uitable copy function allowing for numerical values of mixture of of numbers and chars.

Hi, I have a uitable and have a copy function as below. This works fine if the table is solely numbers

  function CopyTableData(app,src,event) 
           tbl = event.ContextObject;
                   % fieldnames(event)
                   % event.InteractionInformation
           s=tbl.Data;
           size_s = size(s); 
            
            str = ''; 
            for i=1:size_s(1) 
                for j=1:size_s(2)                    
                    str = sprintf ( '%s%f\t', str, s(i,j) );              
                end 
                str = sprintf ( '%s\n', str ); 
           end 
           clipboard('copy',str);
  end   
  

There are instances where my table will be a cell array and I have modified the function (for the str = sprintf ( '%s%f\t', str, s{i,j}) to:

function CopyTableData(app,src,event)

           tbl = event.ContextObject;
                   % fieldnames(event)
                   % event.InteractionInformation
           s=tbl.Data;
           size_s = size(s); 
            
            str = ''; 
            for i=1:size_s(1) 
                for j=1:size_s(2) 
                    try
                       str = sprintf ( '%s%f\t', str, s(i,j) ); 
                    catch
                        str = sprintf ( '%s%f\t', str, s{i,j} ); 
                    end    
                end 
                str = sprintf ( '%s\n', str ); 
           end 
           clipboard('copy',str);
        end   

However, when my data is as below (cellarray) in which the contents of all columns in the cellarray are numbers, but the very last column 'info" are charss

I get this when pasted into excel (notice the last column is incorrect and actually split into 2)

this is how I created my dummy data:

       uit=app.UITableSummary;
                     dx=9437;
                     z1=5795;
                     z2=5775.2
                     dz=z2-z1
                     tilt_mrads=-2.1084 
            A={dx,z1,z2,dz,tilt_mrads,'info'};
            d=uit.Data;                 % Get Current table data
            uit.Data = [d; A];        %Add new row to current data 

I kept on adding this and just manually replaced the 'info' value in the last column ('10', '12', etc..) and the numeric values in the 2nd from last column (-2.11, -1.59, etc..)

thanks for any help

[Stephen23]

https://www.mathworks.com/help/matlab/ref/writetable.html

[Jason]

Thanks Stephen, I'd still like to get the paste function working, sometimes I have other Matlab apps open and want to quickly paste it in rather than save the file and then load it again

[dpb]

Convert the char() column to string() or, if is always numeric, to double().

The problem is the char data is an array of characters which are copied to the clipboard literally and thus show up as the array of equivalent digits in Excel when pasted; MATLAB knows to display the native char() type as text, but Excel can't know that from the clipboard content.

[Jason]

There seems to be an erorr here with sprintf. I've replaced the last column of text with 'info:

str =

'9437.000000	5795.000000	5775.200000	-19.800000	-2.108400	'

>> str2='info'

str2 =

'info'

>>  str3 = sprintf ( '%s%f\t', str, str2) 

str3 =

'9437.000000	5795.000000	5775.200000	-19.800000	-2.108400	105.000000	nfo'

[Jason]

thanks, so I've tried this

    if ischar(s{i,j})==1
             s{i,j}=string( s{i,j})
    end  
    

As shown here:

     function CopyTableData(app,src,event) 
           tbl = event.ContextObject;
                   % fieldnames(event)
                   % event.InteractionInformation
           s=tbl.Data;
           size_s = size(s);
            
            str = ''; 
            for i=1:size_s(1) 
                for j=1:size_s(2) 
                    try
                       str = sprintf ( '%s%f\t', str, s(i,j) ); 
                    catch
                        if ischar(s{i,j})==1
                            s{i,j}=string( s{i,j})
                        end    
                        str = sprintf ( '%s%f\t', str, s{i,j}) 
                    end    
                end 
                str = sprintf ( '%s\n', str ); 
           end 
           clipboard('copy',str);
     end   
     

But now get

NaN's as the last column

[Stephen23]

"There seems to be an erorr here with sprintf."

No, there is no error with SPRINTF. The value 105 is simply the character value of the first element of the character vector that you are providing to SPRINTF. After converting that one element (i.e. character) it continues applying the format string to the remains of the input array/s.

+'i'

ans = 105

sprintf('%f','i')

ans = '105.000000'

Here is a simpler example of what you are doing:

sprintf('%s%f','abc','ijk')

ans = 'abc105.000000jk'

[dpb]

if ischar(s{i,j})==1
  s{i,j}=string( s{i,j});
end

converts every character in the char() arrays to a single string; thus creating an array of strings, not a string representation of the char() string content. See char documentation for how character strings work--they are a remnant of the very first implementation of MATLAB before the introduction even of cell arrays and cellstr(). The char() type is just an array of single characters that, because they are of class char, MATLAB knows to display the character representation of the content of the array as the ASCII text characters instead of as the numeric values that represent the characters, but internally it's just bytes and numbers that have a specific meaning. Carrying on from the example by @Stephen23 in a somewhat different vein,

info=['10'; '12']               % create a sample char() array mimics your info

info = 2x2 char array
    '10'
    '12'

info(1)                         % what is the first element???  Nota bene!!!!

ans = '1'

info(1,:)                       % what is the first string (row)???

ans = '10'

info(2)                         % the second element in linear ordeer

ans = '1'

for i=1:height(info)            % height(x) is more easier than size(x,1) now that supports arrays, too
  cellstr(info(1,:))            % convert each row to a cellstr
end

ans = 1x1 cell array
    {'10'}

ans = 1x1 cell array
    {'10'}

info=cellstr(info)              % but, it's built into cellstr() to convert rows for you

info = 2x1 cell array
    {'10'}
    {'12'}

info=string(info)               % and so does string()

info = 2x1 string array
    "10"
    "12"

So, your above code should be taking that last column alone and converting it.

Without an actual copy of the table, specific code explicitly for your case is harder to write, but I think you should have no need for sprintf() at all, you should be able to select and convert all data from the table by column with a given builtin conversion function based on the type of the data in each column.

The above illustrates the addressing that the char() array is just an array of individual characters, and to reinforce the char() array is just an ordinary array in MATLAB, note the second element in the array in linear addressing mode is also '1'; the array is, like all arrays in MATLAB, stored in column-major order that is, array elements are stored in memory sequentially going down each column, the reason that builtin MATLAB functions operate in a vector fashion by column.

Also note that owing that it is an array, that

info=['10'; '12';'100'] 
Error using vertcat
Dimensions of arrays being concatenated are not consistent. 

errors -- as it says you can't vertically catenate a vector of three things ('100') to an existing array of only two (of course, the prohibition is for any mismatch in sizes, not just longer). That's why there's the specific function just for the purpose--

info=strvcat('10','12','100')
info =
  3×3 char array
    '10 '
    '12 '
    '100'

but, in order to be able to add the longer string into the char() array, all the shorter strings in the array have been blank padded to the overall length. There's where the cellstr() and/or string() differ; they can hold array elements of differing lengths.

So, the question of what is actually in the array can be answered by noting

info=['10'; '12']               % get our original definition back...

info = 2x2 char array
    '10'
    '12'

cast(info,'double')             % convert internal to double to see what is

ans = 2x2

49 48 49 50

<mw-icon class=""></mw-icon>

<mw-icon class=""></mw-icon>

double(info)                    % can use shorthand to get some result(*)

ans = 2x2

49 48 49 50

<mw-icon class=""></mw-icon>

<mw-icon class=""></mw-icon>

char(ans)                       % and convert back to the char() array

ans = 2x2 char array
    '10'
    '12'

The last shows that sprintf has not lied at all, either, when you asked to output a char() string using a numeric format, it obliged and did the same thing as did double() except you get the string representation of the internal values held in memory in the char() array.

This shows the split personality and dual nature of character datatypes mentioned at the beginning; MATLAB knows because of the class that it needs to display the variable info as text so you, the user, see the representation you want, not the internal numeric values that represent that text. Also, numeric classes know how to convert the internal representation of the zeros and ones that make up the actual number in the bytes that hold a double value and spit out those characters to make a readable number.

(*) VBA and many other languages have a specific second paired function ASC() or something similar in name to match CHAR() . MATLAB chose to forego having a redundant function for the char data type.

Answer 1

[dpb]

I'd go at it a little bit differently, although I just discovered the clipboard only takes one line of text so one does have to build a newline delimited text string of the rows in the table.

If you'll use the MATLAB table for the input to the uitable, then it's much easier to manipulate; just return the uitable Data to a table variable and you're almost home.

info='10';
dx=9437;
z1=5795;
z2=5775.2;
dz=z2-z1;
tilt_mrads=-2.1084;
A={dx,z1,z2,dz,tilt_mrads,info};
A=repmat(A,3,1)

A = 3x6 cell array
    {[9437]}    {[5795]}    {[5.7752e+03]}    {[-19.8000]}    {[-2.1084]}    {'10'}
    {[9437]}    {[5795]}    {[5.7752e+03]}    {[-19.8000]}    {[-2.1084]}    {'10'}
    {[9437]}    {[5795]}    {[5.7752e+03]}    {[-19.8000]}    {[-2.1084]}    {'10'}

tA=cell2table(A)

tA = 3x6 table
     A1      A2       A3       A4        A5         A6  
    ____    ____    ______    _____    _______    ______

    9437    5795    5775.2    -19.8    -2.1084    {'10'}
    9437    5795    5775.2    -19.8    -2.1084    {'10'}
    9437    5795    5775.2    -19.8    -2.1084    {'10'}

app.hUIT=uitable(app.hUF,'Data',tA);    % populate your uitable
...                                     % manipulate at heart's content here

To retrieve and do the clipboard thing,

tA=app.hUIT.Data;                                           % retrieve last state of the UItable
tS=convertvars(tA,tA.Properties.VariableNames,'string');    % convert all to strings
C=table2array(tS);                                          % retrieve as a string array
tab=char(9);                                                % a tab charater constant
C=join(join(S,tab),newline);                                % join the columns w/ tab and the rows with newline
clipboard('copy',C);                                        % and put in clipboard

Can't build a uitable here, but I did a small example that mimics yours locally...the result looks like-

>> A={dx,z1,z2,dz,tilt_mrads,info};         % I copied your variables from above...
>> A=repmat(A,3,1)                          % and made a small table from them...
A =
  3×6 cell array
    {[9437.00]}    {[5795.00]}    {[5775.20]}    {[-19.80]}    {[-2.11]}    {'10'}
    {[9437.00]}    {[5795.00]}    {[5775.20]}    {[-19.80]}    {[-2.11]}    {'10'}
    {[9437.00]}    {[5795.00]}    {[5775.20]}    {[-19.80]}    {[-2.11]}    {'10'}
>> tA=array2table(A)                        % and converted to a MATLAB table
tA =
  3×6 table
        A1             A2             A3             A4           A5          A6  
    ___________    ___________    ___________    __________    _________    ______
    {[9437.00]}    {[5795.00]}    {[5775.20]}    {[-19.80]}    {[-2.11]}    {'10'}
    {[9437.00]}    {[5795.00]}    {[5775.20]}    {[-19.80]}    {[-2.11]}    {'10'}
    {[9437.00]}    {[5795.00]}    {[5775.20]}    {[-19.80]}    {[-2.11]}    {'10'}
>> tS=convertvars(tA,tA.Properties.VariableNames,'string')  % turn all into strings
tS =
  3×6 table
      A1        A2         A3         A4          A5         A6 
    ______    ______    ________    _______    _________    ____
    "9437"    "5795"    "5775.2"    "-19.8"    "-2.1084"    "10"
    "9437"    "5795"    "5775.2"    "-19.8"    "-2.1084"    "10"
    "9437"    "5795"    "5775.2"    "-19.8"    "-2.1084"    "10"
>> S=table2array(tS)
S = 
  3×6 string array
    "9437"    "5795"    "5775.2"    "-19.8"    "-2.1084"    "10"
    "9437"    "5795"    "5775.2"    "-19.8"    "-2.1084"    "10"
    "9437"    "5795"    "5775.2"    "-19.8"    "-2.1084"    "10"
>> tab=char(9);                                 % a tab charater constant
cb=join(join(S,tab),newline)                    % join the columns w/ tab and the rows with newline
cb = 
    "9437	5795	5775.2	-19.8	-2.1084	10
     9437	5795	5775.2	-19.8	-2.1084	10
     9437	5795	5775.2	-19.8	-2.1084	10"
>> clipboard('copy',cb)

Should be pretty close, anyways...

[Jason]

Fantastic, thanks dpb, this is a very different and elegant way to do this!

[dpb]

Glad to help, always nice to be able to use the features of MATLAB as MATrix LABoratory to keep things vectorized where possible. The newer string class, while redundant in many ways(*), has some newer features built into it that are handy--the implicit conversion of numerics to their string representation is one handy one I've found, indeed, if one doesn't need a very explicit format. That means one can also do things like:

vnames="Var"+[1:3]

vnames = 1x3 string array
    "Var1"    "Var2"    "Var3"

since the + operator is extended to mean catenation for strings and is enhanced when in an expression with a string class to do the implicit conversion.

Of course, one could always write

'A':'C'

ans = 'ABC'

or

char('A'+[0:2].')

ans = 3x1 char array
    'A'
    'B'
    'C'

because underneath, they're "just bytes" and so MATLAB array syntax still works.

(*) Try the {} indexing on both a cellstr and a string variable and you'll see that both are just char() arrays under the hood.