read file in folder with multiple files with similar names

file_name = dir('*.rpt*'); 

------------------

some where in here would like code that picks the file with latest data

file name 'data_20220804_0211.rpt'

-----------------------------

file = fopen(file_name.name, 'r');
L = textscan(fileID,'%s %s %s %s %s %s %s %s %s %s %s %s %s %s %s %s');
report_size = size(L);

Answer 1

[Walter Roberson]

In the restricted case that the files are being generated dynamically, then

dinfo = dir('*.rpt*');
[~,newest_idx] = max(datetime({dinfo.date}))
newest_filename = dinfo(newest_idx).name;

What this is doing is looking at the modification date information recorded for each of the files, and picking out the newest one.

In the case where the files might have been sitting around for a while, or might have been transferred between directories, then the modification date becomes less reliable, and you have to start processing the date information

dinfo = dir('*.rpt*');
rawnames = {dinfo.name};
datenames = regexprep(rawnames, '^.*_(\d{8})_(\d{4})\..*$', '$1 $2');

%any name that is not exactly 13 characters is a name that did not follow
%the pattern and we cannot parse.
mask = cellfun(@length, datenames) == 13;
rawnames = rawnames(mask);
datenames = datenames(mask);
filetimes = datetime(datenames);
[~, maxidx] = max(filetimes);
newest_filename = rawnames{maxidx};

[Stephen23]

Simpler and more robust:

[~,newest_idx] = max([dinfo.datenum])

[Walter Roberson]

Ah of course. This would be the modification date still, which unfortunately tends to get reset as files are moved.

[Ihaveaquest]

works great! thank you - just for my own learning, what is the diference between

[~,newest_idx] = max(datetime({dinfo.date})) --> [~,newest_idx] = max([dinfo.datenum])

Thank you

[Walter Roberson]

It turns out that in POSIX, file modification times are recorded as time_t which is (full) seconds since a particular time. So resolution down to one second is all that you can get from a file timestamp. Both datetime() and serial date number are able to resolve down to 1 second without problem.

It is potentially possible that the datenum version might vary by about 10 microseconds from the datetime version.

It is possible that the two might differ in interpretation of leap seconds.

It is possible that the two might differ about timezone when formatted.

It is possible that the two might differ about daylight time when formatted.

[Ihaveaquest]

Thank you!