Take a picture of a web page from the command line

webpage-screenshot.rb

#!/usr/bin/env ruby

require 'ferrum'
require 'optparse'
require 'uri'

class WebScreenshotter
  def initialize(options = {})
    @options = {
      width: 1920,
      height: 1080,
      timeout: 30,
      output: 'screenshot.png',
      username: nil,
      password: nil
    }.merge(options)

    @browser = Ferrum::Browser.new(
      headless: true,
      window_size: [@options[:width], @options[:height]],
      timeout: @options[:timeout],
      browser_options: {
        'disable-gpu' => nil,
        'no-sandbox' => nil
      }
    )
  end

  def handle_login(original_url)
    current_url = @browser.current_url

    if current_url.include?('/admin/poobah_session/new')
      username = @options[:username] || ENV['SCREENSHOT_USERNAME']
      password = @options[:password] || ENV['SCREENSHOT_PASSWORD']

      unless username && password
        raise ArgumentError, "Login required but no username/password provided. Use --username and --password options or set SCREENSHOT_USERNAME and SCREENSHOT_PASSWORD environment variables."
      end

      puts "Login page detected. Attempting to authenticate..."

      # Fill in username
      username_field = @browser.at_css('input[type=email][name=email_address]')
      if username_field
        username_field.focus.type(username)
        puts "Username entered"
      else
        raise "Could not find username field"
      end

      # Fill in password
      password_field = @browser.at_css('input[type=password][name=password]')
      if password_field
        password_field.focus.type(password)
        puts "Password entered"
      else
        raise "Could not find password field"
      end

      # Submit the form
      submit_button = @browser.at_css('input[type=submit][name=commit]')
      if submit_button
        submit_button.click
        puts "Login form submitted"
      else
        raise "Could not find submit button"
      end

      # Wait for redirect after login
      @browser.network.wait_for_idle

      # Check if we were redirected to the original URL or if login failed
      current_url_after_login = @browser.current_url
      if current_url_after_login.include?('/admin/poobah_session/new')
        raise "Login failed - still on login page"
      end

      puts "Login successful, redirected to: #{current_url_after_login}"

      # If we weren't redirected to the original URL, navigate there
      unless current_url_after_login == original_url
        puts "Navigating to original URL: #{original_url}"
        @browser.goto(original_url)
        @browser.network.wait_for_idle
      end
    end
  end

  def capture(url)
    unless url =~ URI::DEFAULT_PARSER.make_regexp
      raise ArgumentError, "Invalid URL format: #{url}"
    end

    begin
      puts "Navigating to #{url}..."
      @browser.goto(url)

      # Wait for network requests to finish
      @browser.network.wait_for_idle

      # Check if we were redirected to login page
      handle_login(url)

      # Wait for page to be fully loaded
      @browser.evaluate('document.readyState') == 'complete'

      # Scroll to capture full page content
      @browser.execute(<<~JS)
        window.scrollTo(0, 0);
        window.scrollTo(0, document.body.scrollHeight);
        window.scrollTo(0, 0);
      JS

      puts "Taking screenshot..."
      @browser.screenshot(
        path: @options[:output],
        full: true
      )

      puts "Screenshot saved to #{@options[:output]}"
    rescue Ferrum::NodeNotFoundError => e
      puts "Error: Could not find element on page: #{e.message}"
    rescue Ferrum::TimeoutError => e
      puts "Error: Page load timed out: #{e.message}"
    rescue StandardError => e
      puts "Error: #{e.message}"
    ensure
      @browser.quit
    end
  end
end

# Parse command line options
options = {}
OptionParser.new do |opts|
  opts.banner = "Usage: screenshot.rb [options] URL"

  opts.on("-w", "--width WIDTH", Integer, "Window width (default: 1920)") do |w|
    options[:width] = w
  end

  opts.on("-h", "--height HEIGHT", Integer, "Window height (default: 1080)") do |h|
    options[:height] = h
  end

  opts.on("-t", "--timeout SECONDS", Integer, "Timeout in seconds (default: 30)") do |t|
    options[:timeout] = t
  end

  opts.on("-o", "--output FILE", String, "Output file (default: screenshot.png)") do |o|
    options[:output] = o
  end

  opts.on("-u", "--username USERNAME", String, "Username for login (required for admin pages)") do |u|
    options[:username] = u
  end

  opts.on("-p", "--password PASSWORD", String, "Password for login (required for admin pages)") do |p|
    options[:password] = p
  end

  opts.on("--help", "Show this help message") do
    puts opts
    exit
  end
end.parse!

if ARGV.empty?
  puts "Error: URL is required"
  puts "Use --help for usage information"
  exit 1
end

# Create screenshotter and capture the page
screenshotter = WebScreenshotter.new(options)
screenshotter.capture(ARGV[0])