how to get several numbers from a string in matlab

Question

I'm struggling with getting numbers out of several strings:

'L/22P/93'
'P/8P/48L/3'
'1L/63P/751' (this one is: 1, 63, 75, 1)
'PL/18'
'P/30P/5'
'PP'

I want to get all numbers, so I can use them for calculation.

I have tried using regexp, but I can only get the first number of each string.

Answer 1

One easy way would be to replace all other characters with spaces, then read the string:

function nums = read_numbers(str)
        %// This works only for integer numbers without sign

        not_digit = (str < '0') | (str > '9');
        str(not_digit) = ' ';
        nums = sscanf(str, '%u');
end

As the comment says, the function doesn't take in account signs (+/-), the decimal point or real numbers in scientific notation.

After saving the above code in the file read_numbers.m , you can use it then like in this example:

>> read_numbers('L/22P/93');
        22
        93

Answer 2

While regular expressions can be intimidating, MATLAB's regex documentation is fairly comprehensive and should be sufficient to help solve this problem.

As others have commented there are a couple questions here that need to be answered in order to provide a comprehensive answer to your question:

1. What code have you tried so far that only yields the first number? As @michael_0815 states, the simplest regex call returns the indices to all of the numbers in the string.
2. What is your criteria for a number? Specifically in your third string you say there are 4 number groups when there are only 3. Do you only want a maximum grouping of 2 digits? This affects how the regex is structured.

In the meantime this should return what you've requested using regex , though it assumes your numbers are unsigned integers and you want a maximum grouping of 2 digits.

teststr = '1L/63P/751';
test = str2double(regexp(teststr, '\d{1,2}', 'match'));

Which returns the following array:

test =

 1    63    75     1

I would recommend playing around with an online regex tester to see how your inputs affect the results. My favorite is regex101 . It's geared for other languages but the MATLAB syntax is similar enough for the most part.

Answer 3

Let your data be defined as a cell array of strings:

s = {'L/22P/93'
     'P/8P/48L/3'
     '1L/63P/751'
     'PL/18'
     'P/30P/5'
     'PP'};

Then

y = regexp(s, '\d+', 'match'); %// cell array of cell arrays of strings
y = cellfun(@str2double, y, 'uniformoutput', 0); %// convert to cell array of vectors

gives the result as a cell array of vectors:

y{1} =
    22    93
y{2} =
     8    48     3
y{3} =
     1    63   751
y{4} =
    18
y{5} =
    30     5
y{6} =
     []