splitmime.sh

#!/bin/bash

# a tool to extract data from those MS Winsoze "web-archives" arrived as a ".DOC" file...
# run as ```splitmime.sh < /tmp/where-it-is.doc``` and check for "./C_/" directory.

div=''
typeset -i nl=0
typeset -i ol=0
file_header='yes'
header='no'
file=''
cte=''

qpd() {
	perl -MMIME::QuotedPrint -pe '$_=MIME::QuotedPrint::decode($_);'
}

decode() {
	local cte="$1"
	local name="$2"

	# echo "#####: ($cte) [$name]">/dev/tty

	case "$cte" in
	base64) base64 -d < "${name}.xxx" > "${name}" && rm "${name}.xxx";;
	quoted-printable) qpd < "${name}.xxx" > "${name}" && rm "${name}.xxx";;
	*)	echo "Unsupported '$cte' for '$name'.">/dev/tty; return 1;;
	esac
}

while read line; do
	let nl+=1
	line=$(echo /$line | tr -d '\r' | cut -d/ -f2-)
	# echo "$nl [$line]" > /dev/tty
	if (( nl == 1 )); then
		[[ "/$line" =~ '/MIME-Version:' ]] || { echo "Not a MIME file."; exit 1; }
		echo "$nl: $line" > /dev/tty
		continue
	fi

	if [ -z "$div" ]; then
		if [[ "/$line" =~ '/Content-Type:' ]]; then
			# Content-Type: multipart/related; boundary="----=_NextPart_01D28B89"
			line=$(echo "$line" | tr -s '[ \t]' '_')
			[[ "$line" =~ 'Content-Type:_multipart/related;_boundary=' ]] \
				|| { echo "No boundary."; exit 1; }
			div="--$(echo "$line" | cut -d\" -f2)"
			echo "$nl: Divider=[$div]">/dev/tty
		fi
		continue
	fi

	if [ "/$line" = "/${div}--" ]; then
		echo "$nl: Last divider ($file:$ol)." >/dev/tty
		file_header='no'
		if [ -n "$cte" ] && [ -n "$file" ]; then
			decode "$cte" "$file";
		fi
		file=''; ol=0
		break
	elif [ "/$line" = "/${div}" ]; then
		echo "$nl: Divider ($file:$ol)." >/dev/tty
		file_header='no'
		if [ -n "$file" -a -f "${file}.xxx" ]; then
			decode "$cte" "$file"
		fi
		header='yes'
		file=''; ol=0
		continue
	fi
	[ "$file_header" = 'yes' ] && continue

	if [ "$header" = 'yes' ]; then
		# echo "$nl: [$line]">/dev/tty
		if [ -z "$line" ]; then
			# echo "$nl: header ends.">/dev/tty
			header='no'
		elif [[ "/$line" =~ '/Content-Location:' ]]; then
			file=$(echo "/$line" | tr -d '[ \t]' | cut -d: -f2-)
			if [[ "$file" =~ 'file:///' ]]; then
				file="./$(echo "${file:8}" | tr ':' '_')"
				dirn=$(dirname "$file")
				flnm=$(basename "$file")
				mkdir -p "$dirn" || { echo "Cannot mkdir '$dirn'."; exit 1; }
				echo "$nl: Will save '$flnm' to '$dirn/'">/dev/tty
				touch "${file}.xxx" || { echo "Cannot touch '${file}.xxx'."; exit 1; }
			else
				echo "$nl: Cannot save to '$file'">/dev/tty
				file=''; ol=0;
				break
			fi
		elif [[ "/$line" =~ 'Content-Transfer-Encoding:' ]]; then
			cte=$(echo "/$line" | tr -d '[ \t]' | cut -d: -f2)
			echo "$nl: Content-Transfer-Encoding: [$cte]">/dev/tty
		fi
		continue
	fi
	if [ -n "$file" ]; then
		echo "$line" >> "${file}.xxx"
		let ol+=1
	fi
done

# EOF #