Why do the DIR and LS functions in MATLAB not sort numerical filenames in numerical order?
Show older comments
The DIR and LS functions in MATLAB do not sort numerical filenames in numerical order.
For example, I have filenames of the form:
1-1-1.tif
1-1-2.tif
1-1-3.tif
. . .
1-1-9.tif
1-1-10.tif
1-1-11.tif
However, when I use DIR or LS to list the files using:
files=dir('*.tif')
The results stored in "dir.name" are:
1-1-1.tif
1-1-10.tif
1-1-11.tif
1-1-2.tif
1-1-3.tif
1-1-9.tif
I would like to order these numerically.
Accepted Answer
MathWorks Support Team
on 27 Jun 2009
The DIR or LS functions in MATLAB sort the strings in ASCII dictionary order. This can cause a problem if you want to sort numbered files which do not have leading zeros.
Currently, to work around this problem, you can use the text manipulation and sorting routines in MATLAB.
Here is an example:
files=dir('*.tif');
numfiles = size(files,1); % Find number of files
numdelim = 3; % Number of delimiters
delims = ['-' '.']; % Delimiters used
% Create a matrix of the file list index and the delimiters
filenums=[ [1:numfiles]' zeros(numfiles,3) ];
for i=1:numfiles % Cycles through list of files
rem=files(i).name;
for j=1:numdelim % Cycles through the filename delimiters
[token,rem] = strtok(rem,delims);
filenums(i,j+1) = str2num(token);
end
end
% Sort the matrix by rows
filenums = sortrows(filenums,[2:numdelim+1]);
% Show the filenames in the new order
files(filenums(:,1)).name
% Create a cell array of the filenames in new order
for i=1:numfiles
sortedfiles{i,1} = files(filenums(i,1)).name;
end
4 Comments
"The DIR or LS functions in MATLAB sort the strings in ASCII dictionary order."
Is this order documented anywhere? Neither the dir nor the ls documentation include this information.
Until August 2015 the dir documentation stated "Results appear in the order returned by the operating system":
This means when this answer was written in 2009 either it or the documentation was incorrect. Or both.
"When verifying with a few releases of MATLAB the result of "dir" function is indeed ASCII ordered."
That does not answer my question. I am very curious what possible method you could have used that relies on "verifying with a few releases" that would distinguish between these two cases:
- MATLAB sorting the filenames (which currently is not documented anywhere),
- MATLAB returning the filenames in the order provided by the OS (which was previously documented).
As far as I can tell, simple testing at the MATLAB user interface level will not distinguish between them.
Would you care to give a few details of this "verifying" method?
"The DIR or LS functions in MATLAB sort the strings in ASCII dictionary order."
Easily disproved by actually running LS and DIR on a few very basic filenames:
ls *.txt
dir *.txt
Note that those are different orders.
Different orders means it is not possible for both of them to be "in ASCII dictionary order" as this answer (most likely incorrectly) states that MATLAB sorts them into. One of them does happen to be in ASCII order, but certainly not both of them (unless MST has invented a completely new, undocumented, variable ASCII order).
The R2021a documentation for dir() and ls() do not document any ordering.
ASCII dictionary order is not likely, as the supported operating systems use various forms of unicode.
I just did a test on my Mac;
!touch q R
ls
On my Mac, R was before q -- which is consistent with sorting by codepoint, but is not consistent with ascii dictionary order.
More Answers (1)
Simple solution: download my FEX submission natsortfiles, which sorts filenames into alphanumeric order:
S = dir('*.txt');
S.name
S = natsortfiles(S); % alphanumeric sort by filename
S.name
Categories
Find more on Language Support in Help Center and File Exchange
Products
on 27 Jun 2009
Community Treasure Hunt
Find the treasures in MATLAB Central and discover how the community can help you!
Start Hunting!
