#!/bin/sh

# index -- add ids to headings and print index to stderr

usage() { echo "usage: $0 [-s] [file ...]" 1>&2; exit 1; }

flags=
case "$1" in
-*)	[ x"$1" = x-s ] && { flags=1; shift; } || usage ;;
*)	;;
esac
export flags

source() {
	[ $# -gt 0 ] && cat "$@" || cat
}

source "$@" | awk '
	/<h[0-9]>.*<\/h[0-9]/ {
		heading($0)
		next
	}
	/<h[0-9]>/ {
		open = 1
		s = $0
		next
	}
	open && /<\/h[0-9]>/ {
		open = 0
		s = s "\n" $0
		heading(s)
		next
	}
	open { s = s " " $0; next }
	{ print }

	function heading(s) {
		i = s
		sub("^.*<h[0-9]>", "", i)
		sub("</h[0-9]>.*$", "", i)
		i = toascii(i)
		sub("^<h[0-9]", "& id=\"" i "\"", s)
		print s
		if (!ENVIRON["flags"]) { # if not silent mode
			sub("^<h", "", s)
			sub(" id=\"", "	", s)
			sub("\">", "	", s)
			sub("<\\/h[0-9]>$", "", s)
			print s > "/dev/stderr"
		}
	}

	function toascii(s) {
		s = tolower(s)
		#gsub("[ÀÁÂÃÄÅàáâãäå]", "a", s)
		#gsub("[Ææ]", "ae", s)
		#gsub("[Çç]", "c", s)
		#gsub("[Ðð]", "dh", s)
		#gsub("[ÈÉÊËèéêë]", "e", s)
		#gsub("[ÌÍÎÏìíîï]", "i", s)
		#gsub("[Ññ]", "n", s)
		#gsub("[ÒÓÔÕÖØòóôõöø]", "o", s)
		#gsub("[ß]", "ss", s)
		#gsub("[Þþ]", "th", s)
		#gsub("[ÙÚÛÜùúûü]", "u", s)
		#gsub("[×]", "x", s)
		#gsub("[Ýýÿ]", "y", s)
		gsub("[/]", "-", s)
                gsub("[         \n]+\\+[        \n]+", "+", s)
                gsub("[         \n]+=[  \n]+", "=", s)
                gsub("[         \n]+", "-", s)
		gsub("[^-=+a-z0-9_ 	]", "", s)
		gsub("-\\+-", "+", s)
		gsub("--*", "-", s)
		return s
	}
'