Tweak --ignore-regex to ignore consecutive matches

[peterjc]

Consider this example for ignoring 2 or 3 letter upper case words (typical of acronyms):

codespell --ignore-regex '\W([A-Z]{2,3})\W'

If there are two consecutive matches in a string like "text TE OTU text" only one gets removed - the problem is the space needs to be used twice, the end of TE and the start of OTU as the second but overlapping regex match.

Cross reference #1578.

[peternewman]

That makes sense thanks @peterjc .

Please can you add some tests for it so we won't regress in future (even if we change to a different workaround).

[peterjc]

I wasn't sure if this change was in the spirit of the original regex implementation, but since you like the idea, I agree a test or two is a good idea.

[peterjc]

I've tested this locally now:

$ codespell flag.txt 
flag.txt:1: TE ==> THE, BE, WE, TO
flag.txt:1: OTU ==> OUT
$ codespell --ignore-regex '\W([A-Z]{2,3})\W' flag.txt 
$ codespell --ignore-regex "\W([A-Z]{2,3})\W" flag.txt 
$ codespell --ignore-regex="\W([A-Z]{2,3})\W" flag.txt 
$ codespell --ignore-regex='\W([A-Z]{2,3})\W' flag.txt 
$ codespell --ignore-regex \W([A-Z]{2,3})\W flag.txt 
-bash: syntax error near unexpected token `('
$ codespell --ignore-regex=\W([A-Z]{2,3})\W flag.txt 
-bash: syntax error near unexpected token `('

Looks like the protective quotes are being lost somewhere in the test framework...

[peterjc]

Solved, rather than doing what works at the command line, I needed to match what Python receives from the command line.

Also rebased for a clearer history.