. I'm trying to find out some indices and process a file with rows (comented) with diferent numbers of columns. Any help?

I have a file with data and some comented rows (lat,lon and date). I'm trying to find out the indices of these last ones. Then, these rows have more columns (about 5) than the not comented rows (2). The matlab can't process this way. Need some help. Ex:

% 189° 25° 1980 4 27

3 28,9

4 28,7

5 27,0

6 25,6

[...]

[per isakson]

Does your file look something like this?

    % 189° 25° 1980 4 27
    3 28,9
    4 28,7
    5 27,0
    6 25,6
    ...
    % 189° 25° 1980 4 28
    3 28,9
    4 28,7
    5 27,0
    6 25,6
    ...
    % 189° 25° 1980 4 29
    3 28,9
    4 28,7
    5 27,0
    6 25,6
    ...

comma, ",", as decimal separator? (Matlab has problems with that.)

"find out the indices" &nbsp is only a step on the way to read and parse the file?

Does the entire file fit comfortable in memory?

[Dorfschafer]

Hi Isakson, No, I expressed myself badly.. The file contains dots, not commas. Yes, it looks like your model and fitted well in memory (until then). I need to find out these indices because i'm only interested in certain lats and lons.. so, i need to know where they are and a way to read and parse these particular set of the file.

Answer 1

[per isakson]

• "the indices of these last ones" &nbsp I'm not sure what "indices" refers to.
• "I need to find out these indices because i'm only interested in certain lats and lons" &nbsp If it doesn't cause problems with memory or performance I find it easier to read and parse the entire file in a separate step. Furthermore, that increases the chances that the code can be reused. The function cssm performs that first step.

    >> sas = cssm()
    sas = 
    1x3 struct array with fields:
        sdn
        Lat
        Long
        data

    >> is25 = [sas.Lat] == 25;
    >> datestr( [sas(is25).sdn], 'yyyy-mm-dd' )
    ans =
    1980-04-27
    1980-04-28

where

    function    sas = cssm()

        str = fileread('data.txt');
        deg = char(176);
        pos = '(%[^\n\r]+)'; % matches a substring starting with "%" ending at new-line
        cac = regexp( str, [pos,'[\r\n ]+([\d\.\s ]+)'], 'tokens' );
        len = length( cac );
        sas(1,len) = struct( 'sdn', [], 'Lat',[], 'Long',[], 'data', [] );

        for jj = 1 : len
            buffer1 = sscanf( cac{jj}{1}, ['%*c %d',deg,'%d',deg,'%d%d%d'] ); 
            buffer2 = textscan( cac{jj}{2}, '%f%f', 'CollectOutput', true );
            sas(jj).Long  = buffer1(1);
            sas(jj).Lat   = buffer1(2);
            sas(jj).sdn   = datenum( buffer1(3), buffer1(4), buffer1(5) );
            sas(jj).data  = buffer2{1};
        end 
    end

and where data.txt contains

    % 187° 25° 1980 4 27
    3 28.9
    4 28.7
    5 27.0
    6 25.6
    % 188° 25° 1980 4 28
    3 28.9
    4 28.7
    5 27.0
    6 25.6
    % 189° 30° 1980 4 29
    3 28.9
    4 28.7
    5 27.0
    6 25.6

[Dorfschafer]

By 'indices', i meant, putting into a matrix, the numbers of the rows with lats,lon,etc.. would have. I need to reuse this code several times.

Thanks for answering Isakson, i ll try your code (it doesn't looks like so simple), then ll come back here.