How to grep a specific pattern before match?

user_12238354 0
Viewed 286

I'm currently working on multiple configuration files which use the following format:

[Stanza1]
action.script=1
action.ping=0
action.lookup=1
action.notable.param=0
action.script.filename=script.pl
[Stanza2]
action.script=0
action.ping=0
action.lookup=1
[Stanza3]
action.script=1
action.ping=0
action.lookup=0
action.script.filename=script.pl

I want to know which stanzas include "action.script.filename=script.pl", so the expected result would be 

[Stanza1]
[Stanza3]

Using something like:

grep -B 10 "action.script.filename = script.pl" file

doesn't work for cases where the stanza name is more than 10 lines before the match, and proves quite cumbersome to use.

Any suggestions on how to do this?

grep bash
5 Answers

The following sed command would do the trick :

sed -n '/^\[/h;/^action\.script\.filename=script\.pl$/{x;p}'

You can try it here.

When it encounters a line that starts with "[", it stores it into its hold buffer. When it encounters a "action.script.filename=script.pl" line, it prints the content of the hold buffer.

user_16783620

I'm not sure this can be done purely with grep. I would recommend a small bash script:

while read line
do
    if [[ $line =~ \[.* ]]; then
        # save stanza for later
        stanza=$line
    fi
    if [[ $line =~ action.script.filename=script.pl ]]; then
        echo $stanza
    fi
done < file
user_20829640

With awk

$ awk '/action\.script\.filename=script\.pl/{print h} /^\[/{h=$0}' ip.txt
[Stanza1]
[Stanza3]
  • /^\[/ lines starting with [ character, you can also use something like /Stanza/ as long as it uniquely identifies header lines
    • h=$0 for such lines, save the content ($0) to variable h
  • /action\.script\.filename=script\.pl/ if input line matches the given search criteria
    • print h print the value of h variable
    • if you are matching whole line, then you can also use string match $0 == "action.script.filename=script.pl" instead of regex match
user_40820520

This line of code works for me

grep '^\[Stanza\|^action.script.filename=script.pl$' fileName | grep -B1 'action.script.filename=script.pl' | grep -v 'action.script.filename=script.pl\|\-\-'

Explanation:

grep '^\[Stanza\|^action.script.filename=script.pl$' fileName

matches either [Stanza]* lines or action.script.filename=script.pl ones. Output is something like this

[Stanza1]
action.script.filename=script.pl
[Stanza2]
[Stanza3]
action.script.filename=script.pl

Adding this filter | grep -B1 'action.script.filename=script.pl' will result in this 

[Stanza1]
action.script.filename=script.pl
--
[Stanza3]
action.script.filename=script.pl

Now you just need to clean the output from unwanted parts

| grep -v 'action.script.filename=script.pl\|\-\-'

This is the final output

[Stanza1]
[Stanza3]
user_27161420 
    awk '/^\[.*\]$/{stanza=$0;next} /action.script.filename=script.pl/{print stanza}' filename
    [Stanza1]
    [Stanza3]

You can store each stanza in a variable called stanza and move to next line. Whenever you see the string action.script.filename=script.pl , print the variable stanza.

user_63096010
Related