Capture word between optional hyphens regex

Question

I've following type of strings,

abc - xyz
abc - pqr - xyz
abc - - xyz
abc - pqr uvw - xyz

I want to retrieve the text xyz from 1st string and pqr from 2nd string, `` (empty) from 3rd & pqr uvw. The 2nd hyphen is optional. abc is static string, it has to be there. I've tried following regex,

/^(?:abc) - (.*)[^ -]?/

But it gives me following output,

xyz
pqr - xyz
- xyz
pqr uvw - xyz

I don't need the last part in the second string. I'm using perl for scripting. Can it be done via regex?

@S.Kablar please try not to answer in comments, and more, with an external link. A proper answer, why not with some explanation, would be so much better.. — Kaddath, Feb 13 '18 at 14:09
Perfect! It works. @S.Kablar If you can add answer and some explanation, I can mark it as answer. — Abhishek, Feb 13 '18 at 14:12
Is `abc` dynamic or static? If it is static, just use `/^abc\h*-\h*(\S+)/` — Wiktor Stribiżew, Feb 13 '18 at 14:24
It is static, but based on the regex given by @S.Kablar, I was able to modify it to work for `abc`. But there is only one issue, the regex can't handle `abc - - xyz`. Adding it in question. — Abhishek, Feb 13 '18 at 14:27
What do you expect to get for the `abc - - xyz` string? Do you expect an empty string? Try [`^abc\h*-\h*([^-\s]*)`](https://regex101.com/r/JrVe5x/2) — Wiktor Stribiżew, Feb 13 '18 at 14:34
What about [`^[^-]+-\s\K[^-\s]*`](https://regex101.com/r/WoP9T9/3)? — ctwheels, Feb 13 '18 at 14:40
@WiktorStribiżew the text between hyphen can have spaces. it needn't be single word. — Abhishek, Feb 13 '18 at 15:05

Wiktor Stribiżew · Accepted Answer · 2018-02-13T15:11:02.417

3

Note that (.*) part is a greedily quantified dot and it grabs any 0+ chars other than line break chars, as many as possible, up to the end of the line and the [^ -]?, being able to match an empty string due to the ? quantifier (1 or 0 repetitions), matches the empty string at the end of the line. Thus, pqr - xyz output for abc - pqr - xyz is only logical for the regex engine.

You need to use a more restrictive pattern here. E.g.

/^abc\h*-\h*((?:[^\s-]+(?:\h+[^\s-]+)*)?)/

See the regex demo.

Details

^ - start of a string
abc - an abc
\h*-\h* - a hyphen enclosed with 0+ horizontal whitespaces
((?:[^\s-]+(?:\h+[^\s-]+)*)?) - Group 1 capturing an optional occurrence of
- [^\s-]+ - 1 or more chars other than whitespace and -
- (?:\h+[^\s-]+)* - zero or more repetitions of
  - \h+ - 1+ horizontal whitespaces
  - [^\s-]+ - 1 or more chars other than whitespace and -

edited Feb 13 '18 at 15:11

answered Feb 13 '18 at 14:41

Wiktor Stribiżew

484,719
26
302
397

@Abhishek And what about [`^abc\h*-\h*([^-\n]*[^-\s])`](https://regex101.com/r/JrVe5x/3)? – Wiktor Stribiżew Feb 13 '18 at 15:07
It works, but it takes `\n` into account as well. I just need to take care of spaces, but not new line. – Abhishek Feb 13 '18 at 15:08
@Abhishek It should not now. And it can still return an empty value in Group 1. – Wiktor Stribiżew Feb 13 '18 at 15:13
Yes.. Thank you! – Abhishek Feb 13 '18 at 15:14

PJProudhon · Answer 2 · 2018-02-13T15:07:57.870

1

You could use ^[^-]*-\s*\K[^\s-]*.

Here's how it works:

^       # Matches at the beginning of the line (in multiline mode)
[^-]*   # Matches every non - characters
-       # Followed by -
\s*     # Matches every spacing characters
\K      # Reset match at current position
[^\s-]* # Matches every non-spacing or - characters

Demo.

Update for multiple enclosed words: ^[^-]*-\s*\K[^\s-]*(?:\s*[^\s-]+)*

Last part (?:\s*[^\s-]+)* checks for existence of any other word preceded by space(s).

Demo

edited Feb 13 '18 at 15:07

answered Feb 13 '18 at 14:41

PJProudhon

821
13
14

Edited my answer accordingly. – PJProudhon Feb 13 '18 at 15:09
Thanks.. But updated one, takes \n into account as well. I just need to take care of spaces, but not new line. – Abhishek Feb 13 '18 at 15:09
Did you try with horizontal spacing, as suggested by @wiktor? – PJProudhon Feb 13 '18 at 15:12

score 1 · Answer 3 · answered Feb 14 '18 at 17:02

You could use split:

$answer = (split / \- /, $t)[1];

Where $t is the text string and you want the 2nd split (i.e. [1] as starts from 0). Works for everything except abc - - xyz but if the separator is " - " then it should have 2 spaces in the middle to return nothing. If abc - - xyz is correct then you can do this before the split for all to work:

$t =~ s/\- \-/-  -/;

It simply inserts an extra space so it'll match " - " twice with nothing in-between.

7stud · Answer 4 · 2018-02-14T02:50:20.130

Can it be done via regex?

Yes, with three simple regexes: - and ^\s+ and \s+$.

use strict;
use warnings; 
use 5.020;
use autodie;
use Data::Dumper;

open my $INFILE, '<', 'data.txt';

my @results = map {
    (undef, my $target) = split /-/, $_, 3;
    $target =~ s/^\s+//;  #remove leading spaces
    $target =~ s/\s+$//;  #remove trailing spaces
    $target;
} <$INFILE>;

close $INFILE;

say Dumper \@results;

--output:--
$VAR1 = [
          'xyz',
          'pqr',
          '',
          'pqr uvw'
        ];

Capture word between optional hyphens regex

4 Answers4

Linked