How to use eval(sprintf) without padding?

Dear All,

I inherited a code snippet as part of a larger routine, so I have to use it with minor modifications. With it, a text file containing 3D coordinates of different points in space, points organized in clusters, is read into a corresponding .mat double file containing those 3D coordinates. Here is an example of the text file, each of its lines contains point cluster name, point name (same as cluster) and index, and the 3D (disregard the 4th) coordinates:

MO     MO9  149.529  157.814  215.614  0
MO     MO8  139.855  157.87  215.594  0
MO     MO7  133.018  155.614  210.281  0
MO     MO6  130.482  151.464  201.607  0
MO     MO5  130.692  147.173  192.942  0
MO     MO4  131.176  142.157  183.961  0
MO     MO3  131.3  136.15  176.05  0

I show only a cluster, but the real files have several. Now, the code snipped generates a matrix of 3D coordinates in order of the point index value, however, it adds 0 coordinates for two inexistent MO1 and MO2 points (the points have not originally been numbered starting with 1, long story why). Here is the .mat contents:

0  0  0
0  0  0
131.300000000000  136.150000000000  176.050000000000
131.176000000000  142.157000000000  183.961000000000
130.692000000000  147.173000000000  192.942000000000
130.482000000000  151.464000000000  201.607000000000
133.018000000000  155.614000000000  210.281000000000
139.855000000000  157.870000000000  215.594000000000
149.529000000000  157.814000000000  215.614000000000

Question: How do I write into the .mat file just the existent points MO3-8, without imaginary points? Here is the code:

fid = fopen(['path to file/file.txt'],'r');
line = fgetl(fid);
PointNames = {};
while line ~= -1
  c = textscan(line, '%s %s %f %f %f %f');
  pointName = c{1}{1};
  pos = [c{3:5}];
  pointIdx = str2double(c{2}{1}(length(pointName)+1:end));
  PointNames{size(PointNames,2)+1} = pointName;
  eval(sprintf('%s(%i,:) = [%f %f %f];', pointName, pointIdx,pos));
  line = fgetl(fid);
end
fclose(fid);

I suspect it has to do with '%s(%i,:)', but here I get stuck.

Many thanks, Octavian.

[Image Analyst]

Please attach file.txt so we can try your code. And for next time read this: http://www.mathworks.com/matlabcentral/answers/13205-tutorial-how-to-format-your-question-with-markup (I fixed your formatting this time for you)

[Octavian]

Well taken, here is the file, thank you, Octavian

Answer 1

[Image Analyst]

http://matlab.wikia.com/wiki/FAQ#How_can_I_create_variables_A1.2C_A2.2C....2CA10_in_a_loop.3F

[Octavian]

Thank you,

Again, the issue is not only not to add inexistent points, but to sort the coordinates in the order of increasing point index values, MO3, MO4... MO9, whereas these are scrambled in the original txt file, see file attached in my comment above, thank you, Octavian

[Image Analyst]

Try this:

folder = pwd; % Current folder.
fullFileName = fullfile(folder, 'file.txt');
% Check if file exists.
if ~exist(fullFileName, 'file')
  % File doesn't exist -- didn't find it there.  Check the search path for it.
  fullFileName = baseFileName; % No path this time.
  if ~exist(fullFileName, 'file')
    % Still didn't find it.  Alert user.
    errorMessage = sprintf('Error: %s does not exist in the search path folders.', fullFileName);
    uiwait(warndlg(errorMessage));
    return;
  end
end
% File exists, so open it.
fid = fopen(fullFileName,'r');
thisLine = fgetl(fid);
PointNames = {};
lineNumber = 1;
myArray = zeros(1, 4);
% Scan line by line, building up the array.
while thisLine ~= -1
  c = textscan(thisLine, '%s %s %f %f %f %f');
  myArray(lineNumber, 1) = c{3};
  myArray(lineNumber, 2) = c{4};
  myArray(lineNumber, 3) = c{5};
  myArray(lineNumber, 4) = c{6};
  % Increment the line counter for next time.
  lineNumber = lineNumber + 1;
  % Retrieve the next line:
  thisLine = fgetl(fid);
end
fclose(fid);
% Show in command window:
myArray

[Octavian]

Dear Image Analyst

Thank you for your assist. The problem is a bit more complicated, as I mentioned the output coordinate files must list those in the increasing point index order; if I have, say, MO7,9,8,3,5,4,6,7 successively in the .txt file, the output .mat has to list coordinates of points MO3,4,5,6,7,8,9. The code I posted does that, in that for each line read, calling the eval(sprintf) creates a pointIdx (which is the point index) by 3 double array, where all elements except the bottom row are 0. With each iteration/ line reading, the workspace .mat coordinate file is overwritten, this way the results is always a ordered (point-index-wise) array. Besides, the code generates a separate array per cluster, which is required downstream. Please apply my code on another file I attach with 3 point clusters, to be clear as to what the scope of my quest is. Again, all is well, except that the code generates rows of 0 for incomplete point index series (counting not beginning with 1) and I would like to change this, or delete them (automatically) as the next step in the routine, I tried

eval(sprintf('%s = %s(%s(:,1)~=0& %s(:,2)~=0&%s(:,3)~=0,:);', trodeName));

but it does not work. Thank you,

Octavian.

[Image Analyst]

You don't need eval() - please stop using it. Try this code:

folder = pwd; % Current folder.
fullFileName = fullfile(folder, 'file.txt');
% Check if file exists.
if ~exist(fullFileName, 'file')
  % File doesn't exist -- didn't find it there.  Check the search path for it.
  fullFileName = baseFileName; % No path this time.
  if ~exist(fullFileName, 'file')
    % Still didn't find it.  Alert user.
    errorMessage = sprintf('Error: %s does not exist in the search path folders.', fullFileName);
    uiwait(warndlg(errorMessage));
    return;
  end
end
% File exists, so open it.
fid = fopen(fullFileName,'r');
thisLine = fgetl(fid);
myArray = zeros(1, 4);
% Scan line by line, building up the array.
while thisLine ~= -1
  c = textscan(thisLine, '%s %s %f %f %f %f');
  % Get the single digit after MO.
  secondString = char(c{2});
  theIndex = str2double(secondString(3:end));
  
  myArray(theIndex, 1) = c{3};
  myArray(theIndex, 2) = c{4};
  myArray(theIndex, 3) = c{5};
  myArray(theIndex, 4) = c{6};
  % Retrieve the next line:
  thisLine = fgetl(fid);
end
fclose(fid);
% Show in command window:
myArray
% Find the bad rows that are all zeros.
badRows = all(myArray==0, 2)
% Remove those rows:
myArray(badRows) = []

Answer 2

[Octavian]

Sorry,

but the above does not save individual .mat arrays for each cluster (SO, MO and OP), nor gets coordinates in the order of the point indexes, please try the above file to see. Also, if this does not become too complicated, please run both codes and see the difference in outputs. Thanks, Octavian

Answer 3

[Octavian]

Dear Image Analyst & All,

I found a solution, to follow to my initial code posted; these lines would delete those all 0 rows in all mat files corresponding to different clusters.

for name = PointNames
    name = name{1};
    eval(sprintf('%s = %s(%s(:,1)~=0 |%s(:,2)~=0|%s(:,3)~=0,:);', name,name,name,name,name));
end

Let me know of any suggestions/ comments,

Octavian

[Image Analyst]

You never said anything about mat files. You only said you have a text file. Moreover, your code above doesn't do anything with any mat files. Plus you're using eval() which is virtually never recommended. Note how I deleted all rows where all numbers were 0 in my previous answer:

% Find the bad rows that are all zeros.
badRows = all(myArray==0, 2)
% Remove those rows:
myArray(badRows) = []

I'm pretty sure that all experienced MATLAB programmers would do it my way rather than your way.

Also, how you're dealing with "name" variable is very non-standard. Where did you learn that it's okay to redefine a loop iterator variable inside the loop? Fortunately with MATLAB it won't have any effect because redefining iterators inside the loop is basically ignored when the loop starts again but it's not like that in every language. For example the following loop will execute 10 times, not once:

for k = 1 : 10
  k = 10
end

And finally it seems like you're bucking the recommendation in the FAQ link I gave you about creating variable names in a loop, where it recommends you don't do that . You're doing it anyway despite the recommendation not to. Apparently it's expedient that you use your code (because it works) but soon you should look at this link : http://www.mathworks.com/matlabcentral/answers/8026-best-way-s-to-master-matlab