how to add "wrong" words to dictionary?

[bittorf]

i want to add 'churchs' as a good word (a shoebrand!)
to my private dictionary, but codespell always want
to correct it with churchs->churches

what is the proposed way to do this?

[thdot]

You could use the --exclude-file option to exclude the lines containing the good word explicitly.
Not the most elegant way, I know.

[bittorf]

a more elegant solution would be a comment-statement in the line with e.g.

codespell: churchs

[lucasdemarchi]

@bittorf there isn't a way to do that right now. I think it would be reasonable to add a new option to pass a file with "ignore-words", or to extend the dictionary format to allow passing words as good ones.

For the first option it would be something like:
codespell --ignore-words /path/to/good/dictionary.txt

While for the second we would support in the dictionary a syntax like:
-churchs

So you could use something like thi:
codespell -D - -D /path/to/dictionary/extensions/including/good/words

@anatol @pabs3 @Eric89GXL @thdot thoughts

[bittorf]

@lucasdemarchi that is not the way i like it:

the given example is real ("churchs") but of course i do not want
to loose real occurences of a misspelled "church". think of a large file,
where only 1 thing should be overridden...

[lucasdemarchi]

I don't like the approach of adding comments in code to tweak the behavior of codespell. Even worse that codespell doesn't know what's a comment. I'd prefer to tell people to either use the approaches suggested above, an exclude file. Other option: use a "2 pass approach", excluding that word I the first and using a dictionary with only that word in the second, but in iterative mode. This way codespell will ask for confirmation if the word should be changed.

…

On Jan 4, 2017 11:21 AM, "Bastian Bittorf" ***@***.***> wrote: @lucasdemarchi <https://github.com/lucasdemarchi> that is not the way i like it: the given example is real ("churchs") but of course i do not want to loose real occurences of a misspelled "church". think of a large file, where only 1 thing should be overridden... — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub <#66 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AAB8eHKaS_nu9I0M27pGiypKu53yLUPTks5rO_EhgaJpZM4IPUe3> .

[thdot]

I agree with Lucas, adding comments into the code to influence codespell isn't a sustainable way. Personally I like the idea to extend the dictionary syntax, but a word exclude file is fine for me as well.

@bittorf: if you want to ignore just one occurrence of "churchs" I think the --exclude-file option is a good way to achieve that.

[bittorf]

@thdot - i can work with the --exclude-file approach, but it is not elegant. having statements in the file is an old and good way to talk with a compiler. see e.g. GCC or 'shellcheck': https://github.com/koalaman/shellcheck/wiki/Directive

having extra files describing an issue feels like code duplication. what is if a line moves up or down? i have to tweak my 2nd file also...

[luzpaz]

please integrate a codespell --ignore-words /path/to/good/dictionary.txt

[larsoner]

Fixed by #100

[bittorf]

i still think, that a 'directive' directly above the false positive would be a better solution, e.g.

...
# codespell ignore=churchs
echo "a wrong word: churchs (a shoebrand) and a good word: churches"
...

codespell does not need knowledge about how to detect comments in language XY:
it only must check, if in the line above is a directive "codespell ignore=..."

[luzpaz]

@bittorf
use the -L flag. Try this:
echo "churchs, churches" | codespell -L churchs -
then try
echo "churchs, churches" | codespell -

[bittorf]

The L flag is not enough:
This way you ignore the wrong word "churchs" in the whole file.

Having a "directive" in the file, you can ignore just a single occurrence
and the knowledge about this special situation is directly at the correct place.

[discordianfish]

I agree, a directive would be great! I want to be specific about where to ignore what.