Capturing either one or other group with regex

Question

I have a hard time figuring out if this is possible with regexes. I have the following string (The original string is longer, it is a json string):

... "WorkstationName":"WS-8300E-007","IpAddress":"192.10.10.10" ...

And I want to catch the IpAddress or, if the IpAddress does not exists, the WorkstationName

# IPADDR = 192.10.10.10
... "WorkstationName":"WS-8300E-007","IpAddress":"192.10.10.10" ...

# IPADDR = WS-8300E-007
... "WorkstationName":"WS-8300E-007","IpAddress":"-" ...

I have tried several patterns:

• conditional lookahead
• capturing a backreference
• other tries I forgot

but without success, I need to capture the pattern in a named group (?P<ipaddr>) so that the output can be processed by other software.

The latest regex I ended up with is :

(?:("WorkstationName":)(?=.*IpAddress":"-"))?(?(1)(?:"([^"]+)")?|.*IpAddress":"([^"]+")?)(?P<ipaddr>(?(2)\2|\3))

So, basically, I do:

• check if "WorkstationName" is followed, at some point by an invalid ip ("-")
• if it is, capture the workstation name in \1
• if the group \1 exists, capture the workstation name
• otherwise capture the ip address

The hard time I'm having is using the named group, I have already suceeded in capturing everything in 2 groups, but I absolutely need to be on the same group depending on the string.

I cannot use JSON parsers

Answer 1

This one should suit your needs:

^.*(?:IpAddress(?!":"-)|WorkstationName)":"(?P<ipaddr>[^"]+)

Visualization by Debuggex

Demo on regex101