Sed expression for IP address, getting unexpected result for matching group

Question

I'm trying to extract an IP address from a text, and don't understand the outcome according to the regex I write. Apparantly this:

echo '"IPAddress": "173.14.0.3",' | sed -n -r -e 's/"IPAddress": "(.*)"/\1/p'

returns

173.14.0.3,

Why do I get the , at the end? Doesn't "(.*)" instruct the regex to create a matching group of everything between the last two " ?

Originally I started out with

echo '"IPAddress": "173.14.0.3",' | sed -n -r -e 's/"IPAddress": "([0-9]{1,3}.[0-9]{1,3}.[0-9]{1,3}.[0-9]{1,3})"/\1/p'

, but got same result. I used regex101 , and there I get a different response. Why?

Answer 1

Your input:

input:            "IPAddress": "173.14.0.3",
matched by regex: ^^^^^^^^^^^^^^^^^^^^^^^^^  (note: comma not matched)
captured:                       ^^^^^^^^^^

The matched part is replaced by the captured substring and substituted back into the original string, yielding:

result:       173.14.0.3,
not affected:           ^
replacement:  ^^^^^^^^^^

If you want to get rid of the comma, include it in the match (so it gets substituted by nothing):

s/"IPAddress": "(.*)",/\1/p

regex101 shows the same behavior: https://regex101.com/r/Fy5Lj3/4

Answer 2

General advice: regex101.com does not support the regular expression languages supported by sed . These are explained in the POSIX specification.

While simple things might look similar, expressions differ significantly in detail. Do not use regex101 when working on regular expressions for sed .