HoruxDespachosNuevo/apps/api/scripts/backfill-regimen-fiscal.ts

import 'dotenv/config';
import { Client } from 'pg';

// Backfill NO destructivo de contribuyentes.regimen_fiscal desde la CSF ya
// almacenada en constancias_situacion_fiscal (tabla del tenant). Replica la
// lógica de matching de consultarConstanciaContribuyente para que el resultado
// sea idéntico al de una descarga fresca. Solo toca filas con regimen_fiscal
// NULL o '' (COALESCE semantics). APPLY=1 para escribir; default = dry-run.

const APPLY = process.env.APPLY === '1';
const PG = {
  host: 'localhost',
  port: 5432,
  user: 'postgres',
  password: process.env.PGPASSWORD,
};
const CONTROL_DB = 'horux360';

// Réplica exacta del `norm` de consultarConstanciaContribuyente:
// NFD + strip de diacríticos (U+0300–U+036F) + lowercase + trim.
const DIACRITICS_RE = new RegExp('[̀-ͯ]', 'g');
const norm = (s: string) =>
  s.normalize('NFD').replace(DIACRITICS_RE, '').toLowerCase().trim();

interface CatalogoReg { clave: string; descripcion: string; }

function clavesFromCsfRegimenes(
  regimenes: Array<{ nombre: string; fechaFin?: string | null }> | undefined | null,
  catalogo: CatalogoReg[],
): string[] {
  if (!regimenes?.length) return [];
  const found: string[] = [];
  for (const reg of regimenes) {
    if (reg.fechaFin) continue; // solo activos
    const regNorm = norm(reg.nombre);
    let best: { clave: string; score: number } | null = null;
    for (const r of catalogo) {
      const catNorm = norm(r.descripcion);
      if (regNorm === catNorm || regNorm.includes(catNorm) || catNorm.includes(regNorm)) {
        const score = catNorm.length;
        if (!best || score > best.score) best = { clave: r.clave, score };
      }
    }
    if (best) found.push(best.clave);
  }
  return [...new Set(found)];
}

async function main() {
  const control = new Client({ ...PG, database: CONTROL_DB });
  await control.connect();

  const { rows: catalogo } = await control.query<CatalogoReg>(
    `SELECT clave, descripcion FROM regimenes WHERE activo = true`,
  );

  const { rows: tenants } = await control.query<{ id: string; database_name: string; rfc: string | null }>(
    `SELECT id, database_name, rfc FROM tenants
      WHERE active = true AND database_name IS NOT NULL AND database_name NOT LIKE '%\\_deleted\\_%'
      ORDER BY database_name`,
  );

  let totalContrib = 0;
  let totalUpdated = 0;
  const report: Array<{ db: string; rfc: string; claves: string; applied: boolean }> = [];

  for (const t of tenants) {
    const tdb = new Client({ ...PG, database: t.database_name });
    try {
      await tdb.connect();
    } catch (e: any) {
      console.log(`SKIP ${t.database_name}: no se pudo conectar (${e.message})`);
      continue;
    }
    try {
      const { rows: hasTbl } = await tdb.query(
        `SELECT 1 FROM information_schema.tables WHERE table_schema='public' AND table_name='constancias_situacion_fiscal' LIMIT 1`,
      );
      if (hasTbl.length === 0) continue;

      const { rows: contribs } = await tdb.query<{ entidad_id: string; rfc: string }>(
        `SELECT entidad_id, rfc FROM contribuyentes
          WHERE regimen_fiscal IS NULL OR regimen_fiscal = ''`,
      );
      if (contribs.length === 0) continue;

      for (const c of contribs) {
        totalContrib++;
        const { rows: csfRows } = await tdb.query<{ regimenes: any }>(
          `SELECT datos->'regimenes' AS regimenes FROM constancias_situacion_fiscal
            WHERE UPPER(rfc) = UPPER($1) ORDER BY fecha_consulta DESC LIMIT 1`,
          [c.rfc],
        );
        const regimenes = csfRows[0]?.regimenes;
        const claves = clavesFromCsfRegimenes(regimenes, catalogo);
        if (claves.length === 0) {
          report.push({ db: t.database_name, rfc: c.rfc, claves: '(sin match en CSF)', applied: false });
          continue;
        }
        const csv = claves.join(',');
        let applied = false;
        if (APPLY) {
          const { rowCount } = await tdb.query(
            `UPDATE contribuyentes SET regimen_fiscal = $1
              WHERE entidad_id = $2 AND (regimen_fiscal IS NULL OR regimen_fiscal = '')`,
            [csv, c.entidad_id],
          );
          applied = (rowCount ?? 0) > 0;
          if (applied) totalUpdated++;
        }
        report.push({ db: t.database_name, rfc: c.rfc, claves: csv, applied });
      }
    } catch (e: any) {
      console.log(`ERR ${t.database_name}: ${e.message}`);
    } finally {
      await tdb.end().catch(() => {});
    }
  }

  await control.end();

  console.log(`\n=== ${APPLY ? 'APPLY' : 'DRY-RUN'} regimen_fiscal backfill ===`);
  console.log(`catálogo regimenes activos: ${catalogo.length}`);
  console.log(`contribuyentes con regimen_fiscal vacío: ${totalContrib}`);
  for (const r of report) {
    const tag = APPLY ? (r.applied ? 'UPDATED' : 'skip') : 'would-update';
    console.log(`  [${tag}] ${r.db} ${r.rfc} -> ${r.claves}`);
  }
  if (APPLY) console.log(`\nTOTAL ACTUALIZADOS: ${totalUpdated}`);
  else console.log(`\n(dry-run: ninguna escritura. Corre con APPLY=1 para aplicar.)`);
}

main().then(() => process.exit(0)).catch((e) => { console.error(e); process.exit(1); });