引号之间的Grep模式

Question

I'm trying to grep a code base to find alpha numeric codes between quotes. So, for example my code base might contain the line

some stuff "A234DG3" maybe more stuff

And I'd like to output: A234DG3

I'm lucky in that I know my string is 7 long and only integers and the letters AZ, az.

After a bit of playing I've come up with the following, but it's just not coming out with what I'd like

grep -ro '".*"' . | grep [A-Za-z0-9]{7} | less

Where am I going wrong here? It feels like grep should give me what I want, but am I better off using something else? Cheers!

Answer 1

Using basic or extended POSIX regular expressions there is no way to extract the value between the quotes with grep . Since that I would use sed for a portable solution:

sed -n 's/.*\"\([^"]\+\)".*/\1/p' <<< 'some stuff "A234DG3" maybe more stuff'

However, having GNU goodies, GNU grep will support PCRE expressions with the -P command line option. You can use this:

grep -oP '.*?"\K[^"]+(?=")' <<< 'some stuff "A234DG3" maybe more stuff'

.*" matches everything until the first quote - including it. The \\K option clears the matching buffer and therefore works like a handy, dynamic lookbehind assertion. (I could have used a real lookbehind but I like \\K ). [^"]+ matches the text between the quotes. (?=") is a lookahead assertion the ensure after the match will follow a " - without including it into the match.

Answer 2

The problem is that an RE is pretty much required to match the longest sequence it can. So, given something like:

a "bcd" efg "hij" klm "nop" q

A pattern of ".*" should match: "bcd" efg "hij" klm "nop" (everything from the first quote to the last quote), not just "bcd" .

You probably want a pattern more like "[^"]*" to match the open-quote, an arbitrary number of other things, then a close quote.

Answer 3

So after more playing about I've come up with this which gives me what I'm after:

grep -r -E -o '"[A-Za-z0-9]{7}"' . | less

With the -E allowing the use of the {7} length matcher