How can I access two subset of a string in a table field (to convert into two datetime fields) ?

Hello,

I can not manage to find a vectorized way to deal with a long table string to datetime conversion.

I explain:

I have a 4496835x3 table. My second field (Dates) has been imported into a field containing cells.

This field is composed of uniform strings containing two dates (beggining-ending). Example: '2016.11.07 21:00-2016.11.07 22:00' (a string of 33 characters)

I want to create two other fields in datetime format (Let's say Date1 and Date2).

I can already manage it by looping on each row:

InputFmt='yyyy.MM.dd HH:mm';
        for j=1:height(data)

            data.Date1(j)=datetime(data.Dates{j}(1:16),'InputFormat',InputFmt);
            data.Date2(j)=datetime(data.Dates{j}(18:end),'InputFormat',InputFmt);
        end

which is very long to compute.

I also tried a kind of pre-allocation, using:

        data.Date1(:,1)=DefaultDate;
        data.Date2(:,1)=DefaultDate;

before the loop, where DefaultDate is an odd date in datetime format. But it is still heavy to run...

That's why I am wondering of a way to vectorize this datetime conversion, but I can not find a way to access each 'sub-string' independently. Indeed, I try different combinations of braces/curly braces, indexing, but nothing goes as I wish. I have to admit that struct and cell indexing are still things that I don't catch well either.

Example of codes that I tried to access the "Date1-Part":

 test.Dates{:}(1:16)
 test.Dates{:,1}(1:16)
 {data.Dates(1:16)}
 [data.Dates(1:16)]

and probably some others.

I hope it is clear enough, and thanks if anyone can help.

Answer 1

[Walter Roberson]

splitzees = regexp(data.Dates, '-', 'split');
as_cell = vertcat(splitzees{:});
datetime(as_cell, 'InputFormat', InputFmt)

That would give you an N x 2 array of datetime objects

[Water_RC]

Hello Walter,

Thanks for your answer. I was not aware of this useful function.