Match and replace block using regex with sed on Solaris

Question

This is working on some tools as regex101.com but I cannot make it work with sed.

Blocks:

dn: abcd1,ou=test
aaaaa
bbbb
1111

dn: abcd2,ou=test
33333
ddddd
aaaaa

dn: qwert,ou=test
55555
hhhh
dddd

I want to match and replace with nothing every block that starts with dn: abcd. A block always ends with \n\n.

Regexp: (?s)\b(?:(?!\n\n).)*?\bdn: abcd\b(?:(?!\n\n).)*

Is it possible to achieve with sed?

try using perl... sed doesn't support lookarounds or non-capturing groups and many other features that are available in perl... — Sundeep
– Sundeep, Commented Mar 29, 2018 at 14:32
also, if you could modify given input sample without \n (assuming you have literal newlines and not \ followed by n) - use the {} button in editor to format the sample instead of backticks.. and add complete expected output, it'd help to suggest solutions.. — Sundeep
– Sundeep, Commented Mar 29, 2018 at 14:34
@Sundeep already edited, thanks! I'm not familiar with perl, any idea? — Adami
– Adami, Commented Mar 29, 2018 at 14:37

Gilles Quénot · Accepted Answer · 2018-03-29 17:07:42Z

1

perl is a better choice for this task.

$ perl -00ne 'print if not /^dn: abcd/' file

or

$ perl -ne 'print if not /^dn: abc/ .. /^$/' file

dn: qwert,ou=test
55555
hhhh
dddd

answered Mar 29, 2018 at 14:53

Gilles Quénot

188k43 gold badges232 silver badges229 bronze badges

Sign up to request clarification or add additional context in comments.

Comments

Sundeep · Accepted Answer · 2018-03-29 15:14:03Z

1

Use awk's paragraph mode

$ awk -v RS= '!/^dn: abcd/' ip.txt
dn: qwert,ou=test
55555
hhhh
dddd

-v RS= When RS is set to empty string, one or more consecutive empty lines is used as input record separator
!/^dn: abcd/ to ignore paragraphs starting with dn: abcd

note that default output record separator is single newline, so you might need something like this:

$ awk -v RS= -v ORS='\n\n' '!/^dn: abcd1/' ip.txt
dn: abcd2,ou=test
33333
ddddd
aaaaa

dn: qwert,ou=test
55555
hhhh
dddd

answered Mar 29, 2018 at 15:14

Sundeep

23.9k2 gold badges35 silver badges131 bronze badges

ctac_ · Accepted Answer · 2018-03-29 16:29:57Z

0

You can try this sed

sed '/^dn: abcd/,/^$/d' infile

answered Mar 29, 2018 at 16:29

ctac_

2,5012 gold badges10 silver badges18 bronze badges