Regular expression for unicode in java Dash version

Question

It is possible to improve the performance of the following through a regular expression, the code is functional, but I want to know if there is any way to select the possible dash that exist in the unicode to standardize my dash

Words:

48553−FS002
48553-FS002
48553 FS002
48553-FS002-ESD12

Java

String reference = "48553−FS002";
String separador = reference.replaceFirst ( "\\w+(\\W)?\\w+", "$1" );
if(!separator.equals ( " " )) {
   reference = reference.replaceAll ( separator, "-" );
}

Or you could search for the unicode code, I was reading the following: dash, but i haven't managed to make it work Java Regex Unicode

A dash regex is `\p{Pd}`. Together with a soft hyphen, it is `[\p{Pd}\u00AD]` — Wiktor Stribiżew, Apr 05 '21 at 19:12
@WiktorStribiżew i try but not found in all case for example `48553−FS002` but others found — Gdaimon, Apr 05 '21 at 20:01
Ah, that is a minus sign, `\u2212`, add it. `[\p{Pd}\u00AD\u2212]` — Wiktor Stribiżew, Apr 05 '21 at 20:11
@WiktorStribiżew thanks, I just found it, even a thousand and a thousand thanks and it's the same. `\u2212` — Gdaimon, Apr 05 '21 at 20:28

Wiktor Stribiżew · Accepted Answer · 2021-04-05T20:37:41.610

2

If you need to match any non-word but space, you may use

reference = reference.replaceAll("[^\\w ]", "-");

Or, with character class subtraction:

reference = reference.replaceAll("[\\W&&[^ ]]", "-");

You can use the following pattern to match your hyphen or dash like patterns:

[\p{Pd}\u00AD\u2212]

Here,

\p{Pd} - matches any Punctuation, Dash symbols
\u00AD - matches a soft hyphen
\u2212 - matches a minus symbol.

edited Apr 05 '21 at 20:37

answered Apr 05 '21 at 20:28

Wiktor Stribiżew

607,720
39
448
563

MikeM · Answer 2 · 2021-04-05T20:38:35.000

0

If you know your strings only contain word characters and separators, as seems to be the case, then you can just use

reference = reference.replaceAll("[^ \\w]", "-");

edited Apr 05 '21 at 20:38

answered Apr 05 '21 at 20:32

MikeM

13,156
2
34
47

`\W` pattern is currently used in the OP code. – Wiktor Stribiżew Apr 05 '21 at 20:34
Yes, and that is probably the best approach if the format of the strings is as shown. – MikeM Apr 05 '21 at 20:35
Why? It is easier without any lookaheads, `"[^\\w ]"`. I already added it to my answer, BTW. – Wiktor Stribiżew Apr 05 '21 at 20:38
@WiktorStribiżew Good to see you copying my approach. – MikeM Apr 05 '21 at 20:42
You updated your answer **after** I did that. Literally 2 seconds after. – Wiktor Stribiżew Apr 05 '21 at 20:42

Regular expression for unicode in java Dash version

2 Answers2