New port of readability.js?

[zirkelc]

The current implementation of readability_lxml and the original project readability.js is out of sync, so it would make sense to update the readability_lxml implementation to benefit from potentially newer feature and fixes.

readability.js has a big set of test pages with source.html and expected.html files. As a first step, we could test how readability_lxml compares against this test set?

Based on the results we could check if it's better to fix the differences (if there aren't too many), or if a new port based on the current javascript implementation would make more sense. What do you think?

By the way, readability_lxml looks like it was first ported to ruby and then to python. Besides using lxml, were there any functions added specifically for trafilatura?

[adbar]

The approach looks good!

The Readabillity port was using lxml even before, I just simplified it and integrated it directly since it wasn't maintained anymore.

[prnake]

As far as I know, there is no python implementation with readability.js. It's possible to use go-readability instead of readablity_lxml with C binding, it's faster and with more precision.

[adbar]

Generating binding means additional maintenance work but if someone wants to publish a python package for go-readability I'd be happy to test it.

[zirkelc]

@adbar just wanted to let you know I haven't forgotten about this.... I'm a bit busy at the moment, but this topic is on my list for future quiet weekends 😄