UNIX for Dummies Questions & Answers

Code:

Toby> cat sample1
 
This is some arbitrary text before var1, This IS SOME DIFFERENT ARBITRARY TEXT before var2
 
Toby> sed -e 's/^This .* before //' -e 's/This .* before //' sample1
 
var2

I need to convert the above text in sample1 so that the output becomes

var1, var2

by stripping away two different sets of arbitrary characters that occur before var1 and var2.

My attempt above apparently fails because the use of the wildcard and * will greedily match all the way to the last occurence in the line instead of the first occurence.

Can anyone please give me the sed command that will work for this?

Thanks

Code:

 sed 's/.*\(var1\).*\(var2\)/\1, \2/g' sample1

Or a similar one:

Code:

 grep -oE 'var.{1}' sample1

Thanks for the reply. But your solution does not do what I need. My problem is more general. In my problem the var1 and var2 are not literal strings but are themselves arbitrary and need to have the text before them stripped away. Here is a clearer description of what I need

Toby> cat sample1
This is some arbitrary text before var1, This IS SOME DIFFERENT ARBITRARY TEXT before var2
This is another line with first part before item1, This is the second part before item2

Toby> sed 's/.*\(var1\).*\(var2\)/\1, \2/g' sample1
var1, var2

This is another line with first part before item1, This is the second part before item2

The output I need from sed would then be

var1, var2
item1, item2