Regex: Match until "," but not if "," is inside brackets

Question

Given the following pattern:

group1: hello, group2: world
group1: hello (hello, world) world, group2: world
group1: hello world

of the style <group_name>: <group_value>[, <group_name>: <group_value>[...]].

In general I use the following regex to extract the values:

group1:\s(?P<group1>[^,\n]+)(:?,\sgroup2:\s(?P<group2>[^,\n]+))?\n

which works file unless a , exists inside the group_value.

I know that this toyexample can be solved by something like:

group1:\s(?P<group1>.+?)(?:,\sgroup2:\s(?P<group2>.+?))?\n

However I do want to protect myself agains matching everything accidentally so I would still like to limit my match when it encounters a ,.

Question: Is there a (general) way to match up to , and for that purpose ignore ,s that are in brackets?

Maybe `group1:\s(?P.*?)(?::?,\sgroup2:\s+(?P.*))?$`? Actually, it is a variation of your second solution, what is wrong with that one? — Wiktor Stribiżew, Aug 18 '23 at 08:09
Could you please do post expected output in your question, thank you. — RavinderSingh13, Aug 18 '23 at 08:09
the output i expect is the same as my toyexample. however i dont want to match too much by accident with `.+?`. i would like to at least stop on `,` when i encounter one.. unless inside of ballanced brackets. — Lukas Unterschuetz, Aug 18 '23 at 08:17
Maybe something like `group1:(?P(?>[^,\n()]*(?:($(?:[^()]++|(?1))*$))?)*)(?:, (?P.*))?\n` https://regex101.com/r/rzH1NG/1 — The fourth bird, Aug 18 '23 at 08:18

The fourth bird · Accepted Answer · 2023-08-18T09:05:32.113

Using pcre, you could make use of a recursive pattern for balanced parenthesis with possessive quantifiers.

you define the pattern for group 1, and if the same logic applies for group 2 you can recurse the subpattern defined in group 1.

As you exclude matching a newline in the negated character class, you might use \h to match horizontal whitespace characters instead of using \s

\bgroup1:\h+(?P<group1>(?:[^,\n()]*(?:(\((?:[^()\n]+|(?2))*+\)))?)*+)(?:,\h+group2:\h+(?P<group2>\g<group1>))?\R

\bgroup1:\h+ Match the word group1 and then : and 1+ horizontal whitespace chars
(?P<group1> Named group1
- (?: Non capture group
  - [^,\n()]* Match optional chars other than , newline ( or )
  - (?: Non capture group
    - ($(?:[^()\n]+|(?2))*+$) Match balanced parenthesis recursing group 2
  - )? Close group and make it optional
- )*+ Close the group and optionally repeat with a possessive quantifier (no backtracking)
) Close group1
(?: Non capture group
- ,\h+group2:\h+ Match group2: between horizontal whitespace chars
- (?P<group2>\g<group1>) Named group2, recurse the subpattern in named group1
)? Close the non capture group and make it optional
\R Match a newline

1 Answers1