Read formatted text in MATLAB

Question

I have some text files that I want to access them in MATLAB workspace. MATLAB help says that I can use fscanf, fgetl, and textscan. I chose the last one due to the formatted text. I wrote the below scripts:

filename = 'myFile.txt';
fid = fopen(filename);
myData = textscan(fid, '%u64 %{dd/MM/yyyy}D %{hh:mm:ss.SSS}T %f64 %f64 %u64 %f64 %f64 %f64\r\n', 'HeaderLines', 3)
fclose(fid);

but I get the error:

Error using textscan

Unable to parse the format character vector at position 21 ==> %{HH:mm:ss.SSS}T %f64 %f64 %u64 %f64 %f64 %f64

Date formats must be of the form %T or %{...}T.

The formatted texts are as:


--------------------------------------------------------------------------------------------------
Row     Var1        Var2           Var3    Var4         Var5         Var6    Var7        Var8 
--------------------------------------------------------------------------------------------------
1       08/04/2018  09:56:52.790   020.00  019.999570   1999690178   055.00  010.020000  000.00000  
2       08/04/2018  09:56:52.821   020.00  019.999602   1999690178   055.00  010.020000  000.00000  
3       08/04/2018  09:56:52.852   020.00  019.999580   1999690178   055.00  010.020000  000.00000  
4       08/04/2018  09:56:52.883   020.00  019.999623   1999690179   055.00  010.020000  000.00000  
5       08/04/2018  09:56:52.915   020.00  019.999548   1999690179   055.00  010.020000  000.00000  
6       08/04/2018  09:56:52.946   020.00  019.999602   1999690179   055.00  010.020000  000.00000  
7       08/04/2018  09:56:52.993   020.00  019.999548   1999690179   055.00  010.020000  000.00000  
8       08/04/2018  09:56:53.024   020.00  019.999602   1999690179   055.00  010.020000  000.00000  
9       08/04/2018  09:56:53.055   020.00  019.999548   1999690179   055.00  010.020000  000.00000

Pretty sure the problem is the duration format. I think it wants lower case h instead of H. — Matt

Tommaso Belluzzo Tommaso Belluzzo · Accepted Answer · 2018-04-08T13:31:53

I always tend to avoid using fscanf, importdata, textscan and such functions because they can be tricky to deal with and I think their output sometimes is not easy to manipulate. On the top of that, your file format looks very similar to the one that Matlab uses for displaying tables data... I think that this is nicenly pointing you to the right direction.

I recommend you to use readtable, not only because of the aforementioned reasons but also because tables are very versatile in Matlab:

T = readtable('data.txt', ...
      'Format', '%d %{dd/MM/yyyy}D %{HH:mm:ss.SSS}D %f %f %f %f %f %f', ...
      'HeaderLines', 3)

The final output is:

T =

  9×9 table

    Var1       Var2           Var3        Var4      Var5          Var6       Var7    Var8     Var9
    ____    __________    ____________    ____    _________    __________    ____    _____    ____

    1       08/04/2018    09:56:52.790    20       19.99957    1999690178    55      10.02    0   
    2       08/04/2018    09:56:52.821    20      19.999602    1999690178    55      10.02    0   
    3       08/04/2018    09:56:52.852    20       19.99958    1999690178    55      10.02    0   
    4       08/04/2018    09:56:52.883    20      19.999623    1999690179    55      10.02    0   
    5       08/04/2018    09:56:52.915    20      19.999548    1999690179    55      10.02    0   
    6       08/04/2018    09:56:52.946    20      19.999602    1999690179    55      10.02    0   
    7       08/04/2018    09:56:52.993    20      19.999548    1999690179    55      10.02    0   
    8       08/04/2018    09:56:53.024    20      19.999602    1999690179    55      10.02    0   
    9       08/04/2018    09:56:53.055    20      19.999548    1999690179    55      10.02    0

P.S. = the %{...}T format is probably due to a misleading way of handling format error messages from the part of Matlab, only %{...}D is a valid datetime literal format until at least Matlab 2017A.

Read formatted text in MATLAB

2 Answers