case insensitive completion

case insensitive completion

[sergio]

Hello.

I want to make case insensitive completion for russian language.

>From the manpage:
zstyle ':completion:*' matcher-list 'm:{a-zA-Z}={A-Za-z}

but analog with russian:
zstyle ':completion:*' matcher-list 'm:{а-яА-Я}={А-Яа-я}
doesn't work

Also, this doesn't work too:
zstyle ':completion:*' matcher-list
'm:{абвгдеёжзийклмнопрстуфхцчшщъыьэюя}={АБВГДЕЁЖЗИЙКЛМНОПРСТУФХЦЧШЩЪЫЬЭЮЯ}'

'm:{я}={Я}' doesn't work for letter 'я', but 'm:{z}={Z}' works for
letter 'z'.

This works:
zstyle ':completion:*' matcher-list '+m:{a-zA-Z}={A-Za-z}' '+m:а=А'
'+m:б=Б' '+m:в=В' '+m:г=Г' '+m:д=Д' '+m:е=Е' '+m:ё=Ё' '+m:ж=Ж' '+m:з=З'
'+m:и=И' '+m:й=Й' '+m:к=К' '+m:л=Л' '+m:м=М' '+m:н=Н' '+m:о=О' '+m:п=П'
'+m:р=Р' '+m:с=С' '+m:т=Т' '+m:у=У' '+m:ф=Ф' '+m:х=Х' '+m:ц=Ц' '+m:ч=Ч'
'+m:ш=Ш' '+m:щ=Щ' '+m:ъ=Ъ' '+m:ы=Ы' '+m:ь=Ь' '+m:э=Э' '+m:ю=Ю' '+m:я=Я'
'+m:А=а' '+m:Б=б' '+m:В=в' '+m:Г=г' '+m:Д=д' '+m:Е=е' '+m:Ё=ё' '+m:Ж=ж'
'+m:З=з' '+m:И=и' '+m:Й=й' '+m:К=к' '+m:Л=л' '+m:М=м' '+m:Н=н' '+m:О=о'
'+m:П=п' '+m:Р=р' '+m:С=с' '+m:Т=т' '+m:У=у' '+m:Ф=ф' '+m:Х=х' '+m:Ц=ц'
'+m:Ч=ч' '+m:Ш=ш' '+m:Щ=щ' '+m:Ъ=ъ' '+m:Ы=ы' '+m:Ь=ь' '+m:Э=э' '+m:Ю=ю'
'+m:Я=я'

but this is not very well ):

zsh can lower and uppercase russian letters:
% qwe=йцу
% echo ${(U)qwe}
ЙЦУ
% echo ${(L)${(U)qwe}}
йцу

and in EXTENDED_GLOB there is globbing flag i which works right for
rissian language:

% setopt EXTENDED_GLOB
% ls  (#i)йцу
ЙцУ ЙЦУ

P.S.
Have zsh something for case switching.
It will be great to write something like:

zstyle ':completion:*' matcher-list 'm:$a=swithcase($a)

-- 
sergio

Re: case insensitive completion

[Peter Stephenson]

On Mon, 12 May 2008 16:36:37 +0400
sergio <zsh@sergio.spb.ru> wrote:
> Hello.
> 
> I want to make case insensitive completion for russian language.
> 
> >From the manpage:
> zstyle ':completion:*' matcher-list 'm:{a-zA-Z}={A-Za-z}
> 
> but analog with russian:
> zstyle ':completion:*' matcher-list 'm:{а-яА-Я}={А-Яа-я}
> doesn't work

Yes, this is a known problem.  The code that handles this is pretty hacky
and hasn't been converted to handle multibyte character sets.  This is one
of the big remaining problems with the multibyte support.

-- 
Peter Stephenson <pws@csr.com>                  Software Engineer
CSR PLC, Churchill House, Cambridge Business Park, Cowley Road
Cambridge, CB4 0WZ, UK                          Tel: +44 (0)1223 692070

Re: case insensitive completion

[Andrey Borzenkov]

[-- Attachment #1: Type: text/plain, Size: 566 bytes --]

On Monday 12 May 2008, sergio wrote:
[...]
> but analog with russian:
> zstyle ':completion:*' matcher-list 'm:{а-яА-Я}={А-Яа-я}
> doesn't work
> 
[...]
> This works:
> zstyle ':completion:*' matcher-list '+m:{a-zA-Z}={A-Za-z}' '+m:а=А'
...
> but this is not very well ):
>

it would have been OK if it were not so slow :(

unfortunately as written code heavily depends on character being
exactly one byte long.

so anything outside ASCII characters won't work right now (including
extended characters from ISO-8859-1 in UTF-8 locale).

[-- Attachment #2: This is a digitally signed message part. --]
[-- Type: application/pgp-signature, Size: 197 bytes --]

Re: case insensitive completion

[Peter Stephenson]

On Mon, 12 May 2008 16:36:37 +0400
sergio <zsh@sergio.spb.ru> wrote:
> zstyle ':completion:*' matcher-list 'm:{а-яА-Я}={А-Яа-я}
> doesn't work

I think this really needs fixing a more general way.  You don't care what
the character set is, you just want to make lower and upper case match.  So
really you want to be able to use ctype-style character specifications:

zstyle ':completion:*' matcher-list \
       'm:{[:lower:][:upper:]}={[:upper:][:lower:]}'

More general ctype tests would also be useful in:

  compadd -M 'r:|[.,_-]=* r:|=*'

which could be

  compadd -M 'r:|[[:punct:]]=* r:|=*'

and it's a bit confusing that that doesn't currently work---it's not
spelled out in the completion documentation.

That doesn't leave much for individual multibyte characters in
correspondence classes to do, but it still shouldn't be too difficult to
turn the tables into explicit sets of characters tested the way character
classes in globbing are tested (which is fully general, although based on
the wchar_t ordering, which isn't necessarily Unicode code points).

-- 
Peter Stephenson <pws@csr.com>                  Software Engineer
CSR PLC, Churchill House, Cambridge Business Park, Cowley Road
Cambridge, CB4 0WZ, UK                          Tel: +44 (0)1223 692070

Re: case insensitive completion

[sergio]

Peter Stephenson wrote:

> I think this really needs fixing a more general way.  You don't care what
> the character set is, you just want to make lower and upper case match.
Yes!

> zstyle ':completion:*' matcher-list \
> 'm:{[:lower:][:upper:]}={[:upper:][:lower:]}'
For ascii this doesn't work too :(

-- 
sergio