如何将带引号的字符串分配给 bash 中的变量，并将其用作 awk 中的模式？

Question

The command that I'm using is:

objects=shubham
awk -v myvar="$objects" '($0~myvar),/REPLACE/' a.txt

the a.txt file contains

other-unwanted-content-here
target(shubham)
hello('shubham')
abc
bcd
REPLACE
other-unwanted-content-here

My desired output is:

hello('shubham')
abc
bcd
REPLACE

...but I'm getting target(shubham) as well. How can I make hello('shubham') with the quotes be the place where awk starts matching?

Answer 1

Two approaches, depending on what you want.

• Keep objects defined the way it is, amend your awk variable's assignment

  objects=shubham awk -v myvar="^hello[(]'$objects'[)]\$" '($0~myvar),/REPLACE/' <<'EOF'

  See this running at https://ideone.com/Bto1hB

• Change objects to match only the target line

  objects="^hello[(]'shubham'[)]\$" awk -v myvar="$objects" '($0~myvar),/REPLACE/' a.txt

  See this running at https://ideone.com/nF8MUl

Note that in either case, ~ in awk is a regex operator; since in most regex forms (including POSIX ERE ) ( and ) are syntax, we had to change your string to be a regex that matches the desired line, instead of containing exactly that line itself. (Note that in both cases, the backslash before the $ in the regex is shell syntax, not regex syntax itself, necessary only because we're in double quotes; in single quotes those backslashes would need to be left out).

Answer 2

FWIW I'd do a full string rather than partial regexp comparison for robustness and use a flag instead of a range expression for efficiency and ease of future changes:

$ objects="hello('shubham')"
$ awk -v myvar="$objects" '$0==myvar{f=1} f; $0=="REPLACE"{exit}' a.txt
hello('shubham')
abc
bcd
REPLACE

Answer 3

Providing the pattern s are unique to the relevant lines, combining the variable in your starting line pattern (using && ) with /hello/ should fix the extra line problem:

awk -v myvar="$objects" '($0~myvar && /hello/),/REPLACE/' a.txt

output:

hello('shubham')
abc
bcd
REPLACE

Tested using mawk 1.3.4 on Raspberry Pi 400

Answer 4

a generic awk solution allowing for partial matches that properly escapes all special characters for all awk variants without having to pre-escape them at the shell level:

# one of the cleaner ways to pass single-quoted single quotes 
# to awk since the octal codes aren't being interpreted by the shell  

___='??hello(\47sh{}u*[]bh(+)am\47)'

printf '%s' "other-unwanted-content-here 
             target(shubham)
             ??hello('sh{}u*[]bh(+)am')
             abc
             bcd
             REPLACE
             other-unwanted-content-here" |

 mawk2 -v ___="$___" 'BEGIN { _=(_="\\")_ gsub("[[-_:-/,-@{-~]", "["(__ = "&")"]",___) gsub("["(_)"^]", (_)__; ___), FS = ___ _*=_ } !_<NF, ____==$_' ____='REPLACE'

??hello('sh{}u*[]bh(+)am')
abc
bcd
REPLACE

The pair of regex -escaping gsub() s escapes more than what's needed because the alternative is a backslash nightmare