Change regular expressions with a common sub-expression with different prefix and suffix expressions

Question

I have the following regular expression that has 3 alternations (see full regular expression below), each with its own prefixes and suffixes. I feel that this is being repeated excessively and would like to simplify, if possible. I am matching values ​​in a malformed JSON string to replace values ​​that do not have a key with indexed keys.

Each rotation must correspond to a pair of prefixes and a suffix with a sub-expression. I currently have 3 pairs, but that could change. If I had a few more pairs, the entire regex would turn into a nightmare to change and understand if I need to modify the repeating sub-expression.

Question

How can I shorten the entire regex below without requiring repetition of the subexpression for the listed pairs of suffixes and prefixes?

Subexpression repeated in each alternation

("(?:[^\\"]+|\\.)*") 

prefix / suffix pairs

• {,
• ,
• ,}

Whole regex

 /\{("(?:[^\\"]+|\\.)*")(?=,)|,("(?:[^\\"]+|\\.)*")(?=,)|,("(?:[^\\"]+|\\.)*")(?=\})/g 

Test lines

• {"trailer":"","pallet":"A","date":"11-Dec-15","c","z","a"}
• {"trailer":"","pallet":"A","a","date":"11-Dec-15"}
• {"a","trailer":"","pallet":"A","date":"11-Dec-15"}
• {"a","trailer":"","pallet":"A","date":"11-Dec-15","z\""}
• {"trailer":"","pallet":"A","11-Dec-15"}
• {"trailer\"","pallet":"A","11-Dec\"-15","z\""}

Living example

Please limit your responses to regular expressions, not JSON validation methods, as I am trying to get a better idea of ​​regular expressions, and this is just an example that I use.

Answer 1

While regex can be simplified:

 /\{("(?:[^\\"]+|\\.)*")(?=,)|,("(?:[^\\"]+|\\.)*")(?=,)|,("(?:[^\\"]+|\\.)*")(?=\})/g 

To:

 /{("(?:[^\\"]+|\\.)*")(?=,)|,("(?:[^\\"]+|\\.)*")(?=,)|,("(?:[^\\"]+|\\.)*")(?=})/g 

Removing the escaping { and } , as this is not required for the JavaScript regex mechanism.

It is not possible to remove your explicit duplicate pattern ("(?:[^\\"]+|\\.)*") In JavaScript.

JavaScript does not support all the same regular expression functions that are supported by PCRE (PHP, C ++, Perl, etc.).

For example, in PHP / C ++ you can do this:

 {("(?:[^\\"]+|\\.)*")(?=,)|,((?1))(?=,)|,((?1))(?=}) 

For Perl 5.22, you will need to avoid this again { so that it looks something like this:

 m/\{("(?:[^\\"]+|\\.)*")(?=,)|,((?1))(?=,)|,((?1))(?=})/g 

This (?1) is a subroutine call to match the regular expression within capture group 1 , which in this case is ("(?:[^\\"]+|\\.)*") .