How to use named regex groups in ack output?

Question

Suppose I have a foo.txt file with the following content:

[2010-11-13 12:00:02,656]
[2010-11-13 12:00:02,701]
[2010-11-13 12:00:02,902]

When I ack for the date portion with the following, it works:

ack "(?P<foo>\d{4}-\d{2}-\d{2})" foo.txt --output "\$1"

2010-11-13
2010-11-13
2010-11-13

But when I try to use --output with the named group "foo", I cannot get it to work:

ack "(?P<foo>\d{4}-\d{2}-\d{2})" foo.txt --output "(?P=foo)"

(?=foo)
(?=foo)
(?=foo)

Any help is greatly appreciate it. Thanks so much.

Answer 1

ack "(?P<foo>\d{4}-\d{2}-\d{2})" foo.txt --output "\$+{foo}"

(?P=foo) only works inside the regular expression (it's the named equivalent of \1). $+{foo} is the named equivalent of $1.

Also, with most shells, single quotes will help you avoid extra backslashes:

ack '(?P<foo>\d{4}-\d{2}-\d{2})' foo.txt --output '$+{foo}'