geminawk/geminawk.sh

#!/bin/sh
usage() {
	printf 'usage: geminawk [-R] [-h hostname ] ' >&2
	printf '[-p default-port] [-r max-redirects]\n' >&2
	exit 64
}
Rflag=
PORT="${PORT:-1965}"
REDIRECTS="${REDIRECTS:-5}"
while getopts Rh:p:r: opt; do
	case "$opt" in
	R)	Rflag=1;;
	h)	HOST="$OPTARG";;
	p)	PORT="$OPTARG";;
	r)	REDIRECTS="$OPTARG";;
	*)	usage;;
	esac
done
shift $((OPTIND - 1))
# Try the default nc(1) command
if ! [ "$NC" ] && command -v nc >/dev/null &&
    nc -c 2>/dev/null | head -n1 | grep -q '^usage:'; then
	NC=nc
fi
if ! [ "$NC" ]; then
	# Try to find a nc(1) that supports TLS
	IFS=:
	for dir in ${PATH-/usr/local/bin:/usr/bin:/bin}; do
		if [ -e "$dir/nc" ]; then
			case "$("$dir/nc" -c 2>&1 | head -n1)" in
			usage:*)NC="${NC:-$dir/nc}";;
			esac
		fi
	done
	unset IFS
fi
# Default to openssl(1)
NC="${NC:-openssl s_client}"
# openssl(1) needs the `-quiet' option, otherwise
# it mixes its output with the server's response.
case "$NC" in
*s_client*)
	NCFLAGS="${NCFLAGS:--quiet -no_check_time}";;
esac
NCFLAGS="${NCFLAGS:--c -Tnoverify -Tnoname}"
geminawk() {
	awk -vu="$1" -vR="$Rflag" -vh="$HOST" -vp="$PORT" \
	    -vr="$REDIRECTS"  -vnc="$NC $NCFLAGS" '
	function escape(s) {
		gsub(/'\''/, "'\''\\'\'''\''", s)
		return ("'\''" s "'\''")
	}
	function uriescape(s, c) {
		split(s, uriescape_a, "")
		uriescape_cmd = "printf %s " escape(s) \
		    " | od -An -tx1 | tr a-z A-Z"
		uriescape_hex = ""
		while (uriescape_cmd | getline uriescape_line)
			uriescape_hex = uriescape_hex uriescape_line
		close(uriescape_cmd)
		sub(/^[[:space:]]+/, "", uriescape_hex)
		split(uriescape_hex, uriescape_x, /[[:space:]]+/)
		s = ""
		for (uriescape_i = 1; uriescape_i in uriescape_a;
		    uriescape_i++) {
			uriescape_c = uriescape_a[uriescape_i]
			if (uriescape_c ~ /[-.0-9A-Z_a-z~]/ || \
			    index(c, uriescape_c) > 0)
				s = s uriescape_c
			else
				s = s "%" uriescape_x[uriescape_i]
		}
		delete uriescape_a
		delete uriescape_x
		return (s);
	}
	function command(u) {
		url = u
		sub(/^gemini:\/\//, "", url)
		sub(/^\/+/, "", url)
		host = url
		sub(/^[^\/]*@/, "", host) # userinfo
		port = host
		if (host ~ /^\[/) { # IPv6
			sub(/^\[/, "", host)
			sub(/^\].*/, "", host)
		} else # IPv4/reg-name
			sub(/[\/:].*/, "", host)
		if (h !~ /./ && host == "")
			exit(1)
		sub(/^(\[[^\]]*\]|[^:\/]*)/, "", port) # hostname
		if (port ~ /^:/) { # port
			sub(/^:/, "", port)
			sub(/\/.*/, "", port)
			if (h ~ /./ && (port + 0 != port ||
			    port + 0 < 0))
				exit 1
		} else
			port = p
		u = R ? u : "gemini://" url
		ehost = escape(h ~ /./ ? h : host)
		eport = escape(h ~ /./ ? p : port)
		cmd0 = "printf '\''%s\\r\\n'\'' " escape(u)
		if (nc ~ /s_client/)
			cmd1 = nc " -connect " ehost ":" eport
		else
			cmd1 = nc " -- " ehost " " eport
		cmd = cmd0 " | " cmd1 " 2>/dev/null"
		return (cmd)
	}
	BEGIN {
		r += 0
		for (redir = 0; redir < r; redir++) {
			cmd = command(u)
			cmd | getline
			sub(/\r$/, "")
			if ($1 ~ /^1/) { # INPUT
				prompt = ""
				for (i = 2; i <= NF; i++)
					prompt = prompt \
					    (i > 2 ? " " : "") $i
				print(i > 2 ? prompt : "?")
				getline
				sub(/\?.*/, "", u)
				u = u "?" uriescape($0, \
				    "!$&'\''()*+,/:;=?@")
				redir = -1;
			} else if ($1 ~ /^2/) { # OK
				while (cmd | getline)
					print
				exit
			} else if ($1 ~ /^3/) { # REDIRECT
				if (NF > 2)
					exit 1
				u = $2;
				if ($2 ~ /^\/[^\/]/ && h !~ /./)
					u = host $2
			} else { # Error or unsupported
				print
				exit 1
			}
			close(cmd)
		}
	}'
}
for url in "$@"; do
	geminawk "$url"
done