install_tesseract.py

import os
import subprocess
import sys

def install_tesseract_windows():
    # URL to the Tesseract installer
    tesseract_installer_url = "https://github.com/UB-Mannheim/tesseract/wiki"
    
    # PowerShell command to download the installer
    powershell_command = f"""
    $installerPath = "$env:TEMP\\tesseract-installer.exe"
    Invoke-WebRequest -Uri "{tesseract_installer_url}" -OutFile $installerPath
    Start-Process -FilePath $installerPath -ArgumentList "/S" -Wait
    """

    # Execute the PowerShell command
    subprocess.run(["powershell", "-Command", powershell_command], check=True)

def install_python_dependencies():
    # Install pytesseract and other dependencies
    subprocess.check_call([sys.executable, "-m", "pip", "install", "pytesseract", "opencv-python-headless", "Pillow", "streamlit"])

def verify_tesseract_installation():
    try:
        # Verify Tesseract installation
        output = subprocess.check_output(["tesseract", "--version"], stderr=subprocess.STDOUT)
        print(output.decode())
    except subprocess.CalledProcessError as e:
        print("Tesseract installation failed. Please try installing manually.")
        print(e.output.decode())

if __name__ == "__main__":
    try:
        install_tesseract_windows()
        install_python_dependencies()
        verify_tesseract_installation()
        print("Tesseract OCR and dependencies installed successfully.")
    except Exception as e:
        print(f"An error occurred: {e}")