inspect jar and war files, Grails applications and plugins

jinspect

#!/bin/sh

# needs bash >= 3.2 for regex

jarfile=
file=
fileList=
verbose=0
extract=0
list=0
listFilter=
manifestHeaders=

# check whether zipinfo is available
which zipinfo >/dev/null 2>&1
zipInfoAvailable=$?

ALL_HEADERS="--all--"

options="hHvxXoawmISMWjJrgGcCpPlLdD:"

function usage() {
echo "jinspect [-$options] jarfile [filename ...]"
cat <<EOU
Options:
-h show help
-v be more verbose
-X save specified files (including path) to the current directory instead of printing them 
-l list war contents; specify again to include more information
-x list xml files
-o list properties files
-c list class files
-j list java files
-J list jsp files
-S list js files
-g list groovy files
-G list gsp files
-H list html files
-C list css files
-I list image files
-L list files in lib directory
-M list files in META-INF directory
-W list files in WEB-INF directory
-a show WEB-INF/applicationContext.xml
-w show WEB-INF/web.xml
-r show WEB-INF/grails.xml
-m show META-INF/MANIFEST.MF
-p show plugin.xml
-P show *GrailsPlugin.groovy
-d show manifest headers in nicely formatted way; implies -m
-D <name> display named manifest header; can be used multiple times, implies -m
EOU
}

function addFilter() {
	local expr=$1
	if [ "$list" -eq 0 ]; then
		list=$((list + 1))
	fi
	if [ -n "$listFilter" ]; then
		listFilter="$listFilter|$expr"
	else
		listFilter="$expr"
	fi
}

function addFile() {
	local file=$1
	
	# skip duplicates
	for existingFile in $fileList; do
		if [ "$file" = "$existingFile" ]; then
			return 1
		fi
	done

	if [ -n "$fileList" ]; then
		fileList="$fileList $file"
	else
		fileList="$file"
	fi
}

function addManifestHeader() {
	local header="$1"
	
	if [ -n "$manifestHeaders" ]; then
		manifestHeaders="$manifestHeaders $header"
	else
		manifestHeaders="$header"
	fi
	
	addFile "META-INF/MANIFEST.MF"
}

function displayManifestHeader() {
	awk ' BEGIN { ORS=""; indent="  "; headerCount=0 }; {
		pos=index($0, ":")
		if (pos > 0) {
			# split name from contents
			name=substr($0, 1, pos - 1)
			contents=substr($0, pos + 1, length($0))

			# remove leading and trailing whitespace
			sub(/^[ \t]+/, "", contents)
			sub(/[ \t]+$/, "", contents)

			# header name
			if (headerCount > 0) {
				print "\n"
			}
			print name "\n"
			headerCount++
			
			# iterate over contents, which is separated by ','
			# NOTE: some quoted parts may contain ',' as well, e.g. in version range
			inQuotes=0
			newPart=1
			# for each character
			count = split(contents, parts, "")
			for (p=1; p <= count; p++) {
				# new part
				if (newPart > 0) {
					newPart=0
					print indent
				} 
				# quote character
				if (parts[p] == "\"") {
					if (inQuotes > 0) { 
						inQuotes=0
					}
					else {
						inQuotes=1
					} 
					print parts[p] 
				}
				# comma, which is a spearator, when not within quotes
				else if (parts[p] == ",") {
					if (inQuotes) {
						print parts[p]
					}
					else {
						print "\n"
						newPart=1
					}
				}
				# any other character
				else {
					print parts[p]
				}
			}
		}
		else {
			print $0
		}
	}; END { print "\n" }'
}

function showManifestHeaders() {
	local jarfile=$1
	
	shift
	
	# extract MANIFEST.MF, remove \r and rejoin wrapped lines
	local manifestHeaders=`
		unzip -p "$jarfile" "META-INF/MANIFEST.MF" | tr -d '\r' |
				awk '
BEGIN { ORS="" }
/^[^ ]/{
	if ( NR > 1) {
		print "\n"
	}
	# remove trailing whitespace
	sub(/[ \t]+$/, "", $0)
	print $0
}
/^[ ]/{
	# remove leading whitespace
	sub(/^[ \t]+/, "", $0)
	print $0
}
END { print "\n"; }'`

	# print wanted headers
	for header in $@; do
		if [ "$header" = "$ALL_HEADERS" ]; then
			echo "$manifestHeaders" | displayManifestHeader
		else
			echo "$manifestHeaders" | grep "$header" | displayManifestHeader
		fi
	done
}

# parse options
while getopts "$options" OPT $@; do
	case "$OPT" in
	h) usage; exit 0 ;;
	v) verbose=$((verbose + 1));;
	l) list=$((list + 1));;
	x) addFilter '\.xml';;
	X) extract=1;;
	o) addFilter '\.properties';;
	c) addFilter '\.class';;
	C) addFilter '\.css';;
	j) addFilter '\.java';;
	J) addFilter '\.jsp';;
	S) addFilter '\.js';;
	g) addFilter '\.groovy';;
	G) addFilter '\.gsp';;
	H) addFilter '\.html';;
	I) addFilter '\.jpg'; addFilter '\.png'; addFilter '\.gif'; addFilter '\.ico'; addFilter '\.bmp';;
	M) addFilter 'META-INF/';;
	W) addFilter 'WEB-INF/';;
	L) addFilter 'lib/';;
	m) addFile "META-INF/MANIFEST.MF";;
	a) addFile "WEB-INF/applicationContext.xml";;
	w) addFile "WEB-INF/web.xml";;
	r) addFile "WEB-INF/grails.xml";;
	p) addFile "*plugin.xml";;
	P) addFile "*GrailsPlugin.groovy";;
	d) addManifestHeader "$ALL_HEADERS";;
	D) addManifestHeader "$OPTARG";;
	esac
done

optind=$OPTIND
unset OPTIND
if [ "$optind" -gt 0 ]; then
	# remove options
	shift $((optind - 1))
fi

jarfile="$1"

if [ ! -e "$jarfile" ]; then
	echo >&2 "no such file: $jarfile"
	exit 1
fi

shift

for file in "$@"; do
	addFile "$file"
done

if [ -n "$2" ]; then
	# use file name provided on command line
	file=$2
fi

if [ -n "$fileList" ]; then
	index=0
	if [ "$verbose" -gt 0 ]; then
		echo "=== extracting $fileList from $jarfile"
	fi
	for file in $fileList; do
		if [ "$verbose" -gt 0 ]; then
			if [ "$index" -gt 0 ]; then
				echo ""
			fi
			echo "=== $jarfile:$file"
		fi
		if [ "$extract" -gt 0 ]; then
			unzip "$jarfile" "$file"
		elif [[ "$file" =~ \.xml$ ]]; then
			# xml file, prettify
			unzip -p "$jarfile" "$file" | xmllint --format --recover --nowarning -
		elif [ "$file" = "META-INF/MANIFEST.MF" -a -n "$manifestHeaders" ]; then
			showManifestHeaders "$jarfile" $manifestHeaders
		else
			unzip -p "$jarfile" "$file" 
		fi
		index=$((index + 1))
	done
else
	# no file specified, list contents
	if [ "$list" -eq 0 ]; then
		list=$((list + 1))
	fi
fi

if [ "$list" -gt 0 ]; then
	if [ "$verbose" -gt 0 ]; then
		if [ -n "$listFilter" ]; then
			# remove regex escapes
			displayFilter=`echo "$listFilter" | sed 's/\\\\//g'`
			echo "=== listing contents of $jarfile: $displayFilter"
		else
			echo "=== listing contents of $jarfile"
		fi
	fi
	listing=

	# use unzip -l if long listing is wanted or zip info is not available	
	if [ "$list" -gt 1 -o "$zipInfoAvailable" -gt 0 ]; then
		listing=`unzip -l "$jarfile"`
	else
		listing=`zipinfo -1 "$jarfile"`
	fi
	if [ -n "$listing" ]; then
		if [ -n "$listFilter" ]; then
			echo "$listing" | grep -E $listFilter
		else
			echo "$listing"
		fi
	fi
fi