CODEWITHSUNDEEP
bashfiltergrepcygwin
elp
elp

Reputation: 870

Multiple lines grep

I got a file with a schema like this:

172.18.0.7
172.18.0.9
172.18.0.8
172.18.0.7
172.18.0.9
172.18.0.8
172.18.0.7
172.18.0.9
172.18.0.8
172.18.0.7
172.18.0.9
172.18.0.8

So its 7->9->8->7->9->8->7->9->8->7->9->8->7->9->8 ... at its end.

I want to get the lines where this schema is different. E.g. 7->8->9:

172.18.0.7
172.18.0.8
172.18.0.9

As the file got something about 100000 lines I'd like to use grep to filter them.

I tried something like this:

grep -Pzl "172.18.0.7*\n 172.18.0.9*\n 172.18.0.8*\n"

which did not work out properly. I wanted to find a pattern which fits the schema mentioned first.

Upvotes: 0

Views: 83

Answers (2)

Walter A
Walter A

Reputation: 20002

GNU sed 4.2 supports -z:

sed -z 's/172.18.0.7\n172.18.0.9\n172.18.0.8\n//g' file

This solution will fail when you the first line of a potential set of three is like

some_other_chars_before_172.18.0.7

When you add \n in the beginning of the match, you need to remove the last \n to find 2 sets without something in between, but that would allow the last line ending with

172.18.0.8_and_more_characters

It looks like a deadlock, but you can modify your input to Windows style and match smart:

 sed -rz 's/\n/\r\n/g;s/(\n|^)172.18.0.7\r\n172.18.0.9\r\n172.18.0.8\r//g;s/\r//g' file

Upvotes: 0

oliv
oliv

Reputation: 13249

Using GNU awk:

awk -v RS='\n*[0-9.]+7\n[0-9.]+9\n[0-9.]+8\n' NF file

The record separator RS is set such that it matches 3 lines having digits and dots and that finishing respectively with 7, 9, 8 (in this order).

The output record ORS separator being (the default one) \n, the input script (only NF) prints all non empty lines (that don't match RS).

Upvotes: 2

Related Questions