Default conversions#

This page list the default conversions applied to R objects to convert them to Python objects.

Basic types#

The conversion of basic types is performed directly by the Converter used. Thus, changing the conversion for basic types currently requires creating a custom Converter class. The default SimpleConverter realizes the following conversions:

R object type	Python conversion
builtin function	`rdata.conversion.RBuiltin`.
bytecode	`rdata.conversion.RBytecode`.
char (internal)	`str` or `bytes` (depending on the encoding flags).
closure	`rdata.conversion.RFunction`.
complex	`numpy.ndarray` with 128-bits complex dtype. `numpy.ma.MaskedArray` with 128-bits complex dtype if it contains NA values. `xarray.DataArray` if it contains labeled dimensions.
environment	`rdata.conversion.REnvironment`. There are three special cases: the empty, base and global environments, which are all empty by default. The base and global environments may be supplied to the converter.
expression	`rdata.conversion.RExpression`.
external pointer	`rdata.conversion.RExternalPointer`.
integer	`numpy.ndarray` with 32-bits integer dtype. `numpy.ma.MaskedArray` with 32-bits integer dtype if it contains NA values. `xarray.DataArray` if it contains labeled dimensions.
language	`rdata.conversion.RLanguage`.
list	`list` (if untagged). `dict` (if tagged). Empty lists are considered tagged.
logical (boolean)	`numpy.ndarray` with boolean dtype. `numpy.ma.MaskedArray` with boolean dtype if it contains NA values. `xarray.DataArray` if it contains labeled dimensions.
missing argument	`NotImplemented`.
NULL	`None`.
real	`numpy.ndarray` with 64-bits floating point dtype. `numpy.ma.MaskedArray` with 64-bits floating point dtype if it contains NA values. `xarray.DataArray` if it contains labeled dimensions.
reference	The referenced value, that is, an object already converted.
S4 object	`types.SimpleNamespace`.
special function	`rdata.conversion.RBuiltin`.
string	`numpy.ndarray` with suitable fixed-length string dtype.
symbol	`str`.
vector	`list` (if untagged). `dict` (if tagged). Empty lists are considered tagged.

Custom classes#

In addition, objects containing a “class” attribute are passed to a “constructor function”, if one is available. A dictionary of constructor functions can be supplied to the converter, where the key of each element corresponds to the class name. When the “class” attribute contains several class names, these are tried in order. The default constructor dictionary allows to convert the following R classes:

R class	Python conversion
data.frame	`pandas.DataFrame`.
factor	`pandas.Categorical`.
ordered	`pandas.Categorical` (with ordered categories).
srcfile	`rdata.conversion.SrcFile`.
srcfilecopy	`rdata.conversion.SrcFileCopy`.
srcref	`rdata.conversion.SrcRef`.
ts	`pandas.Series`.