Getting the right dictionary for e-mail and newsgroup messages

Getting the right dictionary for e-mail and newsgroup messages

[Cecil Westerhof]

At the moment I use two languages for my messages: Dutch and English. So
I need to have the dictionary to be set correctly. Therefore I wrote the
following code:
    (defvar gnus-dictionaries
      '(
        ("^nl\\.\\|\\.nl\\.\\|\\.nl$"  . "dutch")
        )
    "A list of group names and dictionaries to use.")

    (defun gnus-set-dictionary ()
      "Determine what dictionary to use based on the current message."
      (let ((dictionary))
        (if (message-news-p)
            (setq dictionary "british")
          (setq dictionary "dutch"))
        (dolist (item gnus-dictionaries)
          (when (string-match (car item) gnus-newsgroup-name)
            (setq dictionary (cdr item))))
        (ispell-change-dictionary dictionary)))

    (add-hook 'message-mode-hook 'gnus-set-dictionary)

Default e-mail is in Dutch and newsgroup messages in English. But
sometimes it could be that it is otherwise. For example Dutch newsgroups
wants Dutch Messages. That is why I use gnus-dictionaries to check
for an exception.

I only have one problem. At the moment I need to use:
    "^nl\\.\\|\\.nl\\.\\|\\.nl$" 
for the regular expression. I would prefer to use something like:
    "[\\.^]nl[\\.$]"

But that does not work. Is there another way to make the regular
expression simpler?

-- 
Cecil Westerhof
Senior Software Engineer
LinkedIn: http://www.linkedin.com/in/cecilwesterhof

Re: Getting the right dictionary for e-mail and newsgroup messages

[Ted Zlatanov]

On Fri, 25 Dec 2009 11:08:27 +0100 Cecil Westerhof <Cecil@decebal.nl> wrote: 

CW> I only have one problem. At the moment I need to use:
CW>     "^nl\\.\\|\\.nl\\.\\|\\.nl$" 
CW> for the regular expression. I would prefer to use something like:
CW>     "[\\.^]nl[\\.$]"

CW> But that does not work. Is there another way to make the regular
CW> expression simpler?

The simplest solution is probably to use split-string, since you'll get
all the path components that way:

(member "nl" (split-string "X.nl.X" "\\."))

Your regex character class of [\\.^] doesn't work because it's matching
the character ^ and not the beginning of the line.  Same for the [\\.$]
class.

If the '.' character is not in your word class (it shouldn't be), you
can use

(string-match "\\bnl\\b" "X.nl.X")

which is probably the best regex-based solution, so it will work with
your existing code.  You could also use \< and \> but that's probably
unnecessary.  Look at the ELisp manual, section "Backslash Constructs in
Regular Expressions" for details.

Ted

Re: Getting the right dictionary for e-mail and newsgroup messages

[Cecil Westerhof]

Ted Zlatanov <tzz@lifelogs.com> writes:

> CW> I only have one problem. At the moment I need to use:
> CW>     "^nl\\.\\|\\.nl\\.\\|\\.nl$" 
> CW> for the regular expression. I would prefer to use something like:
> CW>     "[\\.^]nl[\\.$]"
>
> CW> But that does not work. Is there another way to make the regular
> CW> expression simpler?
> If the '.' character is not in your word class (it shouldn't be), you
> can use
>
> (string-match "\\bnl\\b" "X.nl.X")

This also matches:
    (string-match "\\bnl\\b" "X-nl-X")

But I do not think that is a problem. So I now use "\\bnl\\b". That is a
lot clearer as "^nl\\.\\|\\.nl\\.\\|\\.nl$" and easier to adopt when
another language has to be added.

Thanks.

-- 
Cecil Westerhof
Senior Software Engineer
LinkedIn: http://www.linkedin.com/in/cecilwesterhof

Re: Getting the right dictionary for e-mail and newsgroup messages

[Ted Zlatanov]

On Tue, 05 Jan 2010 21:47:34 +0100 Cecil Westerhof <Cecil@decebal.nl> wrote: 

>> (string-match "\\bnl\\b" "X.nl.X")

CW> This also matches:
CW>     (string-match "\\bnl\\b" "X-nl-X")

CW> But I do not think that is a problem.

I think you can modify the syntax table to accomodate this, making '-' a
member of the word class, but you have to check the manual for the
details.

Ted

Re: Getting the right dictionary for e-mail and newsgroup messages

[Cecil Westerhof]

Ted Zlatanov <tzz@lifelogs.com> writes:

>>> (string-match "\\bnl\\b" "X.nl.X")
>
> CW> This also matches:
> CW>     (string-match "\\bnl\\b" "X-nl-X")
>
> CW> But I do not think that is a problem.
>
> I think you can modify the syntax table to accomodate this, making '-' a
> member of the word class, but you have to check the manual for the
> details.

I would need to put a lot more in the word class, because the original
problem was that before nl needs to be a '.' or the beginning of the
line and after a '.' or the end of the line. So in principal the regular
expression is to lenient, but the chance that it is a real problem is
very small. When that is the case, I need to use my original expression
again. But it is used for parsing newsgroup names, so I think the
expression is good enough.

-- 
Cecil Westerhof
Senior Software Engineer
LinkedIn: http://www.linkedin.com/in/cecilwesterhof