Making forward-word work with curly apostrophes

Making forward-word work with curly apostrophes

[Ian Eure]

If I have a text-mode buffer with the following two words:

1. I've
2. I’ve

forward-word skips over "I've," but treats "I’ve" as _three_ words. I  
seem to recall that forward-char skips over characters with word  
syntax until it finds one with non-word-syntax, but that doesn't seem  
to be the case here. I set ’ to have the same syntax as ' with:

(modify-syntax-entry ?’ "w p")

But forward-word still treats it as three words. All works well if I do:

(skip-syntax-forward "w")

What is forward-word doing, and how can I make it treat ’ as part of  
my word?

  - Ian

Re: Making forward-word work with curly apostrophes

[Leo]

On 2009-08-02 21:19 +0100, Ian Eure wrote:
> If I have a text-mode buffer with the following two words:
>
> 1. I've
> 2. I’ve
>
> forward-word skips over "I've," but treats "I’ve" as _three_ words.

I tested with Emacs 2009-07-26 and it seems there's no difference in
both cases in forward-word.

-- 
Leo's Emacs uptime: 1 day, 22 hours, 21 minutes, 39 seconds

Re: Making forward-word work with curly apostrophes

[Xah Lee]

On Aug 2, 1:19 pm, Ian Eure <i...@digg.com> wrote:
> If I have a text-mode buffer with the following two words:
>
> 1. I've
> 2. I’ve
>
> forward-word skips over "I've," but treats "I’ve" as _three_ words.

in my emacs23, it treats the second one as 2 words, not 3.

My emacs version is:
GNU Emacs 23.0.94.1 (i386-mingw-nt6.0.6002) of 2009-05-28 on
LENNART-69DE564 (patched)

just tested on my carbon emacs which is emacs 22. Same behavior.

i work with text having curly quotes in the past several years, i
don't recall emacs treating it as 3 words with forward-word.

you prob have some customization that does that? tried starting with -
Q?

  Xah
∑ http://xahlee.org/

☄

> I
> seem to recall that forward-char skips over characters with word
> syntax until it finds one with non-word-syntax, but that doesn't seem
> to be the case here. I set ’ to have the same syntax as ' with:
>
> (modify-syntax-entry ?’ "w p")
>
> But forward-word still treats it as three words. All works well if I do:
>
> (skip-syntax-forward "w")
>
> What is forward-word doing, and how can I make it treat ’ as part of
> my word?
>
>   - Ian

Re: Making forward-word work with curly apostrophes

[A.Politz]

On Aug 2, 10:19 pm, Ian Eure <i...@digg.com> wrote:
> If I have a text-mode buffer with the following two words:
>
> 1. I've
> 2. I’ve
>
> forward-word skips over "I've," but treats "I’ve" as _three_ words. I  
> seem to recall that forward-char skips over characters with word  
> syntax until it finds one with non-word-syntax, but that doesn't seem  
> to be the case here. I set ’ to have the same syntax as ' with:
>
> (modify-syntax-entry ?’ "w p")
>
> But forward-word still treats it as three words. All works well if I do:
>
> (skip-syntax-forward "w")
>
> What is forward-word doing, and how can I make it treat ’ as part of  
> my word?
>
>   - Ian



Apart from syntax-class, 2 chars form a word boundary, if they do
not share a common character category. You need to do something
like this:

(let ((latin ?l)
      (other
       (aref (category-set-mnemonics
              (char-category-set ?’)) 0)))
  (add-to-list 'word-combining-categories
               (cons latin other))
  (add-to-list 'word-combining-categories
               (cons other latin)))


For the whole story read

(describe-variable 'word-combining-categories)

and maybe

(info "(elisp)Categories")

.

-ap

Re: Making forward-word work with curly apostrophes

[Ian Eure]

On Aug 2, 2009, at 3:46 PM, A.Politz wrote:
>
> Apart from syntax-class, 2 chars form a word boundary, if they do
> not share a common character category. You need to do something
> like this:
>
> (let ((latin ?l)
>      (other
>       (aref (category-set-mnemonics
>              (char-category-set ?’)) 0)))
>  (add-to-list 'word-combining-categories
>               (cons latin other))
>  (add-to-list 'word-combining-categories
>               (cons other latin)))
>
Perfect, this is what I needed.

I mentioned something like this in a bug report before, but it got  
wontfixed. Basically, a bunch of symbols which are used in latin  
scripts (like directional quotes) are categorized (incorrectly, I  
believe) as CJK.

I'll report it again and see if I can get it fixed.

  - Ian