importing tab delimited text file

Hi,

I am downloading a text file "A" using textscan. I know that the file is table delimited with unknown number of columns and 34000 rows. Some columns are numbers ,others are strings. I used textscan('A.txt','%s') and what I am getting is 1x1 cell called data . This 1 cell data is 34000x1 where each row contains all the data of the respective rows in the original file (i.e not separated to columns and there isn't any space that separates the rows). Any help about how to download the file is appreciated

[Walter Roberson]

textscan('A.txt','%s') is going to give you the cell array {'A.txt'} -- if the first argument to textscan() is a string, then the string itself is considered to be the input to be scanned.

Answer 1

[Fangjun Jiang]

try a=importdata('A.txt') to see what you got. Many times, it will give your well-formatted data.

[Danielle Leblanc]

importdata is nor working. I am receiving the following error

??? Error using ==> importdata at 214

Unable to load file.

Use TEXTSCAN or FREAD for more complex formats.

[Fangjun Jiang]

Then can you show a representative sample of your data in the 34000x1 data? Further processing can be done to get the format you want.

[Zeenat Islam]

importdata ALL THE WAY! Worked like a charm for me

[Walter Roberson]

Note that in more recent versions, importdata by default now returns text as string objects instead of as cell arrays of character vectors. We are seeing people getting caught by that.

Answer 2

[Walter Roberson]

What is the delimiting character? Can any of the strings contain the delimiting character, and if so then how is it indicated that that delimiter is a part of the string rather than marking the end of the string?

[Danielle Leblanc]

the delimiting character is ' ', i.e columns are separated by a space

[Danielle Leblanc]

also teh columns are a mix of strings and numerics. I tried textscan('A.txt','%q') and ('A.txt','%c') but I obtained the same output as ('A.txt','%s')

Answer 3

[Danielle Leblanc]

Hi again,

sorry to bother you with this problem but I am a matlab beginner. I opened one of tmt text files with excel.The data has 33 columns and I am putting the transpose of the columns below as each has a different format (col1 col2 etc.. is the column number and what follows it is the data format):

    col1  5
    col2  6154
    col3  T
    col4  ABN.GG
    col5  ABN
    col6  00077T
    col7  AA2
    col8  N
    col9  N
    col10  4000
    col11  A
    col12  104.61
    col13  +
    col14  7.226596
    col15  A
    col16  20090109
    col17  10:21:00
    col18  0
    col19  @
    col20  A
    col21  Y
    col22  7
    col23  104.61
    col24  +
    col25  7.226596
    col26  104.61
    col27  +
    col28  7.226596
    col29  104.61
    col30  +
    col31  7.226596
    col32  11636
    col33  20090109

excel recognized the file correctly. I have 400 of these text files so importing them from text to excel then to matlab would take a lot of time. How can I import this text file to matlab directly

[Fangjun Jiang]

It's hard to read your data due to format. Can you paste 3 or 4 lines of your text file here and apply the code format to it?

Answer 4

[Walter Roberson]

You continue to have the same problem that I warned about earlier: textscan() with a string as its first argument reads the string, not a file denoted by the string. The older textread() routine expected a filename as the first argument, but textscan() never does.

fid = fopen('A.txt','rt');
data = textscan(fid, '%f %f %s %s %s %s %s %s %s %f %s %f %s %f %s %f %s %f %s %s %s %d %f %s %f %f %s %f %f %s %f %f %f');
fclose(fid);

The spaces within the quoted string are not important and can be left in or removed as desired.

I coded this in such a way that the dates such as 20090109 are read as numbers, but the time such as 10:21:00 is read as a string. textscan() is not able to directly read formatted times as times.

The output, data, will be a cell array, containing one column vector per column of input, so for example data{2} would be a column vector of floating point numbers corresponding to column 2, one entry per line of input.

[Danielle Leblanc]

I copied and pasted the codes as they are and it is not working. I am receiving 3 outputs:

ans=0;

data a 1x33 cell with [] in all cells

and fid=3;

[Danielle Leblanc]

I tried it on other files as well. The output is different but the content of each cell of the cell array never exceeds 2 rows of data although I have 35000 rows per file. for example I am getting:

[95;2722]

[102.680000000000;9730]

<2x1 cell>

<2x1 cell>

<2x1 cell>

<2x1 cell>

<2x1 cell>

<2x1 cell>

<2x1 cell>

[NaN;9000]

<2x1 cell>

[NaN;102.050000000000]

<2x1 cell>

[NaN;8.80793700000000]

<2x1 cell>

[NaN;11]

<2x1 cell>

[NaN;0]

<2x1 cell>

<2x1 cell>

<2x1 cell>

0

NaN

''

NaN

NaN

''

NaN

NaN

''

NaN

NaN

NaN

[Fangjun Jiang]

You need to check the consistency of your text file. As long as the data format is consistent. The code Walter provided should return correct result. I construct three lines of text using your data. Here is the result:

A.txt

5 6154 T ABN.GG ABN 00077T AA2 N N 4000 A 104.61 + 7.226596 A 20090109 10:21:00 0 @ A Y 7 104.61 + 7.226596 104.61 + 7.226596 104.61 + 7.226596 11636 20090109

5 6154 T ABN.GG ABN 00077T AA2 N N 4000 A 104.61 + 7.226596 A 20090109 10:21:00 0 @ A Y 7 104.61 + 7.226596 104.61 + 7.226596 104.61 + 7.226596 11636 20090109

5 6154 T ABN.GG ABN 00077T AA2 N N 4000 A 104.61 + 7.226596 A 20090109 10:21:00 0 @ A Y 7 104.61 + 7.226596 104.61 + 7.226596 104.61 + 7.226596 11636 20090109

data 1x33 5640 cell

>> data{1}

ans =

5

5

5

>> data{33}

ans =

20090109

20090109

20090109

>> data{18}

ans =

0

0

0

Answer 5

[Stephan Koehler]

I wrote a routine for importing tsv files generated by excel. look at http://www.mathworks.com/matlabcentral/fileexchange/32782