Installating R on WinPython¶

This procedure applys for Winpython (Version of December 2015 and after)¶

1 - Downloading R binary¶

In [11]:

import os
import sys
import io

# downloading R may takes a few minutes (80Mo)
try:
    import urllib.request as urllib2  # Python 3
except:
    import urllib2  # Python 2

# specify R binary and (md5, sha1) hash
# R-3.4.3:
r_url = "https://cran.r-project.org/bin/windows/base/old/3.5.0/R-3.5.0-win.exe"
hashes=("d3f579b3aacfdf45a008df3320cb3615","87cf0f72dcd91ff12627178e2438cb5a8d8a13c0")

# specify target location
# tweak change in recent winpython
tool_base_directory=os.environ["WINPYDIR"]+"\\..\\t\\"
if not os.path.isdir(tool_base_directory):
    tool_base_directory=os.environ["WINPYDIR"]+"\\..\\tools\\"




r_installer = tool_base_directory+os.path.basename(r_url)
os.environ["r_installer"] = r_installer

# Download
g = urllib2.urlopen(r_url) 
with io.open(r_installer, 'wb') as f:
    f.write(g.read())
g.close
g = None

#checking it's there
!dir %r_installer%

 Le volume dans le lecteur C n'a pas de nom.
 Le num‚ro de s‚rie du volume est 98F9-A53D

 R‚pertoire de C:\WinP\bd36\buQt5\winp64-3.6.x.0\t

21/10/2018  12:37        83ÿ351ÿ952 R-3.5.0-win.exe
               1 fichier(s)       83ÿ351ÿ952 octets
               0 R‚p(s)  24ÿ943ÿ108ÿ096 octets libres

2 - checking and Installing R binary in the right place¶

In [14]:

# checking it's the official R
import hashlib
def give_hash(of_file, with_this):
    with io.open(r_installer, 'rb') as f:
        return with_this(f.read()).hexdigest()  
print (" "*12+"MD5"+" "*(32-12-3)+" "+" "*15+"SHA-1"+" "*(40-15-5)+"\n"+"-"*32+" "+"-"*40)
print ("%s %s %s" % (give_hash(r_installer, hashlib.md5) , give_hash(r_installer, hashlib.sha1),r_installer))
if give_hash(r_installer, hashlib.md5) == hashes[0] and give_hash(r_installer, hashlib.sha1) == hashes[1]:
   print("looks good!")
else:
   print("problem ! please check")
   assert give_hash(r_installer, hashlib.md5) == hashes[0]
   assert give_hash(r_installer, hashlib.sha1) == hashes[1]

            MD5                                 SHA-1                    
-------------------------------- ----------------------------------------
d3f579b3aacfdf45a008df3320cb3615 87cf0f72dcd91ff12627178e2438cb5a8d8a13c0 C:\WinP\bd36\buQt5\winp64-3.6.x.0\python-3.6.7.amd64\..\t\R-3.5.0-win.exe
looks good!

In [15]:

# preparing Dos variables
os.environ["R_HOME"] = tool_base_directory+ "R\\" 
os.environ["R_HOMEbin"]=os.environ["R_HOME"] + "bin" 

# for installation we need this
os.environ["tmp_Rbase"]=os.path.join(os.path.split(os.environ["WINPYDIR"])[0]  , 't','R' )  
if 'amd64' in sys.version.lower():
    r_comp ='/COMPONENTS="main,x64,translations'
else:
    r_comp ='/COMPONENTS="main,i386,translations'
os.environ["tmp_R_comp"]=r_comp

In [17]:

# let's install it, if hashes do match
assert give_hash(r_installer, hashlib.md5) == hashes[0]
assert give_hash(r_installer, hashlib.sha1) == hashes[1]
# If you are "USB life style", or multi-winpython
#   ==> CLICK the OPTION "Don't create a StartMenuFolder' <== (when it will show up)

!start cmd /C %r_installer% /DIR=%tmp_Rbase% %tmp_R_comp%

During Installation (if you wan't to move the R installation after)¶

Choose non default option "Yes (customized startup"

then after 3 screens, Select "Don't create a Start Menu Folder"

Un-select "Create a desktop icon"

Un-select "Save version number in registery"

3 - create a R_launcher and install irkernel¶

In [18]:

import os
import sys
import io
# let's create a R launcher  
r_launcher = r"""
@echo off
call %~dp0env.bat
rscript %*
"""
r_launcher_bat = os.environ["WINPYDIR"]+"\\..\\scripts\\R_launcher.bat"

# let's create a R init script
# in manual command line, you can use repos = c('http://irkernel.github.io/', getOption('repos'))
r_initialization = r"""
install.packages(c('repr', 'IRdisplay', 'stringr', 'crayon', 'pbdZMQ', 'devtools'), repos = c('http://cran.rstudio.com/', 'http://cran.rstudio.com/'))
devtools::install_github('IRkernel/IRkernel')
library('pbdZMQ')
library('repr')
library('IRkernel')
library('IRdisplay')
library('crayon')
library('stringr')
IRkernel::installspec()
"""
r_initialization_r = os.path.normpath(os.environ["WINPYDIR"]+"\\..\\scripts\\R_initialization.r")


for i in [(r_launcher,r_launcher_bat), (r_initialization, r_initialization_r)]:
    with io.open(i[1], 'w', encoding = sys.getdefaultencoding() ) as f:
        for line in i[0].splitlines():
            f.write('%s\n' %  line  )

In [19]:

#check what we are going to do 
print ("!start cmd /C %WINPYDIR%\\..\\scripts\\R_launcher.bat --no-restore --no-save  " + r_initialization_r)

!start cmd /C %WINPYDIR%\..\scripts\R_launcher.bat --no-restore --no-save  C:\WinP\bd36\buQt5\winp64-3.6.x.0\scripts\R_initialization.r

In [20]:

# Launch Rkernel setup
os.environ["r_initialization_r"] = r_initialization_r
!start cmd /C %WINPYDIR%\\..\\scripts\\R_launcher.bat --no-restore --no-save  %r_initialization_r%  

In [21]:

# make RKernel a movable installation with the rest of WinPython 
from winpython import utils
base_winpython = os.path.dirname(os.path.normpath(os.environ["WINPYDIR"]))
rkernel_json=(base_winpython+"\\settings\\kernels\\ir\\kernel.json")

# so we get "argv": ["{prefix}/../tools/R/bin/x64/R"
utils.patch_sourcefile(rkernel_json, base_winpython.replace("\\","/"), r'{prefix}/..', silent_mode=False)

4- Install a R package via a IPython Kernel¶

In [1]:

%load_ext rpy2.ipython

#vitals: 'dplyr', 'R.utils', 'nycflights13'
# installation takes 2 minutes
%R install.packages(c('dplyr','R.utils', 'nycflights13'), repos='http://cran.rstudio.com/') 

5- Small demo via R magic¶

In [4]:

!echo %R_HOME%

C:/WinP/bd36/buQt5/winp64-3.6.x.0/t/R

In [2]:

%load_ext rpy2.ipython

The rpy2.ipython extension is already loaded. To reload it, use:
  %reload_ext rpy2.ipython

In [ ]:

# avoid some pandas deprecation warning
import warnings
warnings.filterwarnings("ignore", category=DeprecationWarning)

In [22]:

%%R
library('dplyr')
library('nycflights13') 
write.csv(flights, "flights.csv")

In [24]:

%R head(flights)

Out[24]:

	year	month	day	dep_time	sched_dep_time	dep_delay	arr_time	sched_arr_time	arr_delay	carrier	flight	tailnum	origin	dest	air_time	distance	hour	minute	time_hour
0	2013	1	1	517	515	2.0	830	819	11.0	UA	1545	N14228	EWR	IAH	227.0	1400.0	5.0	15.0	2013-01-01 11:00:00-05:00
1	2013	1	1	533	529	4.0	850	830	20.0	UA	1714	N24211	LGA	IAH	227.0	1416.0	5.0	29.0	2013-01-01 11:00:00-05:00
2	2013	1	1	542	540	2.0	923	850	33.0	AA	1141	N619AA	JFK	MIA	160.0	1089.0	5.0	40.0	2013-01-01 11:00:00-05:00
3	2013	1	1	544	545	-1.0	1004	1022	-18.0	B6	725	N804JB	JFK	BQN	183.0	1576.0	5.0	45.0	2013-01-01 11:00:00-05:00
4	2013	1	1	554	600	-6.0	812	837	-25.0	DL	461	N668DN	LGA	ATL	116.0	762.0	6.0	0.0	2013-01-01 12:00:00-05:00
5	2013	1	1	554	558	-4.0	740	728	12.0	UA	1696	N39463	EWR	ORD	150.0	719.0	5.0	58.0	2013-01-01 11:00:00-05:00

In [18]:

%R airports %>% mutate(dest = faa) %>% semi_join(flights) %>% head

Out[18]:

	faa	name	lat	lon	alt	tz	dst	tzone	dest
1	ABQ	Albuquerque International Sunport	35.040222	-106.609194	5355	-7.0	A	America/Denver	ABQ
2	ACK	Nantucket Mem	41.253053	-70.060181	48	-5.0	A	America/New_York	ACK
3	ALB	Albany Intl	42.748267	-73.801692	285	-5.0	A	America/New_York	ALB
4	ANC	Ted Stevens Anchorage Intl	61.174361	-149.996361	152	-9.0	A	America/Anchorage	ANC
5	ATL	Hartsfield Jackson Atlanta Intl	33.636719	-84.428067	1026	-5.0	A	America/New_York	ATL
6	AUS	Austin Bergstrom Intl	30.194528	-97.669889	542	-6.0	A	America/Chicago	AUS

6 - Installing the very best of R pakages (optional, you will start to get a really big directory)¶

In [19]:

# essentials: 'tidyr', 'shiny', 'ggplot2', 'caret' , 'nnet'  
# remaining of Hadley Wickahm "stack" (https://github.com/rstudio)
%R install.packages(c('tidyr', 'ggplot2',  'shiny','caret' , 'nnet'), repos='https://cran.rstudio.com/') 
%R install.packages(c('knitr', 'purrr', 'readr', 'readxl'), repos='https://cran.rstudio.com/')
%R install.packages(c('rvest', 'lubridate', 'ggvis', 'readr','base64enc'), repos='https://cran.rstudio.com/')

# TRAINING = online training book http://r4ds.had.co.nz/ (or https://github.com/hadley/r4ds)

7 - Relaunch Jupyter Notebook to get a R kernel option¶

launch a new notebook of "R" type, and type in it:

library('dplyr')

library('nycflights13')

head(flights)

9 - To Un-install / Re-install R (or other trouble-shooting)¶

launch winpython\t\R\unins000.exe (was formerly winpython\tools\R\unins000.exe)
delete the directory winpython\t\R (was formerly winpython\tools\R)
re-install

In [ ]: