@arela/uploader 1.0.21 → 1.0.23

package/src/document-types/_pedimento-shared-extractors.js

@@ -25,49 +25,118 @@ export const tipoOperacionExtractor = {
   },
 };
 
-// 3) Clave de Pedimento: 2 chars after tipoOperacion
+// 3) Clave de Pedimento: 2 chars after tipoOperacion (multiple layout patterns)
 export const clavePedimentoExtractor = {
   field: 'clavePedimento',
   extract: (source) => {
-    const match = source.match(
-      /\d{2}\s+\d{2}\s+\d{4}\s+\d{7}\s+[A-Z]{3}\s+([A-Z0-9]{2})/,
-    );
-    return new FieldResult('clavePedimento', !!match, match ? match[1] : null);
+    const patterns = [
+      // Standard spaced layout: "22 07 3429 2002089 EXP RT"
+      /\d{2}\s+\d{2}\s+\d{4}\s+\d{7}\s+[A-Z]{3}\s+([A-Z0-9]{2})\b/,
+      // Concatenated 15-digit layout: "260734296013645 EXP RT"
+      /\d{15}\s+[A-Z]{3}\s+([A-Z0-9]{2})\b/,
+      // Fallback: T.OPER keyword followed by 2-char clave
+      /\b(?:EXP|IMP|TRA|TRN)\s+([A-Z][A-Z0-9])\b/,
+      // Explicit label
+      /CVE\.?\s*PED(?:IMENTO)?[^A-Z0-9]{0,60}?\b([A-Z][A-Z0-9])\b/i,
+    ];
+    for (const re of patterns) {
+      const m = source.match(re);
+      if (m) return new FieldResult('clavePedimento', true, m[1]);
+    }
+    return new FieldResult('clavePedimento', false, null);
   },
 };
 
 // 4) Aduana E/S: 3-digit code on the peso-bruto line
+//    Fallback A: allow 2-digit code (some SIMP layouts omit the leading zero).
+//    Fallback B: derive the 2-digit customs-office code from positions 2-3 of
+//    numPedimento (e.g. "260734296013645" → "07"), which is what the
+//    arela_path formula uses after padStart(2,'0').
 export const aduanaEntradaSalidaExtractor = {
   field: 'aduanaEntradaSalida',
   extract: (source) => {
-    const match = source.match(/^\s*\d+\s+[\d,.]+\s+(\d{3})\s*$/m);
-    return new FieldResult(
-      'aduanaEntradaSalida',
-      !!match,
-      match ? match[1] : null,
-    );
+    // Primary: 3-digit aduana code at end of peso-bruto line
+    const m3 = source.match(/^\s*\d+\s+[\d,.]+\s+(\d{3})\s*$/m);
+    if (m3) return new FieldResult('aduanaEntradaSalida', true, m3[1]);
+
+    // Fallback A: 2-digit aduana code at end of peso-bruto line
+    const m2 = source.match(/^\s*\d+\s+[\d,.]+\s+(\d{2})\s*$/m);
+    if (m2) return new FieldResult('aduanaEntradaSalida', true, m2[1]);
+
+    // Fallback B: derive 2-digit customs-office code from numPedimento
+    //   Format: AA BB CCCC DDDDDDD  →  BB (positions 2-3) = aduana
+    const pedMatch = source.match(/\d{2}\s?\d{2}\s?\d{4}\s?\d{7}/);
+    if (pedMatch) {
+      const num = pedMatch[0].replace(/\s/g, '');
+      if (num.length === 15) {
+        return new FieldResult(
+          'aduanaEntradaSalida',
+          true,
+          num.substring(2, 4),
+        );
+      }
+    }
+
+    return new FieldResult('aduanaEntradaSalida', false, null);
   },
 };
 
-// 5) RFC: 12–13 alphanumeric chars on its own line
+// 5) RFC: importer/exporter RFC on its own line.
+//    Strategy A: strict whole-line pattern (3-4 letters + 6 consecutive digits +
+//    3 alphanum). COVE codes like COVE2681B1RX8 naturally fail this because
+//    their digit section is non-consecutive (2681B1 has a letter at pos 5).
+//    Strategy B: RFC as a word within a longer line (handles "RFC: IMS030409FZ0").
+//    Strategy C: loose 12-13 alphanum isolated on its own line — iterate ALL
+//    matches via matchAll() so that a leading COVE code is skipped and the
+//    actual RFC (which appears later in the document) is still found.
 export const rfcExtractor = {
   field: 'rfc',
   extract: (source) => {
-    const match = source.match(/\n\s*([A-Z0-9]{12,13})\s*\n/);
-    return new FieldResult('rfc', !!match, match ? match[1] : null);
+    const RFC_STRICT = /^[A-Z]{3,4}\d{6}[A-Z0-9]{3}$/i;
+    const lines = source
+      .split(/\r?\n/)
+      .map((l) => l.trim())
+      .filter((l) => l);
+
+    // Primary: RFC occupies an entire trimmed line
+    const strictLine = lines.find((line) => RFC_STRICT.test(line));
+    if (strictLine) return new FieldResult('rfc', true, strictLine);
+
+    // Fallback A: RFC embedded in a longer line (word-boundary search)
+    for (const line of lines) {
+      const m = line.match(/\b([A-Z]{3,4}\d{6}[A-Z0-9]{3})\b/i);
+      if (m) return new FieldResult('rfc', true, m[1]);
+    }
+
+    // Fallback B: loose 12-13 alphanum isolated on its own line.
+    // Use matchAll() to iterate ALL occurrences — a leading COVE code must not
+    // short-circuit the search; the RFC typically follows it in the document.
+    for (const m of source.matchAll(/\n\s*([A-Z0-9]{12,13})\s*\n/g)) {
+      if (!/^COVE/i.test(m[1])) return new FieldResult('rfc', true, m[1]);
+    }
+
+    return new FieldResult('rfc', false, null);
   },
 };
 
-// 6) Código de Aceptación: 8 alphanumeric chars on the line right after the RFC
+// 6) Código de Aceptación: 8 alphanumeric chars on the line right after the RFC.
+//    Uses the same RFC-line detection logic as rfcExtractor.
 export const codigoAceptacionExtractor = {
   field: 'codigoAceptacion',
   extract: (source) => {
+    const RFC_STRICT = /^[A-Z]{3,4}\d{6}[A-Z0-9]{3}$/i;
+    const RFC_LOOSE = /^[A-Z0-9]{12,13}$/;
     const lines = source
       .split(/\r?\n/)
       .map((l) => l.trim())
       .filter((l) => l.length > 0);
 
-    const rfcIndex = lines.findIndex((l) => /^[A-Z0-9]{12,13}$/.test(l));
+    // Find RFC line using strict pattern first, then loose (excluding COVE)
+    let rfcIndex = lines.findIndex((l) => RFC_STRICT.test(l));
+    if (rfcIndex < 0) {
+      rfcIndex = lines.findIndex((l) => RFC_LOOSE.test(l) && !/^COVE/i.test(l));
+    }
+
     let code = null;
     if (rfcIndex >= 0 && /^[A-Z0-9]{8}$/.test(lines[rfcIndex + 1] || '')) {
       code = lines[rfcIndex + 1];
@@ -77,24 +146,31 @@ export const codigoAceptacionExtractor = {
 };
 
 // 7) Num. E-Document: collects all 13-char alphanumeric codes following
-//    `NUM. E-DOCUMENT` labels. CoveFact / Pedimento Completo emit one row
-//    per ED clave inside the CLAVE/COMPL. IDENTIFICADOR table.
+//    `NUM. E-DOCUMENT` / `NUMERO DE E-DOCUMENT` labels.
 export const numEDocumentoExtractor = {
   field: 'numEDocumento',
   extract: (source) => {
     const lines = source.split(/\r?\n/);
-    const edocLines = lines.filter((line) => /NUM\.?\s*E-DOCUMENT/i.test(line));
+    const extractedCodes = [];
+    const titlePatterns = [/NUMERO\s+DE\s+E-DOCUMENT/i, /NUM\.?\s*E-DOCUMENT/i];
 
-    if (edocLines.length === 0) {
-      return new FieldResult('numEDocumento', false, null);
-    }
+    for (let i = 0; i < lines.length; i++) {
+      const line = lines[i];
+      const hasTitle = titlePatterns.some((p) => p.test(line));
+      if (!hasTitle) continue;
 
-    const extractedCodes = [];
-    edocLines.forEach((line) => {
-      const afterEdoc = line.replace(/.*NUM\.?\s*E-DOCUMENT\s*/i, '');
-      const codes = afterEdoc.match(/[A-Z0-9]{13}/g) || [];
-      extractedCodes.push(...codes);
-    });
+      // Codes on the title line itself
+      const codesInLine = line.match(/[A-Z0-9]{13}/g) || [];
+      extractedCodes.push(...codesInLine);
+
+      // Codes on the next few lines (e.g. CLAVE/COMPL. table rows)
+      for (let j = 1; j <= 10 && i + j < lines.length; j++) {
+        const nextLine = lines[i + j];
+        if (/NUMERO|OBSERVACIONES/i.test(nextLine)) break;
+        const codesInNextLine = nextLine.match(/[A-Z0-9]{13}/g) || [];
+        extractedCodes.push(...codesInNextLine);
+      }
+    }
 
     if (extractedCodes.length === 0) {
       return new FieldResult('numEDocumento', false, null);
@@ -109,14 +185,37 @@ export const numEDocumentoExtractor = {
 export const paymentDateExtractor = {
   field: 'paymentDate',
   extract: (source) => {
-    let match = source.match(/2\s+PAGO:\s*(\d{2}\/\d{2}\/\d{4})/);
-    if (!match) {
-      match = source.match(/FECHA DE PAGO:\s*(\d{4}\/\d{2}\/\d{2})/);
+    const patterns = [
+      /FECHA\s+DE\s+PAGO:?\s*(\d{2}\/\d{2}\/\d{4})/i, // 0: explicit label DD/MM/YYYY
+      /FECHA\s+DE\s+PAGO:?\s*(\d{4}\/\d{2}\/\d{2})/i, // 1: explicit label YYYY/MM/DD
+      /2\s+PAGO:\s*(\d{2}\/\d{2}\/\d{4})/, // 2: forma simplificada scheduled date ⚠️
+      /(?:^|\n)\s*PAGO\s+(\d{2}\/\d{2}\/\d{4})/i, // 3: PAGO at line start (original)
+      /(?<=\d)PAGO\s+(\d{2}\/\d{2}\/\d{4})/i, // 4: PAGO after digit (pdf-parse artifact)
+      /(\d{2}\/\d{2}\/\d{4})[ \t]+PAGO[ \t]*$/im, // 5: reversed layout — date before PAGO (FECHAS column)
+      // 6: forma simplificada — pdf-parse extracts table cells out of order, so the
+      // label "FECHA DE PAGO:" can appear on its own line and the value (along with
+      // other cells like línea de captura, pedimento, importe) follows several lines
+      // later. Take the FIRST dd/mm/yyyy after the label within a 400-char window.
+      // Safe because `isNoPagado` short-circuits documents without a real payment,
+      // so we won't grab the unrelated ENTRADA date from the "FECHAS:" block above.
+      /FECHA\s+DE\s+PAGO:[\s\S]{1,400}?(\d{2}\/\d{2}\/\d{4})/i,
+      /PRESENTACION:\s*(\d{2}\/\d{2}\/\d{4})/i, // 7: fallback
+    ];
+    // "*** NO PAGADO" is the explicit SAT marker that no payment has been
+    // certified. When present, the bank-certification block is physically
+    // absent, so any date matched by the fallback patterns (e.g.
+    // "2 PAGO:" with a scheduled date, or "PRESENTACION:") would be a false
+    // positive. Return null outright — the document is classified as proforma.
+    const isNoPagado = /\*{3}\s*NO\s+PAGADO/i.test(source);
+    if (isNoPagado) {
+      return new FieldResult('paymentDate', false, null);
     }
-    if (!match) {
-      match = source.match(/PRESENTACION:\s*(\d{2}\/\d{2}\/\d{4})/);
+    for (const pattern of patterns) {
+      const m = source.match(pattern);
+      if (!m) continue;
+      return new FieldResult('paymentDate', true, m[1]);
     }
-    return new FieldResult('paymentDate', !!match, match ? match[1] : null);
+    return new FieldResult('paymentDate', false, null);
   },
 };
 
@@ -152,14 +251,16 @@ export const coveExtractor = {
 };
 
 // 10) Patente: from the PATENTE/PEDIMENTO/ADUANA header table
+//    Fallback A: pago electrónico line "3429 4024126 07" (pedimento_completo).
+//    Fallback B: positions 4-7 of numPedimento (always available when found).
 export const patenteExtractor = {
   field: 'patente',
   extract: (source) => {
+    // Primary: PATENTE:/PEDIMENTO:/ADUANA: header followed by data line
     const lines = source.split(/\r?\n/);
     const patenteHeaderIndex = lines.findIndex((line) =>
       /PATENTE:.*PEDIMENTO:.*ADUANA:/i.test(line),
     );
-
     if (patenteHeaderIndex >= 0) {
       for (let i = patenteHeaderIndex + 1; i < lines.length; i++) {
         const line = lines[i].trim();
@@ -169,6 +270,20 @@ export const patenteExtractor = {
         }
       }
     }
+
+    // Fallback A: pago electrónico line "3429 4024126 07"
+    const pagoMatch = source.match(/(\d{4})\s+\d{7}\s+\d{2}/);
+    if (pagoMatch) return new FieldResult('patente', true, pagoMatch[1]);
+
+    // Fallback B: positions 4-7 of numPedimento
+    const pedMatch = source.match(/\d{2}\s?\d{2}\s?\d{4}\s?\d{7}/);
+    if (pedMatch) {
+      const num = pedMatch[0].replace(/\s/g, '');
+      if (num.length === 15) {
+        return new FieldResult('patente', true, num.substring(4, 8));
+      }
+    }
+
     return new FieldResult('patente', false, null);
   },
 };

package/src/document-types/factura-inter-agencia.js

@@ -0,0 +1,186 @@
+// NOTE: We intentionally do NOT import `FieldResult` from
+// '../document-type-shared.js' to avoid a circular-import TDZ when this
+// module is imported directly (e.g. from unit tests). `FieldResult` is a
+// plain data-class with shape `{ name, found, value }`, so we construct
+// equivalent plain objects locally.
+const fieldResult = (name, found, value) => ({ name, found, value });
+
+/**
+ * Factura Inter-Agencia Document Type Definition
+ *
+ * Detects CFDIs (XML or PDF) issued between customs broker agencies (e.g.,
+ * NORCOM ↔ PALCO). These files are dropped into a pedimento folder by the
+ * broker but they are NOT part of the customs electronic file (expediente
+ * aduanal) — they are inter-agency billing for broker services.
+ *
+ * Detection rules (ALL required):
+ *   1) CFDI markers present (either xml structure or PDF text representation)
+ *   2) Both emisor and receptor RFCs belong to the configured agency pair
+ *      (NAA120215F20 = NORCOM, PCC1008161WA = PALCO) in any direction.
+ *   3) At least one concepto with ClaveProdServ 78141502 (Servicios de
+ *      agentes aduaneros) — confirms the billing is for broker services.
+ *
+ * IMPORTANT: This matcher MUST be registered BEFORE `facturasComerciales`
+ * in document-type-shared.js — both would match a CFDI in a pedimento
+ * folder, but inter-agency invoices must take precedence so they are
+ * filtered out of the Arela push pipeline (see arela-api
+ * NON_PUSHABLE_TYPES_SQL).
+ *
+ * Currently scope-limited to NORCOM↔PALCO. To widen, move INTER_AGENCIA_RFCS
+ * to env config and require ≥2 distinct RFCs from the configured list.
+ */
+
+/**
+ * RFCs of agencies whose mutual invoices should be excluded from the Arela
+ * push pipeline. Order is irrelevant — a match is any pair of distinct RFCs
+ * from this set appearing as emisor and receptor.
+ */
+export const INTER_AGENCIA_RFCS = ['NAA120215F20', 'PCC1008161WA'];
+
+const BROKER_SERVICE_CLAVE_PROD_SERV = '78141502';
+
+const CFDI_XML_MARKERS = [
+  /cfdi:Comprobante/i,
+  /xmlns:cfdi/i,
+  /TipoDeComprobante/i,
+];
+
+/**
+ * Detect that the source represents a CFDI — either as the original XML
+ * structure or as text extracted from a printed CFDI (PDF representation).
+ *
+ * PDF text loses XML tags, so we look for the human-readable equivalents
+ * commonly rendered by SAT-style invoice templates ("Folio Fiscal", "Sello
+ * Digital del CFDI", "Cadena Original ... Certificacion Digital del SAT").
+ */
+function isCfdiContent(source) {
+  const xmlHits = CFDI_XML_MARKERS.filter((re) => re.test(source)).length;
+  if (xmlHits >= 2) return true;
+
+  const pdfMarkers = [
+    /folio\s*fiscal/i,
+    /sello\s*digital\s*del\s*cfdi/i,
+    /cadena\s*original.*certificaci[oó]n\s*digital\s*del\s*sat/i,
+    /representaci[oó]n\s*impresa\s*de\s*un\s*cfdi/i,
+  ];
+  return pdfMarkers.filter((re) => re.test(source)).length >= 2;
+}
+
+/**
+ * Return the subset of INTER_AGENCIA_RFCS that appear in `source`. Matching is
+ * case-insensitive and uses word boundaries so substrings inside larger tokens
+ * (cert/sello base64) don't produce false positives.
+ */
+function findInterAgenciaRfcs(source) {
+  const found = new Set();
+  for (const rfc of INTER_AGENCIA_RFCS) {
+    const re = new RegExp(`\\b${rfc}\\b`, 'i');
+    if (re.test(source)) found.add(rfc.toUpperCase());
+  }
+  return [...found];
+}
+
+export const facturaInterAgenciaDefinition = {
+  type: 'factura_inter_agencia',
+  extensions: ['xml', 'pdf'],
+
+  match: (source) => {
+    if (!isCfdiContent(source)) return false;
+
+    // Need ≥2 distinct configured RFCs present (one as emisor, one as receptor)
+    const rfcsFound = findInterAgenciaRfcs(source);
+    if (rfcsFound.length < 2) return false;
+
+    // Confirm the invoice is for broker services (customs agent services)
+    if (!source.includes(BROKER_SERVICE_CLAVE_PROD_SERV)) return false;
+
+    return true;
+  },
+
+  // Pedimento extraction is optional / informational — these files are
+  // excluded from push, so arela_path is never composed. We still extract
+  // a pedimento number when present (from the "Referencias" / "Pedimento:"
+  // section of the printable CFDI) for auditability.
+  extractNumPedimento: (source, fields) => {
+    return fields?.find((f) => f.name === 'numPedimento')?.value ?? null;
+  },
+
+  extractPedimentoYear: (source, fields) => {
+    const numPedimento = fields?.find((f) => f.name === 'numPedimento')?.value;
+    if (numPedimento && numPedimento.length >= 2) {
+      const yy = parseInt(numPedimento.substring(0, 2), 10);
+      if (!isNaN(yy)) return yy < 50 ? yy + 2000 : yy + 1900;
+    }
+    return null;
+  },
+
+  extractors: [
+    {
+      field: 'rfcEmisor',
+      extract: (source) => {
+        // XML form: <cfdi:Emisor Rfc="..." />
+        const xmlMatch = source.match(
+          /<[^>]*Emisor[^>]*Rfc\s*=\s*["']([A-ZÑ&]{3,4}\d{6}[A-Z0-9]{3})["']/i,
+        );
+        if (xmlMatch) return fieldResult('rfcEmisor', true, xmlMatch[1]);
+
+        // PDF form: "Emisor" section followed by RFC label/value on later lines.
+        // We pick the first INTER_AGENCIA RFC that appears in the document.
+        const rfcs = findInterAgenciaRfcs(source);
+        if (rfcs.length > 0) return fieldResult('rfcEmisor', true, rfcs[0]);
+
+        return fieldResult('rfcEmisor', false, null);
+      },
+    },
+    {
+      field: 'rfcReceptor',
+      extract: (source) => {
+        const xmlMatch = source.match(
+          /<[^>]*Receptor[^>]*Rfc\s*=\s*["']([A-ZÑ&]{3,4}\d{6}[A-Z0-9]{3})["']/i,
+        );
+        if (xmlMatch) return fieldResult('rfcReceptor', true, xmlMatch[1]);
+
+        const rfcs = findInterAgenciaRfcs(source);
+        if (rfcs.length >= 2) {
+          return fieldResult('rfcReceptor', true, rfcs[1]);
+        }
+        return fieldResult('rfcReceptor', false, null);
+      },
+    },
+    {
+      field: 'folio',
+      extract: (source) => {
+        // CFDI Folio attribute
+        const xmlMatch = source.match(/\bFolio\s*=\s*["']([A-Z0-9-]+)["']/i);
+        if (xmlMatch) return fieldResult('folio', true, xmlMatch[1]);
+
+        // PDF: "Numero Folio    012749"
+        const pdfMatch = source.match(/Numero\s+Folio\s+([A-Z0-9-]+)/i);
+        if (pdfMatch) return fieldResult('folio', true, pdfMatch[1]);
+
+        return fieldResult('folio', false, null);
+      },
+    },
+    {
+      field: 'uuid',
+      extract: (source) => {
+        const uuidRe =
+          /[0-9A-F]{8}-[0-9A-F]{4}-[0-9A-F]{4}-[0-9A-F]{4}-[0-9A-F]{12}/i;
+        const m = source.match(uuidRe);
+        return fieldResult('uuid', !!m, m ? m[0].toUpperCase() : null);
+      },
+    },
+    {
+      field: 'numPedimento',
+      extract: (source) => {
+        // Printable CFDI "Pedimento: 3458 6000046 Fecha: ..." — recovers an
+        // 11-digit pedimento (no YY prefix). Useful for auditability only.
+        const m = source.match(/Pedimento:?\s*(\d{4})\s*(\d{7})/i);
+        if (m) {
+          return fieldResult('numPedimento', true, `${m[1]}${m[2]}`);
+        }
+        return fieldResult('numPedimento', false, null);
+      },
+    },
+  ],
+};

package/src/document-types/pedimento-completo-xml.js

@@ -1,10 +1,8 @@
 // VUCEM "consultarPedimentoCompleto" XML matcher.
 //
-// STATUS: implemented but NOT registered in `document-type-shared.js`. To
-// activate, uncomment the import + registration in that file. All downstream
-// code (composeArelaPath, arela-api propagation SQL, IdentifyCommand
-// counters) already includes `pedimento_completo_xml`, so re-enabling is a
-// single-line change.
+// 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
@@ -47,6 +45,22 @@ function pad(value, length) {
   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.
@@ -102,12 +116,17 @@ function yyFromIsoDate(iso) {
   return m ? m[1].substring(2, 4) : null;
 }
 
-// Find <ns2:fechas> block with nested clave==2 and return its <ns2:fecha>.
-function findPaymentDate(source) {
+// 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 === '2') {
+    if (clave === claveValue) {
       const fecha = firstTag(block, 'fecha');
       if (fecha) return fecha;
     }
@@ -115,6 +134,18 @@ function findPaymentDate(source) {
   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 = {
@@ -152,7 +183,7 @@ const aduanaEntradaSalidaExtractor = {
     return new FieldResult(
       'aduanaEntradaSalida',
       !!clave,
-      clave ? pad(clave, 2) : null,
+      aduanaToSeccion(clave),
     );
   },
 };
@@ -165,6 +196,14 @@ const paymentDateExtractor = {
   },
 };
 
+const presentationDateExtractor = {
+  field: 'presentationDate',
+  extract: (source) => {
+    const fecha = findPresentationDate(source);
+    return new FieldResult('presentationDate', !!fecha, fecha);
+  },
+};
+
 const fechaPagoRectificacionExtractor = {
   field: 'fechaPagoRectificacion',
   extract: (source) => {
@@ -257,8 +296,14 @@ export const pedimentoCompletoXmlDefinition = {
 
   /**
    * Compose the 15-digit pedimento number from XML body + filename.
-   *   YY: from rectification fechaPago if present, else from the clave==2
-   *       payment-date fecha; falls back to filename pattern 3.
+   *   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.
@@ -267,15 +312,19 @@ export const pedimentoCompletoXmlDefinition = {
   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) ||
-      (parts && parts.year) ||
       null;
 
     const aduanaField = fields?.find(
@@ -315,6 +364,7 @@ export const pedimentoCompletoXmlDefinition = {
     tipoOperacionExtractor,
     aduanaEntradaSalidaExtractor,
     paymentDateExtractor,
+    presentationDateExtractor,
     fechaPagoRectificacionExtractor,
     coveExtractor,
     numEDocumentoExtractor,

package/src/document-types/pedimento-completo.js

@@ -17,22 +17,55 @@ export const pedimentoCompletoDefinition = {
   type: 'pedimento_completo',
   extensions: ['pdf'],
   match: (source) => {
-    if (/FORMA SIMPLIFICADA DE PEDIMENTO/i.test(source)) return false;
+    // 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) return false;
+      /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;
+    }
 
-    const hasCopyMarker =
-      /ORIGINAL:\s*ADMINISTRACION GENERAL DE ADUANAS/i.test(source) ||
-      /SEGUNDA COPIA/i.test(source) ||
-      /TERCERA COPIA/i.test(source) ||
-      /COPIA\s+(SIMPLIFICAD[AO])?\s*TRANSPORTISTA/i.test(source) ||
-      /DEFINITIVO/i.test(source);
+    // 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,
+    ];
 
-    return hasCopyMarker;
+    const found = clues.filter((clue) => clue.test(source));
+    return found.length > clues.length * 0.25;
   },
 
   /**