@arela/uploader 1.0.21 → 1.0.23

package/tests/unit/pedimento-completo-xml-matcher.test.js

@@ -0,0 +1,271 @@
+/**
+ * Regression tests for the pedimento_completo_xml matcher.
+ *
+ * Covers:
+ *   1) Basic detection + arela_path composition from a VUCEM
+ *      `consultarPedimentoCompletoRespuesta` XML.
+ *   2) **YY truth source** — when the pedimento is opened in one year and
+ *      paid in the next (e.g. presentation 2025-12, payment 2026-01), the
+ *      15-digit pedimento MUST keep the presentation year (`25...`), not the
+ *      payment year (`26...`). This matches what VUCEM stamps in the
+ *      filename and what the PDF matchers produce.
+ *   3) **Aduana padding** — VUCEM returns the aduana code without leading
+ *      zeros (e.g. `70` for Ciudad Juárez instead of the canonical `070`).
+ *      The 2-digit "sección aduanera" prefix used inside the 15-digit
+ *      pedimento is the first 2 digits of the 3-digit form (`70` → `07`).
+ *   4) **numPedimento backfill** — the XML matcher composes numPedimento
+ *      externally via `extractNumPedimento` rather than as a field
+ *      extractor. `extractDocumentFields` must backfill it so that
+ *      `composeArelaPath` can find it.
+ *   5) Resolution to `proforma_completo_xml` when no payment evidence
+ *      exists in the body.
+ */
+import { describe, it, expect } from '@jest/globals';
+
+import { extractDocumentFields } from '../../src/document-type-shared.js';
+import { composeArelaPath } from '../../src/file-detection.js';
+
+// ---------------------------------------------------------------------------
+// Test fixtures
+// ---------------------------------------------------------------------------
+
+/**
+ * Build a minimal VUCEM consultarPedimentoCompletoRespuesta XML.
+ * Only the tags the matcher actually reads are included.
+ */
+function buildXml({
+  rfc = 'CEM090106MU3',
+  pedimento = '5063036',
+  claveDocumento = 'V1',
+  tipoOperacionDesc = 'Exportacion',
+  aduanaClave = '70',
+  presentationDate = '2025-12-01-06:00',
+  paymentDate = '2026-01-07-06:00',
+  rectFechaPago = null,
+  facturas = ['V1-FUJIKURA MEX-202512'],
+  edDocs = [],
+} = {}) {
+  const fechas = [];
+  if (presentationDate) {
+    fechas.push(
+      `<ns2:fechas><ns2:fecha>${presentationDate}</ns2:fecha><ns2:tipo><ns2:clave>5</ns2:clave><ns2:descripcion>FECHA DE PRESENTACION</ns2:descripcion></ns2:tipo></ns2:fechas>`,
+    );
+  }
+  if (paymentDate) {
+    fechas.push(
+      `<ns2:fechas><ns2:fecha>${paymentDate}</ns2:fecha><ns2:tipo><ns2:clave>2</ns2:clave><ns2:descripcion>FECHA DE PAGO</ns2:descripcion></ns2:tipo></ns2:fechas>`,
+    );
+  }
+
+  const rect = rectFechaPago
+    ? `<ns2:rectificacion><ns2:fechaPago>${rectFechaPago}</ns2:fechaPago></ns2:rectificacion>`
+    : '';
+
+  const facturasXml = facturas
+    .map(
+      (num) =>
+        `<ns2:facturas><ns2:numero>${num}</ns2:numero></ns2:facturas>`,
+    )
+    .join('');
+
+  const identificadoresXml =
+    edDocs.length === 0
+      ? ''
+      : `<ns2:identificadores>${edDocs
+          .map(
+            (code) =>
+              `<ns2:identificadores><claveIdentificador><clave>ED</clave></claveIdentificador><complemento1>${code}</complemento1></ns2:identificadores>`,
+          )
+          .join('')}</ns2:identificadores>`;
+
+  return `<?xml version="1.0" encoding="UTF-8"?>
+<S:Envelope xmlns:S="http://schemas.xmlsoap.org/soap/envelope/">
+<S:Body>
+<ns2:consultarPedimentoCompletoRespuesta xmlns:ns2="http://x">
+<ns2:pedimento>
+<ns2:pedimento>${pedimento}</ns2:pedimento>
+<ns2:encabezado>
+<ns2:claveDocumento><ns2:clave>${claveDocumento}</ns2:clave></ns2:claveDocumento>
+<ns2:tipoOperacion><ns2:clave>2</ns2:clave><ns2:descripcion>${tipoOperacionDesc}</ns2:descripcion></ns2:tipoOperacion>
+<ns2:aduanaEntradaSalida><ns2:clave>${aduanaClave}</ns2:clave></ns2:aduanaEntradaSalida>
+</ns2:encabezado>
+<ns2:importadorExportador>
+<ns2:rfc>${rfc}</ns2:rfc>
+${fechas.join('\n')}
+</ns2:importadorExportador>
+${rect}
+${facturasXml}
+${identificadoresXml}
+</ns2:pedimento>
+</ns2:consultarPedimentoCompletoRespuesta>
+</S:Body>
+</S:Envelope>`;
+}
+
+// ---------------------------------------------------------------------------
+// Tests
+// ---------------------------------------------------------------------------
+
+describe('pedimento_completo_xml matcher', () => {
+  it('detects, extracts, and composes arela_path for a basic export pedimento', () => {
+    const xml = buildXml({
+      rfc: 'CEM090106MU3',
+      pedimento: '5063036',
+      aduanaClave: '70',
+      presentationDate: '2025-06-15-06:00',
+      paymentDate: '2025-06-20-06:00',
+    });
+    // 15-digit filename pattern: YY=25 AA=07 PPPP=3429 NNNNNNN=5063036
+    const filePath = '/x/2025/250734295063036_250734295063036.xml';
+
+    const [type, fields, ped, year] = extractDocumentFields(
+      xml,
+      'xml',
+      filePath,
+    );
+
+    expect(type).toBe('pedimento_completo_xml');
+    expect(ped).toBe('250734295063036');
+    expect(year).toBe(2025);
+    expect(fields.find((f) => f.name === 'rfc')?.value).toBe('CEM090106MU3');
+    expect(fields.find((f) => f.name === 'aduanaEntradaSalida')?.value).toBe(
+      '07',
+    );
+    // Backfill check: numPedimento must be exposed as a field so
+    // composeArelaPath can find it.
+    expect(fields.find((f) => f.name === 'numPedimento')?.value).toBe(
+      '250734295063036',
+    );
+
+    const arela = composeArelaPath(type, fields, year, filePath);
+    expect(arela).toBe('CEM090106MU3/2025/3429/07/250734295063036/');
+  });
+
+  it('uses presentation date (not payment date) for YY when payment crosses calendar year', () => {
+    // Pedimento opened Dec 2025, paid Jan 2026 — the YY must be 25.
+    const xml = buildXml({
+      pedimento: '5063036',
+      aduanaClave: '70',
+      presentationDate: '2025-12-01-06:00',
+      paymentDate: '2026-01-07-06:00',
+    });
+    // Use the 3-part filename pattern (no YY in filename) so YY comes from XML body.
+    const filePath = '/x/070-3429-5063036.xml';
+
+    const [type, , ped, year] = extractDocumentFields(xml, 'xml', filePath);
+
+    expect(type).toBe('pedimento_completo_xml');
+    expect(ped).toBe('250734295063036');
+    expect(year).toBe(2025);
+  });
+
+  it('falls back to payment date YY when presentation date is missing', () => {
+    const xml = buildXml({
+      pedimento: '5063036',
+      aduanaClave: '70',
+      presentationDate: null, // No clave=5 block
+      paymentDate: '2026-01-07-06:00',
+    });
+    const filePath = '/x/070-3429-5063036.xml';
+
+    const [, , ped, year] = extractDocumentFields(xml, 'xml', filePath);
+
+    expect(ped).toBe('260734295063036');
+    expect(year).toBe(2026);
+  });
+
+  it('prefers filename YY over body fechas (VUCEM-stamped truth)', () => {
+    // Filename says YY=24 but body has presentation=2025. Filename wins.
+    const xml = buildXml({
+      pedimento: '5063036',
+      aduanaClave: '70',
+      presentationDate: '2025-12-01-06:00',
+      paymentDate: '2026-01-07-06:00',
+    });
+    const filePath = '/x/240734295063036_240734295063036.xml';
+
+    const [, , ped, year] = extractDocumentFields(xml, 'xml', filePath);
+
+    expect(ped).toBe('240734295063036');
+    expect(year).toBe(2024);
+  });
+
+  it('pads VUCEM aduana correctly: 70 -> 07, 750 -> 75, 40 -> 04', () => {
+    const cases = [
+      { aduanaClave: '70', expected: '07', // Cd. Juárez (3-digit canonical: 070)
+        filename: '/x/070-3429-5000001.xml' },
+      { aduanaClave: '750', expected: '75', // Puebla
+        filename: '/x/750-3429-5000002.xml' },
+      { aduanaClave: '40', expected: '04', // Lázaro Cárdenas (canonical: 040)
+        filename: '/x/040-3429-5000003.xml' },
+    ];
+
+    for (const c of cases) {
+      const xml = buildXml({
+        pedimento: c.filename.match(/-(\d{7})\.xml$/)[1],
+        aduanaClave: c.aduanaClave,
+        presentationDate: '2025-06-15-06:00',
+        paymentDate: '2025-06-20-06:00',
+      });
+      const [, fields, ped] = extractDocumentFields(xml, 'xml', c.filename);
+      expect(fields.find((f) => f.name === 'aduanaEntradaSalida')?.value).toBe(
+        c.expected,
+      );
+      // Positions 2-3 of the composed 15-digit pedimento must equal the
+      // aduana prefix.
+      expect(ped.substring(2, 4)).toBe(c.expected);
+    }
+  });
+
+  it('resolves to proforma_completo_xml when no payment evidence exists', () => {
+    const xml = buildXml({
+      pedimento: '5063036',
+      aduanaClave: '70',
+      presentationDate: '2025-12-01-06:00',
+      paymentDate: null, // No payment, no rectificacion
+    });
+    const filePath = '/x/070-3429-5063036.xml';
+
+    const [type] = extractDocumentFields(xml, 'xml', filePath);
+
+    expect(type).toBe('proforma_completo_xml');
+  });
+
+  it('extracts cove and rfc correctly', () => {
+    const xml = buildXml({
+      rfc: 'CEM090106MU3',
+      facturas: ['V1-FUJIKURA MEX-202512', 'INV-2'],
+    });
+    const filePath = '/x/250734295063036_250734295063036.xml';
+
+    const [, fields] = extractDocumentFields(xml, 'xml', filePath);
+
+    expect(fields.find((f) => f.name === 'rfc')?.value).toBe('CEM090106MU3');
+    expect(fields.find((f) => f.name === 'cove')?.value).toBe(
+      '[V1-FUJIKURA MEX-202512,INV-2]',
+    );
+  });
+
+  it('returns null arela_path when filename is unrecognized (no patente)', () => {
+    const xml = buildXml({
+      pedimento: '5063036',
+      aduanaClave: '70',
+      presentationDate: '2025-06-15-06:00',
+      paymentDate: '2025-06-20-06:00',
+    });
+    // Unrecognized filename — no patente derivable.
+    const filePath = '/x/random_name.xml';
+
+    const [type, fields, ped, year] = extractDocumentFields(
+      xml,
+      'xml',
+      filePath,
+    );
+
+    expect(type).toBe('pedimento_completo_xml');
+    expect(ped).toBeNull();
+    expect(year).toBeNull();
+    // composeArelaPath returns null because patente is missing.
+    expect(composeArelaPath(type, fields, year, filePath)).toBeNull();
+  });
+});

package/tests/unit/pedimento-simplificado-matcher.test.js

@@ -0,0 +1,185 @@
+/**
+ * Regression tests for the pedimento_simplificado matcher.
+ *
+ * Covers PDF layouts where:
+ *   - The header reads "FORMA SIMPLIFICADA DEL PEDIMENTO" (with DEL),
+ *     not the canonical "DE PEDIMENTO".
+ *   - The header trio prints "T. OPER" WITHOUT a trailing colon
+ *     (the value sits in a separate table cell).
+ *
+ * Real-world example: REF NQR26-079, Aduana 640 (Querétaro), patente 3458.
+ * Before this regression test, the matcher fell through to
+ * `facturas_comerciales` because the word "FACTURA" appears in the
+ * "OBSERVACIONES" block.
+ */
+import { describe, it, expect } from '@jest/globals';
+
+// Importing only the dispatcher avoids circular-init issues caused by
+// `_pedimento-shared-extractors.js` pulling FieldResult from document-type-shared.
+import { extractDocumentFields } from '../../src/document-type-shared.js';
+
+// REAL pdf-parse output from the NQR26-079 simplificado PDF
+// (CSM9204097Q1, patente 3458, aduana 640).
+// Captured verbatim with `PDFParse({data}).getText()` — pdf-parse extracts
+// table cells out of visual order, so labels and values often live on
+// different lines (see the FECHA DE PAGO block: label appears, then a few
+// unrelated cells, then the date sits on its own line with the importe).
+// This is exactly what the matchers and extractors see in production.
+const SIMP_DEL_NQR26079_TEXT = `A1      CVE. PEDIMENTO: IMP     T. OPER 26 64 3458 6000079      NUM. PEDIMENTO:
+CERTIFICACIONES
+ADUANA E/S:
+DATOS DEL IMPORTADOR / EXPORTADOR
+RFC: CURP:
+CÓDIGO DE
+ACEPTACIÓN
+640
+CSM9204097Q1
+FECHAS:
+17/03/2026
+Ped. 6000079
+CLAVE DE LA SECCION ADUANERA
+DE DESPACHO:
+QUERETARO, EL MARQUES Y
+COLON, QUERETARO.
+640
+DESTINO: 9 PESO BRUTO: 5.350
+MARCAS,NUMEROS Y TOTAL DE BULTOS: 1
+04/03/2026
+        ENTRADA
+PAGO
+3PW4CLHE
+S/M S/N
+CODIGO DE BARRAS
+0326 0132 XMP1 4914 6243 989
+*** PAGO ELECTRONICO ***
+DEPÓSITO REFERENCIADO - LÍNEA DE CAPTURA
+PATENTE:
+NOMBRE DE LA INSTITUCIÓN BANCARIA:
+LÍNEA DE CAPTURA:
+IMPORTE PAGADO:
+NÚMERO DE OPERACIÓN BANCARIA:
+NÚMERO DE TRANSACCIÓN SAT:
+MEDIO DE PRESENTACIÓN:
+MEDIO DE RECEPCIÓN/COBRO:
+OTROS MEDIOS ELECTRÓNICOS (PAGO ELECTRÓNICO)
+EFECTIVO (CARGO A CUENTA)
+PEDIMENTO: ADUANA:
+FECHA DE PAGO:
+0326 0132 XMP1 4914 6243
+6000079 640
+17/03/2026      $989
+Banco Nacional de México, S.A.
+00000000703543
+3458
+40124170320261403012
+NUMERO (GUIA/ORDEN EMBARQUE)/ID: 023-51315051 M 490453269837 H
+NÚMERO DE ACUSE DE VALOR COVE268074HT1
+NÚMERO DE E-DOCUMENT: 0438261DOG9W3 01702619TYEU7
+OBSERVACIONES
+FACTURA DE ACUERDO AL ARTÍCULO 36-A DE LA LEY ADUANERA VIGENTE Y A LA REGLA 3.1.
+8. DE LAS REGLAS
+GENERALES DE COMERCIO EXTERIOR VIGENTES.
+SE TRANSMITE PREVIAMENTE A VENTANILLA DIGITAL CONFORME A LA REGLA 1.9.18. DE LAS
+ REGLAS GENERALES DE
+COMERCIO EXTERIOR VIGENTES.
+SE EFECTÚA LA TRANSMISIÓN DIGITAL DE CONFORMIDAD A LA REGLA 3.1.17. Y 3.1.31. DE
+ LAS REGLAS GENERALES
+DE COMERCIO EXTERIOR VIGENTES.
+LA INFORMACIÓN CONTENIDA EN ESTE PEDIMENTO FUE SUMINISTRADA POR EL IMPORTADOR DE
+ CONFORMIDAD CON EL
+ARTICULO 54 DE LA LEY ADUANERA EN VIGOR.
+SE EXIME NOM-024-SCFI-2013 EN TERMINOS DEL NUMERAL 10, FRACC. X INCISO H, IMPORT
+ACIÓN DEFINITIVA,
+TRATÁNDOSE DE IMPORTADORES QUE CUENTEN CON UN PROSEC.
+SE EXIME NOM-003-SCFI-2014 EN TERMINOS DEL NUMERAL 10, FRACC. X INCISO H, IMPORT
+ACIÓN DEFINITIVA,
+TRATÁNDOSE DE IMPORTADORES QUE CUENTEN CON UN PROSEC.
+JOAQUIN GOMEZ ABAD
+AGENTE ADUANAL, AGENCIA ADUANAL, APODERADO ADUANAL O DE ALMACEN
+NOMBRE O RAZ. SOC.:
+RFC: GAA1003111U6 GOAJ641219HDFMBQ09    CURP:
+e.firma:
+NUMERO DE SERIE DEL CERTIFICADO: 00001000000705949781
+GOAJ641219QT5   RFC:
+DECLARO BAJO PROTESTA DE DECIR VERDAD, EN LOS TERMINOS
+DE LO DISPUESTO ARTICULO 81 DE LA LEY: PATENTE O
+AUTORIZACIÓN: 3458      GOMEZ ABAD ASESORES EN COMERCIO EXTERIOR S.C.
+FORMA SIMPLIFICADA DEL PEDIMENTO
+SEGUNDA COPIA: IMPORTADOR EXPORTADOR DESTINO/ORIGEN: INTERIOR DEL PAÍS
+REF: NQR26-079 Página 1 de 2
+
+-- 1 of 2 --
+
+FORMA SIMPLIFICADA DEL PEDIMENTO
+SEGUNDA COPIA: IMPORTADOR EXPORTADOR DESTINO/ORIGEN: INTERIOR DEL PAÍS
+REF: NQR26-079 Página 1 de 2
+A1      CVE. PEDIMENTO: IMP     T. OPER 26 64 3458 6000079      NUM. PEDIMENTO:
+CURP:
+RFC: CSM9204097Q1
+****** ****** **********        ********** FIN DE PEDIMENTO NUM. TOTAL DE PARTID
+AS: CLAVE PREVALIDADOR: 010     1
+ANEXO DEL PEDIMENTO
+SEGUNDA COPIA: IMPORTADOR EXPORTADOR DESTINO/ORIGEN: INTERIOR DEL PAÍS
+REF: NQR26-079 Página 2 de 2`;
+
+describe('pedimento_simplificado matcher — DEL PEDIMENTO variant', () => {
+  it('dispatcher resolves NQR26-079 (DEL PEDIMENTO) as pedimento_simplificado', () => {
+    const [detectedType, , pedimento] = extractDocumentFields(
+      SIMP_DEL_NQR26079_TEXT,
+      'pdf',
+      '/scans/CSM9204097Q1/NQR26-079.pdf',
+    );
+
+    // Regression: previously this resolved to `factura_comercial` because
+    // (1) the title regex demanded "DE PEDIMENTO" (this PDF says "DEL") and
+    // (2) the header trio required a colon after "T. OPER" (this PDF omits it).
+    expect(detectedType).toBe('pedimento_simplificado');
+    expect(pedimento).toBe('266434586000079');
+  });
+});
+
+describe('pedimento_simplificado matcher — header trio without colon after T. OPER', () => {
+  // Minimal text: title is the canonical "DE PEDIMENTO" so the fast path
+  // does NOT apply; only the fallback that requires the header trio runs.
+  // The trio MUST tolerate "T. OPER" without a trailing colon, because
+  // many printable PDFs render OPER as a column header (value in next cell).
+  const FALLBACK_TEXT = `FORMA SIMPLIFICADA DE PEDIMENTO
+NUM. PEDIMENTO: 22 07 3429 2002089 T. OPER IMP CVE. PEDIMENTO: A1
+DATOS DEL IMPORTADOR
+PATENTE: 3429 PEDIMENTO: 2002089 ADUANA: 070
+FECHA DE PAGO: 01/02/2023`;
+
+  it('resolves via fast-path "FORMA SIMPLIFICADA DE PEDIMENTO" header', () => {
+    const [detectedType] = extractDocumentFields(
+      FALLBACK_TEXT,
+      'pdf',
+      '/scans/SAMPLE/pedimento.pdf',
+    );
+    expect(detectedType).toBe('pedimento_simplificado');
+  });
+});
+
+describe('pedimento_simplificado matcher — title accepts both DE and DEL', () => {
+  // Same minimal body, only the title differs. Both variants are produced
+  // by different prevalidators / agencias in the wild, and BOTH must
+  // resolve to pedimento_simplificado.
+  const body = `
+NUM. PEDIMENTO: 22 07 3429 2002089 T. OPER IMP CVE. PEDIMENTO: A1
+DATOS DEL IMPORTADOR
+PATENTE: 3429 PEDIMENTO: 2002089 ADUANA: 070
+FECHA DE PAGO: 01/02/2023`;
+
+  it.each([
+    ['FORMA SIMPLIFICADA DE PEDIMENTO', 'pedimento_simplificado'],
+    ['FORMA SIMPLIFICADA DEL PEDIMENTO', 'pedimento_simplificado'],
+    ['forma simplificada de pedimento', 'pedimento_simplificado'], // case-insensitive
+    ['FORMA  SIMPLIFICADA   DEL   PEDIMENTO', 'pedimento_simplificado'], // extra spaces
+  ])('title "%s" resolves to %s', (title, expected) => {
+    const [detectedType] = extractDocumentFields(
+      `${title}\n${body}`,
+      'pdf',
+      '/scans/SAMPLE/pedimento.pdf',
+    );
+    expect(detectedType).toBe(expected);
+  });
+});