Casting cell array of cells to something workable

Hi there, I have an Nx1 cell (named 'intervals') which contains cell arrays. For example:

K>> intervals(1:5,1)
ans =
  5×1 cell array

      {'2018-03-2610:25:51 PM'}
      {'2018-03-2610:25:51 PM'}
      {'2018-03-2610:25:51 PM'}
      {'2018-03-2610:25:52 PM'}
      {'2018-03-2610:25:52 PM'}

The cell arrays are supposed to contain dates but as you can see they are improperly formatted - I need to add a space between the day and hour to be able to convert this into something useful. The cell I have is very very long (~1500000x1), so I don't want to do this with a for loop. Is there a smart way to add a space in each cell array at the correct spot? Another option is to cast this cell array of cells into a matrix of strings, but every variation of cell2mat I have tried gives me a 1498393x21 char.

My desired output is an array of datetimes

    2018-03-26 10:25:51 PM
    2018-03-26 10:25:51 PM
    2018-03-26 10:25:51 PM
    2018-03-26 10:25:52 PM
    2018-03-26 10:25:52 PM

My current code will take forever, because the actual text file I'm working with is so huge:

function dates = timestamp_correction()
filename = 'intervals.txt';
fileID = fopen(filename);
intervals = textscan(fileID, '%s', 'delimiter', '\t');
intervals = intervals{1,1};

for i=1:numel(intervals)
    dates(i) = datetime(intervals{i,1}, 'InputFormat', 'yyyy-MM-ddhh:mm:ss a');
end
dates = dates';

end

trying

datetime(intervals{:}, 'InputFormat', 'yyyy-MM-ddhh:mm:ss a');

throws an error, it wants it to be an array first but I can't get that to work :(

I've attached a much smaller text file which is a truncated version of the bigger one I am working with.

Thanks!

[per isakson]

Try

char( intervals(1:5,1) )

[supernoob]

That gives me a 1498393x21 char, which isn't what I want. If it gave an array of strings that would be fine, but I haven't been able to get it to cast to array properly.

[Walter Roberson]

"I need to account for the fact that the day might have one or two digits"

So for example, 2018-03-210:25:51 PM would be possible for 2nd of the month at 10:25:51 PM ?

If so then what happens for times between 1 and 9 ? Is the leading 0 omitted for them, leading to, for example, 2018-03-261:25:51 PM for 1:25:51 PM on the 26th?

If so, then can both end up with leading zeros removed, such as 2018-03-21:25:51 PM for 1:25:51 PM on the 2nd ?

If both can have leading zeros removed then you cannot tell whether 2018-03-211:25:51 PM is 1:25:51 PM on the 21st, or is 11:25:51 PM on the 2nd.

[supernoob]

You know what, you're right. And it turns out the leading zero is not omitted, so luckily this is not a concern. I've edited my question, thanks for pointing this out.

Answer 1

[Walter Roberson]

reformatted_intervals = cellfun(@(S) [S(1:10) ' ' S(11:end)], intervals, 'uniform', 0)

However since your purpose is to convert to another form, you can instead skip the reformatting and use

intervals_dt = datetime(intervals, 'InputFormat', 'yyyy-MM-ddhh:mm:ss a');

[supernoob]

Thanks so much Walter. I'm embarassed I couldn't come up with this on my own. I kept trying to convert the cell array first before calling datetime on it, I should have taken the most straightforward approach! That cellfun but will prove to be handy for other work too. Cheers!