getting "Index exceeds matrix dimensions." Can anyone explain?

function [ coordinates ] = readSTL1( filename,k)

% reads ASCII STL file and gives coordinates of vertices.

%filename-name of file(test.txt),k- number of rows in cell.

fid=fopen(filename);

C=textscan(fid,'%s');

%reads file and generates cell.

m = 11;

i = 1;

coordinates = zeros(4455,1);

while(m < (k-3))

    j = 1;
      while (j < 4)    
       l = 1;
               while(l<4)

                  coordinates(i) = C(m);
                  l = l+1;  % makes sure loop runs thrice.
                  m = m+1;  % access corresponding row from cell'C'.
                  i = i+1;  % row number in output matrix.

               end 

       m = m+1;
       j = j+1;
      end

    m = m+10;

end

end

[siddhesh rane]

error message-

Index exceeds matrix dimensions.

Error in readSTL1 (line 15) coordinates(i) = C(m);

[DGM]

Besides the basic handling of cell arrays and cellchar data, this magic number:

m = 11;

... presumes that the file's header text contains exactly one whitespace-delineated blob. It will result in completely useless jumbled garbage if the header text is empty, or contains multiple words or numbers.

Likewise, the method of blindly extracting a rigidly ordered pattern of blobs from the text will produce similar jumbled garbage if there's any variation in the file structure, such as in a multiobject ASCII STL.

Also, close the file.

At the time, there were multiple ASCII STL decoders on the FEX already.

Answer 1

[Walter Roberson]

Are you sure that is your code? textscan() returns a cell array, so your C is a cell array and thus C(m) is a 1x1 cell array (which has a string inside it), but you cannot assign a cell array into a numeric array (you initialized coordinates as zeros() which is numeric).

Even if you were to change it to C{m} then C{m} would be a string, most likely of multiple characters, and assigning multiple characters into a single numeric location is not going to work.

When you use textscan with a '%s' format, the input will be divided up into (whitespace delimited) strings with line boundaries treated the same as whitespace, so C will contain one location per string. e.g.,

fizz 19 39
bar

would return

{'fizz'; '19'; '39'; 'bar'}

You assume in your code that m starts from 11, so your code that accesses C(m) is going to fail of there were not at least 11 strings in the file.

[siddhesh rane]

Yes..the cell has both numerical and text datas..i wrote this code to separate numerical data from text..numerical data repeats itself in certain sequence.so i have to copy data from those rows and create coordinates matrix .rowise sequence of occurance of data is {[(11,12,13)(15,16,17)(19,20,21)][(32,33,34)(36,37,38)(40,41,42)]}..for accessing thows rows i used three while loops..And as my cell will have atleast 11 columns i guess that wont be a problem.

[Walter Roberson]

At the command line, give the command

dbstop if error

then run the program. When it stops at the error you indicated, examine size(C) and the value of m .

Answer 2

[Iain]

If you know the indices of where the data is, you can instead do:

Indices = [ 11 12 13 15 16 17 19 20 21 32 33 34 36 37 38 40 41 42];

for i = 1:numel(Indices)

coordinates(i) = C{Indices(i)};

end

The error occurs because your "m" is larger than the number of elements in C.