negative with positive lookbehind regex issue

I'm trying to find the location of open parenthesis, '(', that are preceded by a number, but not when that number is an integer also preceded by the characters 'O' or 'S'.

Example:

str = '12()+F34()+O56()';

should return ind = [3,9], i.e. the open parenthesis following the 12 and 34, but not O56

I tried this:

ind = regexp(str,'(?<=((?<![OS])[0-9]+))[\(]');

but it gives me all of them (ind:[3,9,15]). It does however exclude the cases when there is just a single number after 'O' or 'S' (e.g. str = '12()+F34()+O5()'; -> ind:[3,9])

Does anyone know the proper regular expression for this?

Matlab version: 7.13.0.564 (R2011b)

[Stephen23]

@Sebastian: You might like to try using my FEX submission makeregexp:

http://www.mathworks.com/matlabcentral/fileexchange/48930-interactive-regular-expression-maker

which lets you interactively develop regular expressions and see regexp's outputs change as you type.

Answer 1

[Stephen23]

Try this regular expression: |(?<=\d+)(?<![OS]\d+)\(|

It relies on the fact that lookaround operations do not consume any characters: the first lookaround matches some digits, the second then checks that any digits are not preceded by the letters O or S. Here it is tested:

>> str = '12()+F34()+O56()';
>> regexp(str,'(?<=\d+)(?<![OS]\d+)\(')
ans =
     3     9

To help develop this regular expression I used my FEX submission makeregexp:

http://www.mathworks.com/matlabcentral/fileexchange/48930-interactive-regular-expression-maker

which lets you interactively develop regular expressions and see regexp's outputs change as you type.

[Sebastian]

"lookaround operations do not consume any characters"

Of course, how did I forget. That explains everything. Thanks for the nice regex too.

Answer 2

[Azzi Abdelmalek]

str = '12()+F34()+O56()';
ii1=regexp(str,'(?<=[OS]\d+)(\()' )
ii2=regexp(str,'(?<=\d+)(\()' )
out=setdiff(ii2,ii1)

[Stephen23]

Sebastian's "Answer" moved here:

Thanks Azzi, I will consider that solution. Do you know what is wrong with my initial expression?

Right now I'm resorting to

ind = [regexp(strFunc,'(?<=[^OS0-9][0-9]+)[\(]'),regexp(strFunc,'^[0-9]+[\(]','end')]

which seems to get the job done, but is not exactly the prettiest.