Browsh: a modern text-only web browser

Originally published at:


Oxymoron Wednesday?

1 Like

Not really. There are many cases in the modern world where only text is available (sshing into servers, for example). The point is that this is modern browser that supports Javascript rather than something like Lynx or W3m that can’t render most pages properly.


MacOS version link is broken. Bummer.

Technically it’s just rendering a headless Firefox.

Lynx still does surprisingly well:


Especially in a workplace that forbids Web browsing, or so I’m told.


For that there’s always ssh -D 5000:wink: (Assuming you’ve got a shell to connect to external.)


Kind of encourages brutalism.


I downloaded the latest update directly from the GitHub releases page, and it’s working okay for me.

1 Like

Not at all. A text only browser that handles modern javascript heavy web sites (ie, gmail) would be a godsend for anyone who needs to use a screen reader (blind, visually disabled, etc). It’s also going to come in extremely handy for people forced to pay exorbitant rates for data. On a phone screen with a provider that charges by the kilobyte, text-only browsing makes a ton of sense.


From what I see it’s not text only at all and wouldn’t save any data because it’s downloading all the image and video data and using a filter to pixelate them for display.

Like the vlc ascii output filter.


Well that’s… dumb. So it’s basically a mugwump solution - neither one thing nor the other. And thus doomed to irrelevance.

1 Like

But it renders HTML 5 GUYS!

In the comments on the video, he mentions setting it up on a remote server (that doesn’t have the bandwidth problems) and connecting to that via SSH, so you’re only pulling the text output, and the server in the high-bandwidth server farm is the one loading all the heavy stuff. That would introduce some lag though.


It’s like using an image>asci (well, ANSI really) filter, but it does render the text as standard text.
That said, it needs work on it’s image>ansi filter, because the result is basically unrecognisable:

That image is from one of our IP cameras, and the red ‘pixels’ you see in the middle are actually a postman, but there’s no way to tell that from the image.
From the documentation it sounds like they’re picking one pixel, and using that for the colour of the ansi pixel, but it would better I think if it took several pixels, and averaged them out. As you can see in my example, what should be a brown carpet is detected as a rainbow mush (and a flickering one at that) because of noise in the camera’s image which is ignorable in the full resolution image.

There’s also the small problem of installing firefox on whatever machine you’re going to use as a server for this, because firefox depends on a bunch of gui components that I didn’t want to install.


“Text only” browser that has to spend extra processing power to make images look 8-bit-ish??

If you are actually trying to go image-less, a regular browser will do this just fine, no need to install any wrappers.

This topic was automatically closed after 5 days. New replies are no longer allowed.