[SOLVED] Idea : Reader Mode

[user858753257]

Wouldn't it make sense to include in the internal browser a reading mode like chrome/safari ?
Or a general reader mode for all articles

[martinrotter]

Hello. Well, why not. You mean something like readablity?

[martinrotter]

@F3000 ping

[akinokonomi]

Is it like Firefox's built-in Reader view? Which allows to simplify/clean the page's content, and to customize text to some extent (line height, column width, font, colour mode).

If so, I'm not sure if it would fit in RSS reader, to be honest; in RSS Guard it's the work for "skin".

For it to work properly, it would need keep state between all articles, otherwise it doesn't make sense if user would be obliged to toggle and tune the Reader view for each article.

That is about "a general reader mode for all articles".

As to "include in the internal browser", I think it would make some sense if it can be done.

Here is an example of Reader view for this page https://www.w3schools.com/html/html_editors.asp

Some customization which Firefox allows

[martinrotter]

@F3000 Pls, reply.

One more comment: The "reader mode" is not just about "stylling" of the article. AFAIK Readability.js also has some magic which removes "non-article" parts of website, transforms the actual "article" to be layouted better etc.

When the HTML data is parsed with Readability, it returns pure HTML-formatted text, AFAIK stripped from custom CSS classes/formatting so this data can be for example displayed using style from active skin, and this would make sense to have implemented if I decide to implement "reader" mode.

Also, note there are many "readability" libraries based on Node.js:

• https://github.com/mozilla/readability
• https://github.com/postlight/mercury-parser
• https://www.npmjs.com/package/article-parser
• https://github.com/ibrod83/nodejs-web-scraper

Also, there are many libs written in Python, but I would prefer perhaps Node.js (?) because RSS Guard already uses AdBlock component based on Node.js library, I would like to avoid add dependencies on other runtimes etc.

It seems that mozilla's readability is probably the best OSS solution on the market tho.

[martinrotter]

Exactly, that's what I also thought, but what about others?

Did not look really. Not sure there are others.

[akinokonomi]

Why did you chose node.js over python?

[martinrotter]

It was because I was looking for something to replace old C++ adblock implementation with. The only good solution I found was this: https://github.com/ghostery/adblocker/blob/master/packages/adblocker/README.md

Also, there is probably even better nodejs solution: https://github.com/brave/adblock-rust

I did not find any python-based adblock libraries then. Looking now, I see some solutions, I probably just missed them: https://github.com/ArniDagur/python-adblock

So looking at it now, it is actually hard to decide whether node or python is better to integrate...

[akinokonomi]

I would suggest to have modules in RSS GUard user data folder,

Yes, I remembered the way node does things, and this is certainly preferable to installing those packages globally.

... sounds a bit odd, but it can be the way?

That is certainly the way.

And together with this user will just let RSS Guard deal with them. Sounds nice.

[martinrotter]

Yes, I will do that with Adblock to just see how it works in real and we can go from there.

[martinrotter]

Btw: Here is simple node.js script to run readable on HTML content.

var rd = require('@mozilla/readability');
var fs = require('fs');
var jsd = require('jsdom');
var iconvlite = require('iconv-lite');

//var mrc = require('@postlight/mercury-parser');

var data = fs.readFileSync(0);
var encData = iconvlite.decode(data, "win-1250"); // utf-8 should be used, win-1250 just for my testing
var doc = new jsd.JSDOM(encData);

var article = new rd.Readability(doc.window.document).parse();
console.log(article['content']);

Can be called like curl "website" | node "script.js"

Something like this would essentially work on RSS Guard with not great amount of coding.

[akinokonomi]

Oh, indeed! I remember now. (Used to do some few things with puppeteer myself at some point.)

Just tried. Works great, looks fun!

[martinrotter]

@akinokonomi

If so, I'm not sure if it would fit in RSS reader, to be honest; in RSS Guard it's the work for "skin".

Yes. When article from feed is displayed, then yes, formatting is just better to be done by skin as it stands now, but when user clicks "URL" or just loads any other website in internal browser, then "readability.js" could be indeed useful, it could be even subsequently styled with skin's css. I still has this in my mind, did not forget about it.

[martinrotter]

OK, I reimplemented Node.js integration in RSS Guard which is now de-facto automatic. If node/npm binaries are detected (on Linux they are installed to PATH, on Windows the installer also adds them to PATH by default), then RSS Guard automatically installs needed package for the functionality when it is first used.

At this point, the only opt-in node.js package needed is @cliqz/adblocker for adblock. There is new settings category to tweak node.js.

RSS Guard now by default installs packages in a local scope, to a OS-derived subfolder in RSS Guard's data4 user-data folder. This means no admin rights are needed and installed packages are nicely contained in a folder which is easy to by synced etc.

[martinrotter]

Initial tests:

[martinrotter]

Before

After