We need a way to cut out bad scoring elems inside good ones. about python-readability HOT 3 CLOSED

This isn't too bad to add in. So there's the method remove_unlikely_cadidates that cleans out some nodes, but it's not taking into account the link density. The link density is only considered with finding a candidate for the actual document. You'd have to add some work in there to check link density and determine if the node should be dropped.

You can see an example in my port that does this, I track not just candidates, but nodes that should be dropped after we've done the processing so that we only do the checks once.

https://github.com/mitechie/breadability/blob/master/src/breadability/readable.py#L383

and an example of the page parsed:

http://r.bmark.us/view/http%3A%2F%2Fwww.kwqc.com%2Fstory%2F19484184%2Fshots-fired-in-ill-school-2-students-in-custody

from python-readability.

Thank you for the help, your port looks great. Have you totally re-written the core readability parts?

from python-readability.

Yea, I tried to merge up this library with it's forks and decruft but failed and ended up just re-porting straight from the original readability js file from scratch trying to keep up better documentation and tests along the way.

from python-readability.