shell to match a pattern

Shell Programming and Scripting
1

/data}/{bla/${INSTANCE} or /data}/{bla/$INSTANCE

the idea is to match a env varialbe name, so it could be between ${ } or a single $
Perl can do it with: perl -ne '/\$\{?(\w+)\}?/;print $1'
What shell can do?

2

Where is the string you wnat to match? Is it in a file or is it already in a variable?

If it's in a variable:

case $var in
     *\${INSTANCE}*| *\$INSTANCE* )
        echo MATCH;;
     *) echo NO MATCH ;;
esac

If it's in a file or the output of a command, use grep:

grep -e '${INSTANCE}' -e '$INSTANCE'
3

I want to return the word between ${} or after $, but I don't know the word name.

example of Perl result

 
$ echo '/data}/{bla/${INSTANCE}' | perl -ne '/\$\{?(\w+)\}?/;print $1'
INSTANCE
 
$ echo '/data}/{bla/$INSTANCE' | perl -ne '/\$\{?(\w+)\}?/;print $1'
INSTANCE
4

You haven't answered the question:
Where is the string you want to match?

[indent]Is it in a file?
Is it already in a variable?
Is it the output of a command?

Are you looking for it in a single line? Or in a file containing more than one line?
[/indent]

5

It's better if you provide more details but hope this helps:

echo '/data}/{bla/${INSTANCE}' | sed 's/.*${*\([A-Za-z0-9]\{1,\}\)}*.*/\1/'
6

supposing it is a file has 1 line only
/data}/{bla/${INSTANCE} or /data}/{bla/$INSTANCE

7

:b:
Thanks the expression worked for both. Can you expain how it worked?
i don't understand usage of "\1"

8

You can do it entirely in the shell, using no external commands:

read -r line < "$FILE"

var=${line#*$}
var=${var#*{}
var=${var%%[!a-zA-Z_0-9]*}

echo "$var"

Or you can use sed.

sed 's/[^$]*${*\([a-zA-Z_0-9]*\)}*.*/\1/' "$FILE"

Or you can use awk:

awk -F '$' '{
  gsub( /[{}]/,"", $2 )
  gsub( /[^a-zA-Z_0-9].*/,"", $2)
  print $2 }
' "$FILE"

These scripts all find the first variable in the line if there is more than one.

9

Hmmm...I'm not great at explaining but here goes...

Think of \1, \2, \3, ..., \n as "place-holders". This is where results from your searching are placed.

In sed, you can assign a result to a "place holder" by using \(<regexp>\). If you have 3 \(<regexp>\) then you'll have 3 "place-holders" that is if they have a result...

So the \1 means:

\1 - result of this \([A-Za-z0-9]\{1,\}\).

Aaaarrrrggghhhh!!! I just can't explain it enough, man :D. Check the ff. and I hope they give a more detailed explanation:

http://www.grymoire.com/Unix/Sed.html\#uh-4

Regular-Expressions.info - Regex Tutorial, Examples and Reference - Regexp Patterns