matching regex in awk/gawk

Question

Can someone make the awk line below work in awk please. The syntax uses the standard PCRE regex standard (and i need to expect that some non-numeric characters are preceded to the first number, that is the string can look like "++3.59 ± 0.04* "). (note that I tried [0-9] and [:digit:] instead of \d) also note that I did read https://www.gnu.org/software/gawk/manual/gawk.html#Regexp

gawk 'BEGIN{test="3.59 ± 0.04";match(test, /^.*?(\d+?\.\d+?)\s*?±\s*?(\d+?\.\d+?)$/, arr);print arr[1];}'

POSIX regex engine does not support lazy quantifiers. Go on trying with `[0-9]` to match digits, `\d` is not supported. Also, use `[ \t]` or `[[:space:]]` instead of `\s`. — Wiktor Stribiżew, Feb 17 '20 at 15:20
PCRE is not standard at all. Your best bet is to use Perl instead. — tripleee, Feb 17 '20 at 15:21
@WiktorStribiżew And in POSIX standard, is `*` greedy or lazy then? — atapaka, Feb 17 '20 at 15:25

Lutz · Accepted Answer · 2020-02-17T20:59:00.523

0

you add to many ? and I think you need to use [0-9] Also when you start with ^[^0-9] only non-numeric characters are "eaten away". so in summary I think you want:

gawk 'BEGIN{test="3.59 ± 0.04";match(test, /^[^0-9.]*([0-9]+\.[0-9]+)\s*±\s*([0-9]+\.[0-9]+)$/, arr);print arr[1];}

That matches any non-numeric character up to the first numeric one and then captures the first number as group 1.

Thanks @Ed Morton for the corrections. I did miss the + after the first digit in the original regex.

edited Feb 17 '20 at 20:59

answered Feb 17 '20 at 15:31

Lutz

612
3
8

2

The meaning of `*?`, `+?`, and `??` in PCRE is that it switches from greedy to stingy matching. This cannot easily be reimplemented with traditional always-greedy regex like Awk's, though it's not clear how exactly the OP thinks this construct is useful here. – tripleee Feb 17 '20 at 19:14

matching regex in awk/gawk

1 Answers1