[TxMt] Re: Cocoa Autocomplete

[Hans-Joerg Bibiko]

Hi Sam,


On 11 Jan 2007, at 13:31, Sam Halliday wrote:

> I see what your script is doing there and it would be sensible but  
> for the fact that I would require British (not American) English  
> plus my own personal additions to the dictionary.
>
> I could get around the British restraint by using a British file  
> from somewhere (possibly aspell), and my personal additions as  
> located in "Library/Spelling/en_GB" (but uses some odd characters  
> and terminators, so regex definitely needed there).
>

OK. I can see the problem with the bunch of English varieties.

> Perhaps you might want to generalise your script to include the  
> user additions?

What do you mean with to 'generalise'?


The 'only' thing to do is to create your own dict file. The dict file  
is nothing else than a list of words, one word one line.

If you need a British English wordlist you can download such, e.g.,  
via http://packages.debian.org/unstable/text/ (here one can find also  
other languages), look for 'wbritish-huge', download it, use the  
Terminal command 'dpkg-deb -x' to extract the package. After that you  
have a British wordlist of roughly 57000 words. Add your own  
additions to that file and replace '/usr/share/dict/webs2' with the  
location of your own file in the Ruby script. Furthermore you can  
merge this Debian British wordlist with Debian's wamerican-huge list  
and with the ca. 250000 Webster words, remove the duplicates, and at  
the end you have a British/American/Webster wordlist of ca. 280.000  
words.

That means in other words, my approach is based only on a wordlist  
written as a plain text file!


Cheers,

Hans