I’ve been tracking updates/differences in Firefox default settings since around 2010.

It used to be a semi-manual process to extract files like prefs.js from omni.ja files Firefox, and then combine them using cat command.

Recently, more and more settings are embedded inside libxul.so instead of writing them in text file (like prefs.js). In addition, to make it easier to use with trust, these settings were changed to be generated from YAML files when building Firefox binaries. The old methods are not making much sense.

So I changed the process drastically. First, run Firefox headless and marionette mode. And then, run a script using foxr NPM package to get the Firefox defaults directly.

The instruction in the README include the technique for using a fresh clean profile. Refer another post for details.

Create a temporary folder to use a fresh clean profile.

% mkdir /tmp/profile_dir

Run firefox in headless mode.

% /path/to/firefox -profile /tmp/profile_dir -marionette -headless

Execute fetcher.js to obtain all default prefs.

% node fetcher.js > prefall.json

We can run JavaScript snippets on an opened page using foxr, which can be used for things like scraping.

    await page.goto('about:config')

    //await page.evaluate('ShowPrefs()');
    const prefs = await page.evaluate(`(function(){
      const { Services } = ChromeUtils.import("resource://gre/modules/Services.jsm");
      const gPrefBranch = Services.prefs;
      const defaultBranch = gPrefBranch.getDefaultBranch("");
      // ... (snip) ...
      return prefs.map(entry => [entry.name, gTypeStrs[entry.type], entry.value]);

fetcher.js opens about:config and run some snippets. about:config supports firefox private interfaces (used to be available with old extensions). So, fetcher.js is able to access Firefox preference information.

The comparison logic is the same as in the past. I ported from code running in the browser to code running on Node.js. In addition, it was rewritten into a modern grammar. As a result, surveyer.js has also changed a lot.