@arela/uploader 1.1.1 → 1.1.3

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@arela/uploader",
-  "version": "1.1.1",
+  "version": "1.1.3",
   "description": "CLI to upload files/directories to Arela",
   "bin": {
     "arela": "./src/index.js"

package/src/config/config.js

@@ -37,10 +37,10 @@ class Config {
       const __dirname = path.dirname(__filename);
       const packageJsonPath = path.resolve(__dirname, '../../package.json');
       const packageJson = JSON.parse(fs.readFileSync(packageJsonPath, 'utf-8'));
-      return packageJson.version || '1.1.0';
+      return packageJson.version || '1.1.3';
     } catch (error) {
       console.warn('⚠️ Could not read package.json version, using fallback');
-      return '1.1.0';
+      return '1.1.3';
     }
   }
 

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

@@ -145,8 +145,15 @@ export const codigoAceptacionExtractor = {
   },
 };
 
-// 7) Num. E-Document: collects all 13-char alphanumeric codes following
-//    `NUM. E-DOCUMENT` / `NUMERO DE E-DOCUMENT` labels.
+// 7) Num. E-Document: collects all 13-char codes following
+//    `NUM. E-DOCUMENT` / `NUMERO DE E-DOCUMENT` labels. Real eDocument
+//    numbers START WITH A DIGIT (e.g. "0438261FSSN86"); MVE acuse folios
+//    ("MNVA26002Q3V5") are also listed under the same label and kept —
+//    downstream consumers filter them. The old bare [A-Z0-9]{13} also
+//    captured 13-letter words (TRANSPORTISTA), truncated tokens
+//    (IDENTIFICACIO[N], CURP prefixes) and RFCs.
+const EDOC_RE = /(?<![A-Z0-9])(?:\d[A-Z0-9]{12}|MNVA[A-Z0-9]{9})(?![A-Z0-9])/g;
+
 export const numEDocumentoExtractor = {
   field: 'numEDocumento',
   extract: (source) => {
@@ -160,14 +167,14 @@ export const numEDocumentoExtractor = {
       if (!hasTitle) continue;
 
       // Codes on the title line itself
-      const codesInLine = line.match(/[A-Z0-9]{13}/g) || [];
+      const codesInLine = line.match(EDOC_RE) || [];
       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) || [];
+        const codesInNextLine = nextLine.match(EDOC_RE) || [];
         extractedCodes.push(...codesInNextLine);
       }
     }

package/src/file-detection.js

@@ -5,6 +5,15 @@ import { PDFParse } from 'pdf-parse';
 import { extractDocumentFields } from './document-type-shared.js';
 import { classifyDocument } from './scoring/scoring-engine.js';
 
+// pdf.js can reject internal worker promises on corrupt PDFs (e.g.
+// XRefEntryException) that no try/catch around getText() can reach. Track the
+// last PDF handed to the parser so the global unhandledRejection handler can
+// name it; the event fires after finally blocks run, so it is never cleared.
+let currentPdfFile = null;
+function getCurrentPdfFile() {
+  return currentPdfFile;
+}
+
 // Document types that participate in arela_path composition.
 const ARELA_PATH_TYPES = new Set([
   'pedimento_simplificado',
@@ -255,6 +264,7 @@ export class FileDetectionService {
    */
   async extractTextFromPDF(filePath) {
     let parser;
+    currentPdfFile = filePath;
     try {
       const dataBuffer = fs.readFileSync(filePath);
       const uint8Array = new Uint8Array(dataBuffer);
@@ -316,4 +326,4 @@ export class FileDetectionService {
 }
 
 export default FileDetectionService;
-export { composeArelaPath };
+export { composeArelaPath, getCurrentPdfFile };

package/src/index.js

@@ -13,8 +13,21 @@ import watchCommand from './commands/WatchCommand.js';
 import workerCommand from './commands/WorkerCommand.js';
 import appConfig from './config/config.js';
 import ErrorHandler from './errors/ErrorHandler.js';
+import { getCurrentPdfFile } from './file-detection.js';
 import logger from './services/LoggingService.js';
 
+// Exception names thrown by pdf.js (used via pdf-parse) when a PDF is
+// malformed, encrypted, or truncated. Matched by error.name because the
+// classes live inside pdfjs-dist and are not exported by pdf-parse.
+const PDFJS_PARSE_ERRORS = new Set([
+  'XRefEntryException',
+  'XRefParseException',
+  'InvalidPDFException',
+  'MissingPDFException',
+  'PasswordException',
+  'FormatError',
+]);
+
 /**
  * Arela Uploader CLI
  * Professional file uploader with document detection and organization
@@ -660,6 +673,22 @@ class ArelaUploaderCLI {
     process.on('unhandledRejection', (reason, promise) => {
       const error =
         reason instanceof Error ? reason : new Error(String(reason));
+
+      // pdf.js rejects internal worker promises when a PDF is corrupt; those
+      // rejections cannot be caught around the parser call, and one bad file
+      // must not abort the rest of the batch.
+      if (PDFJS_PARSE_ERRORS.has(error.name)) {
+        const file = getCurrentPdfFile();
+        console.warn(
+          `⚠️ Corrupt or unreadable PDF${file ? ` (${file})` : ''}: ${error.message} — skipping`,
+        );
+        this.errorHandler.handleError(error, {
+          context: 'unhandledRejection',
+          file,
+        });
+        return;
+      }
+
       this.errorHandler.handleFatalError(error, {
         context: 'unhandledRejection',
         promise: promise.toString(),