Ruby regex matching a string surrounded by word boundary chars when the string itself ends in a word boundary

Question

My regex: /\b+#{string}\b+/ where string = 'test-'

My test case: #test-#1

I want the regex to match test-. If my test string doesn't end in a word boundary character this works fine but if it does it breaks.

The # chars could be any other word boundary characters also, so another example would be:

-test--1 where I would still only want to match test-

Counter example case that works:

regex: /\b+#{string}\b+/ where string = 'test'

test case: #test#1

If you consider some chars as word boundaries add them as alternatives - `/(?<=#|\b)#{string}(?=#|\b)/` — Wiktor Stribiżew, Mar 22 '17 at 19:41
WiktorStribiżew it's not so much that I consider some chars as word boundaries, as I said the # could be any other word boundary including for example -, I'll update the question. — baarkerlounger, Mar 22 '17 at 19:43
Well, look at my test cases - I want to match a string surrounded by word boundary characters so I need the `\b` - my problem only comes when my interpolated string itself already contains a `\b` character. — baarkerlounger, Mar 22 '17 at 19:50
Sorry my comment didn't make clear, I don't want to match the interpolated string if it's just contained within another word, I want there to be at least 1 `\b` character after it. — baarkerlounger, Mar 22 '17 at 19:52
`\b` doesn't match a position where two non-word characters see each other. In this case `-` and `#`. So normally `test-\b` fails. — revo, Mar 22 '17 at 19:56
@WiktorStribiżew thanks that solves my problem. If you want to post it as an answer I'll close this. — baarkerlounger, Mar 22 '17 at 20:29

Wiktor Stribiżew · Accepted Answer · 2021-12-08T09:47:15.387

Since \b meaning is context-dependent, you need to use lookaround-based boundaries in your regex. As you want to match your keywords in between non-word chars, or start/end of string positions, you may use

/(?<!\w)#{Regexp.escape(string)}(?!\w)/

See a Ruby demo

Details

(?<!\w) - there must not be a word char right before the search string
#{Regexp.escape(string)} - the search string must be escaped so that all special chars inside were treated as literal chars
(?!\w) - there must be no word char right after the search word.

Another idea is to use adaptive dynamic word boundaries:

/(?:(?!\w)|\b(?=\w))#{Regexp.escape(string)}(?:(?<=\w)\b|(?<!\w))/
/(?:\B(?!\w)|\b(?=\w))#{Regexp.escape(string)}(?:(?<=\w)\b|(?<!\w)\B)/

The first one is adaptive word boundaries of Type 1 where no special restrictions are imposed on cases where words start or end with special chars, and the second regex only matches words starting or ending with special chars when they are not preceded / followed with a word char.

Feel free to read more about that in Word boundary with words starting or ending with special characters gives unexpected results and watch this YT video of mine with more explanations and a Python demo.

Ruby regex matching a string surrounded by word boundary chars when the string itself ends in a word boundary

1 Answers1