regex word boundary will not work with diacritic characters

Asked May 09 '16 at 18:44

Active May 09 '16 at 18:44

Viewed 57 times

Assuming you have the following input to check against:

lorem ipsüm dolor ipsum dolor

and the following JavaScript RegExp:

/(\bips[uü]\b)/gmi

What I expect is that nothing will be found as I am searching for a match, either "ipsu" or "ipsü" but with a word boundary. But the word "ipsü" will be matched, even if there is the character "m" at the end and not a word boundary. I don't understand the reason.

Does anyone know the reason?

See this example:

https://regex101.com/r/kM0zW7/2

asked May 09 '16 at 18:44

user3292653

2

`ü` in JS is not considered a word character (cannot be matched with `\w`). Use `\bips[uü](?=\s|$)`. Or create a custum word boundary using a character class. – Wiktor Stribiżew May 09 '16 at 18:48
@WiktorStribiżew Thank you. Do you have a reference for this? – user3292653 May 09 '16 at 18:57
@WiktorStribiżew this is also true for the start of a regex? – user3292653 May 09 '16 at 19:21
For a start of regex, you can use `(?:^|\s)ips[uü](?=\s|$)`. – Wiktor Stribiżew May 09 '16 at 19:28
I think `ü` is extended ascii and considered a word. So `\w\b` won't match `ü`m. But I don't know if JS will do unicode word's. – May 09 '16 at 19:30
1

@WiktorStribiżew That regex will also match a blank before "ipsu", see https://regex101.com/r/zZ0nH6/1 – dude May 09 '16 at 19:35
JS is a crippled engine for many reasons. There is a desire for MS to fix it since they probably wrote the original spec for it. – May 09 '16 at 19:43
@julmot: That is not a problem. One can always use capturing groups. [`(?:^|\s)(ips[uü])(?=\s|$)`](https://regex101.com/r/kM0zW7/4) – Wiktor Stribiżew May 09 '16 at 19:50

regex word boundary will not work with diacritic characters

0 Answers0