bug#71822: Feature request: fullwidth to halfwidth

bug#71822: Feature request: fullwidth to halfwidth

[Dan Jacobson]

M-l runs the command downcase-word.
That's great. But here's a feature request:
Have commands that can change wide to narrow,
with the same ease as M-l.
E.g., change ２０９Ｍ into 209M.
and another command to do the opposite.
OK, I guess they are called fullwidth and halfwidth.

bug#71822: Feature request: fullwidth to halfwidth

[Eli Zaretskii]

> From: Dan Jacobson <jidanni@jidanni.org>
> Date: Fri, 28 Jun 2024 17:06:55 +0800
> 
> M-l runs the command downcase-word.
> That's great. But here's a feature request:
> Have commands that can change wide to narrow,
> with the same ease as M-l.
> E.g., change ２０９Ｍ into 209M.
> and another command to do the opposite.
> OK, I guess they are called fullwidth and halfwidth.

When would such commands be useful?  These characters are deliberately
not the same, and not just different shapes of the same letters
(unlike upper-case and lower-case letters).

bug#71822: Feature request: fullwidth to halfwidth

[Dan Jacobson]

>>>>> "EZ" == Eli Zaretskii <eliz@gnu.org> writes:
>> Have commands that can change wide to narrow,
>> with the same ease as M-l.
>> E.g., change ２０９Ｍ into 209M.
>> and another command to do the opposite.
>> OK, I guess they are called fullwidth and halfwidth.

EZ> When would such commands be useful?

We use them all the time in Chinese,
請由209M號門        Ugly
請由２０９Ｍ號門    Pretty

EZ> These characters are deliberately
EZ> not the same, and not just different shapes of the same letters
EZ> (unlike upper-case and lower-case letters).

$ unicode ２０９Ｍ|grep Deco
Decomposition: <wide> 0032
Decomposition: <wide> 0030
Decomposition: <wide> 0039
Decomposition: <wide> 004D

bug#71822: Feature request: fullwidth to halfwidth

[Eli Zaretskii]

> From: Dan Jacobson <jidanni@jidanni.org>
> Cc: 71822@debbugs.gnu.org
> Date: Sun, 30 Jun 2024 12:04:07 +0800
> 
> >>>>> "EZ" == Eli Zaretskii <eliz@gnu.org> writes:
> >> Have commands that can change wide to narrow,
> >> with the same ease as M-l.
> >> E.g., change ２０９Ｍ into 209M.
> >> and another command to do the opposite.
> >> OK, I guess they are called fullwidth and halfwidth.
> 
> EZ> When would such commands be useful?
> 
> We use them all the time in Chinese,
> 請由209M號門        Ugly
> 請由２０９Ｍ號門    Pretty

That's about use of the characters, not about the use cases for the
commands you proposed.  Or maybe I don't understand the example, in
which case please elaborate.

> EZ> These characters are deliberately
> EZ> not the same, and not just different shapes of the same letters
> EZ> (unlike upper-case and lower-case letters).
> 
> $ unicode ２０９Ｍ|grep Deco
> Decomposition: <wide> 0032
> Decomposition: <wide> 0030
> Decomposition: <wide> 0039
> Decomposition: <wide> 004D

Yes, and..?

I didn't ask whether Emacs has the requisite information, I asked
about the use cases for these commands.  You presented none; instead,
you are lecturing me on basic Unicode attributes of characters and on
the reasons for having the wide ones.  That is not helpful; I already
know all that.  It definitely doesn't provide any rationale for having
such commands in Emacs.

bug#71822: Feature request: fullwidth to halfwidth

[Dan Jacobson]

>>>>> "EZ" == Eli Zaretskii <eliz@gnu.org> writes:
EZ> use cases for these commands

All offline (I. e. all by hand), I start out with English,

1. Her ID number is 0987654321BA.

I rewrite the first part in Chinese,

2. 她證號為 0987654321BA 。

But to get it into its final state,

3. 她證號為０９８...

I need to retype each and every number and letter "all over again" (but
in fullwidth), because there is no M-<something> command to change it
for me. As you see I have given up halfway.

bug#71822: Feature request: fullwidth to halfwidth

[Eli Zaretskii]

> From: Dan Jacobson <jidanni@jidanni.org>
> Cc: 71822@debbugs.gnu.org
> Date: Sun, 30 Jun 2024 16:23:45 +0800
> 
> >>>>> "EZ" == Eli Zaretskii <eliz@gnu.org> writes:
> EZ> use cases for these commands
> 
> All offline (I. e. all by hand), I start out with English,
> 
> 1. Her ID number is 0987654321BA.
> 
> I rewrite the first part in Chinese,
> 
> 2. 她證號為 0987654321BA 。
> 
> But to get it into its final state,
> 
> 3. 她證號為０９８...
> 
> I need to retype each and every number and letter "all over again" (but
> in fullwidth), because there is no M-<something> command to change it
> for me. As you see I have given up halfway.

You are saying that the keyboard and/or the input method you are using
to type Chinese support characters such as 她證號為, but don't support
typing wide digits like ０９８, is that the problem?

bug#71822: Feature request: fullwidth to halfwidth

[Dan Jacobson]

>>>>> "EZ" == Eli Zaretskii <eliz@gnu.org> writes:
EZ> You are saying that the keyboard and/or the input method you are using
EZ> to type Chinese support characters such as 她證號為, but don't support
EZ> typing wide digits like ０９８, is that the problem?

No. It's the difference between needing to type

０９８７６５４３２１ＢＡ

i.e.,

 C-\                     ;; toggle-input-method
 v 0 v                   ;; self-insert-command
 9 v                     ;; self-insert-command
 8 v                     ;; self-insert-command
 7 v                     ;; self-insert-command
 6 v                     ;; self-insert-command
 5 v                     ;; self-insert-command
 4 v                     ;; self-insert-command
 3 v                     ;; self-insert-command
 2 v                     ;; self-insert-command
 1 v                     ;; self-insert-command
 B                       ;; self-insert-command
 v A                     ;; self-insert-command

vs. just M-<one character>.

It's just like to turn this sentence into upper case,
one needs to type each letter again, holding the shift button,
because upcase-region hasn't been invented yet.

bug#71822: Feature request: fullwidth to halfwidth

[Kazuhiro Ito]

> EZ> use cases for these commands
> 
> All offline (I. e. all by hand), I start out with English,
> 
> 1. Her ID number is 0987654321BA.
> 
> I rewrite the first part in Chinese,
> 
> 2. 她證號為 0987654321BA 。
> 
> But to get it into its final state,
> 
> 3. 她證號為０９８...
> 
> I need to retype each and every number and letter "all over again" (but
> in fullwidth), because there is no M-<something> command to change it
> for me. As you see I have given up halfway.

FYI, We have japanese-hankaku[-region] and japanese-zenkaku[-region]
functions, which convert hankaku (halfwidth) and zenkaku (fullwidth)
characters to one another.  It is useful to normalize mixed text
mainly input by someone not me, though these functions don't support
all characters because they are for traditional Japanese character
set.


> japanese-hankaku is an autoloaded native-comp-function in
> ‘japan-util.el’.
>
> (japanese-hankaku OBJ &optional ASCII-ONLY)
>
> Convert argument to ‘hankaku’ and return that.
> The argument may be a character or string.  The result has the same type.
> The argument object is not altered--the value is a copy.
> Optional argument ASCII-ONLY non-nil means to return only ASCII character.

-- 
Kazuhiro Ito

bug#71822: Feature request: fullwidth to halfwidth

[Dan Jacobson]

Indeed,
japanese-hankaku-region
japanese-zenkaku-region
work. So maybe they should
be available as
halfwidth-woed
halfwidth-region
fullwidth-word
fullwidth-region etc.
in emacs -Q.

bug#71822: Feature request: fullwidth to halfwidth

[Stefan Kangas]

Dan Jacobson <jidanni@jidanni.org> writes:

> Indeed,
> japanese-hankaku-region
> japanese-zenkaku-region
> work. So maybe they should
> be available as
> halfwidth-woed
> halfwidth-region
> fullwidth-word
> fullwidth-region etc.
> in emacs -Q.

Do they work for the full set of Chinese characters?

I'm asking because Kazuhiro Ito <kzhr@d1.dion.ne.jp> writes:

> [...] these functions don't support all characters because they are
> for traditional Japanese character set.

bug#71822: Feature request: fullwidth to halfwidth

[Dan Jacobson]

All I am ever concerned with is just ASCII.

All I know is if it hits a Chinese character, it should just leave it 
untouched, no matter if converting to or converting from.

On 2024-07-01 10:13, Stefan Kangas wrote:
> Dan Jacobson <jidanni@jidanni.org> writes:
> 
>> Indeed,
>> japanese-hankaku-region
>> japanese-zenkaku-region
>> work. So maybe they should
>> be available as
>> halfwidth-woed
>> halfwidth-region
>> fullwidth-word
>> fullwidth-region etc.
>> in emacs -Q.
> 
> Do they work for the full set of Chinese characters?
> 
> I'm asking because Kazuhiro Ito <kzhr@d1.dion.ne.jp> writes:
> 
>> [...] these functions don't support all characters because they are
>> for traditional Japanese character set.