A more efficient or compact way to sort strings that contain dates

I have strings that contain dates.

Those strings are in a "random" order, i.e. they are not ordered by following the dates, from 2024/03/01 to 2024/03/31 (i.e. from the 1st of March 2024 to the 31st of March 2024).

Is there a more efficient or compact way to sort the following strings containing dates?

% (1) input (strings containing dates, in a "random" order)
a(1,:) = '123_abc_01_202403020000_202403022359.txt';
a(2,:) = '123_abc_01_202403040000_202403042359.txt';
a(3,:) = '123_abc_01_202403030000_202403032359.txt';
a(4,:) = '123_abc_01_202403050000_202403052359.txt';
a(5,:) = '123_abc_01_202403010000_202403012359.txt';
a

a = 5x40 char array
    '123_abc_01_202403020000_202403022359.txt'
    '123_abc_01_202403040000_202403042359.txt'
    '123_abc_01_202403030000_202403032359.txt'
    '123_abc_01_202403050000_202403052359.txt'
    '123_abc_01_202403010000_202403012359.txt'

% (2) create substrings with ordered dates, that we can use to compare with the unordered strings of the input
for i = 1 : 31
    tmp = [];
    if i <=10
        tmp = sprintf('%02d',i);
    else
        tmp = sprintf('%0d',i);
    end
    b(i,:) = append('_202403',tmp);
end

% sort the unordered strings of the input, by following the substrings that have ordered dates
for i = 1 : 5
    for j = 1 : 31
        if contains(a(i,:),b(j,:))
           which_j(i) = j;
        end
    end
end
sorted_a = sort(a(which_j,:))

sorted_a = 5x40 char array
    '123_abc_01_202403010000_202403012359.txt'
    '123_abc_01_202403020000_202403022359.txt'
    '123_abc_01_202403030000_202403032359.txt'
    '123_abc_01_202403040000_202403042359.txt'
    '123_abc_01_202403050000_202403052359.txt'

Answer 1

[Stephen23]

a = [...
    '123_abc_01_202403020000_202403022359.txt';
    '123_abc_01_202403040000_202403042359.txt';
    '123_abc_01_202403030000_202403032359.txt';
    '123_abc_01_202403050000_202403052359.txt';
    '123_abc_01_202403010000_202403012359.txt'];
b = sortrows(a)

b = 5x40 char array
    '123_abc_01_202403010000_202403012359.txt'
    '123_abc_01_202403020000_202403022359.txt'
    '123_abc_01_202403030000_202403032359.txt'
    '123_abc_01_202403040000_202403042359.txt'
    '123_abc_01_202403050000_202403052359.txt'

Is there are particular reason why you are using a character matrix?

[Sim]

Thanks a lot! Oh gosh!! ...How is it possible that sortrows recognises dates inside the strings?? Is there any magic?

btw, there is no any particular reason why I use a character matrix.. I thought it was an easy way to bring here my example... :-)

[Stephen23]

"How is it possible that sortrows recognises dates inside the strings??"

It doesn't.

"Is there any magic?"

Not really: as long as the dates are written in order from largest unit to smallest unit (i.e. years, months, ... seconds) and use leading zeros to ensure a constant width then a basic character sort will return the dates in chronological order. If those conditions are not met then a character sort will not work, i.e. you will need to parse the dates first.

This is exactly why ISO 8601 specifies timestamps with units going from largest to smallest, and a fixed width:

https://en.wikipedia.org/wiki/ISO_8601

See also:

https://xkcd.com/1179/

[Sim]

Thanks a lot about all these info! Good to know as a good practice for coding and saving files :-)