Split along word into parts with different lengths

problem solved

[José-Luis]

Does it always start with "NCCINS.AttidudePositionINS1" and only differs by the last part? Or does the relative position always change?

If it is static, then it is just a matter of splitting strings by indexing. Otherwise is becomes a rather complicated regex.

[Cedric]

I guess that it is not always starting with the same string, otherwise you wouldn't have to parse it. We need more examples, as well as a clear description of criteria which define sub parts.

[dpb]

Specifically, what's the rule for breaking down the first set of six letters into the two groups of three if that isn't a consistent string. If it is, then just substring referencing of 1:3 and 4:6 takes care of it.

The rest looks like looking for capitals after a lowercase letter.

Answer 1

[Cedric]

Here is a solution based on comments above. Yet, we will need you to answer these comments so we can refine the approach. Assuming

 data = {'NCCINS.AttitudePositionINS1.Pitch', 'NCCINS.AttitudePositionINS1.Roll'} ;

we perform a single call to REGEXP on a comma-separated merger of the original strings, as follows:

 pattern = '(\w{3})(\w{3}).(.[^A-Z]+)(.[^A-Z]+)([^\.]+).([^,]+)' ;
 tokens  = regexp( sprintf('%s,', data{:}), pattern, 'tokens' ) ;

with that, we get

 >> tokens{1}
 ans = 
    'NCC'    'INS'    'Attitude'    'Position'    'INS1'    'Pitch'

 >> tokens{2}
 ans = 
    'NCC'    'INS'    'Attitude'    'Position'    'INS1'    'Roll'

[Cedric]

Can you attach a few example, or one package, or even the 70. The more I have, the more I can refine the approach. The approach above is based on regular expressions and pattern matching. I cannot explain it generally because there would be way too much to develop. Yet, once we have a pattern which suites your case specifically, I can explain it to you.