Repeated words rule gives false positives too often

[evilnick]

Particularly with web pages

e.g. <a href="something/plop">Plop</a> will trigger this rule

[SecondSkoll]

Adding some negative lookahead / lookbehind should clear up some of these issues. If you've got a lot of this happening, would you be able to show some more examples so we can try and address them all?

You can run against this rule specifically with vale --filter '.Name=="Canonical.000-Repeated-words"'.

[SecondSkoll]

So upon investigation and running some tests in Vale studio, it seems like the logic built into the repetition extension point might be broken. Not that I know much Golang, but this line seems like it's designed to catch repeated terms across multiple lines - but it doesn't seem to be functioning.

It seems that the repetition rule turns the token into a capture group after the rule is implemented and runs against the captured content, so we can't implement negative lookahead/behind in the logic for a repetition based rule.

However, we can use the general existence extension to implement our own logic for this rule.

extends: existence
message: "Repeated words '%s'!"
level: error
ignorecase: true
tokens:
  - '([\S]+)( |\n)\1'

This should only provide output in the case of two words (capitalised or not) separated by either a space or a new line.