From 1e06e8f5af834b02fdd7fc48413d3380fa93c974 Mon Sep 17 00:00:00 2001
From: Case Duckworth
Date: Thu, 18 Jun 2020 08:24:01 -0500
Subject: Change implementation of URL array

---
 bollux | 316 +++++++++++++++++++++++++++++++++++++----------------------------
 1 file changed, 182 insertions(+), 134 deletions(-)

diff --git a/bollux b/bollux
index 8f6b596..db47e9d 100755
--- a/bollux
+++ b/bollux
@@ -80,7 +80,7 @@ bollux() {
 
 	log d "BOLLUX_URL='$BOLLUX_URL'"
 
-	run blastoff "$BOLLUX_URL"
+	run blastoff -u "$BOLLUX_URL"
 }
 
 # process command-line arguments
@@ -142,6 +142,8 @@ bollux_config() {
 	: "${C_LIST:=0}"        # list formatting
 	: "${C_QUOTE:=3}"       # quote formatting
 	: "${C_PRE:=0}"         # preformatted text formatting
+	## state
+	UC_BLANK=':?:'
 }
 
 # quit happily
@@ -170,123 +172,167 @@ prompt() { # prompt [-u] PROMPT [READ_ARGS...]
 
 # load a URL
 blastoff() { # blastoff [-u] URL
-	local well_formed=true
-	local proto url
+	local u
+
 	if [[ "$1" == "-u" ]]; then
-		well_formed=false
-		shift
+		u="$(run uwellform "$2")"
+	else
+		u="$1"
 	fi
-	url="$1"
 
-	if $well_formed && [[ "$1" != "$BOLLUX_URL" ]]; then
-		url="$(run transform_resource "$BOLLUX_URL" "$1")"
+	local -a url
+	run utransform url "$BOLLUX_URL" "$u"
+	if ! ucdef url[1]; then
+		run ucset url[1] "$BOLLUX_PROTO"
 	fi
-	[[ "$url" != *://* ]] && url="$BOLLUX_PROTO://$url"
-	url="$(trim_string "$url")"
-	proto="${url%://*}"
-
-	log d "PROTO='$proto' URL='$url'"
 
 	{
-		if declare -Fp "${proto}_request" &>/dev/null; then
-			run "${proto}_request" "$url"
+		if declare -Fp "${url[1]}_request" >/dev/null 2>&1; then
+			run "${url[1]}_request" "$url"
 		else
-			die 99 "No request handler for '$proto'!"
+			die 99 "No request handler for '${url[1]}'"
 		fi
-	} | run normalize |
-		{
-			if declare -Fp "${proto}_response" &>/dev/null; then
-				run "${proto}_response" "$url"
-			else
-				log x "No response handler for '$proto', passing through"
-				passthru
-			fi
-		}
+	} | run normalize | {
+		if declare -Fp "${url[1]}_response" >/dev/null 2>&1; then
+			run "${url[1]}_response" "$url"
+		else
+			log d "No response handler for '${url[1]}', passing thru"
+			passthru
+		fi
+	}
+}
+
+# URLS
+## https://tools.ietf.org/html/rfc3986
+uwellform() {
+	local u="$1"
+	
+	if [[ "$u" != *://* ]]; then
+		u="$BOLLUX_PROTO://$u"
+	fi
+
+	u="$(trim_string "$u")"
+
+	printf '%s\n' "$u"
+}
+
+usplit() { # usplit NAME:ARRAY URL:STRING
+	local re='^(([^:/?#]+):)?(//([^/?#]*))?([^?#]*)(\?([^#]*))?(#(.*))?'
+	[[ $2 =~ $re ]] || return $?
+
+	local scheme="${BASH_REMATCH[2]}"
+	local authority="${BASH_REMATCH[4]}"
+	local path="${BASH_REMATCH[5]}"
+	local query="${BASH_REMATCH[7]}"
+	local fragment="${BASH_REMATCH[9]}"
+
+	# 0=url 1=scheme 2=authority 3=path 4=query 5=fragment
+	local i=1 c
+	for c in scheme authority path query fragment; do
+		if [[ "${!c}" || "$c" == path ]]; then
+			printf -v "$1[$i]" '%s' "${!c}"
+		else
+			printf -v "$1[$i]" "$UC_BLANK"
+		fi
+		((i+=1))
+	done
+	printf -v "$1[0]" "$(ujoin "$1")" # inefficient I'm sure
 }
 
-# transform a URI according to RFC 3986 sec 5.2.2
-transform_resource() { # transform_resource BASE_URL REFERENCE_URL
-	local -A R B T # reference, base url, target
-	eval "$(run parse_url B "$1")"
-	eval "$(run parse_url R "$2")"
-	# A non-strict parser may ignore a scheme in the reference
-	# if it is identical to the base URI's scheme.
-	if ! "${STRICT:-true}" && [[ "${R[scheme]}" == "${B[scheme]}" ]]; then
-		unset "${R[scheme]}"
+ujoin() { # ujoin NAME:ARRAY
+	local -n U="$1"
+
+	if ucdef U[1]; then
+		printf -v U[0] "%s:" "${U[1]}"
 	fi
 
-	# basically pseudo-code from spec ported to bash
-	if isdefined "R[scheme]"; then
-		T[scheme]="${R[scheme]}"
-		isdefined "R[authority]" && T[authority]="${R[authority]}"
-		isdefined R[path] &&
-			T[path]="$(run remove_dot_segments "${R[path]}")"
-		isdefined "R[query]" && T[query]="${R[query]}"
+	if ucdef U[2]; then
+		printf -v U[0] "${U[0]}//%s" "${U[2]}"
+	fi
+
+	printf -v U[0] "${U[0]}%s" "${U[3]}"
+	
+	if ucdef U[4]; then
+		printf -v U[0] "${U[0]}?%s" "${U[4]}"
+	fi
+
+	if ucdef U[5]; then
+		printf -v U[0] "${U[0]}#%s" "${U[5]}"
+	fi
+
+	log d "${U[0]}"
+}
+
+ucdef() { [[ "${!1}" != "$UC_BLANK" ]]; } # ucdef NAME
+ucblank() { [[ -z "${!1}" ]]; }           # ucblank NAME
+ucset() { # ucset NAME VALUE
+	run eval "${1}='$2'"
+	run ujoin "${1/\[*\]}"
+}
+
+utransform() { # utransform TARGET:ARRAY BASE:STRING REFERENCE:STRING
+	local -a B R # base, reference
+	local -n T="$1" # target
+	usplit B "$2"
+	usplit R "$3"
+
+	# initialize T
+	for ((i=1;i<=5;i++)); do
+		T[$i]="$UC_BLANK"
+	done
+
+	# 0=url 1=scheme 2=authority 3=path 4=query 5=fragment
+	if ucdef R[1]; then
+		T[1]="${R[1]}"
+		if ucdef R[2]; then
+			T[2]="${R[2]}"
+		fi
+		if ucdef R[3]; then
+			T[3]="$(pundot "${R[3]}")"
+		fi
+		if ucdef R[4]; then
+			T[4]="${R[4]}"
+		fi
 	else
-		if isdefined "R[authority]"; then
-			T[authority]="${R[authority]}"
-			isdefined "R[authority]" &&
-				T[path]="$(remove_dot_segments "${R[path]}")"
-			isdefined R[query] && T[query]="${R[query]}"
+		if ucdef R[2]; then
+			T[2]="${R[2]}"
+			if ucdef R[2]; then
+				T[3]="$(pundot "${R[3]}")"
+			fi
+			if ucdef R[4]; then
+				T[4]="${R[4]}"
+			fi
 		else
-			if isempty "R[path]"; then
-				T[path]="${B[path]}"
-				if isdefined R[query]; then
-					T[query]="${R[query]}"
+			if ucblank R[3]; then
+				T[3]="${B[3]}"
+				if ucdef R[4]; then
+					T[4]="${R[4]}"
 				else
-					T[query]="${B[query]}"
+					T[4]="${B[4]}"
 				fi
 			else
-				if [[ "${R[path]}" == /* ]]; then
-					T[path]="$(remove_dot_segments "${R[path]}")"
+				if [[ "${R[3]}" == /* ]]; then
+					T[3]="$(pundot "${R[3]}")"
 				else
-					T[path]="$(merge_paths "B[authority]" "${B[path]}" "${R[path]}")"
-					T[path]="$(remove_dot_segments "${T[path]}")"
+					T[3]="$(pmerge B R)"
+					T[3]="$(pundot "${T[3]}")"
+				fi
+				if ucdef R[4]; then
+					T[4]="${R[4]}"
 				fi
-				isdefined R[query] && T[query]="${R[query]}"
 			fi
-			T[authority]="${B[authority]}"
+			T[2]="${B[2]}"
 		fi
-		T[scheme]="${B[scheme]}"
+		T[1]="${B[1]}"
 	fi
-	isdefined R[fragment] && T[fragment]="${R[fragment]}"
-	# cf. 5.3 -- recomposition
-	local r
-	isdefined "T[scheme]" && r="$r${T[scheme]}:"
-	# remove the port from the authority
-	isdefined "T[authority]" && r="$r//${T[authority]%:*}"
-	r="$r${T[path]}"
-	isdefined T[query] && r="$r?${T[query]}"
-	isdefined T[fragment] && r="$r#${T[fragment]}"
-	printf '%s\n' "$r"
-}
-
-# merge URL paths according to RFC 3986 sec 5.2.3
-merge_paths() { # merge_paths BASE_AUTHORITY BASE_PATH REFERENCE_PATH
-	# shellcheck disable=2034
-	local B_authority="$1"
-	local B_path="$2"
-	local R_path="$3"
-	# if R_path is empty, get rid of // in B_path
-	if [[ -z "$R_path" ]]; then
-		printf '%s\n' "${B_path//\/\//\//}"
-		return
+	if ucdef R[5]; then
+		T[5]="${R[5]}"
 	fi
 
-	if isdefined "B_authority" && isempty "B_path"; then
-		printf '/%s\n' "${R_path//\/\//\//}"
-	else
-		if [[ "$B_path" == */* ]]; then
-			B_path="${B_path%/*}/"
-		else
-			B_path=""
-		fi
-		printf '%s/%s\n' "${B_path%/}" "${R_path#/}"
-	fi
+	ujoin T
 }
 
-# remove dot segments in paths according to RFC 3986 sec 5.2.4
-remove_dot_segments() { # remove_dot_segments PATH
+pundot() { # pundot PATH:STRING
 	local input="$1"
 	local output
 	while [[ "$input" ]]; do
@@ -301,7 +347,7 @@ remove_dot_segments() { # remove_dot_segments PATH
 		elif [[ "$input" == . || "$input" == .. ]]; then
 			input=
 		else
-			[[ $input =~ ^(/?[^/]*)(/?.*)$ ]] || log debug NOMATCH
+			[[ $input =~ ^(/?[^/]*)(/?.*)$ ]] || return 1
 			output="$output${BASH_REMATCH[1]}"
 			input="${BASH_REMATCH[2]}"
 		fi
@@ -309,36 +355,28 @@ remove_dot_segments() { # remove_dot_segments PATH
 	printf '%s\n' "${output//\/\//\//}"
 }
 
-# parse a url using the reference regex in RFC 3986 appendix B
-parse_url() { # eval "$(split_url NAME STRING)" => NAME[...]
-	local name="$1"
-	local string="$2"
-	local re='^(([^:/?#]+):)?(//([^/?#]*))?([^?#]*)(\?([^#]*))?(#(.*))?'
-	[[ $string =~ $re ]] || return $?
+pmerge() {
+	local -n b="$1"
+	local -n r="$2"
 
-	local scheme="${BASH_REMATCH[2]}"
-	local authority="${BASH_REMATCH[4]}"
-	local path="${BASH_REMATCH[5]}"
-	local query="${BASH_REMATCH[7]}"
-	local fragment="${BASH_REMATCH[9]}"
+	if ucblank r[3]; then
+		printf '%s\n' "${b[3]//\/\//\//}"
+		return
+	fi
 
-	for c in scheme authority query fragment; do
-		[[ "${!c}" ]] &&
-			run printf '%s[%s]=%q\n' "$name" "$c" "${!c}"
-	done
-	# unclear if the path is always set even if empty but it looks that way
-	run printf '%s[path]=%q\n' "$name" "$path"
+	if ucdef b[2] && ucblank b[3]; then
+		printf '/%s\n' "${r[3]//\/\//\//}"
+	else
+		local bp=""
+		if [[ "${b[3]}" == */* ]]; then
+			bp="${b[3]%/*}"
+		fi
+		printf '%s/%s\n' "${bp%/}" "${r[3]#/}"
+	fi
 }
 
-# is a NAME defined ('set' in bash)?
-isdefined() { [[ "${!1+x}" ]]; } # isdefined NAME
-
-# is a NAME defined AND empty?
-isempty() { [[ ! "${!1-x}" ]]; } # isempty NAME
-
-# work with URLs
 # https://github.com/dylanaraps/pure-bash-bible/
-urlencode() { # urlencode STRING
+uencode() { # uencode URL:STRING
 	local LC_ALL=C
 	for ((i = 0; i < ${#1}; i++)); do
 		: "${1:i:1}"
@@ -355,7 +393,7 @@ urlencode() { # urlencode STRING
 }
 
 # https://github.com/dylanaraps/pure-bash-bible/
-urldecode() { # urldecode STRING
+udecode() { # udecode URL:STRING 
 	: "${1//+/ }"
 	printf '%b\n' "${_//%/\\x}"
 }
@@ -363,19 +401,28 @@ urldecode() { # urldecode STRING
 # GEMINI
 # https://gemini.circumlunar.space/docs/specification.html
 gemini_request() { # gemini_request URL
-	local url port server
-	local ssl_cmd
-	url="$1"
-	port=1965
-	server="${url#*://}"
-	server="${server%%/*}"
+	local -a url
+	usplit url "$1"
+
+	# get rid of userinfo
+	ucset url[2] "${url[2]#*@}"
+
+	local port
+	if [[ "${url[2]}" == *:* ]]; then
+		port="${url[2]#*:}"
+		ucset url[2] "${url[2]%:*}"
+	else
+		port=1965 # TODO variablize
+	fi
 
-	ssl_cmd=(openssl s_client -crlf -quiet -connect "$server:$port")
-	ssl_cmd+=(-servername "$server") # SNI
-	# disable old TLS/SSL versions
-	ssl_cmd+=(-no_ssl3 -no_tls1 -no_tls1_1)
+	local ssl_cmd=(
+		openssl s_client 
+			-crlf -quiet -connect "${url[2]}:$port"
+			-servername "${url[2]}" # SNI
+			-no_ssl3 -no_tls1 -no_tls1_1 # disable old TLS/SSL versions
+	)
 
-	run "${ssl_cmd[@]}" <<<"$url" 2>/dev/null
+	run "${ssl_cmd[@]}" <<<"$url"
 }
 
 gemini_response() { # gemini_response URL
@@ -399,7 +446,7 @@ gemini_response() { # gemini_response URL
 		10) run prompt "$meta" ;;
 		11) run prompt "$meta" -s ;; # password input
 		esac
-		run blastoff "?$(urlencode "$REPLY")"
+		run blastoff "?$(uencode "$REPLY")"
 		;;
 	2*) # OK
 		REDIRECTS=0
@@ -480,7 +527,7 @@ gopher_response() { # gopher_response URL
 	cur_server="${BASH_REMATCH[1]}"
 	type="${BASH_REMATCH[6]:-1}"
 
-	run history_append "$url" "" # TODO: get the title ??
+	run history_append "$url" "" # gopher doesn't really have titles, huh
 
 	log d "TYPE='$type'"
 
@@ -618,9 +665,10 @@ display() { # display METADATA [TITLE]
 		set_title "$title${title:+ - }bollux"
 		less_cmd=(less -R) # render ANSI color escapes
 		mklesskey "$BOLLUX_LESSKEY" && less_cmd+=(-k "$BOLLUX_LESSKEY")
+		local helpline="o:open, g/G:goto, [:back, ]:forward, r:refresh"
 		less_cmd+=(
 			-Pm"$(less_prompt_escape "$BOLLUX_URL") - bollux$" # 'status'line
-			-P='o\:open, g\:goto, [\:back, ]\:forward, r\:refresh$' # helpline
+			-P="$(less_prompt_escape "$helpline")$" # helpline
 			-m # start with statusline
 			+k # float content to the top
 		)
@@ -910,7 +958,7 @@ handle_keypress() { # handle_keypress CODE
 		run blastoff "$BOLLUX_URL"
 		;;
 	53) # G - goto a url (pre-filled with current)
-		prompt -u GO
+		run prompt -u GO
 		run blastoff -u "$REPLY"
 		;;
 	*) # 54-57 -- still available for binding
-- 
cgit 1.4.1-21-gabe81