Hi,
[email protected] wrote:
>> Looking at two of the delta files it seems like they are simply a list
>> off add-on entries to hunspell. Thus even chrome does not seem to have
>> the ability to remove/fix misspelled words from the dictionary
>> content... (unless they get removed by a patch)
>
> Removing misspelled words must always remain under
> control of dictionary maintainer for obvious reasons.
> And also the adding of missing words.
>
>> As for the .bdic file, I have no idea about that one. ^^°
>
> I assume, some internal google format, like aspell's internal
> format. We can safely ignore that.
>
>> I still like the old idea to replace the OOo user-dictionaries by
>> hunspell dictionaries (or at least a format that hunspell can read).
>> But that would require hunspell to read 'negative dictionaries' (often
>> called exception dictionaries) and to provide user-supplied suggestions
>> for those entries.
>
> Hunspell does a good job at suggestions. (I personally never
> use them, but others, who use them, state the above.) Hunspell
> does not need any "bad word list" to list the suggestions,
> it relies on the replacement patterns in the affix file.
> Could you please give an example, to understand what you mean here?
Hunspell may not have need for that, but users have. ^_-
There are two basic usages:
- First: exception dictionaries usually consist of correct words that
you don't like to use in your text or context for some reason.
a) please consider writing a fairy tale for children to read, there are
a lot of words in regular English that you don't want to appear in
there. (Though for that we may better have an English-Child-Safe
dictionary). But it could also be done by a larger exception dictionary.
b) You (or your company) may have a list of words that you are not to
use in your public documents.
Or maybe of two possible and valid choices you still want to use only
one. For example in German, according to the latest spelling reform, we
can either write dolphin as 'Delfin' or 'Delphin' both are valid, but
you don't want them both to appear in a single text. One way to solve
this is to declare one of them as an exception (and to provide the other
as suggestion).
Those words can then be added to an exception dictionary and hence forth
the spell checker should complain about them.
- Second: It allows the user to customize the spelling suggestions.
If for example you tend to make the typo 'rigth' then you could add that
word to an exception dictionary and by providing only a single
suggestion ('right') one would expect the spell checker to return onyl
that one (and none from it's dictionary base) or at least to put that
single word at the top of the suggestion list.
And of course you should be allowed to make more than one suggestion
(OOo currently does not allow for that though), and again the list
should replace the list returned by hunspell or hunspell should add that
word list at the top of the words itself has found.
>>If we then can also have means for a 'Language All'
>> dictionary then we could replace the user-dictionaries by hunspell
>> compatible ones, and that would be a nice thing to do I believe.
>
> Please explain, what do you mean with "language all" dictionary,
> best with some examples.
A 'Language All' dictionary will be a list of words that are correct
that way in ALL languages (usually because they won't get translated).
Common examples are peoples or company names.
E.g.
OpenOffice.org
ASCII
HTML
Thomas
Alva
Edison
If you are writing multilingual documents or if you have a server
installation with a number of multi lingual users, you can add all those
words that would be spelled the same regardless of the texts language in
a single dictionary instead of creating a dictionary for each of those
languages.
And then, for every language and word the spell checker has always to
look up into those dictionaries of 'Language All' as well before
deciding to declare a word as misspelled.
Regards,
Thomas
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
