4.3.1 Enhanced Robj

This example shows an extended Robj type which supports R attribute look up via normal Python attribute access. It also supports the representation given by the R interpreter, and it implements the as_r() method for converting itself to R.

In Python 2.2 you can take advantage of the possibility to subclass types, and the examples can be rewritten in a more powerful way. However, these examples are Python 2.1 and 2.2 compatible.

File erobj.py:

     from rpy import *
     
     class ERobj:
     
         def __init__(self, robj):
             self.robj = robj
     
         def as_r(self):
             return self.robj
     
         def __str__(self):
             a = with_mode(NO_CONVERSION,
                           lambda: r.textConnection('tmpobj', 'w'))()
             r.sink(file=a, type='output')
             r.print_(self.robj)
             r.sink()
             r.close_connection(a)
             str = with_mode(BASIC_CONVERSION,
                             lambda: r('tmpobj'))()
             return '\n'.join(as_list(str))
     
         def __getattr__(self, attr):
             e = with_mode(BASIC_CONVERSION,
                           lambda: r['$'](self.robj, attr))()
             if e:
                 return e
             return self.__dict__[attr]

The __str__ method makes the R interpreter print to the tmpobj R variable. Then, it is retrieved and returned as the string representation of the object. Note the use of the with_mode function for not changing the mode in use. Note, also, the use of the utility functions as_list and r (see Miscellaneous).

An example of use:

     >>> from rpy import *
     >>> from erobj import *
     >>> proc_table[lambda o: 1] = ERobj
     >>> set_default_mode(PROC_CONVERSION)
     >>>
     >>> e = r.t_test([1,2,3])
     >>> e
     <erobj.ERobj instance at 0x8ad4ea4>
     >>> print e
     
             One Sample t-test
     
     data:  c(1, 2, 3)
     t = 3.4641, df = 2, p-value = 0.07418
     alternative hypothesis: true mean is not equal to 0
     95 percent confidence interval:
      -0.4841377  4.4841377
     sample estimates:
     mean of x
             2
     
     >>>
     >>> e.statistic
     {'t': 3.4641016151377548}