How could i search a text for varying patterns

Hello, I am trying to search through an excel file looking through model names and only pulling out the option codes.

For Example the excel sheet with the model names would look like: '1010AB-50K-TB' '1010AB-25K-B' '1010AB-100K'

and I want to create a new array with just the option codes (which in this case is AB). So to look something like this:

'AB' 'AB' 'AB'

and so on... the option codes vary from two letters to three letters and vary from A-Z (AB, AE, AF, AAA, ABD). The only consistency in the model names is the four numbers before the option code and the dash after. How do y'all think the best way to do this would be?

[Paolo]

Do the option codes always start with A?

[Joey Cavale]

They will vary alphabetically, so they will go from AA, AB, AC... BA... and so on all the way to ZZ and then start at AAA and go till ZZZ. I tried using Regexp but cant figure out a good way to identify the pattern

[Joey Cavale]

Here are some examples

[Paolo]

I have edited my regex expression, now the first letter will also range from A to Z. Check out my answer down below.

Answer 1

[Paolo]

For a pattern of two to three letters, starting with any letter (A to Z) and ending with any letter (A to Z) you can use the following code:

 expression = '([A-Z][A-Z]?[A-Z](?=-))';
 model_names = {'A1010ACZ-50K-AAA' '1010AB-25K-B' '1010AT-100K'};
 [tokens] = regexp(cell2mat(model_names),expression,'tokens');

The expression has three elements, all ranging from A to Z (the [A-Z]). The third element is optional (denoted by the question mark). The expression contains a lookahead (?=-) which searches for the dash however does not include it in the tokens.

The code returns:

 tokens{:} = {'ACZ'},{'AB'},{'AT'}

[Joey Cavale]

Thanks! it works well but tends to grab the end letters when faced with a model number like '1211AE-100KN-TB'. In this case it grabs the AE and the KN because of the dash... is there any way I can limit how far it searches into the word, like say I wanted to limit it to only searching to the 10th character or something?

[Paolo]

What output would you require from a model number of '1211AE-100KN-TB?' Just the first term, 'AE'?

[Joey Cavale]

Yeah, it works but there are some rare cases where a '-KN-TB' is there and the code picks up the 'KN'

[Paolo]

Are you working with each model number individually? If so, you can either use:

 expression = '([A-Z][A-Z]?[A-Z](?=-)).*'

or even specify once in the function:

 [tokens] = regexp(model_names,expression,'tokens','once');

Or are you working with a cell array?

[Joey Cavale]

Ahhh.. the 'once' term in the regexp function works wonders, thank you!

expression = '([A-Z][A-Z]?[A-Z](?=-))'; 
[txt] = regexp(MODEL, expression, 'match','once');

Answer 2

[jonas]

Not sure if it's the best way, but the code starts at str(5) and ends at the dash, so this should work

[~,s] = regexp(str, '-', 'match')
out=str(5:s(1)-1)

[Joey Cavale]

MODEL = TXT(:,1); % Model names
for i = 7:length(MODEL)
    if contains(MODEL{i,1},'-') == 0
        continue % Skipping names that don't satisfy argument
    end
    str = MODEL{i,1};
    [~,s] = regexp(str,'-','match');
    out{i,1} = str(5:s(1)-1);
end

Got this to work, had to make it a little janky but it ended up getting the job done, thanks for the help!

[Paolo]

Unsure as to why you selected this answer. The code is limited in application and will not work if for some reason the codes change format, let's say from

 1010AB-50K-TB

to

 10105AB-50K-TB

will result in an incorrect output.

 5AB