:
##########################################################################
# Shellscript:	geturl - get WWW page specified by URL
# Version    :	0.1 (beta)
# Author     :	Heiner Steven (heiner.steven@odn.de)
# Date       :	20.02.1997
# Category   :	Internet
# SCCS-Id.   :	$Id: geturl,v 1.1.1.1 1999/06/15 19:29:05 heiner Exp $
##########################################################################
# Note
#  o  Needs the non-standard program
#	nc	("net cat")
#     to do the actual work
#  o  There should be a way to see the header
##########################################################################

PN=`basename "$0"`			# Program name
VER='0.2 (beta)'

Header="${TMPDIR:=/tmp}/gu$$"		# Temp. file for HTTP response header
CMD=GET					# HTTP request ("GET" or "HEAD")
verbose=no				# yes/no
printheader=no
check=no

Usage () {
    echo >&2 "$PN - get WWW page specified by URL, $VER (hs '97)
usage: $PN [-cv] url [...]
   -c: only check the url, do not get the contents
   -h: only print header, no contents
   -v: verbose mode

An url has the form
	http:[//HOST][:PORT]PATH
i.e.
	http://panic.bintec.de:80/index.html"
    exit 1
}

Msg () {
    for MsgLine
    do echo "$PN: $MsgLine" >&2
    done
}

Fatal () { Msg "$@"; exit 1; }

# Return the parsed URL in the variables
#	Protocol, Host, Port, Path
ParseUrl () {
    [ $# -ge 1 ] || return 2
    U=$1
    # URL-Format: http://HOST/DIR/.../FILE
    Protocol=`expr "$U" : '\([a-zA-Z0-9][a-zA-Z0-9]*\):.*'`
    [ -n "$Protocol" ] || return 1		# Protocol must be specified
    U=`expr "$U" : "$Protocol:\(.*\)"`		# Remove protocol
    H=`expr "$U" : '//\([^/][^/]*\).*'`		# hostname:portnumber
    if [ -n "$H" ]
    then
	Host=`expr "$H" : '\([^:][^:]*\).*'`
	Port=`expr "$H" : '[^:][^:]*:\([0-9][0-9]*\).*'`
	U=`expr "$U" : "//[^/][^/]*\(.*\)"`	# Remove hostname:portnumber
    fi
    Path="$U"
    return 0
}

# Strip header from HTTP response, and write it to a file.
# A header is terminated by an empty line

StripHeader () {
    # Line terminator may be CR LF (instead of LF)
    cat "$@" | (
	    OIFS="$IFS"; IFS=""
	    while read line
	    do
		echo "$line"
		case "$line" in
		    ""|"
")	break;
		esac
	    done > "$Header"
	    IFS="$OIFS"
	    cat)
    return 0
}

getopt hcv "$@" >/dev/null 2>&1 || Usage
set -- `getopt hcv "$@"`
while [ $# -gt 0 ]
do
    case "$1" in
	-c)	CMD=HEAD; check=yes;;		# only check the file
	-v)	verbose=yes;;
	-h)	CMD=HEAD; printheader=yes;;
	--)	shift; break;;
	-h)	Usage;;
	-*)	Usage;;
	*)	break;;			# First file name
    esac
    shift
done

# Remove temporary files at signal or exit
trap 'rm -f "$Header" >/dev/null 2>&1' 0
trap "exit 2" 1 2 3 15

ExitCode=0
for Url
do
    ParseUrl "$Url" || { Msg "malformed URL: $Url"; continue; }
    [ "${Protocol:-INVALID}" = http ] ||
	{ Msg "unknown protocol \"$Protocol\" (only \"http\" allowed)";
	  continue; }

    # Set default values
    : ${CMD:=GET}
    : ${Host:=127.0.0.1}
    : ${Port:=80}
    : ${Path:=/}
    [ $verbose = yes ] && Msg "getting http://$Host:$Port$Path"

    echo "${CMD:-GET} ${Path:-/} HTTP/1.0
\r
"   | nc ${Host:=127.0.0.1} ${Port:=80} | StripHeader ||
	{ Msg "could not connect to $Host:$Port"; continue; }

    if [ $printheader = yes ]
    then
    	cat "$Header"
    fi

    if [ $check = yes ]
    then
	# Search response header for state line "HTTP/1.0 200 OK"
	sed q "$Header" | grep "[ 	]200[ 	]" >/dev/null 2>&1
	ExitCode=$?
	[ $ExitCode -eq 0 ] && Msg "URL o.k." || Msg "URL not o.k."
    fi
done
exit $ExitCode