MATLAB: Can REGEXP be utilized in searching M x N matrix values

blocksMATLABread matrixregexp

The following text file contains numerous lines of data, including a 9 x 9 state covariance matrix:
State Time: 13267.000
State Date: 20120101
State Covariance:
          -0.11000000000000e-01 0.12000000000000e01 0.13000000000000e00 0.14000000000000e-01 0.15000000000000e-01 0.16000000000000e-01 0.17000000000000e-01 0.18000000000000e-01 0.19000000000000e-01
          0.21000000000000e-01 0.22000000000000e-01 0.23000000000000e-01 0.24000000000000e-01 0.25000000000000e-01 0.26000000000000e-01 0.27000000000000e-01 0.28000000000000e-01 0.29000000000000e-01
          0.31000000000000e-01 0.32000000000000e-01 0.33000000000000e-01 0.34000000000000e-01 0.35000000000000e-01 0.36000000000000e-01 0.37000000000000e-01 0.38000000000000e-01 0.39000000000000e-01
          0.41000000000000e-01 0.42000000000000e-01 0.43000000000000e-01 0.44000000000000e-01 0.45000000000000e-01 0.46000000000000e-01 0.47000000000000e-01 0.48000000000000e-01 0.49000000000000e-01
          0.51000000000000e-01 0.52000000000000e-01 0.53000000000000e-01 0.54000000000000e-01 0.55000000000000e-01 0.56000000000000e-01 0.57000000000000e-01 0.58000000000000e-01 0.59000000000000e-01
          0.61000000000000e-01 0.62000000000000e-01 0.63000000000000e-01 0.64000000000000e-01 0.65000000000000e-01 0.66000000000000e-01 0.67000000000000e-01 0.68000000000000e-01 0.69000000000000e-01
          0.71000000000000e-01 0.72000000000000e-01 0.73000000000000e-01 0.74000000000000e-01 0.75000000000000e-01 0.76000000000000e-01 0.77000000000000e-01 0.78000000000000e-01 0.79000000000000e-01
          0.81000000000000e-01 0.82000000000000e-01 0.83000000000000e-01 0.84000000000000e-01 0.85000000000000e-01 0.86000000000000e-01 0.87000000000000e-01 0.88000000000000e-01 0.89000000000000e-01
          0.91000000000000e-01 0.92000000000000e-01 0.93000000000000e-01 0.94000000000000e-01 0.95000000000000e-01 0.96000000000000e-01 0.97000000000000e-01 0.98000000000000e-01 0.99000000000000e-01
State Confidence: -1
I’m looking for a method to extract the 9×9 matrix values and save them for processing. I’m using the following code to extract the matrix:
% Prompt user for file to open
[fn,pn]=uigetfile('*.*','Select SNER Text File');
filename = fullfile(pn,fn);
% Read in entire file
buffer = fileread(filename);
% Parse out the Matrix (9x9) and store as SCD
pattern = '*?State Covariance:\s+([-?\d\.]+\w[-?\d\.]+).([-?\d\.]+\w[-?\d\.]+).([-?\d\.]+\w[-?\d\.]+).([-?\d\.]+\w[-?\d\.]+).([-?\d\.]+\w[-?\d\.]+).([-?\d\.]+\w[-?\d\.]+).([-?\d\.]+\w[-?\d\.]+).([-?\d\.]+\w[-?\d\.]+).([-?\d\.]+\w[-?\d\.]+)';
tokens4 = regexp(buffer, pattern, 'tokens');
SCD = reshape(str2double([tokens4{:}]), 9, []).';
This code, even with the long search pattern, does a great job in finding all nine matrix values in row 1 and saving them. But how would I go about extracting the remaining rows? Is there a more practical approach than using the REGEXP for this task?

Best Answer

If the entire file fits in memory I recommend that you
  • read the entire file as character
  • extracts the "numerical chunks" (an appropriate job for regexp)
  • convert to double
Try
    >> M = cssm();
    >> whos M
      Name      Size             Bytes  Class     Attributes
      M         9x9x3             1944  double
where cssm.m is
    function  M = cssm()
        xpr = '(?<=State Covariance:)[ 0-9\.\+\-e\r\n]++(?=State Confidence:)';
        str = fileread( 'cssm.txt' );
        cac = regexp( str, xpr, 'match' );
        M = nan( 9, 9, length( cac ) ); 
        for jj = 1 : length( cac )
           M( :, :, jj ) = str2num( cac{jj} ); 
        end
    end
and cssm.txt contains copy&paste three times of the data of your question
.
See Lookahead Assertions in Regular Expressions
.
"contains numerous lines of data" made me think your file contains many matrices. However, the code above shall work with one.
Related Question