Read only numbers from string

Given a string, how to read only the numbers? I am trying to sum them up afterward.

This was my solution:

splitted = strsplit(str)
total = 0;
array = [];
for i=1:numel(splitted)
   if ~isnan(str2double(splitted{i}))
       array = [array " " splitted{i}]
   end
end
total = nansum(str2double(array))

However, this is only assuming the numbers are not attached to any characters. Strings like '42hi' would not be counted.

I then thought about something like this:

multiple_digit = [];
single_digit = [];
k=1;
for ii=1:numel(str)
    if isa(str2num(str(ii)), 'double')
        if ~isa(str2num(str(ii+1)), 'double') %if next character isn't a number, then puts that number in
            single_digit(k) = str2num(str(ii)) 
        end
    end
end

I am checking each individual characters in the string to see if it is a number. If it is, then I check to see if the next character is also a number so that I can take a string like '100' as 100, not 1. However, this seems very complicated, as I have to control whether I am going out of the array index or not.

[Stephen23]

"However, this is only assuming the numbers are not attached to any characters. Strings like '42hi' would not be counted."

What does this mean? Do you want to match that 42 or not?

Your question is missing the most important information of all: an actual specification or example of what the string and number substrings are like. It is very difficult to write code to parse strings given no information about exactly what is in the string, how the numbers are to be identified, etc.

"However, this seems very complicated..."

It is very complicated. Read this:

http://xyproblem.info/

[SJ Won]

I was referring to how my code would encode strings like 42hi as 4 and 2, when I want the 42 itself.

Answer 1

[Stephen23]

"Given a string, how to read only the numbers?"

That really depends on what you define as being a "numberˇ: which of these are "numbers" according to your definition?:

• 1
• -1
• +1
• 1.0
• 1e-2
• NaN
• Inf
• 1+2i
• i
• e
• pi
• etc. etc.

If you are only interested in matching integers, then this should get you started:

>> str = '12  3  4';
>> sum(str2double(regexp(str,'\d+','match')))
ans = 19

[SJ Won]

Sorry, I never considered the possibility of other expressions that could represent numbers, so it was quite interesting to see how this question I posed could expand further to other forms. I was solely referring to integers, such as extracting out 42 as 42, not as 4 or 2, and your comment resolves my question. Thank you.