views:

1494

answers:

2

Hi,

There is a very similar question already. One of the solutions uses code like this one:

string.mb_chars.normalize(:kd).gsub(/[^x00-\x7F]/n, '').to_s

Which works wonders, until you notice it also removes spaces, dots, dashes, and who knows what else.

I'm not really sure how the first code works, but could it be made to strip only accents? Or at the very least be given a list of chars to preserve? My knowledge of regexps is small, but I tried (to no avail):

/[^\-x00-\x7F]/n # So it would leave the dash alone

I'm about to do something like this:

string.mb_chars.normalize(:kd).gsub('-', '__DASH__').gsub
  (/[^x00-\x7F]/n, '').gsub('__DASH__', '-').to_s

Atrocious? Yes...

I've also tried:

iconv = Iconv.new('UTF-8', 'US-ASCII//TRANSLIT') # Also tried ISO-8859-1
iconv.iconv 'Café' # Throws an error: Iconv::IllegalSequence: "é"

Help please?

A: 

it also removes spaces, dots, dashes, and who knows what else.

It shouldn't.

string.mb_chars.normalize(:kd).gsub(/[^x00-\x7F]/n, '').to_s

You've mistyped, there should be a backslash before the x00, to refer to the NUL character.

/[^\-x00-\x7F]/n # So it would leave the dash alone

You've put the ‘-’ between the ‘\’ and the ‘x’, which will break the reference to the null character, and thus break the range.

bobince
Oh dear lord... please forgive me :)Thanks!
Ivan
yeah, we've all had a tedious load of debugging confusion caused by the simplest of typos!
bobince
A: 

It's not as neat as Iconv, but does what I think you want:

http://snippets.dzone.com/posts/show/2384

Tiago