Cast R objects into Python ones with rpy2

Question

The question: how to cast R objects to python ones

My case: I need to use the result of cor.test() into a python routine.

correlation = robjects.r('function(x, y) cor.test(x, y)')
corr= correlation(robjects.IntVector(goodtotemp), robjects.IntVector(goodGDPs))
print corr

print corr[3]
print 'coef:',type(corr[3])

outputs, as expected:

     cor 
0.984881 
coef: <class 'rpy2.robjects.vectors.FloatVector'>

Howover, I can't use corr[3] as an python object,

c=corr[3]
print 'c:',c*10., type(c)

look (Here's how I know that I'm doing something wrong!), the output:

c:
Traceback (most recent call last):
  File "./GDPAnalyzer.py", line 234, in <module>
    print 'c:',c*10., type(c)
TypeError: unsupported operand type(s) for *: 'FloatVector' and 'float'

Any hint/help is appreciated!

Answer 1

The segmentation fault is a something that should be reported as an rpy2 bug. Otherwise, try either:

c.ro * 3

or:

c[0] * 3

Answer 2

Python really does not support the concept of casting . It's true that in some limited circumstances it does do something like casting with built-in types.

But beyond that it is up to the programmer to CONVERT objects from one form to another or to create some form of facade or shim object that makes the R object behave like a Python one.

Try print type(corr[3]) to see what Python type is being used. It might just be a string representation of the number, in which case you could use eval() to convert it to a float.