#!/bin/bash
#
# portage-check-files
#
#        -  Ensure that all nominally-installed packages on a portage-based
#           system (e.g. Gentoo) are actually installed.
#           Any missing files will be flagged, and the option to remerge given.
#
# Features: Option of using epm or qpkg (or now even equery)
#
# Initial Version 20031213 - Stuart Shelton [SRCS]
#
# Revision History
# ----------------
# 20031213 SRCS Initial Version (chkpkg)
# 20040405 SRCS Rewrote detection routine to use "read" to fix the spaces problem
# 20040407 SRCS Removed use of pipe from above, due to sync problems
# 20040718 SRCS Renamed to portage-check-files, added logging, added equery support
#
# TODO:
# equery can check MD5 sums, qpkg can list them - we should probably offer to check them
# Command-line parsing is getting a little silly here: Switch to getopt
#
# set -o xtrace
[ -n "$( echo ${*} | grep -E -- "^--debug$|^--debug | --debug$| --debug |-.*d" )" ] && set -o xtrace

NAME="$( echo ${0} | sed "s#^.*/##g" )"
COMMAND=epm
VERBOSE=0
SEARCH=0
NOFILTER=0
NOLOG=0

FIFO=${TMP:-/tmp}/$NAME.fifo
LOG=${TMP:-/tmp}/$NAME.log

function fatalError() {
	[ -e $FIFO ] && rm $FIFO
	echo "$@"
	exit 1
}

function dohelp() {
	echo -e "${NAME}:\n\tCheck that all installed packages for missing files\n"
	echo -e "\tOptions:"
	echo -e "\t--help\t\t\t\t-h\t\t\tDisplay this help text"
	echo -e "\t--debug\t\t\t\t-d\t\t\tPrint debugging information"
	echo -e "\t--verbose\t\t\t-v\t\t\tPrint large amounts of output"
	echo -e "\t--search\t\t\t-s\t\t\tSearch for moved files"
	echo -e "\t--nofilter\t\t\t-f\t\t\tDon't filter out common removals"
	echo -e "\t--nolog\t\t\t\t-l\t\t\tDon't log missing files to $LOG"
	echo -e "\t--command[=]<command_set>\t-c[=]<command_set>\tUse specified command to discover packages"
	echo -e "\n\t... where <command_set> = <epm|qpkg|equery>"
	echo -e "\t    \"epm\" is now the default"
}

# Parse Arguments...
[ -n "$( echo ${*} | grep -Ei -- "^--help$|^--help | --help$| --help |-.*h" )" ] && { dohelp ; exit 0 ; }
[ -n "$( echo ${*} | grep -Ei -- "^--debug$|^--debug | --debug$| --debug |-.*d" )" ] && set -o xtrace
[ -n "$( echo ${*} | grep -Ei -- "^--verbose$|^--verbose | --verbose$| --verbose |-.*v" )" ] && VERBOSE=1
[ -n "$( echo ${*} | grep -Ei -- "^--search$|^--search | --search$| --search |-.*s" )" ] && SEARCH=1
[ -n "$( echo ${*} | grep -Ei -- "^--nofilter$|^--nofilter | --nofilter$| --nofilter |-.*f" )" ] && NOFILTER=1
[ -n "$( echo ${*} | grep -Ei -- "^--nolog$|^--nolog | --nolog$| --nolog |-.*l" )" ] && NOLOG=1
[ -n "$( echo ${*} | grep -Ei -- "^--command[= ]epm$|^--command[= ]epm | --command[= ]epm$| --command[= ]epm |-.*c[= ]epm$|-.*c[= ]epm " )" ] && COMMAND=epm
[ -n "$( echo ${*} | grep -Ei -- "^--command[= ]qpkg$|^--command[= ]qpkg | --command[= ]qpkg$| --command[= ]qpkg |-.*c[= ]qpkg$|-.*c[= ]qpkg " )" ] && COMMAND=qpkg
[ -n "$( echo ${*} | grep -Ei -- "^--command[= ]equery$|^--command[= ]equery | --command[= ]equery$| --command[= ]equery |-.*c[= ]equery$|-.*c[= ]equery " )" ] && COMMAND=equery

LISTCMD="exit 0"
FILECMD="exit 0"
EQUALS=""

case $COMMAND in
	equery)	LISTCMD="equery -q -C list -i \* | sed 's#^.*/##g' | grep -v \"^$\" | cut -d\" \" -f1"
		FILECMD="equery -q -C files"
		EQUALS="="
		;;
	qpkg)	LISTCMD="qpkg -nc -v -I | sed 's#^.*/##g'"
		FILECMD="qpkg -nc -l"
		EQUALS=""
		;;
	epm)	LISTCMD="epm -qa"
		FILECMD="epm -ql"
		EQUALS=""
		;;
	*)	echo "Error: Please specify the command to use with --command"
		exit 1
		;;
esac

if [ -z "$( type -p $COMMAND )" ]; then
	echo "Error: Cannot locate $COMMAND"
	exit 1
fi
(( 1 == VERBOSE )) && echo "Using $( type -p $COMMAND ) to evaluate ebuilds..."

[ -e $FIFO ] && rm -f $FIFO
touch $FIFO
[ -f $FIFO ] || fatalError "Cannot create or read from file $FIFO"

[ -e $LOG ] && rm -f $LOG
touch $LOG
[ -f $LOG ] || fatalError "Cannot create or read from file $LOG"

echo "Building package list..." >&2

# Bulid list of installed packages...
PACKAGES=$( eval $LISTCMD )

echo "Checking ebuild contents..." >&2

(( 1 == VERBOSE )) && echo -e "Package List:\n$PACKAGES\n"

for PACKAGE in $PACKAGES; do

	(( 1 == VERBOSE )) && echo "Processing $PACKAGE:"

	# Now get a list of all of the files installed by this package...
	# NB: Do this in a sub-shell: equery sends EOF :(

	( exec $FILECMD $EQUALS$PACKAGE &>$FIFO ; )

	[ "$COMMAND" = "qpkg" ] && echo $FIFO | cut -d" " -f1-3 | grep -Ev -- "^$|^CONTENTS:$|^.*/$PACKAGE-" >$FIFO

	while read FILE; do

		# Ignore the 2nd half of a symlink
		# (It'd probably make more sense to look at the 2nd half, but this could be relative)
		[ -n "$( echo $FILE | grep -- "->" )" ] && FILE="$( echo $FILE | cut -d" " -f 1 )"

		ls -1d "$FILE" >/dev/null 2>&1
		if (( 0 == ${?} )); then
			(( 1 == VERBOSE )) && echo "Found $FILE"
		else
			FILEMISSING=1
			if (( 0 == NOFILTER )); then
				# The ..'s below are not accidentally unescaped path elements - they should match language files...
				if [ -n "$( echo $FILE | \
				grep -E -- "/\.keep$|man/../man.$|man/.._../man.$|man/../man./[^/]*\..\.gz|man/.._../man./[^/]*\..\.gz|^/usr/src/|^/lib/modules" \
				)" ]; then
					(( 1 == VERBOSE )) && echo "Filtering missing file $FILE"
					FILEMISSING=0
				fi
			fi
			if (( 1 == FILEMISSING )); then
				echo "Error: $PACKAGE missing $FILE"
				(( 0 == NOLOG )) && echo -n "$PACKAGE:$FILE" >> $LOG
				if (( 1 == SEARCH )); then
					FILESRCH="$( echo $FILE | sed "s#^.*/##g" )"
					echo "Finding alternatives:"
					locate $FILESRCH
					RESULT=$?
					if (( 0 == NOLOG )); then
						case $RESULT in
							0)	echo -en "\n" >> $LOG
								;;
							1)	echo " *" >> $LOG
								;;
							*)	echo " !" >> $LOG
								;;
						esac
					fi
					echo
					unset FILESRCH RESULT
				fi
			fi
			unset FILEMISSING
		fi
	done < $FIFO
done

[ -e $FIFO ] && rm -f $FIFO
exit 0