Question

I have an expression

^(((\d{1,2}))|((\d{1,2}\.\d{1,2}))|((\d{1,2}\.\d{1,2}\.\d{1,2})))(;?)$

to validate a text line like these:

9
99
99.9
99.99
99.99.9
99.99.99

Now I need to validate all these possible variants for ;-separated text

When I use an expression:

^(((((\d{1,2}))|((\d{1,2}\.\d{1,2}))|((\d{1,2}\.\d{1,2}\.\d{1,2})))(;?))+?)$

It becomes greedy and allows more than 2 digits.

Sorry guys, I need to add some explanations. This is an example of all possible matches

9;99;99.9;99.99;99.99.9;99.99.99

Answer 1

Your problem isn’t caused by greedy matching, but by the fact that you allow the separating semicolon only optionally. I’d try something like (given dotted_number is regex that matches your original single value, no semicolon)

^(dotted_number;)*dotted_number;?$

In the end, it would look like

^((\d{1,2}\.){0,2}(\d{1,2});)*(\d{1,2}\.){0,2}(\d{1,2});?$

Or you could perenthesize it the other way around, which may be slightly faster

^dotted_number(;dotted_number)*;?$

Even this works:

egrep '^(([[:digit:]]{1,2}\.){0,2}[[:digit:]]{1,2}(;|;?$))+$'