preinstall_cleanup.sh

#!/bin/bash
# ======================================================================
#
# CLEANUP BEFORE DEPLOYING A TGZ ON A SERVER
#
# ----------------------------------------------------------------------
#
# This script compares the content of a tgz with the data in
# the webroot. Additional files will be deleted
#
# SYNTAX:
# with 2 parameters it show infos only
# preinstall_cleanup.sh /var/www/preview.mmmu.iml.unibe.ch/ /var/puppet-tgz/mmmu.tgz
#
# with (any) 3rd parameter deletion will be activated
# preinstall_cleanup.sh /var/www/preview.mmmu.iml.unibe.ch/ /var/puppet-tgz/mmmu.tgz forcedelete
#
# ----------------------------------------------------------------------
# 2013-11-06  axel.hahn@iml.unibe.ch
# 2022-11-25  axel.hahn@unibe.ch      do not check json in target dir; shellfixes
# ======================================================================

# ----------------------------------------------------------------------
# CONFIG
# ----------------------------------------------------------------------

        prjdir=$1
        tgz=$2
        bForceDelete=$3
        filelist=/tmp/keepdirslist_$$
        filetgz=/tmp/list_in_tgz_$$
        fileprj=/tmp/list_in_webroot_$$
        keepfile=".htkeep"


# ----------------------------------------------------------------------
# HEADER
# ----------------------------------------------------------------------

        echo
        echo ===== CLEANUP BEFORE DEPLOYING A TGZ =====
        echo

        if [ $# -lt 2 ]; then
                echo "Syntax is"
                echo "$( basename $0 ) [webroot] [tgz-archive] [[deleteflag]]"
                exit 1
        fi

# ----------------------------------------------------------------------
# CHECKS
# ----------------------------------------------------------------------

        # check $prjdir to prevent deletetion of a complete server
        # the directory must exist, must have a subdir "public_html" or
        # "public" and a part must be "/www/" (i.e. /var/www/)
        cd $prjdir || exit 2

        # see #2235 - for nodejs apps the target dir is different
        # ls -ld $prjdir/public_html/ || ls -ld $prjdir/public/ || exit 2
        # echo $prjdir | fgrep "/www/" || exit 2

        # skip this - aum spider has no json
        # ls -l $prjdir/*.json || exit 2

        # check tgz
        file $tgz | grep -F "gzip compressed" || exit 3

# ----------------------------------------------------------------------
# MAIN
# ----------------------------------------------------------------------


        echo --- scanning for $keepfile in $prjdir ...
        cd $prjdir
        find . -type f -name $keepfile | sort | sed "s#/$keepfile##g" >$filelist
        ls -l $filelist
        cat $filelist
        echo

        echo --- scanning archive $tgz
        # tar -tzf $tgz | sed 's#/$##g' | sort >$filetgz
        tar -tzf $tgz | sed 's#/$##g' | sed "s#^#./#g" | sed "s#^\./\./#./#g" | sort >$filetgz
        ls -l $filetgz
        wc -l $filetgz
        head $filetgz
        echo

        echo --- scanning web $prjdir
        cd $prjdir
        find . -print | sort >$fileprj
        ls -l $fileprj
        wc -l $fileprj
        head $fileprj
        echo

        echo --- compare ...
        # diff $filetgz $fileprj | grep "^>" | cut -f 2 -d " " | while read delfile
        diff $filetgz $fileprj | grep "^>" | cut -f 2 -d " " | grep -vf $filelist | while read delfile
        do
                bDelete=1
                testfile=$delfile
                if [ ! -d $delfile ]; then
                        testfile=$( dirname $delfile )
                fi
                echo $testfile | grep -F -f $filelist >/dev/null
                if [ $? -eq 0 ]; then
                        echo KEEP: $delfile
                else
                        if [ -z $bForceDelete ]; then
                                echo WILL-BE-DELETED: $delfile
                        else
                                echo DELETE: $delfile
                                rm -rf $delfile
                        fi
                fi
        done

# ----------------------------------------------------------------------
# CLEANUP
# ----------------------------------------------------------------------

        rm -f $filelist $filetgz $fileprj
        echo done.