Support for CDL

[martinjuckes]

It would be really useful to be able to read CDL files directly into cfdm, rather than having to first convert to binary netCDF files. Can this be added?

[davidhassell]

Good idea, thanks. This will be ready in the next version, which should be ready in a few days.

[cofinoa]

This would be a interesting feature.

[davidhassell]

All ready in v1.7.8 (https://pypi.org/project/cfdm). See the documentation at https://ncas-cms.github.io/cfdm/function/cfdm.read.html

Try it with test_file.zip

$ unzip test_file.zip
Archive:  test_file.zip
  inflating: test_file.nc          
$ ncdump test_file.nc > test_file.cdl
$ ncdump -h test_file.nc > test_file_h.cdl
$ ncdump -c test_file.nc > test_file_c.cdl

In [10]: import cfdm                                                                                                  

In [11]: f = cfdm.read('test_file.nc')[0]                                                                                                    

In [12]: g = cfdm.read('test_file.cdl')[0]                                                                                                   

In [13]: f.equals(g)                                                                                                                         
Out[13]: True

In [14]: cfdm.read('test_file_h.cdl')[0]                                                                                                     
Out[14]: <Field: eastward_wind(atmosphere_hybrid_height_coordinate(1), grid_latitude(10), grid_longitude(9)) m s-1>

In [15]: cfdm.read('test_file_c.cdl')[0]                                                                                                     
Out[15]: <Field: eastward_wind(atmosphere_hybrid_height_coordinate(1), grid_latitude(10), grid_longitude(9)) m s-1>

[davidhassell]

I am aware that this doesn't work for Python 2. I will gladly fix this if anyone needs it ...

[martinjuckes]

Hi David,

thanks for adding this, it will be very useful.

Trying it out, I found that it fails when a file starts with a comment line (// ........), which I think should still be valid cdl. The error given is raise IOError("Can't determine format of file {}".format(filename)).

Here is the file which triggers the error:

// https://www.unidata.ucar.edu/software/netcdf/examples/sresa1b_ncar_ccsm3-example.cdl
netcdf cal01 {
dimensions:
        time = UNLIMITED ; // (1 currently)
variables:
        double time(time) ;
                time:calendar = "gregorian" ;
                time:standard_name = "time" ;
                time:axis = "T" ;
                time:units = "days since 1980-24-33 12:00:00" ;
                time:long_name = "time" ;
        float mydata(time) ;
                mydata:standard_name = "air_temperature" ;
                mydata:units = "K" ;
// global attributes:
                :Conventions = "CF-1.7" ;

data:

 time = 730135.5 ;

 mydata = 280.;
}

It works when the first line is deleted.

[davidhassell]

Hi Martin, thanks for spotting this - I was only allowing files whose first line started netcdf . I shall update the code to accept leading comment lines, too.