如何在包含特定搜索字符串的两个大括号之间获得多行字符串？

Question

I'm looking for a quick and easy one-liner to extract all brace-delimited text-blocks containing a search string from a text file. I've just about googled myself crazy on this, but everyone seems to be only posting about getting the text between braces without a search string.

I've got a large text file with contents like this:

blabla
blabla {
  blabla
}
blabla
blabla {
  blabla
  blablaeventblabla
}
blabla

The vast majority of bracketed entries do not contain the search string, which is "event".

What I am trying to extract are all text (especially including multi-line matches) between each set of curly braces, but only if said text also contains the search string. So output like this:

blabla {
  blabla
  blablaeventblabla
}

My linux command line is /usr/bin/bash. I've been trying various grep and awk commands, but just can't get it to work:

awk '/{/,/event/,/}/' filepath

grep -iE "/{.*event.*/}" filepath

I was thinking this would be really easy, as it's a common task. What am I missing here?

Answer 1

This gnu-awk should work:

awk -v RS='[^\n]*{|}' 'RT ~ /{/{p=RT} /event/{ print p $0 RT }' file
blabla {
   blabla
   blablaeventblabla
}

RS='[^\\n]*{\\n|}' sets input record separator as any text followed by { OR a } . RT is the internal awk variable that is set to matched text based on RS regex.

Answer 2

User 999999999999999999999999999999 had a nice answer using sed which I really liked, unfortunately their answer appears to have disappeared for some reason.

Here it is for those who might be interested:

sed '/{/{:1; /}/!{N; b1}; /event/p}; d' filepath

Explanation:

/{/ if current line contains { then execute next block { start block :1; label for code to jump to /}/! if the line does not contain then execute next block { start block :1; label for code to jump to /}/! if the line does not contain then execute next block { start block :1; label for code to jump to /}/! if the line does not contain } then execute next block { start block N; add next line to pattern space b1 jump to label 1 }; end block /event/p if the pattern space contains the search string, print it (at this point the pattern space contains a full block of lines from then execute next block { start block N; add next line to pattern space b1 jump to label 1 }; end block /event/p if the pattern space contains the search string, print it (at this point the pattern space contains a full block of lines from then execute next block { start block N; add next line to pattern space b1 jump to label 1 }; end block /event/p if the pattern space contains the search string, print it (at this point the pattern space contains a full block of lines from { to } ) }; end block d delete pattern space ) }; end block d delete pattern space

Answer 3

Here is a modified version of this gem from 'leu' (10x leu for enlighten us). This one is doing something very similarly. Extract everything between which begin with 'DEC::PKCS7[' and ending with ']!':

cat file | sed '/^DEC::PKCS7\[/{s///; :1; /\]\!$/!{N; b1;}; s///;};'
Explanation:
/^DEC::PKCS7\[/             # if current line begins with 'DEC::PKCS7[' then execute next block
{                           # start block
    s///;                       # remove all upto 'DEC::PKCS7['
    :1;                         # label '1' for code to jump to
    /\]\!$/!                     # if the line does not end with ']!' then execute next block
    {                               # start block
        N;                          # add next line to pattern space
        b1;                         # jump to label 1
    };                          # end block
    s///;                       # remove all from ']!' to end of line
};                          # end block

Notes:

• This works on single and multi-line.
• This will have unexpected behavior if you have ']!' in the middle of the input.
• This does not answer the question. It's already answered very well. My intentions are just to help other cases.