MATLAB: How to avoid empty brackets using textscan

empty valueMATLABtextscan

I am having trouble with the textscan function. I have a text file that contains only the information "Study Date: 8/30/2013 10:07:30 PM" (without the quotes). I am trying to use the textscan function to get the date and time information and put each number into a separate cell in an array. When I do this using the code below, for some reason it puts empty square brackets into each cell instead of the numbers. Please advise…
I am using this code to open my file and obtain the mm dd yy hh mm ss variables:
    %%Startup clear data
    clear all;
    close all;
    clc;
    %%Find "event file" and get basic start time.
    % Ask user to locate "date file"
    [c,pathc] ...
    =   uigetfile({'C:\Users\HP\OneDrive\Documents\1 School Stuff\*.txt'} ...
    ,   'Select the Sandman Event List:');
    % Send file info and open the file
    file=[pathc c];
    fid=fopen(file);
    % Read basic date and time data about the file
    % read the time for start of study 
    StudyDate ...
    =   textscan(file,'%*s %*s %d %d %d %d %d %d %*s',1,'delimiter',{':','/'});
    fclose(fid);

Best Answer

The documentation on Whitespace and Delimiter together with space in Formatspec is difficult to interpret (IMO). However, specifying Delimiter overtakes the default value, i.e whitepace. In this example space is needed as a delimiter.
Documentation says:
    textscan interprets repeated delimiter characters as separate delimiters,
    and returns an empty value to the output cell.
However, in my example below "one or more" spaces are obviously interpreted as one delimiter. That's probably because space is included in the values of both Whitespace and Delimiter. The documentation says little (read nothing) about the effect of different values of Whitespace. (And I don't know C.)
Try this example
    str = 'Study Date:  8/30/2013 10:07:30 PM';
    cac = textscan( str,'%*s%*s%d%d%d%d%d%d%*s', 'Delimiter',{':','/',' '} );
    cac
it returns
    cac = 
        [8]    [30]    [2013]    [10]    [7]    [30]
&nbsp
In response to comment
  • I replaced the space in the filename by underscore. Spaces in filenames has caused too many "Cannot open file" over the years.
  • There is a tab after "Date:". This file has two columns, name and value, separated by tab.
    fid = fopen( 'study_date.txt' );
    cac = textscan( fid, '%*s%*s%d%d%d%d%d%d%*s'  ...
                 , 'Delimiter',{':','/',' ','\t'} );
    fclose( fid );
    cac
returns
    cac = 
        [0]    [8]    [30]    [2013]    [10]    [7]
ERROR I'll be back
Of course! &nbsp ...Date:\t has two delimiters in a row. This is too tricky to my taste.
    fid = fopen( 'study_date.txt' );
    cac = textscan( fid,'%*s%*s%d%d%d%d%d%d%*s'     ...
                  , 'Delimiter', {':','/',' ','\t'} ...
                  , 'MultipleDelimsAsOne',  true    );
    fclose( fid );
    cac
returns
    cac = 
        [8]    [30]    [2013]    [10]    [7]    [30]
Conclusion
I have ended up in an overly complicated solution to the literal question. The reason, I think, is that I was concentrating on effects of various values Whitespace and Delimiter and made stepwise fixes.
DONE
&nbsp
In response to the comment by Star Strider
IMO: this is a much better way to read the file
    fid = fopen( 'study_date.txt' );
    cac = textscan( fid,'%s%s', 'Delimiter',{'\t'} );
    fclose( fid );
    vec = datevec( cac{2}, 'mm/dd/yyyy HH:MM:SS PM' )
returns
    vec =
            2013           8          30          22           7          30
    >>
&nbsp
In R2014b textscan( ... 'formatSpec' ... ) has a new feature
    %{fmt}D &nbsp Read a string in the same way as %q above, and then 
    convert to a datetime value. fmt describes the format of the input 
    string. The fmt input is a string of letter identifiers that is a 
    valid value for the Format property of a datetime.
Related Question