@arela/uploader 0.2.1 → 0.2.2

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@arela/uploader",
-  "version": "0.2.1",
+  "version": "0.2.2",
   "description": "CLI to upload files/directories to Arela",
   "bin": {
     "arela": "./src/index.js"

package/src/document-type-shared.js

@@ -1,3 +1,6 @@
+// Import all document type definitions
+import { pedimentoSimplificadoDefinition } from './document-types/pedimento-simplificado.js';
+
 // Document type definitions and extraction utilities
 // Ported from TypeScript to JavaScript for Node.js
 
@@ -10,7 +13,14 @@ export class FieldResult {
 }
 
 export class DocumentTypeDefinition {
-  constructor(type, extensions, match, extractors, extractNumPedimento, extractPedimentoYear) {
+  constructor(
+    type,
+    extensions,
+    match,
+    extractors,
+    extractNumPedimento,
+    extractPedimentoYear,
+  ) {
     this.type = type;
     this.extensions = extensions;
     this.match = match;
@@ -20,9 +30,6 @@ export class DocumentTypeDefinition {
   }
 }
 
-// Import all document type definitions
-import { pedimentoSimplificadoDefinition } from './document-types/pedimento-simplificado.js';
-
 // Registry of all document types
 const documentTypes = [
   pedimentoSimplificadoDefinition,
@@ -44,14 +51,17 @@ export function extractDocumentFields(source, fileExtension, filePath) {
   // Try to match against each document type
   for (const docType of documentTypes) {
     // Check if file extension matches
-    if (fileExtension && !docType.extensions.includes(fileExtension.toLowerCase())) {
+    if (
+      fileExtension &&
+      !docType.extensions.includes(fileExtension.toLowerCase())
+    ) {
       continue;
     }
 
     // Test if content matches this document type
     if (docType.match(source)) {
       console.log(`✅ Matched document type: ${docType.type}`);
-      
+
       // Extract all fields
       const fields = [];
       for (const extractor of docType.extractors) {
@@ -68,8 +78,12 @@ export function extractDocumentFields(source, fileExtension, filePath) {
       }
 
       // Extract pedimento number and year
-      const pedimento = docType.extractNumPedimento ? docType.extractNumPedimento(source, fields) : null;
-      const year = docType.extractPedimentoYear ? docType.extractPedimentoYear(source, fields) : null;
+      const pedimento = docType.extractNumPedimento
+        ? docType.extractNumPedimento(source, fields)
+        : null;
+      const year = docType.extractPedimentoYear
+        ? docType.extractPedimentoYear(source, fields)
+        : null;
 
       return [docType.type, fields, pedimento, year];
     }

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

@@ -33,7 +33,7 @@ export const pedimentoSimplificadoDefinition = {
         return new FieldResult(
           'numPedimento',
           !!match,
-          match ? match[0].replace(/\s/g, '') : null
+          match ? match[0].replace(/\s/g, '') : null,
         );
       },
     },
@@ -50,7 +50,7 @@ export const pedimentoSimplificadoDefinition = {
         return new FieldResult(
           'tipoOperacion',
           !!match,
-          match ? match[1] : null
+          match ? match[1] : null,
         );
       },
     },
@@ -67,7 +67,7 @@ export const pedimentoSimplificadoDefinition = {
         return new FieldResult(
           'clavePedimento',
           !!match,
-          match ? match[1] : null
+          match ? match[1] : null,
         );
       },
     },
@@ -83,7 +83,7 @@ export const pedimentoSimplificadoDefinition = {
         return new FieldResult(
           'aduanaEntradaSalida',
           !!match,
-          match ? match[1] : null
+          match ? match[1] : null,
         );
       },
     },
@@ -93,11 +93,7 @@ export const pedimentoSimplificadoDefinition = {
       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
-        );
+        return new FieldResult('rfc', !!match, match ? match[1] : null);
       },
     },
 
@@ -112,9 +108,7 @@ export const pedimentoSimplificadoDefinition = {
           .filter((l) => l.length > 0);
 
         // 2) find the index of an RFC line (12–13 alnum chars)
-        const rfcIndex = lines.findIndex((l) =>
-          /^[A-Z0-9]{12,13}$/.test(l),
-        );
+        const rfcIndex = lines.findIndex((l) => /^[A-Z0-9]{12,13}$/.test(l));
         let code = null;
 
         // 3) if next line exists and is exactly 8 alnum chars, that's the code
@@ -122,11 +116,7 @@ export const pedimentoSimplificadoDefinition = {
           code = lines[rfcIndex + 1];
         }
 
-        return new FieldResult(
-          'codigoAceptacion',
-          code !== null,
-          code
-        );
+        return new FieldResult('codigoAceptacion', code !== null, code);
       },
     },
 
@@ -175,11 +165,7 @@ export const pedimentoSimplificadoDefinition = {
         if (!match) {
           match = source.match(/PRESENTACION:\s*(\d{2}\/\d{2}\/\d{4})/);
         }
-        return new FieldResult(
-          'paymentDate',
-          !!match,
-          match ? match[1] : null
-        );
+        return new FieldResult('paymentDate', !!match, match ? match[1] : null);
       },
     },
 
@@ -224,11 +210,7 @@ export const pedimentoSimplificadoDefinition = {
       extract: (source) => {
         // Look for the peso bruto value with decimal format
         const match = source.match(/(\d+\.\d+)\d{3}/);
-        return new FieldResult(
-          'pesoBruto',
-          !!match,
-          match ? match[1] : null
-        );
+        return new FieldResult('pesoBruto', !!match, match ? match[1] : null);
       },
     },
 
@@ -268,7 +250,7 @@ export const pedimentoSimplificadoDefinition = {
         return new FieldResult(
           'numeroOperacionBancaria',
           !!match,
-          match ? match[1] : null
+          match ? match[1] : null,
         );
       },
     },
@@ -281,7 +263,7 @@ export const pedimentoSimplificadoDefinition = {
         return new FieldResult(
           'numeroTransaccionSAT',
           !!match,
-          match ? match[1] : null
+          match ? match[1] : null,
         );
       },
     },

package/src/file-detection.js

@@ -1,6 +1,7 @@
 import fs from 'fs';
-import path from 'path';
 import { getTextExtractor } from 'office-text-extractor';
+import path from 'path';
+
 import { extractDocumentFields } from './document-type-shared.js';
 
 const extractor = getTextExtractor();
@@ -10,15 +11,20 @@ const extractor = getTextExtractor();
  * Format: RFC/Year/Patente/Aduana/Pedimento/
  * Example: PED781129JT6/2023/3429/07/3019796/
  */
-function composeArelaPath(detectedType, fields, detectedPedimentoYear, filePath) {
+function composeArelaPath(
+  detectedType,
+  fields,
+  detectedPedimentoYear,
+  filePath,
+) {
   if (detectedType !== 'pedimento_simplificado') {
     return null;
   }
 
-  const rfc = fields?.find(f => f.name === 'rfc')?.value;
-  const patente = fields?.find(f => f.name === 'patente')?.value;
-  const aduana = fields?.find(f => f.name === 'aduanaEntradaSalida')?.value;
-  const pedimento = fields?.find(f => f.name === 'numPedimento')?.value;
+  const rfc = fields?.find((f) => f.name === 'rfc')?.value;
+  const patente = fields?.find((f) => f.name === 'patente')?.value;
+  const aduana = fields?.find((f) => f.name === 'aduanaEntradaSalida')?.value;
+  const pedimento = fields?.find((f) => f.name === 'numPedimento')?.value;
   const year = detectedPedimentoYear;
 
   // All components are required for a valid arela_path
@@ -28,17 +34,17 @@ function composeArelaPath(detectedType, fields, detectedPedimentoYear, filePath)
       year: !!year,
       patente: !!patente,
       aduana: !!aduana,
-      pedimento: !!pedimento
+      pedimento: !!pedimento,
     });
     return null;
   }
 
   // Ensure aduana is padded to 2 digits if needed (07 instead of 7)
   const aduanaFormatted = aduana.toString().padStart(2, '0');
-  
+
   // arela_path should be the folder structure only, without filename
   const arelaPath = `${rfc}/${year}/${patente}/${aduanaFormatted}/${pedimento}/`;
-  
+
   console.log(`✅ Composed arela_path: ${arelaPath}`);
   return arelaPath;
 }
@@ -48,7 +54,6 @@ function composeArelaPath(detectedType, fields, detectedPedimentoYear, filePath)
  * Detects document types and extracts metadata from files
  */
 export class FileDetectionService {
-  
   /**
    * Detect document type from a file
    * @param {string} filePath - Path to the file to analyze
@@ -56,13 +61,16 @@ export class FileDetectionService {
    */
   async detectFile(filePath) {
     try {
-      const fileExtension = path.extname(filePath).toLowerCase().replace('.', '');
+      const fileExtension = path
+        .extname(filePath)
+        .toLowerCase()
+        .replace('.', '');
       const fileName = path.basename(filePath);
-      
+
       console.log(`🔍 Analyzing file: ${fileName} (${fileExtension})`);
 
       let text = '';
-      
+
       // Extract text based on file type
       switch (fileExtension) {
         case 'pdf':
@@ -83,7 +91,7 @@ export class FileDetectionService {
             detectedPedimentoYear: null,
             arelaPath: null,
             text: '',
-            error: `Unsupported file type: ${fileExtension}`
+            error: `Unsupported file type: ${fileExtension}`,
           };
       }
 
@@ -96,16 +104,21 @@ export class FileDetectionService {
           detectedPedimentoYear: null,
           arelaPath: null,
           text: '',
-          error: 'No text could be extracted from file'
+          error: 'No text could be extracted from file',
         };
       }
 
       // Extract document fields and detect type
-      const [detectedType, fields, detectedPedimento, detectedPedimentoYear] = 
+      const [detectedType, fields, detectedPedimento, detectedPedimentoYear] =
         extractDocumentFields(text, fileExtension, filePath);
 
       // Compose arela_path for pedimento_simplificado documents
-      const arelaPath = composeArelaPath(detectedType, fields, detectedPedimentoYear, filePath);
+      const arelaPath = composeArelaPath(
+        detectedType,
+        fields,
+        detectedPedimentoYear,
+        filePath,
+      );
 
       return {
         detectedType,
@@ -114,9 +127,8 @@ export class FileDetectionService {
         detectedPedimentoYear,
         arelaPath,
         text,
-        error: null
+        error: null,
       };
-
     } catch (error) {
       console.error(`❌ Error detecting file ${filePath}:`, error.message);
       return {
@@ -126,7 +138,7 @@ export class FileDetectionService {
         detectedPedimentoYear: null,
         arelaPath: null,
         text: '',
-        error: error.message
+        error: error.message,
       };
     }
   }
@@ -139,13 +151,16 @@ export class FileDetectionService {
   async extractTextFromPDF(filePath) {
     try {
       const buffer = fs.readFileSync(filePath);
-      const text = await extractor.extractText({ 
-        input: buffer, 
-        type: 'file' 
+      const text = await extractor.extractText({
+        input: buffer,
+        type: 'file',
       });
       return text;
     } catch (error) {
-      console.error(`Error extracting text from PDF ${filePath}:`, error.message);
+      console.error(
+        `Error extracting text from PDF ${filePath}:`,
+        error.message,
+      );
       throw new Error(`Failed to extract text from PDF: ${error.message}`);
     }
   }
@@ -157,15 +172,15 @@ export class FileDetectionService {
    */
   async detectFiles(filePaths) {
     const results = [];
-    
+
     for (const filePath of filePaths) {
       const result = await this.detectFile(filePath);
       results.push({
         filePath,
-        ...result
+        ...result,
       });
     }
-    
+
     return results;
   }
 
@@ -186,7 +201,7 @@ export class FileDetectionService {
    * @returns {Array<string>} - Filtered array of supported file paths
    */
   filterSupportedFiles(filePaths) {
-    return filePaths.filter(filePath => this.isSupportedFileType(filePath));
+    return filePaths.filter((filePath) => this.isSupportedFileType(filePath));
   }
 }