Loading tab delimited .txt file with mix of strings and numeric data

I think I'm going crazy, but for the life of me I can't get this .txt file to load into Matlab. All I want is column 2 but I've tried the usual fopen,textscan,fclose route recommended in other posts. I've also tried functions like load, fileread, and some of the customised functions from other users. Nothing is working! I need some help please!

Sample data attached...

Any advice or solutions welcomed!

[Azzi Abdelmalek]

Nothing is attached

Answer 1

[per isakson]

See textscan, Read formatted data from text file or string

.

Later

Try

    fid = fopen( 'Test28_Text_table.txt', 'r', 'ieee-le', 'UCS-2' );
    cac = textscan( fid, '%s%f%s%s', 'Delimiter', '\t', 'Headerlines', 1 );
    fclose( fid );

    >> cac{2}
    ans =
        3.4000
        5.1000
       80.6000
       19.8000
      100.4000
           ...

I get this warning, but it seems to work

    Warning: The encoding 'UTF-16' is not supported.
    See the documentation for FOPEN.

See UCS-2 and its relationship to Unicode (UTF-16)

Notepad++ recognized the encoding of your file.

    >> version
    ans =
    8.1.0.604 (R2013a)

Answer 2

[Azzi Abdelmalek]

ii=fopen('file.txt')
s=textscan(ii,'%s')
fclose(ii)
out=regexp( cellstr(s{1}),'\d+(\.)?(\d+)?','match')
out=cellfun(@(x) str2num(x{1}),out(~cellfun(@isempty,out)))

[Aeronut]

Thanks Azzi,

I've tried your code and while ii is assigned with a value of 3, upon using s=textscan with these parameters, all I get is s = "<1166x1 cell>" of empty fields of the form

{}

{}

{}

{}

{}

...

I think there might be an issue with the type of encoding used to generate the code, which was done on a Windows 8 system?

When using a basic fileread approach:

data = fileread(filename);

...and trying to drag the info out by character position I can see that there are additional characters in front of row 1, column 1 and unnecessary white space, i.e.

ÿþN a m e ...etc

Is there something I should be checking in the file encoding? Or is there something matlab can do to account for this? Is this why textscan produces an array of whitespace?

Regards

Kevin

[Aeronut]

Also, the variable out = Empty matrix: 0-by-1

[per isakson]

See my answer.

Answer 3

[Jos (10584)]

You can skip the columns you do not want:

ValuesInCol2 = textread('Test28_Text_table.txt','%*s%f%*[^\n]','headerlines',1)