从 rpy2 dataframe 到 pandas dataframe 的转换不适用于字符串类型的列

Question

If I try a conversion from pandas to R and back to pandas with a dataframe made only of numeric columns it goes ok, as can be seen below:

import pandas as pd
df_py = pd.DataFrame({'col1':[1,2,3,4,5],
                      'col2':[1.0,2.0,3.0,4.0,5.0]})

from rpy2.robjects.pandas2ri import py2ri
df_r = py2ri(df_py)
ri2py(df_r)

In this case, the result is:

  col1  col2
0   1   1.0
1   2   2.0
2   3   3.0
3   4   4.0
4   5   5.0

Now if I include a third column of string type I get an error. Look below:

import pandas as pd
df_py = pd.DataFrame({'col1':[1,2,3,4,5],
                      'col2':[1.0,2.0,3.0,4.0,5.0],
                      'col3':['a','b','c','d','e']})

from rpy2.robjects.pandas2ri import py2ri
df_r = py2ri(df_py)
ri2py(df_r)

The error is:

/usr/local/lib/python3.6/dist-packages/numpy/core/_asarray.py in asarray(a, dtype, order)
     83 
     84     """
---> 85     return array(a, dtype, copy=False, order=order)
     86 
     87 

ValueError: Buffer for this type not yet supported.

Is this a issue in rpy2 or is there something that I'm doing wrong?

Answer 1

I got the lead for this answer from Parfait comment (see above).

I changed the version of rpy2 in collab to 3.2.2 (latest version).

In this new version, the functions py2ri and ri2py have changed to py2rpy and rpy2py , respectively.

Using them in a code similar to the one presented in my question, the conversion back and forth happened perfectly.

The code that I used is presented below:

!pip install rpy2==3.2.2
import rpy2
import pandas as pd
df_py = pd.DataFrame({'col1':[1,2,3,4,5],
                      'col2':[1.0,2.0,3.0,4.0,5.0],
                      'col3':['a','b','c','d','e']})

from rpy2.robjects.pandas2ri import py2rpy, rpy2py
rpy2.robjects.pandas2ri.activate()

df_r = py2rpy(df_py)    
rpy2py(df_r)