First of all thank you for taking the time to look into this.

Second, apologies for not providing more engine details. I should have known better.

I am using PCRE as you guessed and I am indeed looking for the underlined part in  aaap1bbbp1cccp2ddd (I need p1 and p2 separated but that's easy).

I tried  your pattern on a simple string and altho I don't understand how it works it does seem to work.(I will study it)

FYI p1 is (?<p1>type=\d+, shape=((\d+[ ,])+))

and p2 is  (?<p2>≡⍴2 0⍕B *\d+ (\d+))  (some Unicode chars in there)

The UTF8 text to search includes \n and DOTALL has to be used.

From your pattern and mine above I created the pattern

((?<p1>type=\d+, shape=(?:(?:\d+[ ,])+)))(?:(?!(?&p1)|(?<p2>≡⍴2 0⍕B *\d+ (\d+))).)*((?&p2))

which defines p1 and p2 to be reused and that worked!

Thanks a lot!

/Dan