ecearth3/sources/tm5mp/tools/ttb/bin/mdf.py

#! /usr/bin/env python

"""
MDF - Multipe Data Formats

Description

  Quick access to hdf or netcdf files.
  Since a number of different python modules is available 
  to read this type of files, the mdf routines first try to
  find out which one is installed:
   o for hdf    files : pyhdf
   o for netcdf files : pycdf or netCDF4

Usage:

  mdf.show( 'myfile.nc' )

  vnames = mdf.get_varnames( 'myfile.nc' )
  
  data = mdf.get_var     ( 'myfile.nc', 'mydata' )
  data = mdf.get_var_attr( 'myfile.nc', 'mydata', 'variable_attribute' )
  data = mdf.get_attr    ( 'myfile.nc', 'global_attribute' )

HDF files with non-unique data set names

  HDF files could contain multiple data sets with the same name;
  not very useful, but there is nothing that prevents a user from
  doing it so mdf has to deal with it. The following convention is used:
   o If data is requested with 'get_var' and multiple data sets match
     the name, than all matching data sets are concatenated as if they
     are time records.
   o If a specific sample is needed, add the optional 'sample' argument:
       data = mdf.get_var( 'myfile.nc', 'mydata', sample=1 )
     The sample number is 1-based.
     If only one matching data set is found, this one is returned.
  
History

  2010-08, Arjo Segers
  
"""

# ***


def MyLogger( name ) :
    """
    Setup logger; if not logging configured yet, write to standard output.
    """
    # external:
    import logging
    import sys
    # setup logger:
    logger = logging.getLogger( name )
    # no handlers yet ? then print to standard output:
    if len(logger.handlers) == 0 : logger.addHandler(logging.StreamHandler(sys.stdout))
    # no level set yet ? then set for info and higher:
    if logger.level == 0 : logger.setLevel(logging.INFO)
    # ok:
    return logger
#enddef


# ***


def GetModuleName( fname ) :

    """
    Return a character key that specifies the python module to be used.
    First select between hdf and netcdf given the filename extension.
    Then try if a known module is present to read the file.
    Supported:
      pyhdf
      pycdf
      netCDF4
    """
    
    # get logger:
    logger = MyLogger('mdf')
    
    # set module names for filetype:
    if fname.endswith('.hdf') :
        modnames = ['pyhdf']
    elif fname.endswith('.nc') :
        filetype = 'netcdf'
        modnames = ['netCDF4','pycdf']
    else :
        logger.error( 'could not guess filetype from extension for file : %s' % fname )
        raise ValueError
    #endif
    
    # none found yet:
    modname = None
    # loop over modules to be tested:
    for amodname in modnames :
        try :
            # import :
            testmod = __import__(amodname)
            # no exceptions ? then this one is ok:
            modname = amodname
            break
        except : 
            # try next:
            continue
        #endtry
    #endfor
    # check ..
    if modname == None :
        logger.error( 'could not import any of the modules : %s' % modnames )
        raise ImportError
    #endif
    
    # ok
    return modname
    
#enddef


# ***


def show( fname ) :

    """
    List contents of file.
    """
    
    # modules:
    import os
    
    # get logger:
    logger = MyLogger('mdf')
    
    # check ...
    if not os.path.exists(fname) :
        logger.error( 'file not found : %s' % fname )
        raise IOError
    #endif

    # different modules:
    modname = GetModuleName( fname )
    # import :
    submod = __import__( 'mdf_'+modname )
    # apply:
    submod.show( fname )
    
    # ok
    return

#enddef


# ***


def get_varnames( fname ) :

    """
    Return list with variable names.
    """
    
    # modules:
    import os
    
    # get logger:
    logger = MyLogger('mdf')
    
    # check ...
    if not os.path.exists(fname) :
        logger.error( 'file not found : %s' % fname )
        raise IOError
    #endif

    # different modules:
    modname = GetModuleName( fname )
    # import :
    submod = __import__( 'mdf_'+modname )
    # apply:
    values = submod.get_varnames( fname )
    
    # ok
    return values

#enddef


# ***


def get_var( fname, vname, sample=None ) :

    """
    Read variable from file.
    Sample number (1-based) could be used to select a
    data set from a hdf file which has more than one
    matching data set.
    """
    
    # modules:
    import os
    
    # get logger:
    logger = MyLogger('mdf')
    
    # check ...
    if not os.path.exists(fname) :
        logger.error( 'file not found : %s' % fname )
        raise IOError
    #endif

    # different modules:
    modname = GetModuleName( fname )
    # import :
    submod = __import__( 'mdf_'+modname )
    # apply:
    values = submod.get_var( fname, vname, sample=sample )
    
    # ok
    return values

#enddef


# ***


def get_attr( fname, aname ) :

    """
    Read global attribute from file.
    """
    
    # modules:
    import os
    
    # get logger:
    logger = MyLogger('mdf')
    
    # check ...
    if not os.path.exists(fname) :
        logger.error( 'file not found : %s' % fname )
        raise IOError
    #endif

    # different modules:
    modname = GetModuleName( fname )
    # import :
    submod = __import__( 'mdf_'+modname )
    # apply:
    values = submod.get_attr( fname, aname )
    
    # ok
    return values

#enddef


# ***


def get_var_attr( fname, vname, aname ) :

    """
    Read variable attribute from file.
    """
    
    # modules:
    import os
    
    # get logger:
    logger = MyLogger('mdf')
    
    # check ...
    if not os.path.exists(fname) :
        logger.error( 'file not found : %s' % fname )
        raise IOError
    #endif

    # different modules:
    modname = GetModuleName( fname )
    # import :
    submod = __import__( 'mdf_'+modname )
    # apply:
    values = submod.get_var_attr( fname, vname, aname )
    
    # ok
    return values

#enddef