How to replace newlines between brackets

Question

I have log file similar to this format

Here is the echo command to produce that output

$ echo -e "test {\nseq-cont {\n\t\t\t0,\n\t\t\t67,\n\t\t\t266\n\t\t\t},\n\t\tgrp-id 505\n\t}\n}\ntest{\n\ttest1{\n\t\tval\n\t}\n}\n"

Question is how to remove all whitespace between seq-cont { and the next } that may be multiple in the file.

I want the output to be like this. Preferably use sed to produce the output.

test{seq-cont{0,67,266},
                       grp-id 505
        }
}
test{
        test1{
                val
        }
}

Efforts by OP: Here is the one somewhat worked but not exactly what I wanted:

sed ':a;N;/{/s/[[:space:]]\+//;/}/s/}/}/;ta;P;D' logfile

Why have newlines been removed after `266` but not after `505`, despite both being "between brackets"? Is it only for input "between inner-most brackets" — Bohemian, Oct 06 '21 at 04:11
Yes that's right. Between First matching open and close brackets — Ramanan T, Oct 06 '21 at 04:22
@RamananT, also please do add your tried code as your efforts in your question which is highly encouraged on SO. — RavinderSingh13, Oct 06 '21 at 04:23
Try `echo -e "test {seq-cont {0,67,266},\n\t\tgrp-id 505\n\t}\n}\ntest{\n\ttest1{\n\t\tval\n\t}\n}\n"` — fpmurphy, Oct 06 '21 at 04:33
@RavinderSingh13, Here is the one somewhat worked but not exactly what I wanted: sed ':a;N;/{/s/[[:space:]]\+//;/}/s/}/}/;ta;P;D' — Ramanan T, Oct 06 '21 at 04:53
@RamananT, Thanks for letting know. Its always recommended to add efforts in your question, as of now I have added your tried command, kindly do always add your tried code as your efforts in your question, cheers and happy learning. — RavinderSingh13, Oct 06 '21 at 08:27

score 3 · Accepted Answer · answered Oct 06 '21 at 08:00

It can be done using gnu-awk with a custom RS regex that matches { and closing }:

awk -v RS='{[^}]+}' 'NR==1 {gsub(/[[:space:]]+/, "", RT)} {ORS=RT} 1' file

test {seq-cont{0,67,266},
                grp-id 505
        }
}
test{
        test1{
                val
        }
}

Here:

NR==1 {gsub(/[[:space:]]+/, "", RT)}: For the first record replace all whitespaces (including line breaks) with empty string.
{ORS=RT}: Set ORS to whatever text we captured in RS

PS: Remove NR==1 if you want to do this for entire file.

Wiktor Stribiżew · Answer 2 · 2021-10-06T08:26:10.480

You can do that much easier with perl:

perl -0777 -i -pe 's/\s+(seq-cont\s*\{[^}]*\})/$1=~s|\s+||gr/ge' logfilepath

The -0777 option tells perl to slurp the file into a single string, -i saves changes inline, \s+(seq-cont\s*\{[^}]*\}) regex matches one or more whitespaces, then captures into Group 1 ($1) seq-cont, zero or more whitespaces, and then a substring between the leftmost { and the next } char ([^}]* matches zero or more chars other than }) and then all one or more whitespace character chunks (matched with \s+) are removed from the whole Group 1 value ($1) (this second inner replacement is enabled with e flag). All occurrences are handled due to the g flag (next to e).

See the online demo:

#!/bin/bash
s=$(echo -e "test {\nseq-cont {\n\t\t\t0,\n\t\t\t67,\n\t\t\t266\n\t\t\t},\n\t\tgrp-id 505\n\t}\n}\ntest{\n\ttest1{\n\t\tval\n\t}\n}\n")
perl -0777 -pe 's/\s+(seq-cont\s*\{[^}]*\})/$1=~s|\s+||gr/ge' <<< "$s"

Output:

test {seq-cont{0,67,266},
        grp-id 505
    }
}
test{
    test1{
        val
    }
}

score 1 · Answer 3 · answered Oct 06 '21 at 08:21

With your shown samples, please try following awk program. Tested and written in GNU awk.

awk -v RS= '
match($0,/{\nseq-cont {\n[^}]*/){
  val=substr($0,RSTART,RLENGTH)
  gsub(/[[:space:]]+/,"",val)
  print substr($0,1,RSTART-1) val substr($0,RSTART+RLENGTH)
}
'  Input_file

Explanation: Simple explanation would be, using RS capability to set it to null. Then using match function of awk to match everything between seq-cont { to till next occurrence of }. Removing all spaces, new lines in matched value. Finally printing all the values including newly edited values to get expected output mentioned by OP.

How to replace newlines between brackets

3 Answers3