Remove a part of string in bash

Question

how do i remove refs/heads/ part from refs/heads/hotfix/sub_pag/105 and return the reamining part hotfix/sub_page/105?

The question is clear tbh, doesn’t deserver so many downvotes. But yea if so many people feel that, please specify that `refs/heads/` and the other two are strings or strings assigned to a variable. — Mihir Luthra, Jul 19 '19 at 09:42
@Mihir I think [this](https://duckduckgo.com/?q=stackoverflow+bash+remove+substring&t=ffsb&ia=web) explains the downvotes. That said, the proper course of action would be to close with a duplicate — Aaron, Jul 19 '19 at 09:57
@Aaron, agreed but at the same time its nice to write a short comment explaining reason of downvote. (would help viewers and the OP) — Mihir Luthra, Jul 19 '19 at 10:01
Also it's unclear whether the part to remove is always the same (in which case, [this](https://stackoverflow.com/questions/16623835/remove-a-fixed-prefix-suffix-from-a-string-in-bash) is an appropriate duplicate) or the string must be removed up to the second slash, with varying words in between — Aaron, Jul 19 '19 at 10:01

choroba · Accepted Answer · 2019-07-19T09:46:12.150

3

You can use cut:

echo refs/heads/hotfix/sub_pag/105 | cut -d/ -f3-

-d specifies the delimiter, / in this case
-f specifies which columns (fields) to keep. We want everything from column 3 onwards.

Or, use variables and parameter substitution:

x=refs/heads/hotfix/sub_pag/105
echo ${x#refs/heads/}

# removes the string from the beginning of the variable's value.

edited Jul 19 '19 at 09:46

answered Jul 19 '19 at 09:43

choroba

231,213
25
204
289

1

The answer doesn't involve any regex. – choroba Jul 19 '19 at 09:45
right. i did a quick google search on cut. bash has so many ways of doing the same thing. – Raaz Jul 19 '19 at 09:47

score 1 · Answer 2 · answered Jul 19 '19 at 09:46

1

abc="refs/heads/hotfix/sub_pag/105”
abc=${abc##refs/heads/}

The parameter expansion in the second line would remove that part.

answered Jul 19 '19 at 09:46

Mihir Luthra

6,059
3
14
39

Why `##` and not `#`? – choroba Jul 19 '19 at 09:52
@choroba, came out of habit. The match should be same in both the cases i guess because of fixed pattern. Is the latter faster? – Mihir Luthra Jul 19 '19 at 09:54
If there's no wildcard, it's probably just shorter :-) – choroba Jul 19 '19 at 09:57

score 0 · Answer 3 · answered Aug 07 '20 at 17:44

I've found sed to be effective and obvious in purpose

sed 's/refs\/heads\///g'

Usage example

#!/bin/sh
long_ref="refs/heads/hotfix/sub_pag/105"  # normally from pipeline/Jenkins/etc.
echo "$long_ref"
short_ref=$(echo "$long_ref" | sed 's/refs\/heads\///g')
echo "$short_ref"

output

% sh test.sh
refs/heads/hotfix/sub_pag/105
hotfix/sub_pag/105

Remove a part of string in bash

3 Answers3