Regex to parse command line options

Question

I'm faced with a need for parsing a string into key-value pairs, where the value may be optional. Standard command line parsers are not useful, because all the ones I checked accept a String[] and not a String. Thus, I resorted to regex, and sure enough, faced with the following:

Some people, when confronted with a problem, think "I know, I'll use regular expressions." Now they have two problems.

First, the input string:

"/opt/sensu/embedded/bin/ruby /opt/sensu/embedded/bin/check-graphite-stats.rb " +
"--crit 25 --host 99.99.999.9999:8082 --period -5mins --target 'alias(scale(divideSeries(" +
"summarize(sumSeries(nonNegativeDerivative(transformNull(exclude(" +
"\\\"unknown\\\"), 0))), \\\"30d\\\", \\\"sum\\\", false),summarize(" +
...gigantuous string
\\\"sum\\\", false)), 100), \\\"3pp error rate\\\")' " +
"--unknown-ignore --warn 5"

Next, my regex:

(--(?<option>.+?)\s+(?<value>.+?(?=--))?)+?

the above almost works, but not quite.

Output:

--crit 25 
--host 99.99.999.9999:8082 
--period -5mins 
--target 'gigantuous string' 
--unknown-ignore 
--warn

Why is the value of --warn not picked up?

Your input is a shell command. How did you end up trying to write a shell parser in Java? — that other guy
– that other guy, Commented Nov 8, 2018 at 1:48
@thatotherguy, I can't speak for the poster, but I want to be sure the web API I'm unit testing is invoking a shell command properly. — JohnK
– JohnK, Commented Jan 26, 2023 at 18:31

Harry Cutts · Accepted Answer · 2018-11-08 02:32:18Z

1

Because you're doing a positive lookahead to the next -- at the end of the regex ((?=--)), the value of the last parameter in the string isn't picked up as it's not followed by --. Accepting the end of the string as an alternative ((?:(?=--)|$)) and then filtering values that don't start with -- (by replacing .+? with .(?:[^-].+?)?) should behave in the way you want:

(--(?<option>.+?)\s+(?<value>.(?:[^-].+?)?(?:(?=--)|$))?)+?

(However, as others have mentioned, I'd be very surprised that there isn't a Java argument parsing library that would suit your use case. Even if it means writing the code to split your string into arguments yourself, it might be less brittle.)

edited Nov 8, 2018 at 2:32

answered Nov 8, 2018 at 2:00

Harry Cutts

1,43612 silver badges26 bronze badges

Sign up to request clarification or add additional context in comments.

5 Comments

Abhijit Sarkar Over a year ago

Nop! it picks up --unknown-ignore --warn 5 together.

Harry Cutts Over a year ago

Oh yes, so it does. What about adding [^-] at the start of the value group? ((--(?<option>.+?)\s+(?<value>[^-].+?((?=--)|$))?)+?)

Abhijit Sarkar Over a year ago

No. I appreciate your time, but there are plenty of regex testers online, you should check first; this seems like throwing Spaghetti at the wall hoping it would stick.

Harry Cutts Over a year ago

Ah, sorry, looks like I misconfigured my tester. (--(?<option>.+?)\s+(?<value>.([^-].+?)?((?=--)|$))?)+? should work.

Abhijit Sarkar Over a year ago

Some of the groups were not necessary, I converted them to non-capturing (--(?<option>.+?)\s+(?<value>.(?:[^-].+?)?(?:(?=--)|$))?)+?. Other than that, your answer works for me, so I accepted it. Thanks.

Collectives™ on Stack Overflow

Regex to parse command line options

1 Answer 1

5 Comments

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

1 Answer 1

5 Comments

Your Answer

Sign up or log in

Post as a guest

Related