Martin Kolar?k wrote:
>
> I have two questions of spell checker in XXE:
>
> 1. I tried to make a czech dictionary. I set up *.hints, *.freq and *.wl
> files, all three with ISO8859-2 encoding. I built dictionary with option -cs
> ISO8859_2 and the builder reported a plenty of words, which were rejected --
> these words contained characters scaron, Scaron, zcaron, Zcaron,... These
> character are these one, that are different between ISO8859-2 and Win1250,
> so I tried to convert dictionary files to Win1250. But, some of "bad"
> letters dissapeared, some remained (scaron, Scaron). So I tried to run
> builder WITHOUT -cs option and the builder built dictionary without
> rejecting any word. (I am SURE I have the dictionary files coded properly in
> ISO8859-2. :-) Is this correct behaviour?
>
> 2. I made *.dar file from cs.cdi, default and language and put it into XXE's
> spell directory. But no czech language appeared in spell tool in XXE. Where
> should I this configure or what do I wrong?
Here's the answer of our spell-checker expert:
----------------------------------------------
1. There is visibly an inconsistency between the Hints file and
the word list. The last thing you did is not correct because
in fact your characters are considered ISO8859_1 (the default)
You have to put the -cs option *first*, because it applies to all
files that follow:
C> builder -cs ISO8859_2 -hints cs.hints -freq cs.freq cs.wl -o cs.cdi
[ Sorry for the too short documentation ]
2. To be properly recognized, the contents of your .dar archive
should look like this:
C> jar tf cs.dar
META-INF/
META-INF/MANIFEST.MF
cs/
cs/language
cs/default
cs/base.cdi
If you use:
C> jar cf cs.dar cs
instead of using zip or WinZip, cs.dar will
automatically include the requested META-INF*.
------------------------------------------------