ssb-wrapr-python's Introduction

`ssb-wrapr-python`

wrapr is a python package for using R inside of python. It is built using rpy2, but attempts to be more convient to use. Ideally you should never have to worry about using R objects, instead treating R functions as normal python functions, where the inputs and outputs are python objects.

import wrapr as wr
import pandas as pd
import numpy as np
import pytest


dplyr = wr.library("dplyr")
dt = wr.library("datasets")

dplyr.last(x=np.array([1, 2, 3, 4]))
dplyr.last(x=[1, 2, 3, 4])


iris = dt.iris
df = dplyr.mutate(iris, Sepal = wr.lazily("round(Sepal.Length * 2, 0)"))

To do:

1. Port all test files for SSB-GaussSuppression, and SSBtools
2. Better conversion handling for output
    - Convert ordered dictionaries (which are vectors) to numpy arrays
    - Convert ordered dictionaries (which are lists) to dictionaries
    - Convert lists (which are vectors) to vectors
    - Convert lists (which are lists) to lists
    - S4 CLASSES ARE TRICKY!
3. Better conversion handling for input!
    - S4 CLASSES ARE TRICKY
    - Look at the Matrix library
        - rpy2-Matrix on GitHub (abandoned repo, but might be helpful)
4. Refactor modules, into seperate files 
    - Look at the Matrix library
        - rpy2-Matrix on GitHub (abandoned repo, but might be helpful)
5. Better warning handling (this will likely be tricky)
    - Sometimes we will get datatypes which are incompatible, 
        e.g., warning accompanied by 
6. Add option to return as R-object?
8. Add autcomplete for un-indexed functions?
9. Load some requested functions of import?
10. Better documentation!
11. Add some datatypes?
    - rlist? 
        - it is just a list with a dict[label: str, index: int]
        - boolean indexing?
    - wrapper for R-objects using a intuitive Python-Api, with an option
        to return as python object!

Recommend Projects