new GmailScannerService()