Woman path : adding l10n man pages

Woman path : adding l10n man pages

[Frédéric Perrin]

Hello,

Would it be possible to add to Woman's default manpath the local
manpages ? Something like :

(file-expand-wildcards (format "/usr/share/man/%s*"
                               (substring (getenv "LANG") 0 2)))

I'm adding /usr/share/man/fr* because, at least on my system (Debian
GNU/Linux Sid), I'm finding most French manpages in /u/s/m/fr/, plus
a handful of others in /u/s/m/fr.UTF-8/ and fr.ISO-8859-1/, when my
locale is LANG=fr_FR.UTF-8.

Thanks in advance,

-- 
Frédéric Perrin

Re: Woman path : adding l10n man pages

[Frédéric Perrin]

Le Mon, 31 Dec 2007 14:47:31 +0100, Frédéric Perrin
<frederic.perrin@resel.fr> a écrit :

> Hello,
> 
> Would it be possible to add to Woman's default manpath the local
> manpages ?

I manage to write a patch that does this, by adding /fr*/ to every
path found in manpath.config. Would it be possible to install it ?

Here it is :
 
$ diff woman.el.~1.16.4.29~ woman.el
596,600c596,609
< 		      (add-to-list 'manpath
< 				   (if (match-beginning 1)
< 				       (match-string 1)
< 				     (cons (match-string 2)
< 					   (match-string 3)))))
---
> 		      (if (match-beginning 1)
> 			  (let ((new-path (match-string 1)))
> 			    (add-to-list 'manpath new-path)
> 			    ;; adding localized manpages
> 			    (dolist
> 				(loc-path
> 				 (file-expand-wildcards 
> 				  (format "%s/%s*"
> 					  new-path
> 					  (substring (getenv
> "LANG") 0 2)))) (add-to-list 'manpath
> 					   loc-path)))
> 			(add-to-list 'manpath (cons (match-string 2)
> 						    (match-string
> 3)))))


-- 
Fred

Re: Woman path : adding l10n man pages

[Andreas Schwab]

Frédéric Perrin <frederic.perrin@resel.fr> writes:

>> 					  (substring (getenv
>> "LANG") 0 2)))) (add-to-list 'manpath

LANG can also contain a language alias.  There is no requirement that
the first two character have anything to do with the language code.
Moreover, the language should actually be derived from LC_MESSAGES.

Andreas.

-- 
Andreas Schwab, SuSE Labs, schwab@suse.de
SuSE Linux Products GmbH, Maxfeldstraße 5, 90409 Nürnberg, Germany
PGP key fingerprint = 58CA 54C7 6D53 942B 1756  01D3 44D5 214B 8276 4ED5
"And now for something completely different."

Re: Woman path : adding l10n man pages

[Frédéric Perrin]

Le Tue, 08 Jan 2008 16:38:04 +0100, Andreas Schwab
<schwab@suse.de> a écrit :

> Frédéric Perrin <frederic.perrin@resel.fr> writes:
> 
> >> 					  (substring (getenv
> >> "LANG") 0 2)))) (add-to-list 'manpath
> 
> LANG can also contain a language alias.  There is no requirement
> that the first two character have anything to do with the language
> code. Moreover, the language should actually be derived from
> LC_MESSAGES.
> 
> Andreas.

I didn't know that. Changing LANG into LC_MESSAGES is trivial, but I
cannot parse it so as to get the user's language's two letter code. I
don't know the possible aliases, and even if I did, I wouldn't be
able to do it in Lisp.

Could someone do that ? I would be very nice to have woman display
localised man pages, just like man(1) does.

-- 
Fred

Re: Woman path : adding l10n man pages

[Frédéric Perrin]

Le Tue, 08 Jan 2008 16:38:04 +0100, Andreas Schwab <schwab@suse.de> a
écrit :

> Frédéric Perrin <frederic.perrin@resel.fr> writes:
> 
> >> 					  (substring (getenv
> >> "LANG") 0 2)))) (add-to-list 'manpath
> 
> LANG can also contain a language alias.  There is no requirement
> that the first two character have anything to do with the language
> code. Moreover, the language should actually be derived from
> LC_MESSAGES.
> 
> Andreas.

I'm sorry I have to insist, but is somebody willing to do it
correctly, i.e. adding localised man pages based on LC_MESSAGES'
value, and parsing it correctly ?

I'm afraid I'm not able to do it; nevertheless it would be very nice
for users to have man pages in their own languages. If it is not
done, well, I'll keep my own patch for myself.

-- 
Fred

Re: Woman path : adding l10n man pages

[Glenn Morris]

Andreas Schwab wrote:

> LANG can also contain a language alias.  There is no requirement that
> the first two character have anything to do with the language code.
> Moreover, the language should actually be derived from LC_MESSAGES.

Please try this patch. I have no idea if `woman-expand-locale' is
returning a sensible list of alternatives.


Index: mule-cmds.el
===================================================================
RCS file: /sources/emacs/emacs/lisp/international/mule-cmds.el,v
retrieving revision 1.320
diff -c -c -w -r1.320 mule-cmds.el
*** mule-cmds.el	1 Feb 2008 16:01:17 -0000	1.320
--- mule-cmds.el	4 Feb 2008 19:51:39 -0000
***************
*** 2422,2427 ****
--- 2422,2440 ----
  ;; too, for setting things such as calendar holidays, ps-print paper
  ;; size, spelling dictionary.
  
+ (defun locale-translate (locale)
+   "Expand LOCALE according to `locale-translation-file-name', if possible.
+ For example, translate \"swedish\" into \"sv_SE.ISO8859-1\"."
+   (if locale-translation-file-name
+       (with-temp-buffer
+         (set-buffer-multibyte nil)
+         (insert-file-contents locale-translation-file-name)
+         (if (re-search-forward
+              (concat "^" (regexp-quote locale) ":?[ \t]+") nil t)
+             (buffer-substring (point) (line-end-position))
+           locale))
+     locale))
+ 
  (defun set-locale-environment (&optional locale-name frame)
    "Set up multi-lingual environment for using LOCALE-NAME.
  This sets the language environment, the coding system priority,
***************
*** 2491,2506 ****
        (setq locale mac-system-locale))
  
      (when locale
! 
!       ;; Translate "swedish" into "sv_SE.ISO8859-1", and so on,
!       ;; using the translation file that many systems have.
!       (when locale-translation-file-name
! 	(with-temp-buffer
! 	  (set-buffer-multibyte nil)
! 	  (insert-file-contents locale-translation-file-name)
! 	  (when (re-search-forward
! 		 (concat "^" (regexp-quote locale) ":?[ \t]+") nil t)
! 	    (setq locale (buffer-substring (point) (line-end-position))))))
  
        ;; Leave the system locales alone if the caller did not specify
        ;; an explicit locale name, as their defaults are set from
--- 2504,2510 ----
        (setq locale mac-system-locale))
  
      (when locale
!       (setq locale (locale-translate locale))
  
        ;; Leave the system locales alone if the caller did not specify
        ;; an explicit locale name, as their defaults are set from
***************
*** 2508,2515 ****
        ;; want to set them to the same value as LC_CTYPE.
        (when locale-name
  	(setq system-messages-locale locale)
! 	(setq system-time-locale locale))
  
        (setq locale (downcase locale))
  
        (let ((language-name
--- 2512,2527 ----
        ;; want to set them to the same value as LC_CTYPE.
        (when locale-name
  	(setq system-messages-locale locale)
! 	(setq system-time-locale locale)))
! 
!     (setq woman-locale
!           (or system-messages-locale
!               (let ((msglocale (getenv "LC_MESSAGES")))
!                 (if (zerop (length msglocale))
!                     locale
!                   (locale-translate msglocale)))))
  
+     (when locale
        (setq locale (downcase locale))
  
        (let ((language-name
Index: woman.el
===================================================================
RCS file: /sources/emacs/emacs/lisp/woman.el,v
retrieving revision 1.67
diff -c -c -w -r1.67 woman.el
*** woman.el	12 Jan 2008 23:21:13 -0000	1.67
--- woman.el	4 Feb 2008 19:56:37 -0000
***************
*** 603,610 ****
        (setq path (cdr path)))
      (nreverse manpath)))
  
  (defcustom woman-manpath
!   (or (woman-parse-colon-path (getenv "MANPATH"))
        '("/usr/man" "/usr/share/man" "/usr/local/man"))
    "List of DIRECTORY TREES to search for UN*X manual files.
  Each element should be the name of a directory that contains
--- 603,661 ----
        (setq path (cdr path)))
      (nreverse manpath)))
  
+ ;;;###autoload
+ (defcustom woman-locale nil
+   "String specifying a manual page locale, or nil.
+ If a manual page is available in the specified locale, it will be
+ offered in preference to the default version.  Normally,
+ `set-locale-environment' sets this at startup."
+   :type '(choice string (const nil))
+   :group 'woman-interface
+   :version "23.1")
+ 
+ (defun woman-expand-locale (locale)
+   "Expand a locale into a list suitable for man page lookup.
+ Expands a locale of the form LANGUAGE_TERRITORY.CHARSET into the list:
+ LANGUAGE_TERRITORY.CHARSET LANGUAGE_TERRITORY LANGUAGE.CHARSET LANGUAGE.
+ The TERRITORY and CHARSET portions may be absent."
+   (string-match "\\([^._]*\\)\\(_[^.]*\\)?\\(\\..*\\)?" locale)
+   (let ((lang (match-string 1 locale))
+         (terr (match-string 2 locale))
+         (charset (match-string 3 locale)))
+     (delq nil (list locale
+                     (and charset terr (concat lang terr))
+                     (and charset terr (concat lang charset))
+                     (if (or charset terr) lang)))))
+ 
+ (defun woman-manpath-add-locales (manpath)
+   "Add locale-specific subdirectories to the elements of MANPATH.
+ MANPATH is a list of the form of `woman-manpath'.  Returns a list
+ with those locale-specific subdirectories specified by the action
+ of `woman-expand-locale' on `woman-locale' added, where they exist."
+   (if (zerop (length woman-locale))
+       manpath
+     (let ((subdirs (woman-expand-locale woman-locale))
+           lst dir)
+       (dolist (elem manpath (nreverse lst))
+         (dolist (sub subdirs)
+           (when (file-directory-p
+                  (setq dir
+                        ;; Use f-n-a-d because parse-colon-path does.
+                        (file-name-as-directory
+                         (expand-file-name sub (substitute-in-file-name
+                                                (if (consp elem)
+                                                    (cdr elem)
+                                                  elem))))))
+             (add-to-list 'lst (if (consp elem)
+                                   (cons (car elem) dir)
+                                 dir))))
+         ;; Non-locale-specific has lowest precedence.
+         (add-to-list 'lst elem)))))
+ 
  (defcustom woman-manpath
!   ;; Locales could be added in woman-expand-directory-path.
!   (or (woman-manpath-add-locales
!        (woman-parse-colon-path (getenv "MANPATH")))
        '("/usr/man" "/usr/share/man" "/usr/local/man"))
    "List of DIRECTORY TREES to search for UN*X manual files.
  Each element should be the name of a directory that contains

Re: Woman path : adding l10n man pages

[Frédéric Perrin]

Le Mon, 04 Feb 2008 14:58:34 -0500, Glenn Morris <rgm@gnu.org> a
écrit :

> Andreas Schwab wrote:
> 
> > LANG can also contain a language alias.  There is no requirement
> > that the first two character have anything to do with the
> > language code. Moreover, the language should actually be derived
> > from LC_MESSAGES.
> 
> Please try this patch. I have no idea if `woman-expand-locale' is
> returning a sensible list of alternatives.
> 
> [patch snipped]

As far as I can tell (but then again, I don't necessarily have the
required expertise), Morris' patch works as expected, with « M-x woman
RET su RET » allowing me to see the French manpage for su (in fact,
it gives me the choice between /u/s/man/fr/man1/su.1.gz
and /u/s/man/man1/su.1.gz). Similarly, starting emacs with
« LC_MESSAGES=spanish ./emacs/src/emacs » (all other $LANG and
$LC_* variables being set to fr_FR.UTF-8) gives me manpages in Spanish
(well, for the handful of them which are present on my system).

I suggest installing this patch.

Please CC me in further discussions, as I'm no longer subscribed to
the mailing list.

-- 
Fred

Re: Woman path : adding l10n man pages

[Glenn Morris]

Frédéric Perrin wrote:

> I suggest installing this patch.

Thanks for testing it; it's installed.

Re: Woman path : adding l10n man pages

[Juri Linkov]

>> I suggest installing this patch.
>
> Thanks for testing it; it's installed.

Setting `woman-locale' in mule-cmds.el doesn't seem like the right thing.
Is it possible to move setting `woman-locale' to woman.el?

-- 
Juri Linkov
http://www.jurta.org/emacs/

Re: Woman path : adding l10n man pages

[Glenn Morris]

Juri Linkov wrote:

> Setting `woman-locale' in mule-cmds.el doesn't seem like the right thing.

Other things like ps-paper-type are already set there. This seems similar.
Indeed, I think it _has_ to stay there since in principle I'm supposed
to be able to use `(set-locale-environment locale-name)' to set up all
my locale-specific stuff.

Re: Woman path : adding l10n man pages

[Juri Linkov]

>> Setting `woman-locale' in mule-cmds.el doesn't seem like the right thing.
>
> Other things like ps-paper-type are already set there. This seems similar.
> Indeed, I think it _has_ to stay there since in principle I'm supposed
> to be able to use `(set-locale-environment locale-name)' to set up all
> my locale-specific stuff.

This doesn't justify leaving the woman-specific logic in mule-cmds.el.
I think set-locale-environment can just set the general locale
variables, and other packages can use them for their purposes
(e.g. woman-manpath-add-locales can have the same code that currently
sets woman-locale in mule-cmds.el).

Also I don't understand why woman-locale is defcustom?  A defcustom
option is supposed to be customized by the user, but the customized
value of woman-locale always gets overridden at startup and on every
call to `set-locale-environment'.  Perhaps the code that uses the locale
in woman.el should prefer the customized value if woman-locale is non-nil?

-- 
Juri Linkov
http://www.jurta.org/emacs/

Re: Woman path : adding l10n man pages

[Glenn Morris]

Juri Linkov wrote:

> (e.g. woman-manpath-add-locales can have the same code that
> currently sets woman-locale in mule-cmds.el).

The technical obstacle is that `locale', the fall-back value, isn't
available outside set-locale-environment (AFAICS).

> Also I don't understand why woman-locale is defcustom?  A defcustom
> option is supposed to be customized by the user, but the customized
> value of woman-locale always gets overridden at startup 

I had assumed that `set-locale-environment' runs before processing of
customization options, and by experiment this is the case.

LC_MESSAGES=en_US emacs --no-site-file
  -> woman-locale = en_US.ISO8859-1

Customize woman-locale to nil, save and exit

LC_MESSAGES=en_US emacs --no-site-file
  -> woman-locale = nil

> and on every call to `set-locale-environment'.

Arguably that is what should happen.

> Perhaps the code that uses the locale in woman.el should prefer the
> customized value if woman-locale is non-nil?

I see what you're saying and perhaps you're right. There could be a
general locale variable which woman should use, unless woman-locale is
non-nil, in which case it should be preferred.

But AFAICS there is no such general locale variable at present (no
Lisp variable, including system-messages-locale, seems to be set based
on LC_MESSAGES), so I'd have to add one, which would only be used
by... woman.

Re: Woman path : adding l10n man pages

[Juri Linkov]

>> Perhaps the code that uses the locale in woman.el should prefer the
>> customized value if woman-locale is non-nil?
>
> I see what you're saying and perhaps you're right. There could be a
> general locale variable which woman should use, unless woman-locale is
> non-nil, in which case it should be preferred.
>
> But AFAICS there is no such general locale variable at present (no
> Lisp variable, including system-messages-locale, seems to be set based
> on LC_MESSAGES), so I'd have to add one, which would only be used
> by... woman.

I think the best course of action in the long term is to add such
a general locale variable that could be later used by other packages
besides woman.

-- 
Juri Linkov
http://www.jurta.org/emacs/

Re: Woman path : adding l10n man pages

[Glenn Morris]

Juri Linkov wrote:

> I think the best course of action in the long term is to add such
> a general locale variable that could be later used by other packages
> besides woman.

OK, agreed. The usual stupid problem - what should it be called? There
is already system-messages-locale, which is set and used in ways I
don't understand, eg:

LC_MESSAGES=en_US.iso88591 emacs -Q
  -> system-messages-locale = nil

Re: Woman path : adding l10n man pages

[Juri Linkov]

>> I think the best course of action in the long term is to add such
>> a general locale variable that could be later used by other packages
>> besides woman.
>
> OK, agreed. The usual stupid problem - what should it be called? There
> is already system-messages-locale, which is set and used in ways I
> don't understand, eg:
>
> LC_MESSAGES=en_US.iso88591 emacs -Q
>   -> system-messages-locale = nil

As I see the current locale system in Emacs is quite a mess.  I think we
should carefully redesign it and also get rid of language environments
that are in fact a redundant duplicate of the locale system.

-- 
Juri Linkov
http://www.jurta.org/emacs/

Re: Woman path : adding l10n man pages

[Miles Bader]

Juri Linkov <juri@jurta.org> writes:
> As I see the current locale system in Emacs is quite a mess.  I think we
> should carefully redesign it and also get rid of language environments
> that are in fact a redundant duplicate of the locale system.

I suppose it depends on what you mean by "the locale system" --
obviously the _mechanism_ use by the traditional locale system, with
environment variables etc, doesn't fit Emacs very well; Emacs needs
things to be more dynamic and flexible.

However it might be nice to have more coordination between the two, for
emacs-external program interfaces.  [There is already some, e.g.,
starting Emacs with LANG set usually does the right thing; I don't know
how pervasive this is, though...]

-Miles

-- 
Admiration, n. Our polite recognition of another's resemblance to ourselves.

Re: Woman path : adding l10n man pages

[Juri Linkov]

>> As I see the current locale system in Emacs is quite a mess.  I think we
>> should carefully redesign it and also get rid of language environments
>> that are in fact a redundant duplicate of the locale system.
>
> I suppose it depends on what you mean by "the locale system" --
> obviously the _mechanism_ use by the traditional locale system, with
> environment variables etc, doesn't fit Emacs very well; Emacs needs
> things to be more dynamic and flexible.

There are two related tasks in etc/TODO that I hope will be implemented:

** Improve language environment handling so that Emacs can fit
  better to a users locale.  Currently Emacs uses utf-8 language
  environment for all utf-8 locales, thus a user in ja_JP.UTF-8 locale
  are also put in utf-8 lang. env.  In such a case, it is
  better to use Japanese lang. env. but prefer utf-8 coding system.

** Enhance locale handling:  handle language, territory and charset
  orthogonally and de-emphasize language environments.  Use the locale
  to set up more things, such as fontsets, the default Ispell
  dictionary, diary format, calendar holidays and display, quoting
  characters and phrase boundaries, sentence endings, collation for
  sorting (at least for unicodes), HTTP Accept-language, patterns for
  directory listings and compilation messages, yes-or-no replies,
  common menu items when the toolkit supports it ...  `locale-info'
  needs extending for LC_COLLATE &c.  [fx started on this.]

-- 
Juri Linkov
http://www.jurta.org/emacs/

Re: Woman path : adding l10n man pages

[Miles Bader]

Juri Linkov <juri@jurta.org> writes:
> ** Improve language environment handling so that Emacs can fit
>   better to a users locale.  Currently Emacs uses utf-8 language
>   environment for all utf-8 locales, thus a user in ja_JP.UTF-8 locale
>   are also put in utf-8 lang. env.

This seems to be wrong:  If I do "LANG=ja_JP.UTF-8 emacs -Q -nw"
the resulting language environment is "Japanese".

-Miles

-- 
Barometer, n. An ingenious instrument which indicates what kind of weather we
are having.