path: root/bollux

#!/usr/bin/env bash
# bollux: a bash gemini client or whatever
# Author: Case Duckworth <acdw@acdw.net>
# License: MIT
# Version: -0.7

# set -euo pipefail              # strict mode

### constants ###
PRGN="${0##*/}"                # program name
DLDR="${BOLLUX_DOWNDIR:=.}"    # where to download
LOGL="${BOLLUX_LOGLEVEL:=3}"   # log level
MAXR="${BOLLUX_MAXREDIR:=5}"   # max redirects
PORT="${BOLLUX_PORT:=1965}"    # port number
PROT="${BOLLUX_PROTO:=gemini}" # protocol
RDRS=0                         # redirects
VRSN=-0.7                      # version number

# shellcheck disable=2120
bollux_usage() {
	cat <<END_USAGE >&2
	$PRGN ($VRSN): a bash gemini client
	usage:
		$PRGN [-h]
		$PRGN [-L LVL] [URL]
	options:
		-h	show this help
		-L LVL	set the loglevel to LVL.
			Default: $BOLLUX_LOGLEVEL
			The loglevel is between 0 and 5, with
			lower levels being more dire.
	parameters:
		URL	the URL to navigate view or download
END_USAGE
	exit "${1:-0}"
}

# LOGLEVELS:
# 0 - application fatal error
# 1 - application warning
# 2 - response error
# 3 - response logging
# 4 - application logging
# 5 - diagnostic

### utility functions ###
# a better echo
put() { printf '%s\n' "$*"; }

# conditionally log events to stderr
# lower = more important
log() { # log [LEVEL] [<] MESSAGE
	case "$1" in
	-)
		lvl="-1"
		shift
		;;
	[0-5])
		lvl="$1"
		shift
		;;
	*) lvl=4 ;;
	esac

	output="$*"
	if ((lvl < LOGL)); then
		if (($# == 0)); then
			while IFS= read -r line; do
				output="$output${output:+$'\n'}$line"
			done
		fi
		printf '\e[34m%s\e[0m:\t%s\n' "$PRGN" "$output" >&2
	fi
}

# halt and catch fire
die() { # die [EXIT-CODE] MESSAGE
	case "$1" in
	[0-9]*)
		ec="$1"
		shift
		;;
	*) ec=1 ;;
	esac

	log 0 "$*"
	exit "$ec"
}

# ask the user for input
ask() { # ask PROMPT [READ_OPT...]
	prompt="$1"
	shift
	read </dev/tty -e -r -p "$prompt> " "$@"
}

# fail if something isn't installed
require() { hash "$1" 2>/dev/null || die 127 "Requirement '$1' not found."; }

# trim a string
trim() { sed -e 's/^[[:space:]]*//' -e 's/[[:space:]]*$//'; }

# stubs for when things aren't implemented (fully)
NOT_IMPLEMENTED() { die 200 "NOT IMPLEMENTED!!!"; }
NOT_FULLY_IMPLEMENTED() { log 1 "NOT FULLY IMPLEMENTED!!!"; }

### gemini ###
# url functions
# normalize a path from /../ /./ /
normalize_path() { # normalize_path <<< PATH
	gawk '{
	if ($0 == "" || $0 ~ /^\/\/[^\/]/) {
		return -1
	}
	split($0, path, /\//)
	for (c in path) {
		if (path[c] == "" || path[c] == ".") {
			continue
		}
		if (path[c] == "..") {
			sub(/[^\/]+$/, "", ret)
			continue
		}
		if (! ret || match(ret, /\/$/)) {
			slash = ""
		} else {
			slash = "/"
		}
		ret = ret slash path[c]
	}
	print ret
	}'
}

# split a url into the URL array
split_url() {
	gawk '{
	if (match($0, /^[A-Za-z]+:/)) {
		arr["scheme"] = substr($0, RSTART, RLENGTH)
		$0 = substr($0, RLENGTH + 1)
	}
	if (match($0, /^\/\/[^\/?#]+?/) || (match($0, /^[^\/?#]+?/) && scheme)) {
		arr["authority"] = substr($0, RSTART, RLENGTH)
		$0 = substr($0, RLENGTH + 1)
	}
	if (match($0, /^\/?[^?#]+/)) {
		arr["path"] = substr($0, RSTART, RLENGTH)
		$0 = substr($0, RLENGTH + 1)
	}
	if (match($0, /^\?[^#]+/)) {
		arr["query"] = substr($0, RSTART, RLENGTH)
		$0 = substr($0, RLENGTH + 1)
	}
	if (match($0, /^#.*/)) {
		arr["fragment"] = substr($0, RSTART, RLENGTH)
		$0 = substr($0, RLENGTH + 1)
	}
	for (part in arr) {
		printf "URL[\"%s\"]=\"%s\"\n", part, arr[part]
	}
	}'
}

# example.com => gemini://example.com/
_address() { # _address URL
	addr="$1"

	[[ "$addr" != *://* ]] && addr="$PROT://$addr"
	trim <<<"$addr"
}

# return only the server part from an address, with the port added
# gemini://example.com/path/to/file => example.com:1965
_server() {
	serv="$(_address "$1")" # normalize first
	serv="${serv#*://}"
	serv="${serv%%/*}"
	if [[ "$serv" != *:* ]]; then
		serv="$serv:$PORT"
	fi
	trim <<<"$serv"
}

# request a gemini page
# by default, extract the server from the url
request() { # request [-s SERVER] URL
	case "$1" in
	-s)
		serv="$(_server "$2")"
		addr="$(_address "$3")"
		;;
	*)
		serv="$(_server "$1")"
		addr="$(_address "$1")"
		;;
	esac

	log 5 "serv: $serv"
	log 5 "addr: $addr"

	sslcmd=(openssl s_client -crlf -ign_eof -quiet -connect "$serv")
	# use SNI
	sslcmd+=(-servername "${serv%:*}")
	log "${sslcmd[@]}"
	"${sslcmd[@]}" <<<"$addr" 2>/dev/null
}

# handle the response
# cf. gemini://gemini.circumlunar.space/docs/spec-spec.txt
handle() { # handle URL < RESPONSE
	URL="$1"
	while read -d $'\r' -r head; do
		break # wait to read the first line
	done
	code="$(gawk '{print $1}' <<<"$head")"
	meta="$(gawk '{for(i=2;i<=NF;i++)printf "%s ",$i;printf "\n"}' <<<"$head")"

	log 5 "[$code]	$meta"

	case "$code" in
	1*) # INPUT
		log 3 "Input"
		RDRS=0 # this is not a redirect
		ask "$meta" QUERY
		bollux "$URL?$QUERY"
		;;
	2*) # SUCCESS
		log 3 "Success"
		RDRS=0 # this is not a redirect
		case "$code" in
		20) log 5 "- OK" ;;
		21) log 5 "- End of client certificate session" ;;
		*) log 2 "- Unknown response code: '$code'." ;;
		esac
		display "$meta"
		;;
	3*) # REDIRECT
		log 3 "Redirecting"
		case "$code" in
		30) log 5 "- Temporary" ;;
		31) log 5 "- Permanent" ;;
		*) log 2 "- Unknown response code: '$code'." ;;
		esac
		((RDRS += 1))
		((RDRS > MAXR)) && die "$code" "Too many redirects!"
		bollux "$meta"
		;;
	4*) # TEMPORARY FAILURE
		log 2 "Temporary failure"
		RDRS=0 # this is not a redirect
		case "$code" in
		41) log 5 "- Server unavailable" ;;
		42) log 5 "- CGI error" ;;
		43) log 5 "- Proxy error" ;;
		44) log 5 "- Rate limited" ;;
		*) log 2 "- Unknown response code: '$code'." ;;
		esac
		exit "$code"
		;;
	5*) # PERMANENT FAILURE
		log 2 "Permanent failure"
		RDRS=0 # this is not a redirect
		case "$code" in
		51) log 5 "- Not found" ;;
		52) log 5 "- No longer available" ;;
		53) log 5 "- Proxy request refused" ;;
		59) log 5 "- Bad request" ;;
		*) log 2 "- Unknown response code: '$code'." ;;
		esac
		exit "$code"
		;;
	6*) # CLIENT CERT REQUIRED
		log 2 "Client certificate required"
		RDRS=0 # this is not a redirect
		case "$code" in
		61) log 5 "- Transient cert requested" ;;
		62) log 5 "- Authorized cert required" ;;
		63) log 5 "- Cert not accepted" ;;
		64) log 5 "- Future cert rejected" ;;
		65) log 5 "- Expired cert rejected" ;;
		*) log 2 "- Unknown response code: '$code'." ;;
		esac
		exit "$code"
		;;
	*) # ???
		die "$code" "Unknown response code: '$code'."
		;;
	esac
}

# display the page
display() { # display META < DOCUMENT
	case "$1" in
	*\;*)
		mime="$(cut -d\; -f1 <<<"$1" | trim)"
		charset="$(cut -d\; -f2 <<<"$1" | trim)"
		;;
	*) mime="$(trim <<<"$1")" ;;
	esac
	[ -z "$mime" ] && mime="text/gemini"
	if [ -z "$charset" ]; then
		charset="utf-8"
	else
		charset="${charset#*=}"
	fi

	log 5 "mime=$mime; charset=$charset"

	case "$mime" in
	text/gemini)
		lc="/tmp/bollux-currentpage.gmi" # link copy
		lfn | typeset_gemini | tee "$lc" | less -R ||
			cat "$lc" # TODO list out links on success
		# lesskey:
		# o #> Open a link (quit 1)
		# q #> Quit (quit 0)
		# cf. also prompt & filename
		;;
	text/*) lfn ;;
	*) download "$URL" ;;
	esac
}

# normalize line endings to \n (LF)
lfn() {
	gawk 'BEGIN{RS="\n\n"}{gsub(/\r\n?/,"\n");print;print ""}'
}

# typeset text
typeset_gemini() { # typeset_gemini < INPUT
	gawk '
	BEGIN { pre = 0 }
	/^###/ { sub(/^#+[[:space:]]*/, ""); 
		printf " \033[3m%s\033[0m\n", $0
	next }
	/^##/  { sub(/^#+[[:space:]]*/, ""); 
		printf " \033[1m%s\033[0m\n", $0
	next }
	/^#/   { sub(/^#+[[:space:]]*/, ""); 
		printf " \033[1;4m%s\033[0m\n", $0
	next }
	/^=>/  { 
		sub(/=>[[:space:]]*/, "")
		url = $1; desc = ""
		for (w=2;w<=NF;w++) 
			desc = desc (desc?" ":"") $w
		printf " \033[1m[%s]\033[0m \033[4m%s\033[0m \033[36m%s\033[0m\n", 
			(++ln), desc, "(" url ")"
	next }
	# /^\*/  { sub(/\*[[:space:]]*/, ""); }
	/```/  { pre = !pre; next }
	{ sub(/^/, " "); print }
	'
}

download() { # download URL < FILE
	tn="$(mktemp)"
	dd status=progress >"$tn"
	fn="$DLDR/${URL##*/}"
	if [[ -f "$fn" ]]; then
		log - "Saved '$tn'."
	else
		if mv "$tn" "$fn"; then
			log - "Saved '$fn'."
		else
			log 0 "Error saving '$fn'."
			log - "Saved '$tn'."
		fi
	fi
}

### main entry point ###
bollux() {
	OPTIND=0
	process_cmdline "$@"
	shift $((OPTIND - 1))

	if (($# == 1)); then
		URL="$1"
	else
		ask GO URL
	fi

	log 5 "URL : $URL"

	request "$URL" | handle "$URL"
}

process_cmdline() {
	while getopts :hL: OPT; do
		case "$OPT" in
		h) bollux_usage ;;
		L) LOGL="$OPTARG" ;;
		:) die 1 "Option -$OPTARG requires an argument" ;;
		*) die 1 "Unknown option: -$OPTARG" ;;
		esac
	done
}

bollux_setup() {
	trap bollux_cleanup INT QUIT EXIT
}

bollux_cleanup() {
	exit $?
}

if [[ "${BASH_SOURCE[0]}" == "$0" ]]; then
	# requirements here -- so they're only checked once
	require gawk
	require dd
	require mv
	require openssl
	require sed

	bollux_setup

	bollux "$@"

	bollux_cleanup
fi