Using Matlab regexp on a cell array to return specific filenames

Question

I have a folder in which there are many files and I want to create a matrix that holds filenames with a specific pattern. For example: The folder contains files with names starting with a subject number (e.g. 03T1A.xxx.nii, 03T1A.yyy.nii) as well as filenames with specific patterns in the middle (e.g. 03T1A.c100.nii, 03T1A.c200.nii, 03T1A.c300.nii). In this specific case I am looking to extract all the filenames with the pattern c1 and c2 in the middle (e.g. 03T1A.c100.nii and 03T1A.c200.nii but not 03T1A.c300.nii).

To this point I have used the following code to create a pattern matching variable in 'pattern' which I would like to apply to the cell array of filenames I have extracted into the variable 'all_files' via the dir call.

func_path = char(strcat(input_dir, '/', subs(files), '/Func'));
pattern = 'c[12]*.nii'
all_files = dir(func_path); 
all_files = {all_files.name};

I'd like to use (read. practice) regexp and doing it with string input seems easy but I am 100% stumped as to how to do it with cell input. I started trying to do something like this:

files = all_files(cellfun(@(x)regexp(x, pattern));

But it doesn't work, obviously. Could someone help me figure out what to do here if my ultimate goal is to get a matrix output with just the relevant filenames? I've been searching MATLAB answers and other Stack Overflow posts but part of my problem is I don't understand what's happening in their code snippets. I took the above line (or at the least the beginning of it) from another post but I don't know what, for example, 'x' is (an output variable?) or what's going on in the larger command such as

fin = cellfun(@(x)regexp(x, '\.', 'split'), res, 'UniformOutput', false)

Which I found in another thread. So basically, can someone help me figure out a command that will work while explaining it to me?

regexp works natively on cell arrays, there is no need to use cellfun. See also: Regex 101 as a playground for building the expression. — sco1
– sco1, Commented Jan 27, 2017 at 14:30
I had originally tried to use the command: x = regexp(all_files, pattern, 'match') But it returns an empty cell array of the same size as 'all_files.' — chainhomelow
– chainhomelow, Commented Jan 27, 2017 at 14:37
Holy crap, you were absolutely right. If I had files c1004.EXAMPLE.nii and c2004.EXAMPLE.nii, for some reason the pattern 'c[12]*.nii' returns nothing but the pattern 'c[12]' returns both correctly. Is there any reason why this is? — chainhomelow
– chainhomelow, Commented Jan 27, 2017 at 14:43
@chainhomelow Yes, because c[12]*.nii will only match a "c" followed by only 1's and 2's before the .nii. You would need c[12].*\.nii so that you can match everything else in the string between c1 or c2 and the extension — Suever
– Suever, Commented Jan 27, 2017 at 14:53

Suever · Accepted Answer · 2017-01-27 15:08:56Z

2

A couple of recommendations for doing this sort of thing

Do not use strcat and '/' characters to construct file paths. strcat trims whitespace from all inputs prior to concatenation (filenames may have actual leading or trailing whitespace) and also rather than hard-coding a file path separator such as '/' , use filesep or better yet use fullfile to construct the path to ensure that it will work on various platforms without problems.
```
func_path = fullfile(input_dir, subs(files), 'Func');
```

regexp works directly on cell arrays therefore you can simply do:

all_files = dir(func_path); 

% Search for the pattern in all filenames
matches = regexp({all_files.name}, pattern);

% Get the filenames of those that matched
all_files = {all_files(~cellfun('isempty', matches)).name};

Your pattern isn't matching any files because it currently would match only strings that contain a "c" with only zero or more 1's or 2's before the file extension. Instead, you'll want to use .* to match anything between the "c1" or "c2" and the filename. Also you'll want to not use a * after [12] since that will actually match c3 since that has zero 1's or 2's. Also you'll want to escape the . in .nii so that it's not treated like a wildcard. For your pattern I would use something like
```
pattern = 'c[12].*\.nii';
```
If you really don't want to work with regular expressions, you could avoid all of this by simply using wildcards in your dir call
```
c1_files = dir(fullfile(func_path, '*c1*.nii'));
c2_files = dir(fullfile(func_path, '*c2*.nii'));
```

answered Jan 27, 2017 at 15:08

Suever

65.6k14 gold badges91 silver badges104 bronze badges

Sign up to request clarification or add additional context in comments.

1 Comment

chainhomelow Over a year ago

Thank you very much, this does the trick for sure. I was looking at the regexp usage and yeah the wildcard thing threw me off. Apparently you can also use 'c[12]\S*nii' as well but the syntax change threw me. The strcat comment was very helpful as well, and the 'isempty' line works for completeness to give me exactly the new array I need. Really appreciate it.

Collectives™ on Stack Overflow

Using Matlab regexp on a cell array to return specific filenames

1 Answer 1

1 Comment

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

1 Answer 1

1 Comment

Your Answer

Sign up or log in

Post as a guest

Related