inner_text joins text without whitespace

[DougPuchalski]

Joining <span>one</span><span>two</span> inner_text elements without whitespace results in onetwo rather than one two. Votes for a single space when joining?

https://github.com/tenderlove/nokogiri/commit/be67bd3eb17bda8b08a14a512980983691376c65#lib/nokogiri/xml/node_set.rb-P7

[tenderlove]

I don't think this is a good idea. Would this only impact span tags? What if the tags already have a space between them? What about two spaces? How do you differentiate the inner_text between tags separated by spaces and tags without?

There's too much ambiguity, not to mention it would break backwards compatibility.

[DougPuchalski]

That was only an example. Is it the intention for inner_text to concatenate content from separate tags without delineation?

[tenderlove]

Ya, it literally just recursively joins the inner_text from each tag. If you want to do text indexing, it's probably best to start at the leaf nodes of the tree and walk back up, only asking for text in that node.

@flavorjones you have any ideas?

[DougPuchalski]

Wouldn't want to break anything, but I'm not imagining a use-case for munging text like that. Should the method accept a block or options perhaps?

Webkit appears to use newlines, at least in the console.

[flavorjones]

You may want to use the Loofah gem (I wrote it). It's got a #to_text method that inserts whitespace after an inline element, and a newline after a block element.

[DougPuchalski]

Excellent, looks like just what I need. Thanks flavor.

[feliperaul]

@tenderlove Please reconsider reopening this one. It's a bug. When I ask for the text of an element, you can't change it. If I have <span>Experts</span><span>exchange</span>', .text output is expertsexchange... it's actually changing the text!

Regarding the ambiguity, you just add one whitespace and that's all. If there's already other whitespaces there, it's up to the user use .strip or something like that. But please don't concatenate strings that belongs to different HTML elements.

Had a nasty bug because of this.

[feliperaul]

In the meantime, a workaround is using this:
Nokogiri::HTML(str).xpath('//text()').map(&:text).join(' ')
source: http://stackoverflow.com/questions/28449761/replace-html-tags-with-whitespaces

[knu]

Whitespace is a language-specific choice for a word separator. There are languages in the world such as Japanese that do not use space for separating words or sentences.

As you may know, the XML/HTML and DOM specifications are clear about how to deal with whitespaces, so unless this argument is based on some standards and backward compatibility is fully taken into account, we could never be able to buy it.

I, for one, have never seen a DOM API that complements implicit whitespaces between nodes, and I think an XPath equivalent for inner_text() is string(.), which simply concatenates texts without any seperator:

Nokogiri::HTML(str).xpath('string(.)')  #=> "onetwo"

[ar7max]

Firefox
JavaScript

D = div element of  '<span>one</span><span>two</span>'
D.innerText returns 'onetwo'
D.textContent returns 'some whitespaces onetwo some whitespaces'

So, whats a problem? What did you expect from inner_text?