reading mixed format csv data with empty value '-'

The data is mixed formatted. I just list the float number parts to show you my problem.

    1 1 100.3 -45000 -0.23 0.2555 600000
    2 1 100.3 -45000 -0.23 0.2555 800000
    3 1 100.3 -45000 -0.23 0.2555 800000
    4 1 -     -      -     -      -
    5 1 -     -      -     -      -

I can not delete the empty lines since I want to know their location. But when I use

   textscan (fid,'%f %f %f %f %f %f %f),

I have trouble with class of every column. And If I use

    'TreatAsEmpty', '-'

inside textscan, all the negative value will be read as wrong. Does anyone have any suggestions?

Answer 1

[per isakson]

If the file together with the parsed result fits in memory try

    >> out = cssm('cssm.txt')
    out =
       1.0e+05 *
        0.0000    0.0000    0.0010   -0.4500   -0.0000    0.0000    6.0000
        0.0000    0.0000    0.0010   -0.4500   -0.0000    0.0000    8.0000
        0.0000    0.0000    0.0010   -0.4500   -0.0000    0.0000    8.0000
        0.0000    0.0000       NaN       NaN       NaN       NaN       NaN
        0.0001    0.0000       NaN       NaN       NaN       NaN       NaN
    >>

where

    function out = cssm( filespec )
        str = fileread( filespec );
        str = regexprep( str, '(?<=\s)\-(?=\s)', 'nan' );
        out = cell2mat(textscan(str,'%f%f%f%f%f%f%f','CollectOutput',true ));
    end

and where cssm.txt contains

    1 1 100.3 -45000 -0.23 0.2555 600000
    2 1 100.3 -45000 -0.23 0.2555 800000
    3 1 100.3 -45000 -0.23 0.2555 800000
    4 1 -     -      -     -      -
    5 1 -     -      -     -      -

&nbsp

This was a demo of Lookahead Assertions in Regular Expressions

&nbsp

cssm.m above fails if the last character of the file is -. To fix that replace

    str = regexprep( str, '(?<=\s)\-(?=\s)', 'nan' );

by

    str = regexprep( str, '(?<=\s)\-(?=\s|$)', 'nan' );

to match the character, -, when it is the last character of the string.