/rcppcnpy

Rcpp bindings for NumPy files

Primary LanguageC++GNU General Public License v2.0GPL-2.0

RcppCNPy: Rcpp bindings for NumPy files

CI License CRAN Dependencies Downloads Last Commit
status DOI

About

This package uses the cnpy library written by Carl Rogers to provide read and write facilities for files created with (or for) the NumPy extension for Python. Vectors and matrices of numeric types can be read or written to and from files as well as compressed files. Support for integer files is available if the package has been built with -std=c++11 which is the default starting with release 0.2.3 following the release of R 3.1.0, and available on all platforms following the release of R 3.3.0 with the updated 'Rtools'.

Example

The following Python code

>>> import numpy as np
>>> fm = np.arange(12).reshape(3,4) * 1.1
>>> fm
array([[  0. ,   1.1,   2.2,   3.3],
       [  4.4,   5.5,   6.6,   7.7],
       [  8.8,   9.9,  11. ,  12.1]])
>>> np.save("fmat.npy", fm)
>>> 
>>> im = np.arange(12).reshape(3,4)
>>> im
array([[ 0,  1,  2,  3],
       [ 4,  5,  6,  7],
       [ 8,  9, 10, 11]])
>>> np.save("imat.npy", im)
>>> 

saves two matrices in floating-point and integer representation.

With this R code we can read and assign the files:

R> library(RcppCNPy)
R> fmat <- npyLoad("fmat.npy")
R> fmat
     [,1] [,2] [,3] [,4]
[1,]  0.0  1.1  2.2  3.3
[2,]  4.4  5.5  6.6  7.7
[3,]  8.8  9.9 11.0 12.1
R> 
R> imat <- npyLoad("imat.npy", "integer")
R> imat
     [,1] [,2] [,3] [,4]
[1,]    0    1    2    3
[2,]    4    5    6    7
[3,]    8    9   10   11
R> 

Going the opposite way by saving in R and reading in Python works equally well. An extension not present in CNPy allows reading and writing of gzip-compressed files.

The package has been tested and used on several architecture, and copes correctly with little-vs-big endian switches.

More details are available in the package vignette.

Installation

The package is on CRAN and can be installed per:

R> install.packages("RcppCNPy")

Status

On CRAN, stable and mostly feature-complete.

Alternative: reticulate

The reticulate package can also provide easy and comprehensive access to NumPy data; see the additional vignette in RcppCNPy for examples and more details.

Feedback

Contributions are welcome, please use the GitHub issue tracker for bug reports, feature requests or general discussions before sending pull requests.

Author

Dirk Eddelbuettel and Wush Wu

License

GPL (>= 2)