Output table padding formatter goes wrong with UTF8 characters

[darkskiez]

The padding of the |'s goes wrong for the chinese input example.

Feature: Multilingual

  Cucumber output pads columns example

  Scenario Outline: Language Example
    Given dummy step <language>

  Examples:
        | language  | labelled  |
        | English   | OK        |
        | 日本語    | 确定      |
        | Español   | Aceptar   |

[mattwynne]

Which version of Ruby is this with?

[darkskiez]

MRI 1.9.3p0

[dkowis]

So I'm not sure how one expects Multibyte characters that don't fit the width of an ASCII Character to line up.
If all the characters were the same language, of course they'd line up, but I'm not sure there's any way to make them line up differently. I'll investigate a bit, but I'm sceptical.

Also, I assume this is for the pretty formatter output?

[dkowis]

Yeah I'm of the mind that we shouldn't spend time trying to fix this. Anything we could do would be complicated. We'd have to some how compute the size of the character being displayed, not simply the amount of characters being displayed.

@mattwynne, thoughts?

[os97673]

I'm second to @dkowis let'd do not spend time on this. It looks like a nice exercise in UTF area but adds nothing to Cucumber.
Though I will be happy to review PR for it ;)

@mattwynne @tooky do you agree?

[aslakhellesoy]

Some multibyte characters have width != 1:

XY
确定

Knowing whether a character has double width or not requires separate code: http://stackoverflow.com/questions/3634627/how-to-know-the-preferred-display-width-in-columns-of-unicode-characters

Bikeshedding springs to mind :-)

[os97673]

So, I'm closing the ticket.

[mattwynne]

A simple solution might be to offer a setting on the API that allows the user to specify double-width characters. That way if someone suffers from this, and they care enough about it, they can fix the problem by explicitly configuring Cucumber to display those characters as double-width.

e.g. in features/support/env.rb

Cucumber.configure do
  config.determine_console_character_width do |character|
    ['日', '本', '語'].include?(character) ? 2 : 1
  end
end

If nobody has called this configuration block, we assume everything is a 1.

It would be trivial to change Cucumber to use this lookup. That would also open up for someone to write a plugin that used one of the C libraries @aslakhellesoy referenced.

If anyone wants to write a PR for this, that's what I'd recommend they do.

[aslakhellesoy]

I thought widths could be fractional, like 1.4.

[mattwynne]

@aslakhellesoy ah, so that would make our implementation non-trivial. Is that what you're saying?

[aslakhellesoy]

Yes. See the SO I linked to

[dkowis]

Yeah, when I looked at it, it would end up being a fractional width, which is basically impossible to do on the console. The way I think, if you want to do mixed languages, you should probably output to something that deals with them better, rather than the console. The HTML formatter springs to mind. The pretty formatter is good enough in a pinch, but won't ever be perfect for stuff like this, IMHO.

[lock]

This thread has been automatically locked since there has not been any recent activity after it was closed. Please open a new issue for related bugs.