16th November 2005, 1 min read

Grep is just not for matching lines anymore

Thanks to Owen Kaser, I’ve learned that grep can return just the match, and not the whole line.

These examples say it all:

$ grep Ab phone
505-837-2938 Abby Abbott
212-940-2039 Abel Baker
301-302-3030 Abigail Adams


$ grep –o 'Ab[^ ]*' phone
Abby
Abbott
Abel
Abigail

So, you can get all email addresses in an HTML file using the following syntax (correct me if this is wrong):
grep -o "[a-zA-Z\.]@[a-zA-Z\.]" somefile

Under Windows, you can use the Win32 port of grep which has this feature. Tim Charron’s port of grep doesn’t have it. Tim Charron’s version is especially convenient since it supports a findutils-like function (grep -S “pattern” *.txt will go in subdirectories) whereas the findutils port to Windows clashes with the Microsoft find command. (Don’t tell me to use cygwin, I already do.)

Grep would be so much more useful however if it supported reluctant quantifiers.