Replacements that are themselves mistakes

[EdwardBetts]

Some of the suggested replacements in dictionary.txt are listed as mistakes themselves.

Examples:

mimimise->minimise
minimise->minimize
orginal->orignal
orignal->original
passtime->pasttime
pasttime->pastime
shrinked->shrunk
srinkd->shrinked
srinked->shrinked
tranfering->transfering
transfering->transferring

Code to find these mistakes:

bad_words, new_words = {}, set()
lines = [line[:-1] for line in open('dictionary.txt')]

for line in lines:
    orig, replacements = line.split('->')
    bad_words[orig] = {word.strip() for word in replacements.split(',')}
    new_words.update(bad_words[orig])

highlight = {word for word in new_words if word in bad_words and word not in bad_words[word]}
for line in lines:
    if any(word in line for word in highlight):
        print(line)

[Gelma]

Thanks a lot! Very smart.

Now I apply this also for my PR.

[peternewman]

Is it worth adding this as a Travis check too?

[EdwardBetts]

I think it would make sense to add this as a Travis check. I'm not sure how to make that happen.

[larsoner]

You could adapt your code to work with test_dictionary_formatting, which added some sanity checks within the last couple of months

https://github.com/codespell-project/codespell/blob/master/codespell_lib/tests/test_basic.py#L389

[peternewman]

The dictionary checks are now here @EdwardBetts :
https://github.com/codespell-project/codespell/blob/master/codespell_lib/tests/test_dictionary.py

Do you fancy giving it a go?

[larsoner]

Should be fixed in #705