#!/bin/sh -
#
#   Shell script for FreeBSD (requires fetch, lynx or mget).
#
#   It downloads the current directory of HTML docs for NetPBM
#   and converts them to text files suitable as "cat" pages.
#   They're stored in the current directory.  Therefore,
#   please run this script in a separate directory that's not
#   writable for anybody else.  Also, there's no reason to run
#   this script as root.
#
#   As an admin, jst move the files to /usr/local/man/cat1.
#   
#   As a normal user, create a directory man/cat1 in your
#   home directory and move the files there (you might have
#   to add $HOME/man to your MANPATH environment variable
#   so the man command finds them).
#
#   Copyright (C) 2006-2007 by Oliver Fromme, Munich, Germany
#   All rights reserved.  Contact:  olli@fromme.com
#   Standard 2-clause BSD license and disclaimer applies.
#

set -Cefu
unset LC_CTYPE LC_ALL LANG

COMPRESS=true

HEAD="NetPBM Manual"
BASE_URL="http://netpbm.sourceforge.net/doc"

FETCH="fetch -q -o -"
# FETCH="lynx -source"
# FETCH="mget"

rm -f tmp.html

Tagline()
{
	CENTER="$1"
	TAG="$2"
	SPACE=$(( (79 - ${#CENTER}) / 2 ))
	printf '%-*s%s%*s\n' $SPACE "$TAG" "$CENTER" $SPACE "$TAG"
}

$FETCH "$BASE_URL"/directory.html \
| tr '<\n' '\n ' \
| sed -n 's/^[]*[Aa][ 	][ 	]*[Hh][Rr][Ee][Ff]=["'\'']*//p' \
| sed 's/["'\''> 	].*$//' \
| sort -u \
| while read PAGE; do
	case "$PAGE" in
		'#'*)	continue ;;
	esac
	NAME="${PAGE%.html}"
	if [ -f "$NAME".1 -o -f "$NAME".1.gz ]; then
		echo "File ${NAME}.1[.gz] already exists, skipping!" >&2
		continue
	fi
	UPNAME=$(echo "$NAME" | tr '[a-z]' '[A-Z]')
	if ! $FETCH "$BASE_URL/$PAGE" > tmp.html ; then
		echo "Cannot fetch $PAGE, skipping!" >&2
		rm -f tmp.html
		continue
	fi

	DATE=$(
		egrep '(Updated|Created):' tmp.html \
		| head -1 \
		| sed 's/^.*[UC][pr][de]ated:[ 	]*//;s/<.*$//'
	)

	(
		Tagline "$HEAD" "${UPNAME}(1)"

		if grep -q '[Hh][Rr][Ee][Ff].*Table Of Contents' tmp.html; then
			STOP="Table Of Contents"
		else
			STOP="<[Bb][Rr]>"
		fi

		sed '/<[Hh][Rr]>/,$d' tmp.html \
		| sed '/<[Hh]1>/,/'"$STOP"'/d' \
		| lynx -dump -nolist -force_html -with_backspaces /dev/stdin

		echo ""
		Tagline "$DATE" "NetPBM"
	) \
	| unexpand -a \
	| cat -s > "$NAME".1

	if $COMPRESS; then
		gzip -9 "$NAME".1
		ls -l "$NAME".1.gz
	else
		ls -l "$NAME".1
	fi

	rm tmp.html
done

#--