Support dashes in default.extend-identifiers

[cbachhuber]

Thank you for typos, this is an amazing app!

I'd like to be able to have identifiers with dashes in default.extend-identifiers. I need it for gcc's -Wno-hsa and clang-tidy's clang-analyzer-optin. (Note: Bare keys with dashes are supported by toml)

Does a dash separate two identifiers or is it part of one identifier, similar to an underscore? Should we add this info to the docs? I can't find this anywhere.

Minimal example:

.typos.toml:

[default.extend-identifiers]
wrongWrod = 'wrongWrod'
wrong_wrod = 'wrong_wrod'
wrong-wrod = 'wrong-wrod'  # doesn't have any effect and doesn't warn

example.txt:

wrongWrod
wrong_wrod
wrong-wrod

$ typos example.txt
error: `wrod` should be `word`
  --> file.txt:3:7
  |
3 | wrong-wrod
  |       ^^^^
  |

[hofbi]

Maybe the solution in #575 works here as well

[epage]

I've updated the docs with the definition of an identifier (uncode XID_Continue) and word (split identifier).

Yes, these are cases where identifiers are split by a dash. The challenge is knowing when to treat them as identifiers or not and how to do so efficiently.

[cbachhuber]

Maybe the solution in #575 works here as well

I don't think so: in #575, @epage suggests using extend-identifiers instead of extend-words. We're already working with extend-identifiers in this issue.

I've updated the docs with the definition of an identifier (uncode XID_Continue) and word (split identifier).

Thank you for this! While being correct, I find the unicode reference hard to read and to derive common cases from. For example, within 5 minutes, I wasn't able to find that a dash separates two identifiers. What do you think about #648?

The challenge is knowing when to treat them [dashes] as identifiers or not and how to do so efficiently.

Generally, dashes split identifiers, right? In that case, and if it stays this way, we can close this issue with #648 imo.

[epage]

59c4713 handled the docs side of things and #400 is the issue for other word joiners besides _, so I think its safe to close this issue out. If there is something I missed, let me know.

[tinmarino]

Duplicates: #400.