Add: PDf to DOCX conversion function

Author: Bansnetsajak007

convertTodocx.ps1

@@ -0,0 +1,45 @@
+param (
+    [string]$inputPath,
+    [string]$outputPath,
+    [string]$keepActive
+)
+
+# Convert $keepActive to a boolean
+$keepActiveBoolean = $keepActive -eq 'true'
+
+# Check if the input file exists
+if (-not (Test-Path $inputPath)) {
+    Write-Error "Input file does not exist: $inputPath"
+    exit 1
+}
+
+# Add COM object for Microsoft Word
+Add-Type -AssemblyName "Microsoft.Office.Interop.Word"
+
+# Create a new Word application instance
+$word = New-Object -ComObject Word.Application
+
+# Set visibility of the Word application
+if (-not $keepActiveBoolean) {
+    $word.Visible = $false
+}
+
+try {
+    # Open the PDF document
+    $document = $word.Documents.Open($inputPath)
+    
+    # Save the document as DOCX
+    $document.SaveAs([ref]$outputPath, [ref]16)  # 16 represents the default DOCX format
+    
+    # Close the document
+    $document.Close()
+    
+    Write-Host "Conversion completed: $inputPath to $outputPath"
+} catch {
+    Write-Error "An error occurred during conversion: $_"
+} finally {
+    # Quit the Word application if keepActive is false
+    if (-not $keepActiveBoolean) {
+        $word.Quit()
+    }
+}

convertTodocx.sh

@@ -0,0 +1,31 @@
+#!/bin/bash
+
+inputFilePath=$1
+outputFilePath=$2
+keepActive=$3
+
+# Check if input file exists
+if [ ! -f "$inputFilePath" ]; then
+    echo "Input file does not exist: $inputFilePath"
+    exit 1
+fi
+
+# Extract text from PDF using pdftotext (requires Poppler installed)
+tempTextFilePath="$(mktemp).txt"
+pdftotext -layout "$inputFilePath" "$tempTextFilePath"
+
+if [ ! -f "$tempTextFilePath" ]; then
+    echo "Failed to extract text from PDF: $inputFilePath"
+    exit 1
+fi
+
+# Convert extracted text to DOCX using pandoc (requires pandoc installed)
+if [ -z "$outputFilePath" ]; then
+    outputFilePath="${inputFilePath%.pdf}.docx"
+fi
+
+pandoc "$tempTextFilePath" -o "$outputFilePath"
+
+# Clean up temporary text file
+rm "$tempTextFilePath"
+echo "Conversion complete: $outputFilePath"

index.js

@@ -22,6 +22,22 @@ function windows(inputPath, outputPath, keepActive) {
 
     execSync(command);
   }
+
+  // New PDF to DOCX conversion functions for Windows and macOS hell yeahhh
+function windowsPdfToDocx(inputPath, outputPath, keepActive = false) {
+  if (!inputPath) {
+      console.error('Input path is not provided.');
+      return;
+  }
+
+  const scriptPath = path.resolve(__dirname, 'convertTodocx.ps1');
+  const inputFilePath = path.resolve(inputPath);
+  const outputFilePath = path.resolve(outputPath);
+
+  const command = `powershell -File "${scriptPath}" "${inputFilePath}" "${outputFilePath}" ${keepActive ? 'true' : 'false'}`;
+
+  execSync(command);
+}
 
 
 /**
@@ -111,6 +127,7 @@ module.exports = {
     convert,
     resolvePaths,
     windows,
+    windowsPdfToDocx,
     macos,
     packageVersion,
   };