As many inappropriate edits add bad words to articles, many filters¹ are created to check if a user added an expression which was not previously in the page, like this:
bad_words := 'ba+d|real+y\s*bad|not?\s*goo+d|...'; something & added_lines irlike bad_words & !( removed_lines irlike bad_words )
However, this approach causes false negatives such as this: a user can remove "baaad" while he adds "really bad", and the edit will not be matched. This is more frequent if the regex for bad_words contains many alternatives, because then the user can remove one of them and add any/all of the others while going undetected.
An approach to fix that would be to write the filter as
bad_word1 := 'ba+d'; bad_word2 := 'real+y\s*bad'; bad_word3 := 'not?\s*goo+d'; bad_word... := '...'; something & ( added_lines irlike bad_word1 & !( removed_lines irlike bad_word1 ) | added_lines irlike bad_word2 & !( removed_lines irlike bad_word2 ) | added_lines irlike bad_word3 & !( removed_lines irlike bad_word3 ) | ... )
but this is unnecessarily repetitive, makes the filter very long (and maybe increase the condition count more than it should?). It should be possible to just check if a user is adding some bad thing without having all this trouble... Maybe a new function could be added, which would allow something like this:
bad_word_regexes := [ 'ba+d', 'real+y\s*bad', 'not?\s*goo+d', '...' ]; something & irlike_added_any( added_lines, removed_lines, bad_word_regexes )
(the name and syntax is just an example, feel free to suggest something better)