#!/bin/bash

# browser-cleanup 1.1.8
# by Jon Jensen
# jon@swelter.net

# Removes all Netscape and Mozilla history, cookie, and cache files.
# When run as root, deletes these files from all users' home directories,
# otherwise only for current user's home directory.

# Options:
# -i  interactive delete (except cache, because it has so many files)
# -q  quiet (default)
# -v  verbose
# -vv very verbose
# -l  ignore lock files
# -p  ignore running processes

# History:
# 1999 April 18 v1.0.0 - netscape-cleanup initial version
# 1999 June 7   v1.0.1 - Stopped deleting cache directory (just contents)
#   to avoid annoying startup error message first launch afterwards.
# 1999 Dec 5    v1.0.2 - Start out with a check for running Netscape processes.
# 2001 Feb 4    v1.1.0 - Rename to browser-cleanup. Now also clean out Mozilla
#   folder, genericize utility paths.
# 2001 Feb 5    v1.1.1 - Update message verbosity levels, now quiet by default.
#   Mozilla profiles are somewhat complex; still working on variations.
# 2001 Feb 9    v1.1.2 - Parse out history from Mozilla localstore.rdf if a Perl
#   interpreter is available; otherwise, simply delete it. Mostly you would just
#   lose window positions and sizes, etc., but I want to avoid that where possible.
# 2002 May 4    v1.1.3 - Make .mozilla directory handling smarter, stop trying to
#   process all users unless run as root
# 2002 Oct. 22  v1.1.4 - Clean Phoenix files (same as Mozilla, really)
# 2002 Nov. 5   v1.1.5 - Clean Galeon files (also Mozilla-based)
# 2003 March 5  v1.1.6 - Clean Konqueror & Opera files
# 2003 May 27   v1.1.7 - Clean Mozilla Firebird formhistory.dat and URL.tbl
# 2005 Nov. 5   v1.1.8 - Clean Epiphany files and check for running Firefox

# To do:
# allow multiple options behind a single hyphen (-ql, etc.)

# Unix flatfile database locations
passwdfile=/etc/passwd
shellsfile=/etc/shells

# Unix utility configuration (usually not necessary)
#shopt -s expand_aliases
#alias egrep=/bin/egrep
#alias grep=/bin/grep
#alias head=/usr/bin/head
#alias ls=/bin/ls
#alias perl=/usr/bin/perl
#alias ps=/bin/ps
#alias pwd=/bin/pwd
#alias rm=/bin/rm
## useful for testing:
##alias rm='echo /bin/rm'
#alias sed=/bin/sed
#alias sort=/bin/sort
#alias tail=/usr/bin/tail
#alias uniq=/usr/bin/uniq
#alias whoami=/usr/bin/whoami

# parse command-line options
for arg; do
	case "$arg" in
	-i)  ask="$arg" ;;
	-q)  verbose="" ;;
	-v)  verbose="$arg" ;;
	-vv) verbose="$arg"
	     rmverbose="-v" ;;
	-l)  ignorelock="$arg" ;;
	-p)  ignoreproc="$arg" ;;
	-*)  echo "`head -18 $0 | tail -17 | sed 's/^# //'`"
	     echo
	     exit 0 ;;
	esac
done

# running as root or normal user?
iamuser=`whoami`
if [ $verbose ]; then echo "Running as user $iamuser"; fi
if [ "$iamuser" = "root" ]; then
	psargs="auxww"
else
	psargs="uxww"
fi

# print all running browser processes
processes=`ps $psargs | egrep '(netscape|mozilla|phoenix|galeon|firefox|epiphany)' | grep -v grep | grep -v $0`
if [ -n "$processes" ] && [ -z "$ignoreproc" ]; then
	if [ $verbose ]; then
		echo "The following browser processes appear to be running:"
		echo "$processes"
	fi
	netscape_processes=`echo "$processes" | grep netscape | awk '{print $1}' | sort | uniq`
	mozilla_processes=`echo "$processes" | egrep '(mozilla|phoenix|galeon|firefox|epiphany)' | awk '{print $1}' | sort | uniq`
fi

# get list of all users who have a valid shell
for user in `awk -F : '{print $1}' $passwdfile` ; do
	shell=`grep "^$user:" $passwdfile | awk -F : '{print $7}'`
	if [ -n "$shell" ] && [ -n "`grep $shell $shellsfile`" ]; then
		users="$users $user"
	fi
done

# remember current directory and my script pathname for later
orig_cwd=`pwd`
if [ ${0:0:1} = / ]; then
	self_cmd="$0"
else
	self_cmd="`pwd`/$0"
fi

# now go through each user's directory if root, current user's only otherwise
for user in $users; do
	
	home_dir=`grep "^$user:" $passwdfile | awk -F : '{print $6}'`

	if [ "$iamuser" = "root" ]; then
		if [ "$verbose" = "-vv" ]; then echo "Working on $home_dir for user $user"; fi
	else
		[ "$iamuser" != "$user" ] && continue
	fi

	# Netscape loop
	this_dir="$home_dir/.netscape"
	if [ -d $this_dir ]; then
		cd $this_dir
		if [ -z "$ignorelock" ] && ( [ -e lock ] || [ -L lock ] ); then
			if [ $verbose ]; then echo "Skipping $this_dir because lock file exists"; fi
		else
			if [ -z "$ignoreproc" ]; then
				if [ -n "`echo "$netscape_processes" | grep $user`" ]; then
					if [ $verbose ]; then echo "Skipping $this_dir because process is running"; fi
				else
					if [ $verbose ]; then echo "Removing junk files from $this_dir"; fi
					for del_file in history.dat history.db history.list cookies; do
						if [ -f $del_file ]; then
							rm $rmverbose $ask $del_file
						fi
					done
					for del_dir in cache; do
						if [ -d $del_dir ]; then
							rm -rf $rmverbose $del_dir/*
						fi
					done
				fi
			fi
		fi
	else
		if [ "$verbose" = "-vv" ]; then
			echo "No directory $this_dir"
		fi
	fi

	# remove some Galeon-specific files
	this_dir="$home_dir/.galeon"
	if [ -d $this_dir ]; then
		if [ $verbose ]; then echo "Removing junk files from $this_dir"; fi
		pushd $this_dir >/dev/null \
			&& rm $rmverbose $ask -r history.xml session_crashed.xml host_favicons.keys favicons favicons.keys \
			&& popd >/dev/null
	fi

	# Mozilla loop
	for this_dir in \
		`find $home_dir/.mozilla -mindepth 1 -maxdepth 1 -type d 2>/dev/null` \
		`find $home_dir/.phoenix -mindepth 1 -maxdepth 1 -type d 2>/dev/null` \
		`find $home_dir/.galeon -mindepth 1 -maxdepth 1 -type d 2>/dev/null` \
		`find $home_dir/.gnome2/epiphany -mindepth 1 -maxdepth 1 -type d 2>/dev/null`
	do
		if [ -d $this_dir ]; then
			if [ -z "$ignoreproc" ]; then
				if [ -n "`echo "$mozilla_processes" | grep $user`" -a -z "$ignorelock" ]; then
					if [ $verbose ]; then echo "Skipping $this_dir because process is running"; fi
				else
					cd $this_dir
					profile_dirs=`echo *`
					if [ -d $this_dir/$user ]; then
						profile_dirs="$profile_dirs `echo $this_dir/$user/*`"
					fi
					for profile_dir in $profile_dirs ; do
						if [ -d $profile_dir ]; then
							cd $profile_dir
							if [ $verbose ]; then echo "Removing junk files from `pwd`"; fi
							for del_file in history.dat formhistory.dat URL.tbl history.list old\\ history.dat cookies.txt; do
								if [ -f $del_file ]; then
									rm $rmverbose $ask $del_file
								fi
							done
							for del_dir in Cache; do
								if [ -d $del_dir ]; then
									rm -rf $rmverbose $del_dir/*
								fi
							done
							rdf_file=localstore.rdf
							if [ -f $rdf_file ]; then
								if [ -n "`perl -v`" ]; then
									cp -p $rdf_file $rdf_file.bak && \
									perl -x $self_cmd $rdf_file.bak > $rdf_file && \
									rm -f $rdf_file.bak
								else
									rm $rmverbose $ask $rdf_file
								fi
							fi
							cd ..
						fi
					done
				fi
			fi
		else
			if [ "$verbose" = "-vv" ]; then
				echo "No directory $this_dir"
			fi
		fi
	done

	# Opera loop
	this_dir="$home_dir/.opera"
	if [ -d $this_dir ]; then
		if [ $verbose ]; then echo "Removing junk files from $this_dir"; fi
		pushd $this_dir >/dev/null \
			&& rm $rmverbose $ask global.dat cookies4.dat autopera.win vlink4.dat \
			&& rm -f $rmverbose acpo/* cache4/* \
			&& popd >/dev/null
	fi

	# Konqueror loop
	this_dir="$home_dir/.kde/share"
	if [ -d $this_dir ]; then
		if [ $verbose ]; then echo "Removing junk files from $this_dir"; fi
		pushd $this_dir >/dev/null \
			&& rm $rmverbose $ask apps/konqueror/konq_history apps/kcookiejar/cookies config/konq_history \
			&& popd >/dev/null
	fi

done

cd $orig_cwd

if [ $verbose ]; then echo "All done."; fi

exit 0


# Below is a short Perl program to clean out the Mozilla localstore.rdf file.
# If a Perl interpreter is not in the path, the .rdf file will simply be deleted.

#!/usr/bin/perl -w

# slurp entire file
undef $/;
$_ = <>;

# remove URL bar history
s{[\t ]*<\s*RDF:Seq.*?nc:urlbar-history.*?<\s*/RDF:Seq\s*>[\t ]*\n?}{}sg;

# remove all URL info caching (except internal Mozilla URLs)
s{[\t ]*<\s*RDF:Description\s*about="\s*(?!chrome).*?<\s*/RDF:Description\s*>[\t ]*\n?}{}sg;

# remove all FTP directory cache info
s{[\t ]*<\s*RDF:Description\s*about="\s*.*?directory.xul.*?<\s*/RDF:Description\s*>[\t ]*\n?}{}sg;

print;

__END__