Dear useRs, Is it possible to get the R package usage statistics? That is, does R contain any tools to estimate which packages were used and how often? I am going to temporary change the workplace and packing the data and their processing scripts on my computer in order to continue my projects. During my work on the current workplace I periodically have had installed new R packages, have investigated them and used them in my work or did not used them, depending on their functionality. Now I am thinking about writing an R script which will automatically download and install everything I need from the R repository. So, I need a list of packages I have used in R. The first solution in my head is to scan all disks for R scripts and .Rhistory files, extract calls for "library" from them and save names of loaded packages. I would appreciate other variants. --- Best regards, Vladimir mailto:wl at eimb.ru
Here is the perl script with some comments
<pre>
#!/bin/perl -w
use File::Find;
# we use the standard Perl module.
# its procedure will scan the directory tree and put all package names to the
hash
# along with counting the number of their loadings.
%pkgs=("base"=>-1, # won't print packages
installed by default
"datasets"=>-1,
"grDevices"=>-1,
"graphics"=>-1,
"grid"=>-1,
"methods"=>-1,
"splines"=>-1,
"stats"=>-1,
"stats4"=>-1,
"tcltk"=>-1,
"tools"=>-1,
"utils"=>-1,
"MASS"=>-1
);
sub wanted { # this subroutine is used by the File::Find
procedure
# it adds package names to the hash above
return if($_!~/\.[Rr]$/ && $_!~/\.[Rr]history$/); # do nothing if
this file
doesn't contain R commands
open IN,"< ".$File::Find::name or die("cannot open file
$!");
while(<IN>){
if(/library\((.*)\)/){ # looking for
library(...) calls
$pkgname=$1;
next if(! -d "C:\\Program Files\\R\\library\\$pkgname"); #
don't do
anything if the package directory doesn't exist
# simple
protection against typos
if(exists $pkgs{$pkgname}) {
$pkgs{$pkgname}=$pkgs{$pkgname}+1; # here we assume that
basic packages are not loaded
}else{ # with
"library()"
$pkgs{$pkgname}=1;
}
}
}
close(IN);
}
sub getdepends { # this subroutine resolves the package dependencies
$pkgname=$_[0]; # its argument is a package name. It finds the packages
the current one depends on
# and adds them to the hash above
open IN, "< C:\\Program Files\\R\\library\\$pkgname\\DESCRIPTION"
or return;
#do {print ("cannot open file C:\\Program
Files\\R\\library\\$pkgname\\DESCRIPTION\n $!");
while(<IN>){
if($_=~/^Imports: (.*)/ || $_=~/^Depends: (.*)/) {
@deplist=split(/,/,$1);
for(@deplist) {
next if(/R \(.*\)/); # exclude dependencies on R version
s/\s//g;
if(/(.*)\(.*\)/) {
$pkgname=$1;
}else{
$pkgname=$_;
}
if(exists $pkgs{$pkgname}) {
$pkgs{$pkgname}=$pkgs{$pkgname}+1 if($pkgs{$pkgname}>0); #
don't add
basic packages
}else{
$pkgs{$pkgname}=1;
}
}
}
}
close(IN);
}
# now the main loop. hope, it is self-describing
print "Searching for R commands...";
find({ wanted => \&wanted, no_chdir => 1 }, '.');
print "done!\n";
print "Now resolving dependencies...";
for $p (keys %pkgs) {
#print "$p\n";
getdepends($p);
}
print "done!\n";
open OUT,"> install.pkgs.r" or die("cannot create file
install.pkgs.r");
print OUT "install.packages(\n";
foreach(keys %pkgs){
print OUT " $_,\n" if($pkgs{$_}>0);
}
print OUT " ask=FALSE)\n";
close(OUT);
</pre>
You could just check what the last time is that the file system was accessed on each package listing out the packages in order of last time accessed. From the Windows console: cd \Program Files\R\R-2.3.1pat\library dir /od or the analogous cd and ls -lt commands on UNIX On 9/26/06, Vladimir Eremeev <wl at eimb.ru> wrote:> Dear useRs, > > Is it possible to get the R package usage statistics? > That is, does R contain any tools to estimate which packages were > used and how often? > > I am going to temporary change the workplace and packing the data > and their processing scripts on my computer in order to continue my > projects. > > During my work on the current workplace I periodically have had installed > new R packages, have investigated them and used them in my work or did > not used them, depending on their functionality. > > Now I am thinking about writing an R script which will automatically > download and install everything I need from the R repository. > So, I need a list of packages I have used in R. > > The first solution in my head is to scan all disks for R > scripts and .Rhistory files, extract calls for "library" from them > and save names of loaded packages. > > I would appreciate other variants. > > > --- > Best regards, > Vladimir mailto:wl at eimb.ru > > ______________________________________________ > R-help at stat.math.ethz.ch mailing list > https://stat.ethz.ch/mailman/listinfo/r-help > PLEASE do read the posting guide http://www.R-project.org/posting-guide.html > and provide commented, minimal, self-contained, reproducible code. >