wman

#!/bin/sh
#
# Fetch man pages for selected OS from a man page mirror.
# Copyright © 2004  Emre Sevinc <fz@ileriseviye.org>
# Tweaked by KIVILCIM "Sundance" Hindistan <sundance@fazlamesai.net>
# Lots of changes by Recai Oktaþ <roktas@debian.org>
# Fixes from Wei Dai <x@wei23.net>
#
# Licensed under the GNU General Public License, version 2.
# See the file `http://www.gnu.org/copyleft/gpl.txt'.


# ------------------------------------------------------------------------------
# Script constants
# ------------------------------------------------------------------------------

NAME=wman
THIS=${0##*/}

SYSTEM_CONF=/etc/${NAME}rc
USER_CONF=$HOME/.${NAME}rc


# ------------------------------------------------------------------------------
# Default setting for target OSes
# ------------------------------------------------------------------------------
DEFAULT_OPENBSD='
OSNAME=OpenBSD
HOST=http://www.openbsd.org
CGI=cgi-bin/man.cgi?query=$page&sektion=$sect&format=troff
FORMAT=raw
FAILURE=[Nn]o [Mm]anual [Ee]ntry
'

DEFAULT_FREEBSD='
OSNAME=FreeBSD
HOST=http://www.freebsd.org
CGI=cgi/man.cgi?query=$page&sektion=$sect&format=ascii
FORMAT=raw
FAILURE=^[Ss]orry
'

DEFAULT_NETBSD='
OSNAME=NetBSD
HOST=http://netbsd.gw.com
CGI=cgi-bin/man-cgi?$page+$sect
FORMAT=html
FILTER=DEFAULT_NETBSD_FILTER
FAILURE=[Mm]anual [Pp]age was not found
'
DEFAULT_NETBSD_FILTER() {
	sed '/^-----/,$d'
}

DEFAULT_DRAGONFLY='
OSNAME=Dragonfly BSD
HOST=http://leaf.dragonflybsd.org
CGI=cgi/web-man?command=$page&section=$sect
FORMAT=html
FAILURE=[Mm]anual [Pp]age could not be found
'

DEFAULT_LINUX='
OSNAME=GNU/Linux
HOST=http://www.freebsd.org
CGI=cgi/man.cgi?query=$page&sektion=$sect&manpath=Red+Hat+Linux%2Fi386+9&format=ascii
FORMAT=raw
FAILURE=^[Ss]orry
'

# ------------------------------------------------------------------------------
# Helpers
# ------------------------------------------------------------------------------

# Portable which(1).
pathfind() {
    ifs_save="$IFS"
    IFS=:
    for _p in $PATH; do
        if [ -x "$_p/$*" ] && [ -f "$_p/$*" ]; then
            IFS="$OLDIFS"
            return 0
        fi
    done
    IFS="$ifs_save"
    return 1
}

dumper_raw() {
	for p in wget lynx w3m curl links w3c; do
		if pathfind $p; then
			dumper=$p
			break
		fi
	done
	# Setup proper options.
	case "$dumper" in
	wget)  opt="-O-" ;;
	lynx)  opt="-source" ;;
	w3m)   opt="-dump_source" ;;
	curl)  opt="" ;;
	links) opt="-source" ;;
	w3c)   opt="-n -get" ;;
	"")    printf "Couldn't locate a program to fetch files from net " >&2
	       printf "(e.g. wget, w3m, lynx, w3c, or curl).\n" >&2
	       exit 1 ;;
	esac
	$dumper $opt "$@"
}

dumper_html() {
	# TODO w3c
	for p in w3m links elinks lynx; do
		if pathfind $p; then
			dumper=$p
			break
		fi
	done
	# Setup proper options.
	case "$dumper" in
	w3m)    opt="-dump" ;;
	links)  opt="-dump" ;;
	elinks) opt="-dump" ;;
	lynx)   opt="-dump" ;;
	"")     printf "Couldn't locate a program to fetch files from net " >&2
	        printf "(e.g. lynx, w3m, lynx or w3c).\n" >&2
	        exit 1 ;;
	esac
	$dumper $opt "$@"
}

is_cache_valid() {
	cache="$1"
	failstamp="$2"
	if [ $(sed '/^[[:blank:]]*$/d' $cache | wc -l) -lt 20 ]; then
		[ -z "$DEBUG" ] || echo >&2 "Fetched file is too small!"
		return 1
	fi
	if sed '/^[[:blank:]]*$/d;{3q}' $cache | grep -q "$failstamp"; then
		[ -z "$DEBUG" ] || echo >&2 "Failure stamp '$failstamp' seen!"
		return 1
	fi
	return 0
}


# ------------------------------------------------------------------------------
# Init
# ------------------------------------------------------------------------------

LC_ALL=C
export LC_ALL

for conf in $SYSTEM_CONF $USER_CONF; do
	# Read user conf file.
	if [ -r $conf ]; then
		case $- in
		*e*) ;;
		*)   set -e; noexitonerr=yes ;;
		esac
		trap "Error in configuration file $conf; aborting." QUIT INT
		. $conf
		trap - QUIT INT
		if [ -n "$noexitonerr" ]; then
			set +e
		fi
	fi
done

OS=
DEBUG=${DEBUG-}
CACHE_RENEW=
while getopts o:dhr opt; do
	case $opt in
	o) OS="$OPTARG" ;;
	d) DEBUG=yes ;;
	r) CACHE_RENEW=yes ;;
	h) cat >&2 <<- EOF
		Usage: $THIS [options] [section] page
		Options:
		  -o os_name   operating system (e.g. openbsd, netbsd)
		  -d           print debug info
		  -n           renew cache

		  -h           this screen
	EOF
	   exit 2 ;;
	esac
done
shift $(($OPTIND - 1))

if [ -z "$OS" ] && ! [ "$THIS" = "$NAME" ]; then
	OS=$(echo "$THIS" | sed 's/^man[-_+]//')
else
	DEFAULT_OS=${DEFAULT_OS:-$(uname)}
	OS=${OS:-$DEFAULT_OS}
fi

if [ -z "$OS" ]; then
	echo "Couldn't determine OS; aborting"
	exit 1
fi

OS=$(echo $OS | tr 'a-z' 'A-Z')

# Defaults for some global settings.
PAGER=${PAGER-less}
DEBUG=${DEBUG:-}
CACHEROOT=${CACHEROOT-$HOME/.$THIS}

if [ $# -eq 0 ]; then
	echo >&2 "What manual page do you want?"
	exit 1
fi

sect=
case "$1" in
[1-9])
	sect=$1
	shift

	if [ -z "$1" ]; then
		echo >&2 "What manual page do you want from section ${sect}?"
		exit 1
	fi
	;;
esac

page="$1"

NEWLINE='
'
OLDIFS="$IFS"
_expand_variable() {
	condensed=${1:?'variable empty!'}

	eval "this=\$$condensed"

	# Expand page and sect.
	eval "this=\"$this\""

	IFS="$NEWLINE"
	for line in $this; do
		IFS="$OLDIFS"

		var="${line%%=*}"
		val="${line#[!=]*=}"
		case "$var" in
		'OSNAME'|'HOST'|'CGI'|'FORMAT'|'FILTER'|'FAILURE')
			var="man_$(echo $var | tr 'A-Z' 'a-z')"
			eval "$var=\${$var:-$val}" ;;
		*)
			continue ;;
		esac
	done
}

# User settings.
_expand_variable "$OS"

# Builtin default settings.
_expand_variable "DEFAULT_$OS"

man_url="$man_host/$man_cgi"
man_dumper="dumper_$man_format"
man_osname="${man_osname:-$(echo $OS | tr 'A-Z' 'a-z')}"

[ -z "$DEBUG" ] || cat >&2 <<- EOF
	Fetch parameters:
	    OS:        $man_osname
	    Page:      $page
	    Section:   $sect
	    Host:      $man_host
	    CGI:       $man_cgi
	    Format:    $man_format
	    Dumper:    $man_dumper
	    Filter:    $man_filter
	    Failure:   '$man_failure'

	Resulting url: $man_url

EOF

( eval $man_dumper </dev/null >/dev/null 2>&1 )
if [ $? -eq 127 ]; then
	echo >&2 "No such '$man_format' dumper for '$OS'; aborting"
	exit 1
fi

if [ -n "$man_filter" ]; then
	( eval $man_filter </dev/null >/dev/null 2>&1 )
	if [ $? -eq 127 ]; then
		echo >&2 "No such '$man_filter' for '$OS'; aborting"
		exit 1
	fi
else
	man_filter=cat
fi

# ... now cache variables
cachedir=$CACHEROOT/$(echo $OS | tr 'A-Z' 'a-z')
man_cache=$cachedir/$page.${sect:-0} # use 0 to catch the default man page


# ------------------------------------------------------------------------------
# Main
# ------------------------------------------------------------------------------

if [ -n "$CACHE_RENEW" ] || {
	[ -f $man_cache ] && ! is_cache_valid $man_cache "$man_failure"
}; then
	rm -f $man_cache
fi

what="$man_osname manual entry for $page"
if [ -n "$sect" ]; then
	what="$what in section $sect"
fi

if ! [ -f $man_cache ]; then
	# Initiate cache if it doesn't exist.
	(umask 0077 && mkdir -p $cachedir)

	tmpfile="$(mktemp $cachedir/$THIS.XXXXXXXX)" || exit 1
	# Remove partial files.
	trap 'exitcode=$?
	      [ -z "$tmpfile" ] || rm -rf "$tmpfile"
	      exit $exitcode' 0 1 2 3 13 15

	printf "Querying $what... " >&2
	if $man_dumper "$man_url" 2>/dev/null | sed '/./,$!d' >$tmpfile; then
		if is_cache_valid $tmpfile "$man_failure"; then
			echo "success."
			mv -f $tmpfile $man_cache && tmpfile=
		else
			echo "failed."
		fi >&2
	else
		echo >&2 "fetch failed."
	fi
fi

if ! [ -f $man_cache ]; then
	echo >&2 "No $what."
	exit 1
fi

# Attempt to reduce cache space whenever a section is explicitly given.
# Note that we avoid to extract the section info from cache file, since
# this might be unreliable for some formats and also it would increase
# the code complexity.
if [ "$sect" != "0" ]; then
	man_fallback=$cachedir/$page.0
	if [ -f $man_fallback ] && ! [ -L $man_fallback ]; then
		(
		cd $cachedir
		# Search for an idantic file.
		found=
		for f in $page.[1-9]; do
			if cmp -s $f $man_fallback; then
				found=$f
				break
			fi
		done
		# If found, symlink it to sectioned file to reduce space.
		[ -z "$found" ] || ln -sf $found $man_fallback
		)
	fi
fi

$man_filter <$man_cache | $PAGER