ADMBrowser

Author: Alan D Moore (http://www.alandmoore.com, e-mail me_AT_alandmoore_DOT_com)

Contributors:

• Isaac "hunternet93" Smith ([email protected])

Description

ADMBrowser is a browser specifically for use on web kiosks. It's based on PyQt5 and QtWebEngine (Chromium/Blink), and is designed to make lock-down very simple and painless.

It was originally conceived for use in library catalog terminals, when it became clear that browsers with ever-growing feature lists like Firefox and Chrome were too much work to lock down correctly and completely. It was also designed to be easily configurable using a simple text file that can be hand-edited in a terminal over SSH across a slow WAN, so no databases, XML, or crazy binaries here.

ADMBrowser is a fork of WCGBrowser, with these key differences:

• WCGBrowser was based on QtWebKit, ADMBrowser is based on QtWebEngine
• WCGBrowser supports Python 2 and Python 3. ADMBrowser requires Python 3.
• WCGBrowser runs with Qt4 or Qt5. ADMBrowser requires Qt5 (5.5 or higher).
• The command line and configuration syntax may differ slightly.

ADMBROWSER IS CURRENTLY BETA QUALITY

Many features that were in WCGBrowser are broken or unavailable simply because QtWebEngine does things differently than QtWebKit.

Features

• Blink (Chrome/Chromium) rendering
• Text-based, YAML configuration
• (Optional) Inactivity timeout
• Popups/open-in-new-window can be disabled
• Minimal, no-clutter interface simple for the general public.
• Configurable navigation bar with bookmarks
• Configurable handling of external MIME-types (PDF, etc)
• (Optional) Whitelisting of hosts & domains
• 'Screensaver' mode to display a specified URL when idle
• Printing support
• Custom error screens
• Much, much more...

Requirements

• Python 3.6 or later - You need the dataclasses library if you're using < 3.7
• PyQt5 (v.5.5 or higher)
• Python YAML library (http://pyyaml.org)

It should work on any platform, but it's only been tested on Arch Linux, Debian, and Ubuntu.

Usage

The included admbrowser.yaml file shows a documented example configuration. To use it, copy it to /etc/admbrowser.yaml, ~/.admbrowser.yaml, or specify it with the -c (--config-file) switch. You can make the browser.py executable, or launch it using python, like so:

python admbrowser.py

At a minimum, you need to specify a start_url using either the config file or the -l switch, or else the browser isn't much use. Advanced configuration is probably best done in the configuration file, but many basic features can be enabled or disabled at the command line using these switches:

ADMBrowser also accepts the built-in Qt command-line arguments, which provide some low-level overrides. Documentation of these switches can be found at https://doc.qt.io/qt-5/qapplication.html#QApplication.

Configuration File

The sample configuration file is fully commented, and should be pretty easy to configure if you just read through it. In case you just want to start from scratch, here are the current configuration options available for the application.

Bookmarks

Bookmarks are created in a YAML list called "bookmarks" with this format:

bookmarks:
  1:
    name: "Bookmark Name"
    url: "http://bookmark.url/"
    description: "A short description of the bookmark, for the tooltip"

  2:
    name: "Another bookmark name":
    url: "http://example.com/some_bookmark"
    description: "A short description of this bookmark"

Bookmark names can include an ampersand to specify an accelerator key. You can also specify bookmark entries like so:

bookmarks:
  "Bookmark Name":
    url: "http://bookmark.url/"
    description: "A short description of the bookmark, for the tooltip"

This is more compact, but the downside is that you have no control over the order of the bookmarks (they are ordered by key, so it'll be alphabetical by name). This mode is really for backwards compatibility, but if you have a lot of bookmarks that you want alphabetized and want to save some typing, this may be the way to go.

Content Handlers

If you're allowing external content to be launched, the "content_handlers" array allows you to specify in which programs the external content will open by MIME type. The syntax looks like this:

content_handlers:
  "application/pdf": "xpdf"
  "application/vnd.oasis.opendocument.text":"libreoffice"

ADMBrowser will download the file to a temp directory and pass it as an argument to whatever command you specify in the second column. Be aware of this, as in some cases you might want to write a wrapper script of some sort to deal with some types of files or programs that don't properly deal with arguments.

Navigation Layout

The "navigation_layout" parameter is a list of items to place on the navigation bar, if it's showing. You can choose from the following:

• "back", "forward", "refresh", "stop": the traditional browser navigation buttons.
• "print": a button to open the print dialog for the main page.
• "zoom_in", "zoom_out": the zoom buttons
• "bookmarks": your bookmark buttons
• "quit": your "I'm finished" button
• "separator": A vertical line to separate sections
• "spacer": an expanding spacer to push widgets around

The list can be specified in any valid YAML list format, but I recommend enclosing it in square braces and separating with commas. "separator" and "spacer" can be used as many times as you wish, the others should only be used once each.

Whitelist

The whitelist feature is added as a convenience to help lock down your kiosk when you don't have complete control over all the links on your kiosk pages and want to prevent users from going off to strange sites. It's not a firewall or content filter, and may not behave exactly how you expect it to; so if you plan to use it, please read a bit about what it does and what it does not do.

If you don't want to use the whitelist feature, just comment it out, leave the list empty, or give it a value of "False".

What the whitelist does

You give the whitelist a list of domains or hosts, like this:

whitelist: ["somehost.example.com", "some-local-host", "mydomain.org"]

Whenever the user clicks a link or otherwise tries to navigate to a page, the hostname is extracted from the requested URL and matched against the whitelist. If there's a match, the page is displayed; if not, the error text is shown.

Some things are automatic:

• The start_url host is automatically whitelisted
• Bookmark hosts are automatically whitelisted
• Subdomains are also automatically whitelisted. Thus, if you whitelist "example.com", then "foo.example.com" will be whitelisted as well (though "foo-example.com" will not, since that's actually a different domain).

If you just want to whitelist the start_url and bookmark urls and nothing else, you can just do this in the configuration file:

whitelist: True

When relying on the automatic whitelisting, it's important to understand that the complete host string of these URLs is whitelisted. So for example, if your start_url is "http://example.com", "example.com" will be added to the whitelist (and thus all subdomains of example.com, such as foo.example.com, bar.example.com, etc.). If you specify "http://www.example.com" as the start_url, though, "www.example.com" is added to the whitelist. Thus, "foo.example.com" would not be whitelisted.

Also note that if you whitelist a URL that just forwards you to another host, you need to specify both hosts in the whitelist.

What the whitelist does NOT do

• The whitelist does not block content on a whitelisted page from being displayed, regardless of where the content is hosted. As long as the page's URL is acceptable, all the content is displayed. So, for example, if you have your images and scripts (or ads!) on a separate content delivery network, you don't need to whitelist that server. You only need to whitelist hosts/domains of URLs to which the user is explicitly navigating (via hyperlink, bookmark, JavaScript forward, etc) -- in other words, the URL that would show up in a normal browser's location bar.
• The whitelist cannot take an actual path or filename, nor does it check the port, protocol, username, or any other component of the URL other than the host or domain. Sorry.
• If you whitelist a host, its IP will not be automatically whitelisted (and vice-versa); nor will a fully-qualified hostname in the whitelist automatically whitelist the hostname by itself (or vice-versa). A URL is only allowed when its literal hostname matches a whitelist entry.

Screensaver Mode

The screensaver mode is a special timeout mode that lets you display a given URL only while the browser is idle. Consider a configuration like this:

start_url: 'http://example.com/kiosk'
timeout: 1800
timeout_mode: 'screensaver'
screensaver_url: 'http://example.com/slides'

This configuration would do the following:

• The browser will start on http://example.com/kiosk
• After 30 minutes of no user activity (mouse/keyboard/touchscreen/etc), the navigation bar will hide and http://example.com/slides will be displayed.
• As soon as a user steps up and generates activity (moves a mouse, touches the screen, etc), the navigation bar (if configured) will reappear, and the browser will load http://example.com/kiosk.

The screensaver_url could be, for example, an image rotator, a page with ads, a welcome message, etc. It doesn't really matter, but keep in mind the user can't actually interact with the screensaver page, because as soon as they touch a mouse or keyboard, the start_url will load.

Proxy Server

ADMBrowser will allow you to set a host (name or IP) and port number for an HTTP proxy. HTTPS, FTP, SOCKS, or authenticated proxy is not currently supported. You can set the proxy settings one of three ways:

• The environment variable http_proxy is respected
• The CLI switch --proxy_server
• The configuration file option proxy_server

To set the proxy server, use the format host:port, as in these examples:

proxyserver.mynetwork.local:3128
localhost:8080
192.168.1.1:8880

If you neglect to include a port, and just put an IP address or hostname, the port 8080 will be used by default.

NOTE This feature may not work on some OS. It currently relies on setting the http_proxy environment variable (regardless of which method you use to configure it), which may not be respected on all systems. It definitely works on Linux, and probably on any unix-like system.

Print Settings

ADMBrowser supports configuring default printer settings and allows printing without showing a dialog box. Options are set with the print_settings variable. For example:

print_settings:
    silent: True
    margins: [5, 5, 3, 3]
    orientation: "landscape"

The following options are supported:

Bugs and Limitations

The following are known limitations:

• There is no password dialog when a page requests authentication. You can set a single user/password set in the config file to be sent whenever a site does request it, or provide authentication credentials in the URL (in a bookmark/start_url).
• Only one popup window can exist at a time (if they're enabled at all).

The following issues showed up with the port from QtWebKit to QtWebEngine:

• Custom 404/Network Error pages don't work. This is due to a QtWebkitEngine limitation which will hopefully change someday.
• Sometimes ADMBrowser crashes with (C++) memory allocation errors.
• Probably much more that hasn't been tested yet.

If you find bugs, please report them as an "issue" at the project's github page: http://github.com/alandmoore/admbrowser/issues. If your "bug" is really a feature request, see below.

Raspberry Pi Support

ADMBrowser can work on a Raspberry Pi, in theory, BUT NOT ON RASPBIAN 10. ADMBrowser requires QtWebEngine, which (as of January 2020) is not packaged for Raspbian 10.

In order to use ADMBrowser, you need to use a distribution that provides a working version of QtWebEngine. As of January 2020, various solutions have been tested with the following results (tested on Pi 3b+):

See issue #16 for ongoing discussion of Raspberry Pi support.

Contributing

Contributions are welcome, so long as they are consistent with the spirit and intent of the browser -- that is, they are features useful in a kiosk, signage, or other lock-down situation, and keep the browser simple to configure. I would also prefer that changes to features or behavior are opt-in (require a switch to enable them), unless it just makes no sense to do it that way.

Coding Standards

If you're contributing code, please follow these best practices:

• Follow PEP8; use a linter/checker like pyflakes, pep8, or pylint and make sure your code doesn't generate errors.
  • This includes the 79 character limit. Yes, I'm like that.
  • Use snake_case variables, not camelCase (except for PyQt stuff we can't change)
  • Use f-strings rather than format() or old-style % variable substitution
  • Remember that code should work in Python 3.6 and PyQt5.
• Please document per PEP257; functions & classes need a docstring.
• Fork the project on GitHub, make your changes, and submit a pull request. You will probably be asked to change or fix some things, that's just how it goes.

Making Feature Requests

If there are features you'd like to see supported in this project, you have three options to see them implemented:

• Write the code (or have it written by someone else) and submit it to the project as a pull request.
• Contact me and offer to sponsor the development of the feature. My rates are reasonable and negotiable.
• Keep your fingers crossed and hope that somebody else does one of the previous two things for the feature you want.

License

ADMBrowser is released under the terms of the GNU GPL v3.