Reading specific rows and columns of numerical values from a text file

I am trying to read specific numerical values from this text file into matlab. I want the values from columns RetTime[min] and Area[pA*s]. Originally I was trying to use the dlmread function, but was running into an error. The error message was "Mismatch between file and format character vector". I believe this is from the non-numerical data at the beginning of the file. After doing some reading the textscan function may be better for this case, but I am struggling with formatting aspect of this function. I don't understand how to read a specific row and column using the textscan function. However, I am unsure if this would even be the best function to use in this case. Using the textscan function, how would I specify a certain column and row to read? Or would there be a better function to use in this case?

[per isakson]

Which release do you use? FixedWidthImportOptions, Import options object for fixed-width text files was Introduced in R2017a

[Walter Roberson]

Note: the file is written as UTF16LE with Byte Order Marker.

per's mention of fixed width is a good point: some of the columns are empty, and there is no delimiter such as comma or tab between the columns.

Answer 1

[Image Analyst]

Because you have three separate chunks of data separated, before and after, by some other non-matrix stuff, I think you may need to write your own reader to handle the peculiarities of this type of file format. For example, you may want a structure array where each structure has the numerical data along with non-numerical data in fields. So for that file you'd have 3 structures in your structure array, while for another data file there may be 4 or 12 or whatever.

Answer 2

[per isakson]

Here is an implementation along the lines suggested by @ImageAnalyst. The code to read the text file is copied from @Walter Roberson. The code is tested on R2016a.

Run cssm

>> out = cssm
out = 
1x3 struct array with fields:
    Header
    Peak
    RetTime
    Type
    Width
    Area
    pArea
    Name

and inspect the result (I use the transpose-blip to save screen area)

>> out(1).RetTime'
ans =
  Columns 1 through 8
    1.4520    2.4040    3.2870    5.1550    5.5540    6.6860    6.7760    7.0180
  Columns 9 through 16
    7.8000    7.8550    7.9550    7.9920    8.0480    8.3350    9.1410    9.2370
  Columns 17 through 18
   10.3690   10.7530
>> out(2).Type'
ans = 
    ''    'BB'    'BV'    'VB'    ''    'BB'    ''    ''    'BB'    'BB'    ''    ''
>> >> out(3).Area'
ans =
  165.0026         0   42.9935         0         0  674.3498         0

where

function  out = cssm()
    warning( 'off', 'MATLAB:iofun:UnsupportedEncoding' )
    fid = fopen( 'h:\m\cssm\report2.txt', 'rt', 'n', 'UTF16LE' );
    warning( 'on', 'MATLAB:iofun:UnsupportedEncoding' )
    [~] = fread( fid, 2, '*uint8' );
    str = fread( fid, [1,inf], '*char' );
    fclose( fid );
    str(str==char(13))=[];  % Unexpected: "\r\n" despite "rt" in fopen
    xpr = '(?m-s)^Signal \d: [A-Z]{3}\d [A-Z].+$';  % matches first line of block
    [ body, header ] = strsplit( str, xpr, 'DelimiterType','RegularExpression' );
    body(1) = [];
    clh = {'Peak','RetTime','Type','Width','Area','pArea','Name'};
    spc = {'%f','%f','%s','%f','%f','%f','%s'};
    out = struct( 'Header', header, clh{1},{[]}, clh{2},{[]}, clh{3},{''}   ...
                , clh{4},{[]}, clh{5},{[]}, clh{6},{''}, clh{7},{''}        );
    %
    delimiter = ',';
    frmtspec = '%4c%8c%7c%8c%11c%9c%21c%*[^\n]';
    for jj = 1 : length( body )
        tok = strsplit( body{jj}, 'Totals' );
        tok = strtrim( tok{1} );
        buf = textscan( tok, frmtspec, 'Headerlines',3 ...
                      , 'Whitespace','', 'Delimiter','' );
        for ii = 1 : length( clh )
            str = cat( 2, char(buf(ii)), repmat( delimiter, [size(buf{ii},1),1] ) );
            str = permute( str, [2,1] );
            cac = textscan( str, spc{ii}, 'Delimiter',delimiter );
            if strcmp( spc(ii), '%s' )
                out(jj).(clh{ii}) = strtrim( cac{:} );
            else
                out(jj).(clh{ii}) = cac{:};
            end
        end
    end
end

... left as an exercise ... ;-)