Search for a file using edit text with a part of the file name

Hello everyone,

Does anybody know how to find a file in a folder with a certain part of the file name? So whether someone's leaves out the extension or a part of it, it will still find the correct file? Now I have this code that allows the user to find a file in 'folder name', but the user would have to enter the entire filename instead of part of it. I have seen things like dir(.) but I cannot seem to make it work

Thanks for the help!

% --- Executes on button press in findfile1.
function findfile1_Callback(hObject, eventdata, handles)
% hObject    handle to findfile1 (see GCBO)
% eventdata  reserved - to be defined in a future version of MATLAB
% handles    structure with handles and user data (see GUIDATA)
foldername = handles.foldername;
files = dir(fullfile(foldername, '*.csv'));
files = {files.name}';
for k = 1: length(files);
    fname = fullfile(foldername,files{k});
end

filename = get(handles.edit1, 'String');
filename = char(filename); 
if 2 == exist(fullfile(foldername,filename), 'file') 
    %csvread(files);
    Filename = sprintf('file exists:\n%s', filename);
    uiwait (msgbox(Filename));
else 
    warningMessage = sprintf ('Warning: file does not exist:\n%s' , filename);
    uiwait(msgbox(warningMessage));
end

[Stephen23]

See also earlier question:

https://www.mathworks.com/matlabcentral/answers/394702-finding-file-in-a-before-selected-folder

[Stephen23]

"Does anybody know how to find a file in a folder with a certain part of the file name? So whether someone's leaves out the extension or a part of it, it will still find the correct file?"

Which part of the filename will the user provide? Which part of the file extension will they provide? If the user enters "cat", which file/s should this match:

1. catherine.txt
2. ca.txt
3. name.cat
4. cat.csv

??

[Debbie Oomen]

All the files have the same layout for the name: ID103915sec.csv or ID028315sec.csv. The ID, 15sec and .csv is always the same. So if the user would only input ID0283 or ID1039 or 1039 or 0283, then it would already be enough to specify the file

[Stephen23]

"then it would already be enough to specify the file"

Not always. If a user enters a number ending in 15 then this could be ambiguous, for example

      3415        % user enters this 
  ID123415sec.csv % filename match

[dpb]

The Win directory pattern-matching is pretty feeble; it wouldn't be perfect but you could pass

d=dir('ID*15sec.csv');

and retrieve all files with whatever the other four (or N) digits are. Unfortunately, while Win implements the '?' wildcard matching a single character, MATLAB doesn't pass it through to the builtin dir; you'd have to shell to the OS for

d=dir('ID????15sec.csv');

to be effective.

Answer 1

[Stephen23]

You could experiment with dir:

str = '0283';
fmt = sprintf('*%s*.csv',str);
S = dir(fmt);
filename = S.name

Or perhaps:

fmt = sprint('ID%s15sec.csv',str);

One flexible solution would be to let the user define as much of the filename as they want, find all matches, and then filter for only those that match the required filename pattern:

str = '0283';
S = dir(sprintf('*%s*',str));
N = {S(~[S.isdir]).name};
idx = ~cellfun('isempty',regexp(N,'^ID\d{4}15sec\.csv$'));
filename = N{idx}

Note:

• you will need to add the filepath.
• you will need to add checks to see if zero/multiple files match what the user gives. If you omit this it will only cause problems later!

As an alternative, you could simply use uigetfile.

[Debbie Oomen]

Where would I need to put the code? I assume it would have to be in this part?

filename = get(handles.edit1, 'String');
filename = char(filename);

[Stephen23]

@Debbie Oomen: that would seem to be about the right place. I would recommend that experiment with that code at the command line first, to ensure that it does what you want.

[Debbie Oomen]

Yes this works perfectly! You are a genius! Only problem now is that I have no clue how to implement that code into the GUI. The handles.edit1 is the edittextbox. Would I need to implement the code where 'String' now is?

filename = get(handles.edit1, 'String');
filename = char(filename);

[Stephen23]

@Debbie Oomen: If you picked the regexp-based code you can do this:

str = char(get(handles.edit1, 'String'));
... the code that you want to use
filename = N{idx};

Remember that you will need to ensure that exactly one file has been matched (hint: use numel or nnz).

[Debbie Oomen]

% --- Executes on button press in findfile1.
function findfile1_Callback(hObject, eventdata, handles)
% hObject    handle to findfile1 (see GCBO)
% eventdata  reserved - to be defined in a future version of MATLAB
% handles    structure with handles and user data (see GUIDATA)
foldername = handles.foldername;

str = char(get(handles.edit1, 'String'));
fmt = sprintf('*%s*',str);
S = dir(fmt);
N = {S.name};
idx = ~cellfun('isempty',regexp(N,'^ID\d{4}15sec.csv$'));
filename = S.name;

if 2 == exist(fullfile(foldername,filename), 'file') 
    %csvread(files);
    Filename = sprintf('file exists:\n%s', filename);
    uiwait (msgbox(Filename));
else 
    warningMessage = sprintf ('Warning: file does not exist:\n%s' , filename);
    uiwait(msgbox(warningMessage));
end

This is the code now, but it says that : Insufficient number of outputs from right hand side of equal sign to satisfy assignment.

Error in whatever>findfile1_Callback (line 181) filename = S.name;

I believe this has something to do with number or nnz but I do not know what you mean by this..

[Stephen23]

@Debbie Oomen: if you are using the solution with regexp then you will need to use the index to select the filename (did you notice that you left idx unused? And that the MATLAB editor warned you about this too?), so instead of this:

filename = S.name;

you need:

filename = N{idx};

But first you should use nnz(idx) to check how many filenames matched: it is up to you to decide what to do if there are zero/multiple files.

[Debbie Oomen]

Now I tried this, I am not sure how to do the nnz{idx}. It still gives the same error: unable to perform assignment with 0 elements on the right-hand side. And also, is it possible to input more than or less than the 4 numbers? For example if a file would be: ID32115sec

str = char(get(handles.edit1, 'String'));
fmt = sprintf('*%s*',str);
S = dir(fmt);
N = {S.name};
idx = ~cellfun('isempty',regexp(N,'^ID\d{4}15sec.csv$'));
nnz{idx};
filename = N{idx};

[Stephen23]

@Debbie Oomen: just try exactly the code that I gave you: nnz(idx). Note that I used parentheses, not curly braces. The value that it gives you is how many files matched what the user gave. It is your GUI, so you need to decide how to deal with the cases when it matches zero/multiple files.

"is it possible to input more than or less than the 4 numbers? For example if a file would be: ID32115sec"

I wrote in my answer that the second method I gave you (using regexp) will "...let the user define as much of the filename as they want...". Try it, and see what happens.

Bonus: to be extra robust you could ensure that any folders are excluded:

N = {S(~[S.isdir]).name};

[Debbie Oomen]

So I tried the following code on the command line once again and it seems to be working. However, in the GUI code it keeps on giving me the error: Insufficient number of outputs from right hand side of equal sign to satisfy assignment at filename = S.name

This is the code now:

% --- Executes on button press in findfile1.
function findfile1_Callback(hObject, eventdata, handles)
% hObject    handle to findfile1 (see GCBO)
% eventdata  reserved - to be defined in a future version of MATLAB
% handles    structure with handles and user data (see GUIDATA)
foldername = handles.foldername;

str = char(get(handles.edit1, 'String'));
fmt = sprintf('*%s*',str);
S = dir(fmt);
N = {S(~[S.isdir]).name};
idx = ~cellfun('isempty',regexp(N,'^ID\d{4}15sec.csv$'));
nnz(idx);
filename = S.name;

if 2 == exist(fullfile(foldername,filename), 'file') 
    %csvread(files);
    Filename = sprintf('file exists:\n%s', filename);
    uiwait (msgbox(Filename));
else 
    warningMessage = sprintf ('Warning: file does not exist:\n%s' , filename);
    uiwait(msgbox(warningMessage));
end

I don't understand why it keeps giving me that error while it does work on the command line..

[Stephen23]

The regexp-based code does not use filename = S.name;. The correct code is:

filename = N{idx};

Note that in your code you currently have this line, not allocated to anything:

nnz(idx);

Because its value (which is just a number) it not allocated to any variable its is simply discarded. If you want to check for how many files have matched the user input then you will need to do something with this value, not just calculate it and then throw it away.

[Debbie Oomen]

Sorry I didn't mention it before but I have also tried that and then it gives me the following error: Unable to perform assignment with 0 elements on right hand side. I think I see the problem, I tried it again on the command line and looked at the values given for the idx command and the output gives me a 0. I believe this has something to do with ID\d{4}. This means that the user HAS to enter 4 digits. Is it possible to let this be a variable amount of digits? Also, now the GUI does not give an error if the file does not exist. It just skips the everything after else if the N{idx} = 0

idx = ~cellfun('isempty',regexp(N,'^ID\d{4}15sec.csv$'));

[Stephen23]

"This means that the user HAS to enter 4 digits"

Not at all. It actually means that the filenames in the directory must have four digits, which is what you explained is the filename format. That regular expression is not used to parse what the user enters: the user could enter one digit, four digits, or the complete filename: this input is used to match filenames in the directory, it is these filenames that are then parsed by the regular expression. Here is an explanation:

str = '0283';                    % user input: can be *ANY* char vector.
S = dir(sprintf('*%s*',str));    % find filename/s that match user input.
N = {S(~[S.isdir]).name};        % put filenames in a cell array.
idx = ~cellfun('isempty',regexp(N,'^ID\d{4}15sec\.csv$')); % indices of filenames which match regular expression.
assert(nnz(idx)==1,'Oh no!')     % ensure 1 filename match.
filename = N{idx}                % get that filename.

I just tried it myself again. Given two files named ID123415sec.csv and ID678915sec.csv:

>> str = '23'; % two digits matching *one* filename.
>> S = dir(sprintf('*%s*',str));
>> N = {S(~[S.isdir]).name};
>> idx = ~cellfun('isempty',regexp(N,'^ID\d{4}15sec\.csv$'));
>> assert(nnz(idx)==1,'Oh no!')
>> filename = N{idx}
filename = ID123415sec.csv

So this correctly matched the expected file. Then I tested for if the user enters something with multiple matches:

>> str = '15'; % two digits matching *both* filenames.
>> S = dir(sprintf('*%s*',str));
>> N = {S(~[S.isdir]).name};
>> idx = ~cellfun('isempty',regexp(N,'^ID\d{4}15sec\.csv$'));
>> assert(nnz(idx)==1,'Oh no!')
error: Oh no!

This correctly threw an error because multiple files matched the user input.

PS: Note that I made a very small improvement to the regular expression.

[Debbie Oomen]

Yes, this does seem to work! Thank you! Now, would it be possible to let this code work for filenames with more than just 4 numbers between ID .... and 15 sec? I tried leaving the curly brackets empty but that gave me an error in foldername?

[Stephen23]

To match "filenames with more than just 4 numbers...":

   '^ID\d{4,}15sec\.csv$'

Or if you require one or more digits in that location:

'^ID\d+15sec\.csv$'

See: https://www.mathworks.com/help/matlab/matlab_prog/regular-expressions.html#f0-43073

[Debbie Oomen]

@Stephen thank you so very much! I am truly grateful!

[Debbie Oomen]

@Stephen, would you mind looking at my other question I just posted? Would you have any idea how to do this?