CODEWITHSUNDEEP
bashsed
eli123
eli123

Reputation: 1068

Format results in Bash

For example, I have test.txt with the following line:

L1~00~00~00~00~00~Test~122113~00~L2~This~Is~A~Sample~Data~L1~00~00~00~00~00~Test1~123456~00

I want to get "Test" and "Test1", both are after L1~00~00~00~00~00~ with the following format.

Test, Test1

I already have this line in my bash script:

grep -oP 'L1(?:.[\w\s]*){5}.(\K[\w\s]*)' < test.txt

But it returns a different format:

Test
Test1

How can I achieve this by adding sed in my script? I'm still a newbie. I hope somebody could help me. Thanks

Upvotes: 2

Views: 77

Answers (4)

Ed Morton
Ed Morton

Reputation: 204731

Wth GNU awk for multi-char RS and RT:

$ awk -v RS='L1~00~00~00~00~00~' -F~ 'NF{ORS=(RT?", ":"\n"); print $1}' file
Test, Test1

The above just splits each line into records that contain whatever is between L1~00~00~00~00~00~s, and splits each record into fields between ~s and then prints the first field of each (which is the text that comes between each L1~00~00~00~00~00~ and the next ~) followed by , if it's not the last record and \n if it is.

Upvotes: 1

NarūnasK
NarūnasK

Reputation: 4950

If you are not inclined using perl regex - you can cling to sed alone:

sed -rn 's#(L1.)((\w+.){5})(\w+)(.*\1\2)(\w+)(.*)#\4, \6#p' < test.txt

Upvotes: 1

tripleee
tripleee

Reputation: 189958

Of course, if you are using Perl regex anyway, you might as well use Perl directly.

perl -nle '@m = m/L1(?:.[\w\s]*){5}.([\w\s]*)/g; print(join(",", @m)) if @m' test.txt

This collects matches into @m, then prints them joined by a comma if there are matches in @m. The -l option is a convenience to supply the trailing newline on the print, and the -n option makes Perl loop over the input lines one at a time, like sed.

Upvotes: 2

anubhava
anubhava

Reputation: 786359

You can use:

grep -oP 'L1(?:.[\w\s]*){5}.(\K[\w\s]*)' test.txt | sed 'N;s/\n/, /'
Test, Test1

Upvotes: 1

Related Questions