ansible-roles/R/templates/update_r_packages.sh.j2

#!/bin/bash
#
# TODO: kill an old process if it is running from too much time (12 hours?)
#       using something like ps -o etimes= -p "$PROCNUM"
# 

export PATH="$PATH:/bin:/sbin:/usr/bin:/usr/sbin:/usr/local/bin:/usr/local/sbin"

RETVAl=
PARAMS=$#
ACTION=$1
PROCNUM=$$
OLDPROC=
OLDPROC_RUNNING=
LOCKDIR=/var/run
LOCK_FILE=$LOCKDIR/.update_r_pkgs.lock
TMP_FILES_DIR=/var/tmp/r_pkgs_update
# We cannot answer questions
DEBIAN_FRONTEND=noninteractive
R_CRAN_MIRROR={{ r_cran_mirror_site }}
R_PKGS_FROM_SVN={{ r_package_updater_via_subversion }}
R_PKGS_SVN_DIR=RPackagesManagement
R_PKGS_SVN_URL={{ r_package_updater_subversion_repo }}
R_PKGS_SVN_BASE_DIR={{ r_packages_svn_base_dir }}
R_PKGS_FILES_PREFIX={{ r_packages_svn_files_prefix }}
SVN_UPDATE_STATUS=
# In seconds. 60*60*6=21600s (6h)
UPDATER_PROCESS_MAX_RUNTIME=21600
# - debian packages list format:
# one package per line
DEB_PKGS_SKIP=0
DEBIAN_PKGS_LIST_URL={{ r_debian_packages_list_url | default('') }}
PKGS_LIST=
# - R packages list format:
# name[:mirror]
CRAN_PKGS_SKIP=0
R_PKGS_LIST_URL={{ r_cran_packages_list_url | default('') }}
R_PKGS_LIST=
# - R packages from github list format:
# - owner/package
GITHUB_PKGS_SKIP=0
R_PKGS_FROM_GITHUB_LIST_URL={{ r_github_packages_list_url | default('') }}
R_PKGS_GITHUB=

trap "logger 'update_r_packages: trap intercepted, exiting.' ; cleanup" SIGHUP SIGINT SIGTERM

function cleanup() {
    logger "update_r_packages: cleaning up"
    rm -f $LOCK_FILE
    rm -fr $TMP_FILES_DIR
}

function usage() {
    if [ $PARAMS -ne 1 ] ; then
        echo "Need at least an argument: 'upgrade' or 'install'."
        echo "- 'upgrade' installs new packages and upgrades the existin ones when needed."        
        echo "- 'install' installs new packages."        
        cleanup
        exit 1
    fi
}

function get_args() {
    if [ "$ACTION" != "upgrade" -a "$ACTION" != "install" ] ; then
        usage
    fi
}

function fail() {
    logger "update_r_packages: Something went wrong, exiting."
    cleanup
    exit 1
}

function init_env() {
    if [ -f $LOCK_FILE ] ; then
        OLDPROC=$( cat $LOCK_FILE )
        OLDPROC_RUNNING=$( ps auwwx |  grep -v grep | grep $OLDPROC  | awk '{ print $2 }' )
        RETVAL=$?
        if [ ! -z "$OLDPROC_RUNNING" ] ; then
            logger "update_r_packages: pid of the already running process: $OLDPROC_RUNNING"
            OLDPROC_RUNNING_TIME=$( ps -o etimes= -p ${OLDPROC_RUNNING} )
            if [ $OLDPROC_RUNNING_TIME -gt $UPDATER_PROCESS_MAX_RUNTIME ] ; then
                logger "update_r_packages: process $OLDPROC_RUNNING was running for $OLDPROC_RUNNING_TIME seconds. Got stuck, killing it"
                kill -9 $OLDPROC_RUNNING
                cleanup
            else
                logger "update_r_packages: another process is running, exiting."
                exit 0
            fi
        else
            logger "update_r_packages: lock file exist but the process not. Continuing."
            rm -fr $TMP_FILES_DIR
        fi
    else
        logger 'update_r_packages: no other jobs running, proceeding.'
    fi
    RETVAL=
    echo "$PROCNUM" > $LOCK_FILE
    mkdir -p $TMP_FILES_DIR
}

function get_data_files() {
    logger "update_r_packages: get the single files from http."
    # Get the packages list
    if [ -z $DEBIAN_PKGS_LIST_URL ] ; then
        DEB_PKGS_SKIP=1
        logger "update_r_packages: the debian packages list is not available."
    else
        PKGS_LIST=$( mktemp $TMP_FILES_DIR/rdebs.XXXXXXX )
        logger "update_r_packages: getting the debian packages list."
        wget -q -o /dev/null -O $PKGS_LIST $DEBIAN_PKGS_LIST_URL
    fi
    if [ -z $R_PKGS_LIST_URL ] ; then
        CRAN_PKGS_SKIP=1
        logger "update_r_packages: the CRAN packages list is not available."
    else
        R_PKGS_LIST=$( mktemp $TMP_FILES_DIR/rpkgs.XXXXXXX )
        logger "update_r_packages: getting the R packages list that will be installed from CRAN"
        wget -q -o /dev/null -O $R_PKGS_LIST $R_PKGS_LIST_URL
    fi
    if [ -z $R_PKGS_FROM_GITHUB_LIST_URL ] ; then
        GITHUB_PKGS_SKIP=1
        logger "update_r_packages: the Github packages list is not available."
    else
        R_PKGS_GITHUB=$( mktemp $TMP_FILES_DIR/rpkgsgithub.XXXXXXX )
        logger "update_r_packages: getting the R packages list that will be installed from github"
        wget -q -o /dev/null -O $R_PKGS_GITHUB $R_PKGS_FROM_GITHUB_LIST_URL
    fi
}

function get_data_files_from_svn() {
    logger "update_r_packages: files from a SVN repo."
    if [ -d $R_PKGS_SVN_BASE_DIR/$R_PKGS_SVN_DIR ] ; then
        logger "update_r_packages: SVN update"
        cd $R_PKGS_SVN_BASE_DIR/$R_PKGS_SVN_DIR
        SVN_CLEANUP_OP=$( svn cleanup )
        SVN_UPDATE_OP=$( svn update | tail -1 | grep Updated  >/dev/null 2>&1 )
        SVN_UPDATE_STATUS=$?
    else
        cd $R_PKGS_SVN_BASE_DIR
        logger "update_r_packages: first SVN checkout."
        svn co $R_PKGS_SVN_URL >/dev/null 2>&1
    fi
    PKGS_LIST=$R_PKGS_SVN_BASE_DIR/$R_PKGS_SVN_DIR/${R_PKGS_FILES_PREFIX}r_deb_pkgs.txt
    R_PKGS_LIST=$R_PKGS_SVN_BASE_DIR/$R_PKGS_SVN_DIR/${R_PKGS_FILES_PREFIX}r_cran_pkgs.txt
    R_PKGS_GITHUB=$R_PKGS_SVN_BASE_DIR/$R_PKGS_SVN_DIR/${R_PKGS_FILES_PREFIX}r_github_pkgs.txt
}

function debian_pkgs() {
    if [ $DEB_PKGS_SKIP -eq 0 ] ; then
        # Update the apt cache and install the packages in non interactive mode
        logger "update_r_packages: Installing the debian dependencies"
        if [ -z "$(find /var/cache/apt/pkgcache.bin -mmin -360)" ]; then
            apt-get update -q >/dev/null 2>&1
        else
            logger "update_r_packages: APT cache not updated"
        fi
        >/var/log/update_r_debs.log
        while read deb_pkg ; do
            apt-get install ${deb_pkg} -q -y -o Dpkg::Options::="--force-confdef" -o Dpkg::Options::="--force-confold" >>/var/log/update_r_debs.log 2>&1
        done < $PKGS_LIST
        apt-get autoremove -y -o Dpkg::Options::="--force-confdef" -o Dpkg::Options::="--force-confold" >> /var/log/update_r_debs.log 2>&1
    else
        logger "update_r_packages: skipping the debian packages installation"
    fi
}

function remove_r_install_packages_lock_files() {
    # install.packages leaves lock files around if the process crashes
    rm -fr {{ r_sitelib_path }}/00LOCK-*
    }

function r_cran_pkgs() {
    if [ $CRAN_PKGS_SKIP -eq 0 ] ; then
        logger "update_r_packages: Installing R packages from CRAN"
        for l in $( cat $R_PKGS_LIST ) ; do
            pkg=$( echo $l | cut -d : -f 1 )
            is_mirror_ret=
            is_mirror=$( echo $l | grep ':' )
            is_mirror_ret=$?
            if [ $is_mirror_ret -eq 0 ] ; then
                mirror=$( echo $l | cut -d : -f 2- )
            else
                mirror=$R_CRAN_MIRROR
            fi
            if [ "$ACTION" == "upgrade" ] ; then
                Rscript --slave --no-save --no-restore-history -e "install.packages(pkgs='$pkg', repos=c('$mirror/'));"
            else
                Rscript --slave --no-save --no-restore-history -e "if (! ('$pkg' %in% installed.packages()[,'Package'])) { install.packages(pkgs='$pkg', repos=c('$mirror/')); }"
            fi
        done
    else
        logger "update_r_packages: skipping the R CRAN packages installation"
    fi
}

function r_github_pkgs() {
    if [ $GITHUB_PKGS_SKIP -eq 0 ] ; then
        logger "update_r_packages: Installing R packages from Github"
        for l in $( cat $R_PKGS_GITHUB ) ; do
            pkg=$( echo $l | cut -d "/" -f 2 )
            if [ "$ACTION" == "upgrade" ] ; then
                #Rscript --slave --no-save --no-restore-history -e "require(devtools); require(methods); install_github('$l');"
                Rscript --slave --no-save --no-restore-history -e "require(devtools); require(methods); require(jsonlite) ; package_to_install <- '$l' ; refs <- jsonlite::read_json(sprintf('https://api.github.com/repos/%s/releases', package_to_install)) ; ref_to_install <- 'master'; if(length(refs)>0) { ref_to_install <- refs[[1]][['tag_name']] } ; devtools::install_github(package_to_install, ref = ref_to_install)"

            else
                #Rscript --slave --no-save --no-restore-history -e "if (! ('$pkg' %in% installed.packages()[,'Package'])) { require(devtools); require(methods) ; install_github('$l'); }"
                Rscript --slave --no-save --no-restore-history -e "if (! ('$pkg' %in% installed.packages()[,'Package'])) { require(devtools); require(methods); require(jsonlite) ; package_to_install <- '$l' ; refs <- jsonlite::read_json(sprintf('https://api.github.com/repos/%s/releases', package_to_install)) ; ref_to_install <- 'master'; if(length(refs)>0) { ref_to_install <- refs[[1]][['tag_name']] } ; devtools::install_github(package_to_install, ref = ref_to_install) }"
            fi
        done
    else
        logger "update_r_packages: skipping the R GitHub packages installation"
    fi
}

#########
# Main
#

usage
get_args
init_env
if [ $R_PKGS_FROM_SVN == 'True' ] ; then
    get_data_files_from_svn
    if [ $SVN_UPDATE_STATUS -ne 0 -a "$ACTION" == "install" ] ; then
        logger "update_r_packages: nothing new to install from SVN, exiting"
        cleanup
        exit 0
    fi
else
    get_data_files
fi
debian_pkgs
remove_r_install_packages_lock_files
r_cran_pkgs
r_github_pkgs
cleanup
exit 0