@arela/uploader 1.1.4 → 1.2.0

package/.claude/worktrees/agent-multi-profile/src/document-types/pedimento-completo-xml.js

@@ -0,0 +1,372 @@
+// VUCEM "consultarPedimentoCompleto" XML matcher.
+//
+// Registered in `document-type-shared.js`. Downstream code
+// (composeArelaPath, arela-api propagation SQL, IdentifyCommand counters)
+// also includes `pedimento_completo_xml`.
+//
+// Filename patterns recognized (try in order — patente extraction):
+//   1) VU_PATENTE_ADUANA_PEDIMENTO.xml          → e.g. VU_3429_070_5016101.xml
+//   2) ADUANA-PATENTE-PEDIMENTO.xml             → e.g. 670-3806-2002487.xml
+//   3) {15-digit}[_{15-digit}].xml              → e.g. 260734296016642_260734296016642.xml
+// If none match, arela_path is left null and propagation fills it from a
+// sibling PDF in the same directory.
+//
+// numPedimento is composed from XML body + filename because the body never
+// carries the 15-digit form: YY|AA|PPPP|NNNNNNN.
+import { FieldResult } from '../document-type-shared.js';
+
+// --------------------------- helpers ---------------------------------------
+
+function firstTag(source, tag) {
+  // Match <ns2:tag>value</ns2:tag> or unprefixed <tag>value</tag>
+  const re = new RegExp(
+    `<(?:[a-z0-9]+:)?${tag}>([^<]*)</(?:[a-z0-9]+:)?${tag}>`,
+    'i',
+  );
+  const m = source.match(re);
+  return m ? m[1].trim() : null;
+}
+
+function allTagBlocks(source, tag) {
+  const re = new RegExp(
+    `<(?:[a-z0-9]+:)?${tag}>([\\s\\S]*?)</(?:[a-z0-9]+:)?${tag}>`,
+    'gi',
+  );
+  const out = [];
+  let m;
+  while ((m = re.exec(source)) !== null) {
+    out.push(m[1]);
+  }
+  return out;
+}
+
+function pad(value, length) {
+  if (value == null) return null;
+  return String(value).padStart(length, '0');
+}
+
+/**
+ * Convert a VUCEM `aduanaEntradaSalida.clave` (e.g. "70", "750", "40") to the
+ * 2-digit "sección aduanera" prefix used inside the 15-digit pedimento number.
+ *
+ *   VUCEM strips leading zeros from the canonical 3-digit SAT aduana code,
+ *   so `070` (Ciudad Juárez) arrives as `70`. The pedimento prefix is the
+ *   first 2 digits of the 3-digit code:
+ *     `70`  → `070` → `07`   (Cd. Juárez)
+ *     `750` → `750` → `75`   (Puebla)
+ *     `40`  → `040` → `04`   (Lázaro Cárdenas)
+ */
+function aduanaToSeccion(claveValue) {
+  if (claveValue == null) return null;
+  return pad(claveValue, 3).substring(0, 2);
+}
+
+/**
+ * Try the three known filename patterns and return {patente, aduana, pedimento}
+ * with any subset of the fields populated. Returns null if no pattern matches.
+ */
+function parseFilenameParts(filePath) {
+  if (!filePath) return null;
+  const fileName = filePath.split(/[\\/]/).pop();
+  if (!fileName) return null;
+
+  let m;
+
+  // 1) VU_PATENTE_ADUANA_PEDIMENTO.xml
+  m = fileName.match(/^VU_(\d{4})_(\d{3})_(\d{7})\.xml$/i);
+  if (m) {
+    return {
+      patente: m[1],
+      aduana: m[2].substring(0, 2),
+      pedimento: m[3],
+      year: null,
+    };
+  }
+
+  // 2) ADUANA-PATENTE-PEDIMENTO.xml
+  m = fileName.match(/^(\d{3})-(\d{4})-(\d{7})\.xml$/i);
+  if (m) {
+    return {
+      patente: m[2],
+      aduana: m[1].substring(0, 2),
+      pedimento: m[3],
+      year: null,
+    };
+  }
+
+  // 3) 15-digit form YY|AA|PPPP|NNNNNNN
+  m = fileName.match(/^(\d{15})(?:_\d{15})?\.xml$/i);
+  if (m) {
+    const fifteen = m[1];
+    return {
+      year: fifteen.substring(0, 2),
+      aduana: fifteen.substring(2, 4),
+      patente: fifteen.substring(4, 8),
+      pedimento: fifteen.substring(8, 15),
+    };
+  }
+
+  return null;
+}
+
+// Capture the YY year from an ISO date string like "2026-05-08-06:00".
+function yyFromIsoDate(iso) {
+  if (!iso) return null;
+  const m = iso.match(/^(\d{4})-/);
+  return m ? m[1].substring(2, 4) : null;
+}
+
+// Find <ns2:fechas> block whose nested <clave> matches `claveValue` and
+// return its <ns2:fecha>. Works for both shapes:
+//   <fechas><clave>N</clave><fecha>...</fecha></fechas>
+//   <fechas><fecha>...</fecha><tipo><clave>N</clave></tipo></fechas>
+// (firstTag finds the FIRST <clave> in the block — both layouts expose only
+//  one clave per fechas entry.)
+function findFechaByClave(source, claveValue) {
+  const fechasBlocks = allTagBlocks(source, 'fechas');
+  for (const block of fechasBlocks) {
+    const clave = firstTag(block, 'clave');
+    if (clave === claveValue) {
+      const fecha = firstTag(block, 'fecha');
+      if (fecha) return fecha;
+    }
+  }
+  return null;
+}
+
+// Fecha de pago de las contribuciones (tipo.clave == 2).
+function findPaymentDate(source) {
+  return findFechaByClave(source, '2');
+}
+
+// Fecha de presentacion (tipo.clave == 5). This is the authoritative source
+// for the pedimento's YY prefix — a pedimento opened in Dec-2025 but paid in
+// Jan-2026 keeps the `25` prefix, matching what VUCEM stamps in the filename.
+function findPresentationDate(source) {
+  return findFechaByClave(source, '5');
+}
+
+// --------------------------- extractors ------------------------------------
+
+const rfcExtractor = {
+  field: 'rfc',
+  extract: (source) => {
+    const value = firstTag(source, 'rfc');
+    return new FieldResult('rfc', !!value, value);
+  },
+};
+
+const clavePedimentoExtractor = {
+  field: 'clavePedimento',
+  extract: (source) => {
+    // <ns2:claveDocumento><ns2:clave>R1</ns2:clave></ns2:claveDocumento>
+    const blocks = allTagBlocks(source, 'claveDocumento');
+    const clave = blocks.length > 0 ? firstTag(blocks[0], 'clave') : null;
+    return new FieldResult('clavePedimento', !!clave, clave);
+  },
+};
+
+const tipoOperacionExtractor = {
+  field: 'tipoOperacion',
+  extract: (source) => {
+    const blocks = allTagBlocks(source, 'tipoOperacion');
+    const desc = blocks.length > 0 ? firstTag(blocks[0], 'descripcion') : null;
+    return new FieldResult('tipoOperacion', !!desc, desc);
+  },
+};
+
+const aduanaEntradaSalidaExtractor = {
+  field: 'aduanaEntradaSalida',
+  extract: (source) => {
+    const blocks = allTagBlocks(source, 'aduanaEntradaSalida');
+    const clave = blocks.length > 0 ? firstTag(blocks[0], 'clave') : null;
+    return new FieldResult(
+      'aduanaEntradaSalida',
+      !!clave,
+      aduanaToSeccion(clave),
+    );
+  },
+};
+
+const paymentDateExtractor = {
+  field: 'paymentDate',
+  extract: (source) => {
+    const fecha = findPaymentDate(source);
+    return new FieldResult('paymentDate', !!fecha, fecha);
+  },
+};
+
+const presentationDateExtractor = {
+  field: 'presentationDate',
+  extract: (source) => {
+    const fecha = findPresentationDate(source);
+    return new FieldResult('presentationDate', !!fecha, fecha);
+  },
+};
+
+const fechaPagoRectificacionExtractor = {
+  field: 'fechaPagoRectificacion',
+  extract: (source) => {
+    const rectBlocks = allTagBlocks(source, 'rectificacion');
+    if (rectBlocks.length === 0) {
+      return new FieldResult('fechaPagoRectificacion', false, null);
+    }
+    const fechaPago = firstTag(rectBlocks[0], 'fechaPago');
+    return new FieldResult('fechaPagoRectificacion', !!fechaPago, fechaPago);
+  },
+};
+
+const coveExtractor = {
+  field: 'cove',
+  extract: (source) => {
+    // Collect <ns2:numero> values that live inside <ns2:facturas> blocks.
+    const facturas = allTagBlocks(source, 'facturas');
+    const numeros = [];
+    facturas.forEach((block) => {
+      const numero = firstTag(block, 'numero');
+      if (numero) numeros.push(numero);
+    });
+    if (numeros.length === 0) {
+      return new FieldResult('cove', false, null);
+    }
+    const unique = [...new Set(numeros)];
+    return new FieldResult('cove', true, `[${unique.join(',')}]`);
+  },
+};
+
+const numEDocumentoExtractor = {
+  field: 'numEDocumento',
+  extract: (source) => {
+    // For each <ns2:identificadores> block, check the nested
+    // <ns2:claveIdentificador><ns2:clave> value. If it equals 'ED',
+    // collect the sibling <ns2:complemento1> value.
+    const blocks = allTagBlocks(source, 'identificadores');
+    const codes = [];
+    blocks.forEach((block) => {
+      const claveBlocks = allTagBlocks(block, 'claveIdentificador');
+      const clave =
+        claveBlocks.length > 0 ? firstTag(claveBlocks[0], 'clave') : null;
+      if (clave === 'ED') {
+        const complemento = firstTag(block, 'complemento1');
+        if (complemento) codes.push(complemento);
+      }
+    });
+    if (codes.length === 0) {
+      return new FieldResult('numEDocumento', false, null);
+    }
+    const unique = [...new Set(codes)];
+    return new FieldResult('numEDocumento', true, `[${unique.join(',')}]`);
+  },
+};
+
+// Composed numPedimento — needs the filename + already-extracted fields.
+// We expose it as the LAST extractor so `aduanaEntradaSalida` is available
+// via the `fields` array if the runner exposes it. To stay compatible with
+// the existing extractor signature (which only receives `source`), we
+// re-derive aduanaEntradaSalida inside this extractor and read the filename
+// from a closure populated by `match()` via `extractNumPedimento` below.
+// (See `extractNumPedimento` — that is the canonical place numPedimento is
+// composed for XML.)
+
+// --------------------------- definition ------------------------------------
+
+export const pedimentoCompletoXmlDefinition = {
+  type: 'pedimento_completo_xml',
+  extensions: ['xml'],
+
+  match: (source) => {
+    return /consultarPedimentoCompletoRespuesta/i.test(source);
+  },
+
+  /**
+   * Resolve to `pedimento_completo_xml` only if payment evidence exists:
+   *   - When a <rectificacion> block is present, require fechaPagoRectificacion.
+   *   - Otherwise require paymentDate.
+   */
+  resolveType: (fields) => {
+    const hasRect = !!fields?.find(
+      (f) => f.name === 'fechaPagoRectificacion' && f.found,
+    );
+    if (hasRect) return 'pedimento_completo_xml';
+
+    const paymentDate =
+      fields?.find((f) => f.name === 'paymentDate' && f.found)?.value ?? null;
+    return paymentDate ? 'pedimento_completo_xml' : 'proforma_completo_xml';
+  },
+
+  /**
+   * Compose the 15-digit pedimento number from XML body + filename.
+   *   YY: priority order (most authoritative first):
+   *     1) Filename pattern 3 (`{15-digit}.xml`) — VUCEM stamps the correct
+   *        prefix at export time.
+   *     2) Fecha de presentacion (<fechas><clave>5) — the year the pedimento
+   *        was opened. Authoritative for the YY prefix even when payment
+   *        crosses calendar year (e.g. opened Dec-2025, paid Jan-2026 → YY=25).
+   *     3) Rectification fechaPago (only when no presentation date exists).
+   *     4) Payment date (last-resort fallback).
+   *   AA: from <aduanaEntradaSalida><clave> padded to 2.
+   *   PPPP: from the filename (any of the three patterns).
+   *   NNNNNNN: from <pedimento> padded to 7.
+   * Returns null if any component cannot be resolved.
+   */
+  extractNumPedimento: (source, fields, filePath) => {
+    const parts = parseFilenameParts(filePath);
+
+    const presentation = fields?.find(
+      (f) => f.name === 'presentationDate' && f.found,
+    )?.value;
+    const rect = fields?.find(
+      (f) => f.name === 'fechaPagoRectificacion' && f.found,
+    )?.value;
+    const pay = fields?.find((f) => f.name === 'paymentDate' && f.found)?.value;
+
+    let yy =
+      (parts && parts.year) ||
+      yyFromIsoDate(presentation) ||
+      yyFromIsoDate(rect) ||
+      yyFromIsoDate(pay) ||
+      null;
+
+    const aduanaField = fields?.find(
+      (f) => f.name === 'aduanaEntradaSalida',
+    )?.value;
+    let aduana = aduanaField || (parts && parts.aduana) || null;
+
+    let patente = parts && parts.patente ? parts.patente : null;
+
+    const pedimentoBody = firstTag(source, 'pedimento');
+    let pedimento = pedimentoBody
+      ? pad(pedimentoBody, 7)
+      : parts && parts.pedimento
+        ? parts.pedimento
+        : null;
+
+    if (!yy || !aduana || !patente || !pedimento) return null;
+
+    return `${pad(yy, 2)}${pad(aduana, 2)}${pad(patente, 4)}${pedimento}`;
+  },
+
+  extractPedimentoYear: (source, fields, filePath) => {
+    // Reuse extractNumPedimento; the year is its leading 2 digits.
+    const num = pedimentoCompletoXmlDefinition.extractNumPedimento(
+      source,
+      fields,
+      filePath,
+    );
+    if (!num) return null;
+    const year = parseInt(num.substring(0, 2), 10);
+    return year < 50 ? year + 2000 : year + 1900;
+  },
+
+  extractors: [
+    rfcExtractor,
+    clavePedimentoExtractor,
+    tipoOperacionExtractor,
+    aduanaEntradaSalidaExtractor,
+    paymentDateExtractor,
+    presentationDateExtractor,
+    fechaPagoRectificacionExtractor,
+    coveExtractor,
+    numEDocumentoExtractor,
+  ],
+};

package/.claude/worktrees/agent-multi-profile/src/document-types/pedimento-completo.js

@@ -0,0 +1,108 @@
+import {
+  pedimentoYearFromFields,
+  sharedPedimentoExtractors,
+} from './_pedimento-shared-extractors.js';
+
+/**
+ * "Pedimento Completo" matcher — the 7-page DEF / SEGUNDA / TERCERA copy
+ * (and the "CoveFact" variant). This is distinct from `pedimento_simplificado`
+ * which uses the "FORMA SIMPLIFICADA DE PEDIMENTO" header.
+ *
+ * Match strategy: require the structural fields that the long-form pedimento
+ * always carries (`NUM. PEDIMENTO:`, `CVE.PEDIMENTO:`, `T.OPER:`) plus at
+ * least one of the printed copy markers, while explicitly excluding any
+ * document that already declares itself as a "FORMA SIMPLIFICADA".
+ */
+export const pedimentoCompletoDefinition = {
+  type: 'pedimento_completo',
+  extensions: ['pdf'],
+  match: (source) => {
+    // Hard exclude: "FORMA SIMPLIFICADA [DE|DEL] PEDIMENTO" is handled by
+    // pedimento_simplificado.
+    if (/FORMA\s+SIMPLIFICADA\s+DEL?\s+PEDIMENTO/i.test(source)) return false;
+
+    // Hard exclude: "AVISO CONSOLIDADO" shares the header trio but is a
+    // different document type handled by aviso_consolidado.
+    if (/AVISO\s+CONSOLIDADO/i.test(source)) return false;
+
+    // The colon after "T. OPER" is optional — see note in pedimento-simplificado.js.
+    const hasHeaderFields =
+      /NUM\.?\s*PEDIMENTO:/i.test(source) &&
+      /CVE\.?\s*PEDIMENTO:/i.test(source) &&
+      /T\.?\s*OPER:?/i.test(source);
+    if (hasHeaderFields) {
+      const hasCopyMarker =
+        /ORIGINAL:\s*ADMINISTRACION GENERAL DE ADUANAS/i.test(source) ||
+        /SEGUNDA\s+COPIA/i.test(source) ||
+        /TERCERA\s+COPIA/i.test(source) ||
+        /COPIA\s+(SIMPLIFICAD[AO])?\s*TRANSPORTISTA/i.test(source) ||
+        /DEFINITIVO/i.test(source) ||
+        /ANEXO\s+DEL\s+PEDIMENTO/i.test(source) ||
+        /\*+FIN\s+DE\s+PEDIMENTO\s*\*+/i.test(source);
+      if (hasCopyMarker) return true;
+    }
+
+    // Fallback clue-counting heuristic for exotic layouts.
+    const clues = [
+      /\bPEDIMENTO\s*\n.*NUM\.\s*PEDIMENTO:/i,
+      /NUM\.\s*PEDIMENTO:\s*T\.OPER:\s*CVE\.PEDIMENTO:\s*REGIMEN:/i,
+      /\d{2}\s+\d{2}\s+\d{4}\s+\d{7}\s+[A-Z]{3}\s+[A-Z]{3}/i,
+      /CERTIFICACIONES/i,
+      /DESTINO\/ORIGEN\s+TIPO\s+CAMBIO\s+PESO\s+BRUTO\s+ADUANA\s+E\/S/i,
+      /MEDIOS\s+DE\s+TRANSPORTE/i,
+      /DATOS\s+DEL\s+IMPORTADOR\/EXPORTADOR/i,
+      /RFC:\s+NOMBRE,\s+DENOMINACION\s+O\s+RAZON\s+SOCIAL:/i,
+      /CUADRO\s+DE\s+LIQUIDACION/i,
+      /\*\*\*\s+PAGO\s+ELECTRONICO\s+\*\*\*/i,
+      /PATENTE:\s+PEDIMENTO:\s+ADUANA:/i,
+      /LINEA\s+DE\s+CAPTURA:/i,
+      /DATOS\s+DEL\s+PROVEEDOR\s+O\s+COMPRADOR/i,
+      /CLAVE\/COMPL\.\s+IDENTIFICADOR/i,
+      /ANEXO\s+DEL\s+PEDIMENTO/i,
+      /\*+FIN\s+DE\s+PEDIMENTO\s+\*+NUM\.\s+TOTAL\s+DE\s+PARTIDAS:/i,
+      /DECLARO\s+BAJO\s+PROTESTA\s+DE\s+DECIR\s+VERDAD/i,
+      /PEDIMENTO\s+ELABORADO\s+DE\s+CONFORMIDAD/i,
+    ];
+
+    const found = clues.filter((clue) => clue.test(source));
+    return found.length > clues.length * 0.25;
+  },
+
+  /**
+   * Resolve the final document type after fields have been extracted.
+   * Mirrors the simplificado logic:
+   *   - R1 rectifications require fechaPagoRectificacion
+   *   - Everything else requires paymentDate
+   * No payment evidence ⇒ proforma_completo.
+   *
+   * Exception: the "CoveFact" annex (…-CoveFact.pdf) doesn't always print the
+   * payment section, so the payment-based demotion misfires and the file gets
+   * stuck as proforma_completo (NON_PUSHABLE) forever — the CLI never
+   * re-identifies rows that already have a detected_type. The filename is the
+   * payment-independent signal for this variant.
+   */
+  resolveType: (fields, filePath) => {
+    if (/covefact/i.test(filePath ?? '')) return 'pedimento_completo';
+    const clavePedimento =
+      fields?.find((f) => f.name === 'clavePedimento')?.value ?? null;
+    const paymentDate =
+      fields?.find((f) => f.name === 'paymentDate' && f.found)?.value ?? null;
+    const fechaPagoRectificacion =
+      fields?.find((f) => f.name === 'fechaPagoRectificacion' && f.found)
+        ?.value ?? null;
+
+    if (clavePedimento === 'R1') {
+      return fechaPagoRectificacion
+        ? 'pedimento_completo'
+        : 'proforma_completo';
+    }
+    return paymentDate ? 'pedimento_completo' : 'proforma_completo';
+  },
+
+  extractNumPedimento: (source, fields) => {
+    return fields?.find((f) => f.name === 'numPedimento')?.value ?? null;
+  },
+  extractPedimentoYear: (source, fields) => pedimentoYearFromFields(fields),
+
+  extractors: sharedPedimentoExtractors,
+};

package/.claude/worktrees/agent-multi-profile/src/document-types/pedimento-simplificado.js

@@ -0,0 +1,76 @@
+import {
+  pedimentoYearFromFields,
+  sharedPedimentoExtractors,
+} from './_pedimento-shared-extractors.js';
+
+export const pedimentoSimplificadoDefinition = {
+  type: 'pedimento_simplificado',
+  extensions: ['pdf'],
+  match: (source) => {
+    // Hard exclude: "AVISO CONSOLIDADO" shares the header trio but is a
+    // different document type handled by aviso_consolidado.
+    if (/AVISO\s+CONSOLIDADO/i.test(source)) return false;
+
+    // Fast path: the literal title appears on standard SIMP layouts.
+    // Some prevalidators print "FORMA SIMPLIFICADA DEL PEDIMENTO" (with DEL).
+    if (/FORMA\s+SIMPLIFICADA\s+DEL?\s+PEDIMENTO/i.test(source)) return true;
+
+    // Some PDFs (single-page anchors) lack that title but still carry the
+    // three pedimento header fields. Treat them as simplificado UNLESS they
+    // have the multi-page copy markers that uniquely identify a completo.
+    // NOTE: the colon after "T. OPER" is optional — many printable layouts
+    // render OPER as a table-header label with the value in the next cell.
+    const hasHeaderFields =
+      /NUM\.?\s*PEDIMENTO:/i.test(source) &&
+      /CVE\.?\s*PEDIMENTO:/i.test(source) &&
+      /T\.?\s*OPER:?/i.test(source);
+    if (!hasHeaderFields) return false;
+
+    const hasCompletoCopyMarker =
+      /ORIGINAL:\s*ADMINISTRACION GENERAL DE ADUANAS/i.test(source) ||
+      /SEGUNDA\s+COPIA/i.test(source) ||
+      /TERCERA\s+COPIA/i.test(source) ||
+      /COPIA\s+(SIMPLIFICAD[AO])?\s*TRANSPORTISTA/i.test(source) ||
+      /ANEXO\s+DEL\s+PEDIMENTO/i.test(source) ||
+      /\*+FIN\s+DE\s+PEDIMENTO\s*\*+/i.test(source);
+    if (hasCompletoCopyMarker) return false;
+
+    // Exclude COVE/eDocument forms that may reference a pedimento in their body.
+    if (/COMPROBANTE\s+DE\s+VALOR\s+ELECTR[ÓO]NICO/i.test(source)) return false;
+    if (/\bCOVE\b\s*:/i.test(source) && !/PAGO/i.test(source)) return false;
+
+    return true;
+  },
+
+  /**
+   * Resolve the final document type after fields have been extracted.
+   * A document is a "pedimento_simplificado" only if it has been paid:
+   *   - For clavePedimento = 'R1': must have fechaPagoRectificacion
+   *   - Otherwise: must have paymentDate
+   * If no payment evidence is found, it's a "proforma".
+   *
+   * @param {import('../document-type-shared.js').FieldResult[]} fields
+   * @returns {string} - 'pedimento_simplificado' or 'proforma'
+   */
+  resolveType: (fields) => {
+    const clavePedimento =
+      fields?.find((f) => f.name === 'clavePedimento')?.value ?? null;
+    const paymentDate =
+      fields?.find((f) => f.name === 'paymentDate' && f.found)?.value ?? null;
+    const fechaPagoRectificacion =
+      fields?.find((f) => f.name === 'fechaPagoRectificacion' && f.found)
+        ?.value ?? null;
+
+    if (clavePedimento === 'R1') {
+      return fechaPagoRectificacion ? 'pedimento_simplificado' : 'proforma';
+    }
+    return paymentDate ? 'pedimento_simplificado' : 'proforma';
+  },
+
+  extractNumPedimento: (source, fields) => {
+    return fields?.find((f) => f.name === 'numPedimento')?.value ?? null;
+  },
+  extractPedimentoYear: (source, fields) => pedimentoYearFromFields(fields),
+
+  extractors: sharedPedimentoExtractors,
+};

package/.claude/worktrees/agent-multi-profile/src/document-types/proforma.js

@@ -0,0 +1,29 @@
+import { FieldResult } from '../document-type-shared.js';
+import { pedimentoSimplificadoDefinition } from './pedimento-simplificado.js';
+
+/**
+ * Proforma Document Type Definition
+ *
+ * A proforma is essentially the same document as a pedimento simplificado,
+ * but it has NOT been paid yet. It shares the same extractors and structure.
+ *
+ * Classification logic:
+ * - If clavePedimento = 'R1': needs fechaPagoRectificacion to be pedimento_simplificado
+ * - Otherwise: needs paymentDate to be pedimento_simplificado
+ * - If neither payment field is found → it's a proforma
+ */
+export const proformaDefinition = {
+  type: 'proforma',
+  extensions: ['pdf'],
+
+  // Same content marker as pedimento simplificado (accepts "DE" or "DEL").
+  match: (source) => {
+    return /FORMA\s+SIMPLIFICADA\s+DEL?\s+PEDIMENTO/i.test(source);
+  },
+
+  extractNumPedimento: pedimentoSimplificadoDefinition.extractNumPedimento,
+  extractPedimentoYear: pedimentoSimplificadoDefinition.extractPedimentoYear,
+
+  // Reuse all extractors from pedimento simplificado
+  extractors: pedimentoSimplificadoDefinition.extractors,
+};