Improving hangul input module

Features
 - Support unicode hangul input
 - Improve input method for korean hanja, symbol
 - Implement hangul automata
 - Support some hangul keyboard layouts(Hangul 2-Bulsik,
Hangul
3-Bulsik final, Hangul 3-Bulsik 390)

In article
<9d644d9b0805310151l54e50721g821232e37aff8e2mail.gmail.com>, "Jihyun Cho"
<jihyun.jogmail.com> writes:

> Features
>  - Support unicode hangul input
>  - Improve input method for korean hanja, symbol
>  - Implement hangul automata
>  - Support some hangul keyboard layouts(Hangul
2-Bulsik, Hangul
> 3-Bulsik final, Hangul 3-Bulsik 390)

Thank you.  I'm now checking the code.  I'll commit the
change next week.

---
Kenichi Handa
handani.aist.go.jp

In article <E1K2lfm-0000Or-LJetlken.m17n.org>,
Kenichi Handa <handam17n.org> writes:

> In article
<9d644d9b0805310151l54e50721g821232e37aff8e2mail.gmail.com>, "Jihyun Cho"
<jihyun.jogmail.com> writes:
> > Features
> >  - Support unicode hangul input
> >  - Improve input method for korean hanja, symbol
> >  - Implement hangul automata
> >  - Support some hangul keyboard layouts(Hangul
2-Bulsik, Hangul
> > 3-Bulsik final, Hangul 3-Bulsik 390)

> Thank you.  I'm now checking the code.  I'll commit
the
> change next week.

I've just committed your code.  I put hanja-util.el under
lisp/language, and moved the calls of register-input-method
in hangul.el to leim-ext.el.

I also added these comments in hangul.el and hanja-util.el.

;; NEED COMMENT. ...
;; NEED ADJUTMENT. ...

Jihyun, please check them and add/adjust comments/docstring
appropriately.

---
Kenichi Handa
handani.aist.go.jp

2008/6/3 Kenichi Handa <handam17n.org>:
> In article <E1K2lfm-0000Or-LJetlken.m17n.org>,
Kenichi Handa <handam17n.org> writes:
>
>> In article
<9d644d9b0805310151l54e50721g821232e37aff8e2mail.gmail.com>, "Jihyun Cho"
<jihyun.jogmail.com> writes:
>> > Features
>> >  - Support unicode hangul input
>> >  - Improve input method for korean hanja,
symbol
>> >  - Implement hangul automata
>> >  - Support some hangul keyboard layouts(Hangul
2-Bulsik, Hangul
>> > 3-Bulsik final, Hangul 3-Bulsik 390)
>
>> Thank you.  I'm now checking the code.  I'll commit
the
>> change next week.
>
> I've just committed your code.  I put hanja-util.el
under
> lisp/language, and moved the calls of
register-input-method
> in hangul.el to leim-ext.el.
>
> I also added these comments in hangul.el and
hanja-util.el.
>
> ;; NEED COMMENT. ...
> ;; NEED ADJUTMENT. ...
>
> Jihyun, please check them and add/adjust
comments/docstring
> appropriately.
>
> ---
> Kenichi Handa
> handani.aist.go.jp
>

Thank you. I will adjust it.

2008/6/3 Kenichi Handa <handam17n.org>:
> In article <E1K2lfm-0000Or-LJetlken.m17n.org>,
Kenichi Handa <handam17n.org> writes:
>
>> In article
<9d644d9b0805310151l54e50721g821232e37aff8e2mail.gmail.com>, "Jihyun Cho"
<jihyun.jogmail.com> writes:
>> > Features
>> >  - Support unicode hangul input
>> >  - Improve input method for korean hanja,
symbol
>> >  - Implement hangul automata
>> >  - Support some hangul keyboard layouts(Hangul
2-Bulsik, Hangul
>> > 3-Bulsik final, Hangul 3-Bulsik 390)
>
>> Thank you.  I'm now checking the code.  I'll commit
the
>> change next week.
>
> I've just committed your code.  I put hanja-util.el
under
> lisp/language, and moved the calls of
register-input-method
> in hangul.el to leim-ext.el.
>
> I also added these comments in hangul.el and
hanja-util.el.
>
> ;; NEED COMMENT. ...
> ;; NEED ADJUTMENT. ...
>
> Jihyun, please check them and add/adjust
comments/docstring
> appropriately.
>
> ---
> Kenichi Handa
> handani.aist.go.jp
>

I have adjusted comments and docstring.
My English is not good.
If you can not be understood, let me know it.

In article
<9d644d9b0806040914v46839e87v48afdff983c3980email.gmail.com>, "Jihyun Cho"
<jihyun.jogmail.com> writes:

> I have adjusted comments and docstring.
> My English is not good.
> If you can not be understood, let me know it.

Thank you I've just committed your change with some
improvement.  Please check if my understanding is correct
with the latest code.

But there are a few places I still don't understand.

In hangul.el...

(1)

;; Current input character buffer. Store separated hangul
character.
;; First and second index of vector stored
"Choseong".
;; Third and forth index of vector stored
"Jungseong".
;; Fifth and sixth index of vector stored
"Jongseong".
(defvar hangul-queue
  (make-vector 6 0))

I don't understand 2nd to 4th lines.

(2)

(defsubst hangul2-input-method-jaum (char)
  "Store hangul jamo indice CHAR in `hangul-queue'.
Maybe, it is a Hangul 2-Bulsik jaum.
This function processes a Hangul 2-Bulsik jaum.
Unless the function inserts CHAR to current input queue,
commit current character and then start next character.
The Hangul 2-Bulsik is composed of `jaum' and `moum'."

"Maybe" is strange.  If CHAR is not a "Hangul
2-Bulsik
jaum", what it will be.

In what condition, it doesn't insert CHAR to the queue.

What does "start next character" mean?  Where is
the "next
character"?

The same for hangul2-input-method-moum,
hangul3-input-method-cho, hangul3-input-method-jung,
hangul3-input-method-jong.

In hanja-util.el...

(1)

;; List of current conversion status.
(defvar hanja-conversions
  nil)

What format does it have?

(2)

(defun hanja-filter (condp lst)
  "Return list.
It contains true elements to evaluate CONDP with each
element from LST."

I don't understand it.

By the way, why do you use a hash-table for hanja-table?
Isn't a char-table more efficient?  And, the value is now
"(CHAR) or (CHAR . STRING)", but most of elements
are
(CHAR).  I think it's a waste of cons cells.  How about
changing the format to "CHAR or (CHAR . STRING)"? 
It seems
easy to handle the changed format in hangul-to-hanja-char.

---
Kenichi Handa
handani.aist.go.jp

2008/6/5 Kenichi Handa <handam17n.org>:
> In article
<9d644d9b0806040914v46839e87v48afdff983c3980email.gmail.com>, "Jihyun Cho"
<jihyun.jogmail.com> writes:
>
>> I have adjusted comments and docstring.
>> My English is not good.
>> If you can not be understood, let me know it.
>
> Thank you I've just committed your change with some
> improvement.  Please check if my understanding is
correct
> with the latest code.
>
> But there are a few places I still don't understand.
>
> In hangul.el...
>
> (1)
>
> ;; Current input character buffer. Store separated
hangul character.
> ;; First and second index of vector stored
"Choseong".
> ;; Third and forth index of vector stored
"Jungseong".
> ;; Fifth and sixth index of vector stored
"Jongseong".
> (defvar hangul-queue
>  (make-vector 6 0))
>
> I don't understand 2nd to 4th lines.
>
> (2)
>
> (defsubst hangul2-input-method-jaum (char)
>  "Store hangul jamo indice CHAR in
`hangul-queue'.
> Maybe, it is a Hangul 2-Bulsik jaum.
> This function processes a Hangul 2-Bulsik jaum.
> Unless the function inserts CHAR to current input
queue,
> commit current character and then start next
character.
> The Hangul 2-Bulsik is composed of `jaum' and
`moum'."
>
> "Maybe" is strange.  If CHAR is not a
"Hangul 2-Bulsik
> jaum", what it will be.
>
> In what condition, it doesn't insert CHAR to the
queue.
>
> What does "start next character" mean?  Where
is the "next
> character"?
>
> The same for hangul2-input-method-moum,
> hangul3-input-method-cho, hangul3-input-method-jung,
> hangul3-input-method-jong.
>
> In hanja-util.el...
>
> (1)
>
> ;; List of current conversion status.
> (defvar hanja-conversions
>  nil)
>
> What format does it have?
>
> (2)
>
> (defun hanja-filter (condp lst)
>  "Return list.
> It contains true elements to evaluate CONDP with each
element from LST."
>
> I don't understand it.
>
> By the way, why do you use a hash-table for
hanja-table?
> Isn't a char-table more efficient?  And, the value is
now
> "(CHAR) or (CHAR . STRING)", but most of
elements are
> (CHAR).  I think it's a waste of cons cells.  How
about
> changing the format to "CHAR or (CHAR .
STRING)"?  It seems
> easy to handle the changed format in
hangul-to-hanja-char.
>
> ---
> Kenichi Handa
> handani.aist.go.jp
>

Oh. I don't know that a char-table.
I modified a structure of hanja-table and adjusted
comments.

"Maybe" in hangul2-input-method-jaum,
hangul2-input-method-moum,
hangul3-input-method-cho, hangul3-input-method-jung,
hangul3-input-method-jong are deleted. It is checked by a
calling
function. Therefore, always satisfies the conditions.

The diff file is too large. So I attached .el files.

2008/6/6 Jihyun Cho <jihyun.jogmail.com>:
> 2008/6/5 Kenichi Handa <handam17n.org>:
>> In article
<9d644d9b0806040914v46839e87v48afdff983c3980email.gmail.com>, "Jihyun Cho"
<jihyun.jogmail.com> writes:
>>
>>> I have adjusted comments and docstring.
>>> My English is not good.
>>> If you can not be understood, let me know it.
>>
>> Thank you I've just committed your change with
some
>> improvement.  Please check if my understanding is
correct
>> with the latest code.
>>
>> But there are a few places I still don't
understand.
>>
>> In hangul.el...
>>
>> (1)
>>
>> ;; Current input character buffer. Store separated
hangul character.
>> ;; First and second index of vector stored
"Choseong".
>> ;; Third and forth index of vector stored
"Jungseong".
>> ;; Fifth and sixth index of vector stored
"Jongseong".
>> (defvar hangul-queue
>>  (make-vector 6 0))
>>
>> I don't understand 2nd to 4th lines.
>>
>> (2)
>>
>> (defsubst hangul2-input-method-jaum (char)
>>  "Store hangul jamo indice CHAR in
`hangul-queue'.
>> Maybe, it is a Hangul 2-Bulsik jaum.
>> This function processes a Hangul 2-Bulsik jaum.
>> Unless the function inserts CHAR to current input
queue,
>> commit current character and then start next
character.
>> The Hangul 2-Bulsik is composed of `jaum' and
`moum'."
>>
>> "Maybe" is strange.  If CHAR is not a
"Hangul 2-Bulsik
>> jaum", what it will be.
>>
>> In what condition, it doesn't insert CHAR to the
queue.
>>
>> What does "start next character" mean? 
Where is the "next
>> character"?
>>
>> The same for hangul2-input-method-moum,
>> hangul3-input-method-cho,
hangul3-input-method-jung,
>> hangul3-input-method-jong.
>>
>> In hanja-util.el...
>>
>> (1)
>>
>> ;; List of current conversion status.
>> (defvar hanja-conversions
>>  nil)
>>
>> What format does it have?
>>
>> (2)
>>
>> (defun hanja-filter (condp lst)
>>  "Return list.
>> It contains true elements to evaluate CONDP with
each element from LST."
>>
>> I don't understand it.
>>
>> By the way, why do you use a hash-table for
hanja-table?
>> Isn't a char-table more efficient?  And, the value
is now
>> "(CHAR) or (CHAR . STRING)", but most of
elements are
>> (CHAR).  I think it's a waste of cons cells.  How
about
>> changing the format to "CHAR or (CHAR .
STRING)"?  It seems
>> easy to handle the changed format in
hangul-to-hanja-char.
>>
>> ---
>> Kenichi Handa
>> handani.aist.go.jp
>>
>
> Oh. I don't know that a char-table.
> I modified a structure of hanja-table and adjusted
comments.
>
> "Maybe" in hangul2-input-method-jaum,
hangul2-input-method-moum,
> hangul3-input-method-cho, hangul3-input-method-jung,
> hangul3-input-method-jong are deleted. It is checked by
a calling
> function. Therefore, always satisfies the conditions.
>
> The diff file is too large. So I attached .el files.
>

I found a bug.
That bug was occurred when the program could not find a
character in
hanja-table.
It was occurring after hanja-table was modified, and I fixed
it.

In article
<9d644d9b0806051257n25d43302p3c2e187fb48e387mail.gmail.com>, "Jihyun Cho"
<jihyun.jogmail.com> writes:

> Oh. I don't know that a char-table.
> I modified a structure of hanja-table and adjusted
comments.

> "Maybe" in hangul2-input-method-jaum,
hangul2-input-method-moum,
> hangul3-input-method-cho, hangul3-input-method-jung,
> hangul3-input-method-jong are deleted. It is checked by
a calling
> function. Therefore, always satisfies the conditions.

> The diff file is too large. So I attached .el files.

I installed them with slight modifications (fix arguments
to
make-char-table, use mapc instead of mapcar, improve
several
docstrings).

> I found a bug.
> That bug was occurred when the program could not find a
character in
> hanja-table.
> It was occurring after hanja-table was modified, and I
fixed it.

?? You change is only to hanja-init-load (when to show
message), and doesn't seem to solve the problem you wrote
above.

Anyway, please make a diff always with the latest CVS code.

---
Kenichi Handa
handani.aist.go.jp

2008/6/9 Kenichi Handa <handam17n.org>:
> In article
<9d644d9b0806051257n25d43302p3c2e187fb48e387mail.gmail.com>, "Jihyun Cho"
<jihyun.jogmail.com> writes:
>
>> Oh. I don't know that a char-table.
>> I modified a structure of hanja-table and adjusted
comments.
>
>> "Maybe" in hangul2-input-method-jaum,
hangul2-input-method-moum,
>> hangul3-input-method-cho,
hangul3-input-method-jung,
>> hangul3-input-method-jong are deleted. It is
checked by a calling
>> function. Therefore, always satisfies the
conditions.
>
>> The diff file is too large. So I attached .el
files.
>
> I installed them with slight modifications (fix
arguments to
> make-char-table, use mapc instead of mapcar, improve
several
> docstrings).
>
>> I found a bug.
>> That bug was occurred when the program could not
find a character in
>> hanja-table.
>> It was occurring after hanja-table was modified,
and I fixed it.
>
> ?? You change is only to hanja-init-load (when to show
> message), and doesn't seem to solve the problem you
wrote
> above.
>
> Anyway, please make a diff always with the latest CVS
code.

OK. I'm sorry.
The error message is "hanja-filter: Wrong type
argument: sequencep,
0", when "(make-char-table 'char-table-extra-slots
0)".
So, I modified "(make-char-table nil)", because
hanja-filter's second
argument is list type.