Allow for regex in dictionaries

[spetrosi]

I am considering using codespell to search for non-inclusive terms in our code. codespell has a built-in dictionary in https://github.com/codespell-project/codespell/blob/main/codespell_lib/data/dictionary_usage.txt.
However, codespell cannot search words from dictionary as a substring. For example, cosider this line:
timemaster is a program that uses ptp4l and phc2sys in combination with chronyd or ntpd to synchronize the system clock to NTP and PTP time masters.

Codespell would report the use of masters but it ignores timemaster.
I wonder if it is possible to allow passing regular expressions in the dictionary. So that I can add .*master.* to a dictionary and expect codespell to catch all strings that match that regex.

[DimitriPapadopoulos]

Codespell processes words. The regex that splits text into words is documented here:

codespell/codespell_lib/_codespell.py

Lines 479 to 489 in faf0c8d

	parser.add_argument(
	"-r",
	"--regex",
	action="store",
	type=str,
	help="regular expression that is used to find words. "
	"By default any alphanumeric character, the "
	"underscore, the hyphen, and the apostrophe are "
	"used to build words. This option cannot be "
	"specified together with --write-changes.",
	)

[spetrosi]

@DimitriPapadopoulos I didn't get what --regex does exactly when was reading it today. Do you know how I can use this option to get what I need?

[spetrosi]

The best way to use --regex that I found is like this:

$ codespell tasks/main.yml --builtin usage --regex "master|slave|blacklist|dummy|whitelist" -C0

• with --regex I specify all the words that I want to search for separated by regex's OR - |
• -C0 to print the line because the word found may be different from what codespell outputs.

I can also get all the words to --regex from the usage dictionary into a variable:

$ myregexvar=$(sed 's/->.*//g' /home/spetrosi/.local/lib/python3.12/site-packages/codespell_lib/data/dictionary_usage.txt | sed 's/$/|/' | tr -d '\n')
$ codespell tasks/main.yml --builtin usage --regex $myregexvar -C0

This works, but is not a clean solution. The initial request of this issue was the ability to add regular expressions to dictionaries. For example, to be able to add /.*master.*/ to my dictionary so that codespell search not for this word but for this regex.
I do not have permissions to re-open this issue.

[DimitriPapadopoulos]

The answer is that it's currently not possible to add regexes to dictionaries. I can reopen this as an enhancement request, no problem.

But then, given the requirement of matching .*master.*, why not use plain old grep/sed?

[spetrosi]

I thought that it's possible in some way because you closed as completed. Having it open as enhancement request works for me. Will it be hard to implement?

Do not want to use grep to have all the functionality around the check - to be able to provide dicts, set which files to test and to ignore, use inline comments to ignore a specific rule etc.

[DimitriPapadopoulos]

I think it will be very hard to implement. Using grep is your best bet.