Does anyone have a Matlab code that would delete part of filenames in many directories?

I have a large number of waveform records that have been saved as individual files in many directories.

The directory structure is \RawData\Cell01, \RawData\Cell02, \RawData\Cell03...\RawData\Cell99

The data files in each of these subdirectories are named Cellxx_00000_xxxxx.xxx except the first file which is Cellxx_00000.dat

For example I have a subdirectory \RawData\Cell01 that may contain several thousand files named Cell01_00000_00001.dat, Cell01_00000_00002.dat; Cell01_00000_00003.dat… Cell01_00000_04300.dat.

I would like to:

1. remove the characters "_00000" from these files and rename them as: Cell01_00001.dat, Cell01_00002.dat, Cell01_00003.dat ... Cell01_04300.dat, and
2. copy the renamed files to subdirectory \RawData\Renamed\Cell01. The file Cell01_00000.dat must not be changed.

Does anyone have a code that could perform this task?

[Vahe Ghazikhanian]

Walter, shouldn't there be two loops in the code? One for the files in a given directory and one for subdirectories in the RawData?

[Walter Roberson]

No need for multiple loops. The Cell* in the directory specification causes it to look for all the .dat files in all subdirectories whose name begins with Cell. With R2016b and later, the folder field tells you which directory each file was actually found in. Then the first fileparts breaks the name into directory + filename, and the fileparts run on that gives you the parent directory and the directory itself, which will be the Cell* directory name. That directory name is then used relative to the "Renamed" directory, getting RawData/Renamed/Cell<whatever>/<filename.dat>

[Vahe Ghazikhanian]

Thanks for your explanation; however when I modified and ran the code it gives errors.

I have attached the modified code and the screenshots of the command window as well as the Workspace. The code does generate a new directory but stops short of converting filenames.

I have also included a screenshot of the first 27 files in one of the directories to be renamed. Please note that the first file of each directory does not contain the string which is being renamed.

I also tried to include a wild card character instead of the last "0" of '00000_' which is being deleted from the file name (i.e., I replaced the '00000_' with '0000(\w)_' so any number or character appearing in that spot would be replaced, but that did not work (error recognizing the character). Any idea how I could do this?

[Walter Roberson]

I have edited the code to fix the reference to filename

[Vahe Ghazikhanian]

Great...Thanks a lot. It is working now.

Answer 1

[Walter Roberson]

Given a filename

newfilename = regexprep(filename, '000000_', '');

The rest is just finding files and copyfile()

The below needs R2016b or later

projectdir = 'RawData';
outdir = fullfile(projectdir, 'Renamed');
if ~exist(outdir, 'dir'); mkdir(outdir); end

dinfo = dir( fullfile(projectdir, 'Cell*', '*.dat') );
fullnames = fullfile( {dinfo.folder}, {dinfo.name} );

for K = 1 : length(fullnames)
    thisfile = fullnames{K};
    [fulldir, basename, ext] = fileparts(thisfile);
    [parentdir, basedir] = fileparts(fulldir);
    newfilename = [regexprep(basename, '000000_', '', 'once'), ext];
    newdir = fullfile(outdir, basedir);
    if ~exist(newdir, 'dir'); mkdir(newdir); end
    newfilename = fullfile(newdir, newfilename);
    copyfile(thisfile, newfilename)
end

[Vahe Ghazikhanian]

Thanks for the code. What does the string 'first" in the regexprep(filename, '000000_', '', 'first'), do? I could not find this under the possible options for regexprep().

[Walter Roberson]

Sorry, should be 'once' not 'first'