Possible source of misspellings

[ccoVeille]

https://github.com/crate-ci/typos/blob/HEAD/crates%2Ftypos-dict%2Fassets%2Fwords.csv

https://github.com/crate-ci/typos

The project is licensed under MIT license (or Apache 2).

The readme is clear about it, I'll open a PR to fix the way github computes the license.

So this list could be used. It could supersede the current en-common-mispelling dictionary that is released under CC BY-SA license, cf our previous discussions.

[Jason3S]

@ccoVeille,

They are in violation of the CC BY-SA license. I don't think it is a good idea to pretend that they are not.

See: https://github.com/crate-ci/typos/blob/master/crates/wikipedia-dict/assets/dictionary.txt
They just used some code to populate words.csv.

[ccoVeille]

But the PR that people are doing are valuable, and under MIT license, no ?

take a look at https://github.com/crate-ci/typos/commits/HEAD/crates/typos-dict/assets/words.csv

here is an example

crate-ci/typos@f715ca8 related to crate-ci/typos#1290

But I agree, the root import they did is a violation of the wikipedia license

[Jason3S]

I think the PR's are under MIT. Do you think they have a lot of additions that are not already included here?

[ccoVeille]

Here is just a count of the line

$ wc -l typos-crate-ci/crates/typos-dict/assets/words.csv cspell-dicts/dictionaries/en-common-misspellings/dict-en.yaml
  71530 typos-crate-ci/crates/typos-dict/assets/words.csv
   4197 cspell-dicts/dictionaries/en-common-misspellings/dict-en.yaml

Also, if we consider their words.csv can contain multiple misspelled words for one words.

The number is closer than 75k (estimation)

[ccoVeille]

Also this list of target spelling, could be used to complete the English dictionaries

$ cat typos-crate-ci/crates/typos-dict/assets/words.csv | sed 's/.*,//' | sort -u | wc -l
17612

Some are missing in cspell

[Jason3S]

I have create PR #4538

[ccoVeille]

I feel like you replied in the wrong thread, even if it's the right discussion.

This thread was about finding candidate words in the one typos-cli tool suggests and so add them to English dictionaries or other software ones.

But the PR is great

[Jason3S]

I feel like you replied in the wrong thread, even if it's the right discussion.

Yes, wrong thread (it was the first box I could type into).

Also this list of target spelling, could be used to complete the English dictionaries

That might be possible, only I don't trust their targets. It is a mix of British and American English and coding terms.

[ccoVeille]

Yes. I noticed, but the list is pretty short when I checked. Around 200-500 words

The real issue is indeed the software terms. Based on cspell dictionary organization, it would be difficult to know to what software dictionary (npm, rust ...) it should go for cspell

[Jason3S]

In the PR, my approach was to take suggestions that were found in the respective dictionaries:

I generated a list of suggestion by looking up each target word in the following dictionaries: en_US, en_GB, and sofwareTerms.

• If it was found in both en_US and en_GB, then it was added to the en file.
• Otherwise it was added into the respective en_US or en_GB file.
• Suggestions found in softwareTems, but not en_US were added to the softwareTerms suggested list.

[ccoVeille]

That's a huge improvement to cspell