Merge branch 'master' of github.com:profullstack/generate-pdf-api

Author: Preshy

README.md

@@ -401,6 +401,3 @@ node scripts/test-pdf-generation.js
 ```
 
 This script will generate a test PDF and output the detected Chrome path.
-
-test2
-

package.json

@@ -30,6 +30,7 @@
     "@supabase/supabase-js": "^2.49.4",
     "chalk": "^5.3.0",
     "commander": "^12.0.0",
+    "docx": "^9.5.0",
     "dotenv": "^16.5.0",
     "dotenv-flow": "^4.1.0",
     "form-data": "^4.0.2",
@@ -38,11 +39,9 @@
     "jsdom": "^26.1.0",
     "mailgun.js": "^12.0.1",
     "marked": "^15.0.8",
-    "pptxgenjs": "^3.12.0",
     "puppeteer": "^22.5.0",
     "qrcode": "^1.5.4",
     "stripe": "^14.22.0",
-    "turndown": "^7.2.0",
     "uuid": "^11.1.0",
     "ws": "^8.18.1",
     "xlsx": "^0.18.5"

pnpm-lock.yaml

@@ -81,7 +81,8 @@ export class ApiClient {
    * @param {boolean} store - Whether to store the document in Supabase
    * @returns {Promise<Blob>} - Word document blob
    */
-  static async htmlToDoc(html, filename = 'document.doc', store = false) {
+  static async htmlToDoc(html, filename = 'document.docx', store = false) {
+    console.log('Sending HTML to DOC conversion request with filename:', filename);
     return this.fetchBinaryResponse(`${this.baseUrl}/html-to-doc`, { 
       html, 
       filename, 

public/js/api-client.js

@@ -126,7 +126,7 @@ function updateFilenameExtension(documentType) {
   // Add new extension based on document type
   const extensions = {
     pdf: '.pdf',
-    doc: '.doc',
+    doc: '.docx',  // Updated from .doc to .docx
     excel: '.xlsx',
     ppt: '.pptx',
     epub: '.epub',
@@ -182,7 +182,9 @@ async function convertDocument() {
         break;
 
       case 'doc':
+        console.log('Converting to DOCX');
         result = await ApiClient.htmlToDoc(html, filename, storeDocument);
+        console.log('DOCX conversion successful');
         break;
 
       case 'excel':

public/js/views/dashboard.js

@@ -10,17 +10,20 @@ import { errorUtils } from '../utils/error-utils.js';
  */
 export async function htmlToDocHandler(c) {
   try {
-    const { html, filename = 'document.doc', store = false } = c.get('body');
+    const { html, filename = 'document.docx', store = false } = c.get('body');
 
-    // Generate Word document from HTML
-    const docBuffer = docService.generateDoc(html);
+    // Ensure filename has the correct extension
+    const finalFilename = filename.endsWith('.docx') ? filename : filename.replace(/\.doc$|$/, '.docx');
+    
+    // Generate Word document from HTML - now returns a Promise
+    const docBuffer = await docService.generateDoc(html);
 
     // Store the document in Supabase if requested
     if (store) {
       try {
         // Extract metadata from the request
         const metadata = {
-          contentType: 'application/msword',
+          contentType: 'application/vnd.openxmlformats-officedocument.wordprocessingml.document',
           userAgent: c.req.header('user-agent'),
           timestamp: new Date().toISOString()
         };
@@ -29,7 +32,7 @@ export async function htmlToDocHandler(c) {
         const userEmail = c.get('userEmail');
 
         // Store the document with user association and original HTML content
-        const result = await storageService.storeDoc(docBuffer, filename, metadata, userEmail, html);
+        const result = await storageService.storeDoc(docBuffer, finalFilename, metadata, userEmail, html);
 
         // Add storage information to the response headers
         c.header('X-Storage-Path', result.path);
@@ -39,13 +42,14 @@ export async function htmlToDocHandler(c) {
       }
     }
 
-    // Set response headers
-    c.header('Content-Type', 'application/msword');
-    c.header('Content-Disposition', `attachment; filename="${filename}"`);
+    // Set response headers for .docx format
+    c.header('Content-Type', 'application/vnd.openxmlformats-officedocument.wordprocessingml.document');
+    c.header('Content-Disposition', `attachment; filename="${finalFilename}"`);
 
     // Return the document buffer
     return c.body(docBuffer);
   } catch (error) {
+    console.error('Error in HTML to DOC conversion:', error);
     return errorUtils.handleError(error, c);
   }
 }

src/routes/html-to-doc.js

@@ -13,7 +13,7 @@ export async function htmlToExcelHandler(c) {
     const { html, filename = 'document.xlsx', sheetName = 'Sheet1', store = false } = c.get('body');
 
     // Generate Excel spreadsheet from HTML tables
-    const excelBuffer = excelService.generateExcel(html, sheetName);
+    const excelBuffer = await excelService.generateExcel(html, sheetName);
 
     // Store the Excel file in Supabase if requested
     if (store) {

src/routes/html-to-excel.js

@@ -1,27 +1,60 @@
+import { exec } from 'child_process';
+import fs from 'fs';
+import path from 'path';
+import { promisify } from 'util';
+import { v4 as uuidv4 } from 'uuid';
+import os from 'os';
+
+const execPromise = promisify(exec);
+
 /**
  * Service for generating Word documents from HTML content
  */
 export const docService = {
   /**
-   * Generate a Word document from HTML content
+   * Generate a Word document from HTML content using pandoc
    * @param {string} html - The HTML content to convert to Word format
-   * @returns {Buffer} - A buffer containing the Word document data
+   * @returns {Promise<Buffer>} - A buffer containing the Word document data
    */
-  generateDoc(html) {
-    // Create Word document from HTML
-    const header = `
-      <html xmlns:o='urn:schemas-microsoft-com:office:office' 
-            xmlns:w='urn:schemas-microsoft-com:office:word' 
-            xmlns='http://www.w3.org/TR/REC-html40'>
-      <head><meta charset='utf-8'></head><body>`;
-    const footer = `</body></html>`;
-    const sourceHTML = header + html + footer;
-
-    // Add BOM (Byte Order Mark) for proper encoding in Word
-    const bomPrefix = Buffer.from([0xEF, 0xBB, 0xBF]);
-    const htmlBuffer = Buffer.from(sourceHTML);
-    const docBuffer = Buffer.concat([bomPrefix, htmlBuffer]);
-    
-    return docBuffer;
+  async generateDoc(html) {
+    try {
+      // Create temporary files for input and output
+      const tempDir = os.tmpdir();
+      const inputId = uuidv4();
+      const outputId = uuidv4();
+      const inputPath = path.join(tempDir, `${inputId}.html`);
+      const outputPath = path.join(tempDir, `${outputId}.docx`);
+      
+      // Write HTML to temporary file
+      await fs.promises.writeFile(inputPath, html, 'utf8');
+      
+      // Use pandoc to convert HTML to DOCX
+      const command = `pandoc -f html -t docx "${inputPath}" -o "${outputPath}"`;
+      console.log(`Executing pandoc command: ${command}`);
+      
+      await execPromise(command);
+      
+      // Read the generated DOCX file
+      const docBuffer = await fs.promises.readFile(outputPath);
+      
+      // Clean up temporary files
+      try {
+        await fs.promises.unlink(inputPath);
+        await fs.promises.unlink(outputPath);
+      } catch (cleanupError) {
+        console.warn('Error cleaning up temporary files:', cleanupError);
+      }
+      
+      return docBuffer;
+    } catch (error) {
+      console.error('Error generating Word document with pandoc:', error);
+      
+      // If pandoc fails, provide a detailed error message
+      if (error.stderr) {
+        console.error('Pandoc error output:', error.stderr);
+      }
+      
+      throw new Error(`Failed to generate Word document: ${error.message}`);
+    }
   }
 };

src/services/doc-service.js

@@ -9,10 +9,10 @@ export const excelService = {
    * Generate an Excel spreadsheet from HTML content containing tables
    * @param {string} html - The HTML content containing tables
    * @param {string} sheetName - The name for the worksheet (default: 'Sheet1')
-   * @returns {Buffer} - A buffer containing the Excel file data
+   * @returns {Promise<Buffer>} - A buffer containing the Excel file data
    * @throws {Error} - If no tables are found in the HTML content
    */
-  generateExcel(html, sheetName = 'Sheet1') {
+  async generateExcel(html, sheetName = 'Sheet1') {
     // Create a DOM from the HTML
     const dom = new JSDOM(html);
     const document = dom.window.document;