Replacing part of a string with sed

Question

I know that some form of this question has been asked multiple times over, but I can't seem to get any of the solutions I've found to work. I have a line in a file that looks something like:

<package foo="bar" unique-identifier="NEEDS_TO_BE_REPLACED" version="foobar">

I need to replace the part indicated above, but I seem to always change unwanted information as well. The text that needs to be replaced is different from file to file, so currently I'm using a wildcard in it's place. Below are the things I've tried.

sed 's/unique\-identifier\=\".*\"/unique\-identifier\=\"NEW_TEXT\"/g' $file >> $newFile

sed 's/\bunique\-identifier\=\".*\"\b/unique\-identifier\=\"NEW_TEXT\"/g' $file >> $newFile

sed 's/\<unique\-identifier\=\".*\"\>/unique\-identifier\=\"NEW_TEXT\"/g' $file >> $newFile

sed 's/[[:<:]]unique\-identifier\=\".*\"[[:>:]]/unique\-identifier\=\"NEW_TEXT\"/g' $file >> $newFile

All of those are various solutions that I've found, the last being Mac OSX specific. Ideally I'd like to be able to replace just what's in the double-quotes, but my regex skills are lacking.

Answer 1

You can try:

sed 's/unique\-identifier\=\"[^"]*\"/unique\-identifier\=\"NEW_TEXT\"/g' $file >> $newFile
                             ↑

The reason your first attempt failed is because .* is greedy and matches as much as it can provided that the rest of the expression can be matched as well. In this case it matched also " version="foobar, and the rest of the expression (\") matched the last character.

Replacing .* with [^"] makes this part match everything that is not ".

You could also remove repetitions and unnecessary escapes in your expression to make it more readable, like this:

sed 's/\(unique-identifier="\)[^"]*/\1NEW_TEXT/g' $file >> $newFile