Do not allow for a word to start or end with punctuation symbols

[yarikoptic]

The inspired for me to look use case

• https://github.com/INTERSECT-SDK/python-sdk/pull/19/files/33da9ff31d6162caa0dfc1a1155f321e6d68b1cc#diff-10380fd6e5ecb84c1ae11e135982739946c5aff1a50499378db397cf5034f54e

And then I found the issue this

• Closes Detection of string delimiters #3305
  • added disallowance of trailing - or apostrophes -- or should they be allowed ?

Although may be I am missing the use-cases/problems @DimitriPapadopoulos and @mdeweerd discussed back then

Edits:

• I had to partially go back and change so there are two alternative word captures quoted
• Allow for only trailing, but not leading quotes happen words were not in quotes to start with.

After I pushed, I realized that we have a use case where we are not covering ``LaTeX'' way to quote. So were and keep missing those. Do you think I should add regex for them too?

some gory details on me discovering were' and other "typos" in dictionaries

ok -- tests fail due to the typo:

codespell_lib/data/dictionary_code.txt:were'->we're

and apparently it is not a single one like that left:

codespell_lib/data/dictionary_code.txt:were'->we're
codespell_lib/data/dictionary.txt:aircrafts'->aircraft's
codespell_lib/data/dictionary.txt:arent'->aren't
codespell_lib/data/dictionary.txt:cant'->can't
codespell_lib/data/dictionary.txt:cnat'->can't
codespell_lib/data/dictionary.txt:couldnt'->couldn't
codespell_lib/data/dictionary.txt:didnt'->didn't
codespell_lib/data/dictionary.txt:doesent'->doesn't
codespell_lib/data/dictionary.txt:doesn'->doesn't
codespell_lib/data/dictionary.txt:doesnt'->doesn't
codespell_lib/data/dictionary.txt:dont'->don't
codespell_lib/data/dictionary.txt:dosent'->doesn't
codespell_lib/data/dictionary.txt:hasnt'->hasn't
codespell_lib/data/dictionary.txt:havent'->haven't
codespell_lib/data/dictionary.txt:isnt'->isn't
codespell_lib/data/dictionary.txt:packges'->packages'
codespell_lib/data/dictionary.txt:shouldnt'->shouldn't
codespell_lib/data/dictionary.txt:thats'->that's
codespell_lib/data/dictionary.txt:wasnt'->wasn't
codespell_lib/data/dictionary.txt:wouldnt'->wouldn't

but some of those IMHO make no sense to list ' if correction is also with ' which is AFAIK is not a part of the word, i.e. I think following should be simply removed (replaced with ones with '):

codespell_lib/data/dictionary.txt:gaus'->Gauss'
codespell_lib/data/dictionary.txt:guas'->Gauss'
codespell_lib/data/dictionary.txt:guass'->Gauss'

First I wondered if that is the case worth striving for fixing: since were is a legit word, it could have also been forgotten ' somewhere long before, e.g. in a

var = stay as you were'

which would be programming language gotcha, not a typo.

FWIW were' was added originally in

• f32b84d " Typos from Perl5 (2014-2019) Grand Finale (Typos from Perl5 (2014-2019) Grand Finale #1106)" by @Fourchaux
• more similar ones (like files') at some point existed but then were removed as well e.g. in 8804b93

In leftover cases it boils down to

' is a part of the word, and thus could be present in the typo in alternative location"

(I would still argue to exclude were')..

[larsoner]

I haven't looked deeply but am I right that this single-line comment:

# blah blah 'blah blah' blah

would get handled differently from this multi-line one

# blah blah 'blah
# blah' blah

? To me these are essentially the same code comment so it's weird that codespell would treat them differently.

It's even a bit weird to try to keep track of the quotation level in any way as it seems brittle, especially given people can forget start and end quotes from time to time

[yarikoptic]

since no space should be in "word regex", we should be robust to such examples -- those should be separate words (on one line or not)