重击：替换包含任何字符类型的引号之间的子字符串

Question

I am trying to edit part of a string using a Bash (version 3.2) script.

For example, in $line

line='<Coordinate text1="0" coordinateIndex="78?907??" anotherID="9098" yetanoherID="1.2.3" xyz:text="abc"/>'

I need to edit the contents of coordinateIndex (which can have any character/any length). My last attempt (below) doesn't give an error but doesn't solve the problem either:

echo "${line/coordinateIndex=\"\[(.*)\]\"/coordinateIndex="124"/line}"

I also tried using ")" instead of "]"; also .+, among many others.

The output I am looking for is:

line='<Coordinate text1="0" coordinateIndex="124" anotherID="9098" yetanoherID="1.2.3" xyz:text="abc"/>'

I tried solutions based on

Regex Match any string powershell

http://unix.ittoolbox.com/groups/technical-functional/shellscript-l/shell-script-to-replace-string-within-double-quotes-4107915

https://superuser.com/questions/515421/using-sed-get-substring-between-two-double-quotes

but I am still unable to solve this.

Any help appreciated, thank you!

Answer 1

It can be done easily with perl :

#!/bin/bash

str=$(cat << EOF
line='<Coordinate text1="0" coordinateIndex="78?907??" anotherID="9098" yetanoherID="1.2.3" xyz:text="abc"/>'
EOF
)

echo "$str" |perl -pe 's|(coordinateIndex=)".*?"|$1"abc"|g'

Output:

bash test.sh 
line='<Coordinate text1="0" coordinateIndex="abc" anotherID="9098" yetanoherID="1.2.3" xyz:text="abc"/>

Answer 2

You can do it with Bash regex matching.

var=coordinateIndex
value=124
if [[ $line =~ $var=\"([0-9|\?]+)\" ]]; then
    echo ${line/$var=\"${BASH_REMATCH[1]}\"/$var=\"$value\"}
fi

They key here is to know which type of characters can be found in between the quotes after coordinateIndex= . If you just use * , which matches any character, you'll end up matching and replacing everything up to the final " in the variable line .