A Server that Returns Website Screenshots Using Headless Chromium with Bottle and Gevent

This is the code for a web application that takes a URL included as a parameter in the request URL, requests it with a headless Chrome, takes a screenshot, and returns the response.

The service is provided using Python, Bottle, and gevent.

from gevent import monkey
monkey.patch_all()

import hashlib
import os
import subprocess

from bottle import route, run, request, abort, HTTPResponse

thumbnail_dir = '/tmp/web-thumbnail'

veil_file_dir = os.path.join(os.path.dirname(__file__), 'static/veil.png')


def get_screenshot_filename(url):
    h = hashlib.sha256(url.encode('utf-8', 'ignore'))
    return os.path.join(thumbnail_dir, '{}.png'.format(h.hexdigest()))


def take_shapshot(url):
    """
    Generates a snapshot using headless Chromium
    """
    command = [
        '/usr/bin/chromium-browser',
        '--no-sandbox',
        '--headless',
        '--disable-gpu',
        '--no-zygote',
        # '--disable-software-rasterizer',
        '--screenshot={}'.format(get_screenshot_filename(url)),
        '--hide-scrollbars',
        '--window-size=400,400',
        '--lang=ja-JP',
        url,
    ]

    return subprocess.check_call(command)


@route('/')
def index():
    """
    Thumbnail generation page
    """
    url = request.query.u
    if not url:
        abort(400, 'Please specify parameter u')
        return
    append_headers = []
    screenshot_filename = get_screenshot_filename(url)
    if os.path.exists(screenshot_filename):
        append_headers.append(
            ('X-CacheFileResult', f'Cache file found:{screenshot_filename}'))
    else:
        ret = take_shapshot(url)
        append_headers.append(('X-TakeResult', str(ret)))

    if os.path.exists(screenshot_filename):
        with open(screenshot_filename, 'rb') as fp:
            data = fp.read()
    else:
        # Could not create the file
        with open(veil_file_dir, 'rb') as fp:
            data = fp.read()
        append_headers.append(
            ('X-CacheFileResult', f'NotFound:{screenshot_filename}'))
    r = HTTPResponse(status=200, body=data)
    r.set_header('Content-type', 'image/png')
    for header_name, header_value in append_headers:
        r.set_header(header_name, header_value)
    return r


run(host='0.0.0.0', port=8080, server='gevent')