correctly split with multiple under score

I have a cell array with several underscore. I was wondering how to correctly split the string when you have underscores. My input looks like this:

(a) MK_0334_300151IT0005274805/MD_01082027_3_103 I want only 300151IT0005274805

(b) MK_0334_700720ES01653860_1_103 . I want only 700720ES01653860_1

© CAB001_0578_SK09 1100 0000 i want SK09 1100 0000.

(d) CR_07_C_59163230 i want only C_5916323

i have tried to split the data using : regexp({my_data}, '_', 'split'); then selecting the cell 3 and then split it using

regexp({cell 3}, '/MD', 'split');. however, that doensn't work for (b) and (d) data type.

[Dyuman Joshi]

The results you want don't have uniform logic. Are you expecting a single logic for all 4 options?

[Sheikh Omar Bah]

Hello Joshi,

yes, i was looking for a logic that can work for all.

[the cyclist]

The hard part is not writing the MATLAB code. The hard part is understanding what single "rule" or logic is that results in what you want.

Tell us the rule.

[Adam Danz]

To illustrate what others have mentioned above, explain why you don't want to return "SK09 1100" in your 3rd example or why you don't want to return "07" in your 4th example?

Answer 1

[Walter Roberson]

S = {'MK_0334_300151IT0005274805/MD_01082027_3_103', 'MK_0334_700720ES01653860_1_103', 'CAB001_0578_SK09 1100 0000', 'CR_07_C_59163230' }

S = 1×4 cell array
    {'MK_0334_300151IT0005274805/MD_01082027_3_103'}    {'MK_0334_700720ES01653860_1_103'}    {'CAB001_0578_SK09 1100 0000'}    {'CR_07_C_59163230'}

regexp(S, '(?<=^[^_]+_[^_]+_)[^/_]+(_[^/_]+)?', 'match', 'once')

ans = 1×4 cell array
    {'300151IT0005274805'}    {'700720ES01653860_1'}    {'SK09 1100 0000'}    {'C_59163230'}

The rules used:

• discard up to and including the second _
• scan forward from there, stopping just before the first / or the second _ if present

[the cyclist]

Using the Roberson Clairvoyance Toolbox™ again, I see. Still waiting on the public beta for that.

[Sheikh Omar Bah]

Thanks Walter. That was very helpful.

I just have to add some tweak. some of the data had '/' which were important to keep.

MK_0458_002/001/30/210/E/K_1. where it was necessary to keep 002/001/30/210/E/K_1.

so i instead use

Split_data = regexp({s}, '(?<=^[^_]+_[^_]+_)[^_]+(_[^/_]+)?', 'match', 'once');

A = regexp(Split_data, '/MD', 'split');

then select the first cell with a loop.