From 8a8c5deaf3e1680ba91ae663496db18880bb7ba8 Mon Sep 17 00:00:00 2001
From: Case Duckworth <acdw@acdw.net>
Date: Sat, 30 May 2020 16:00:12 -0500
Subject: [PATCH] Update typesetting of text/gemini

---
 bollux            | 217 ++++++++++++++++++++++++++++++++--------------
 typeset_gemini.sh | 172 ------------------------------------
 2 files changed, 152 insertions(+), 237 deletions(-)
 delete mode 100644 typeset_gemini.sh

diff --git a/bollux b/bollux
index 87c35fb..6f121b0 100755
--- a/bollux
+++ b/bollux
@@ -104,6 +104,7 @@ bollux_config() {
 		log debug "Can't load config file '$BOLLUX_CONFIG'."
 	fi
 
+	## behavior
 	: "${BOLLUX_DOWNDIR:=.}"                   # where to save downloads
 	: "${BOLLUX_LOGLEVEL:=3}"                  # log level
 	: "${BOLLUX_MAXREDIR:=5}"                  # max redirects
@@ -112,6 +113,20 @@ bollux_config() {
 	: "${BOLLUX_LESSKEY:=/tmp/bollux-lesskey}" # where to store binds
 	: "${BOLLUX_PAGESRC:=/tmp/bollux-src}"     # where to save the page source
 	: "${BOLLUX_URL:=}"                        # start url
+	## typesetting
+	: "${T_MARGIN:=4}"      # left and right margin
+	: "${T_WIDTH:=0}"       # width of the viewport -- 0 = get term width
+	# colors -- these will be wrapped in \e[ __ m
+	C_RESET='\e[0m'         # reset
+	: "${C_SIGIL:=35}"      # sigil (=>, #, ##, ###, *, ```)
+	: "${C_LINK_NUMBER:=1}" # link number
+	: "${C_LINK_TITLE:=4}"  # link title
+	: "${C_LINK_URL:=36}"   # link URL
+	: "${C_HEADER1:=1;4}"   # header 1 formatting
+	: "${C_HEADER2:=1}"     # header 2 formatting
+	: "${C_HEADER3:=3}"     # header 3 formatting
+	: "${C_LIST:=0}"        # list formatting
+	: "${C_PRE:=0}"         # preformatted text formatting
 }
 
 prompt() {
@@ -399,74 +414,146 @@ mklesskey() {
 }
 
 normalize_crlf() {
-	while read -r line; do
-		printf '%s\n' "${line//$'\r'?($'\n')/}"
+	while IFS= read -r; do
+		printf '%s\n' "${REPLY//$'\r'?($'\n')/}"
 	done
 }
 
 typeset_gemini() {
-	gawk '
-	BEGIN {
-		pre = 0
-		margin = margin ? margin : 4
-		txs = ""	
-		lns = "\033[1m"	
-		lus = "\033[36m"	
-		lts = "\033[4m"	
-		pfs = ""	
-		h1s = "\033[1;4m"	
-		h2s = "\033[1m"	
-		h3s = "\033[3m"	
-		lis = ""	
-		res = "\033[0m"	
-		ms = "\033[35m"
-	}
-	/```/ {
-		pre = ! pre
-		next
-	}
-	pre {
-		mark = "```"
-		fmt = pfs "%s" res
-		text = $0
-	}
-	/^#/ {
-		match($0, /#+/)
-		mark = substr($0, RSTART, RLENGTH)
-		sub(/#+[[:space:]]*/, "", $0)
-		level = length(mark)
-		if (level == 1) {
-			fmt = h1s "%s" res
-		} else if (level == 2) {
-			fmt = h2s "%s" res
-		} else {
-			fmt = h3s "%s" res
-		}
-	}
-	/^=>/ {
-		mark = "=>"
-		sub(/=>[[:space:]]*/, "", $0)
-		desc = $1
-		text = ""
-		for (w = 2; w <= NF; w++) {
-			text = text (text ? " " : "") $w
-		}
-		fmt = lns "[" (++ln) "]" res " " lts "%s" res "\t" lus "%s" res
-	}
-	/^\*[[:space:]]/ {
-		mark = "*"
-		sub(/\*[[:space:]]*/, "", $0)
-		fmt = lis "%s" res
-	}
-	{
-		mark = mark ? mark : mark
-		fmt = fmt ? fmt : "%s"
-		text = text ? text : $0
-		desc = desc ? desc : ""
-		printf ms "%" (margin-1) "s " res fmt "\n", mark, text, desc
-		mark = fmt = text = desc = ""
-	}
-	'
+	local pre=false
+	local ln=0 # link number
+
+	if ((T_WIDTH == 0)); then
+		shopt -s checkwinsize
+		(
+			:
+			:
+		) # XXX this doesn't work!?
+		log d "LINES=$LINES; COLUMNS=$COLUMNS"
+		T_WIDTH=$COLUMNS
+	fi
+
+	WIDTH=$((T_WIDTH - (T_MARGIN * 2)))
+	((WIDTH < 0)) && WIDTH=80 # default if dumb
+
+	log d "T_WIDTH=$T_WIDTH"
+	log d "WIDTH=$WIDTH"
+
+	while IFS= read -r; do
+		case "$REPLY" in
+		'```')
+			if $pre; then
+				pre=false
+			else
+				pre=true
+			fi
+			continue
+			;;
+		=\>*)
+			: $((ln += 1))
+			gemini_link "$REPLY" $pre "$ln"
+			;;
+		\#*) gemini_header "$REPLY" $pre ;;
+		\**) gemini_list "$REPLY" $pre ;;
+		*) gemini_text "$REPLY" $pre ;;
+		esac
+	done
+}
+
+gemini_link() {
+	local re="^(=>)[[:blank:]]*([^[:blank:]]+)[[:blank:]]*(.*)"
+	local s t a l # sigil, text, annotation(url), line
+	if ! ${2-false} && [[ "$1" =~ $re ]]; then
+		s="${BASH_REMATCH[1]}"
+		a="${BASH_REMATCH[2]}"
+		t="${BASH_REMATCH[3]}"
+		if [[ -z "$t" ]]; then
+			t="$a"
+			a=
+		fi
+
+		printf "\e[${C_SIGIL}m%-${T_MARGIN}s${C_RESET}" "$s"
+		printf -v l "\e[${C_LINK_NUMBER}m[%d]${C_RESET} \
+			\e[${C_LINK_TITLE}m%s${C_RESET} \
+			\e[${C_LINK_URL}m%s${C_RESET}\n" \
+			"$3" "$t" "$a"
+		fold_line "$WIDTH" "$l"
+	else
+		gemini_pre "$1"
+	fi
+}
+
+gemini_header() {
+	local re="^(#+)[[:blank:]]*(.*)"
+	local s t a l # sigil, text, annotation(lvl), line
+	if ! ${2-false} && [[ "$1" =~ $re ]]; then
+		s="${BASH_REMATCH[1]}"
+		a="${#BASH_REMATCH[1]}"
+		t="${BASH_REMATCH[2]}"
+		local hdrfmt
+		hdrfmt="$(eval echo "\$C_HEADER$a")"
+
+		printf "\e[${C_SIGIL}m%$((T_MARGIN - 1))s ${C_RESET}" "$s"
+		printf -v l "\e[${hdrfmt}m%s${C_RESET}\n" "$t"
+		fold_line "$WIDTH" "$l"
+	else
+		gemini_pre "$1"
+	fi
+}
+
+gemini_list() {
+	local re="^(\*)[[:blank:]]*(.*)"
+	local s t a l # sigil, text, annotation(n/a), line
+	if ! ${2-false} && [[ "$1" =~ $re ]]; then
+		s="${BASH_REMATCH[1]}"
+		t="${BASH_REMATCH[2]}"
+
+		printf "\e[${C_SIGIL}m%-${T_MARGIN}s" "$s"
+		printf -v l "\e[${C_LIST}m%s${C_RESET}\n" "$t"
+		fold_line "$WIDTH" "$l"
+	else
+		gemini_pre "$1"
+	fi
+}
+
+gemini_text() {
+	if ! ${2-false}; then
+		printf "%${T_MARGIN}s" ' '
+		fold_line "$WIDTH" "$1"
+	else
+		gemini_pre "$1"
+	fi
+}
+
+gemini_pre() {
+	printf "\e[${C_SIGIL}m%-${T_MARGIN}s" '```'
+	printf "\e[${C_PRE}m%s${C_RESET}\n" "$1"
+}
+
+fold_line() { # fold_line WIDTH TEXT
+	local width="$1"
+	local margin="${2%%[![:space:]]*}"
+	if [[ "$margin" ]]; then
+		margin="${#margin}"
+	else
+		margin="$T_MARGIN"
+	fi
+	local ll=0 wl plain
+	# shellcheck disable=2086
+	set -- $2 # TODO: is this the best way?
+
+	for word; do
+		plain="${word//$'\x1b'\[*([0-9;])m/}"
+		wl=$((${#plain} + 1))
+		if (((ll + wl) >= width)); then
+			printf "\n%${margin}s" ' '
+			ll=$wl
+		else
+			ll=$((ll + wl))
+		fi
+		printf '%s ' "$word"
+	done
+	printf '\n'
 }
 
 handle_keypress() {
@@ -500,7 +587,7 @@ select_url() {
 }
 
 extract_links() {
-	gawk -F$'\t' '
+	gawk '
 	/^=>/ {
 		sub(/=>[[:space:]]*/,"")
 		if ($2) 
diff --git a/typeset_gemini.sh b/typeset_gemini.sh
deleted file mode 100644
index 2f9abbc..0000000
--- a/typeset_gemini.sh
+++ /dev/null
@@ -1,172 +0,0 @@
-typeset_gemini() {
-	local pre=false
-	local ln=0
-
-	while read -r line; do
-		case "$line" in
-		'```')
-			flip pre
-			continue
-			;;
-		=\>*)
-			: $((ln += 1))
-			gemini_link "$line" $pre
-			;;
-		\#*) gemini_header "$line" $pre ;;
-		\**) gemini_list "$line" $pre ;;
-		*) gemini_text "$line" $pre ;;
-		esac
-	done
-}
-
-gemini_link() {
-	local re="^(=>)[[:blank:]]*([^[:blank:]]+)[[:blank:]]*(.*)"
-	local s t a # sigil, text, annotation(url)
-	if ! ${2-false} && [[ "$1" =~ $re ]]; then
-		s="${BASH_REMATCH[1]}"
-		t="${BASH_REMATCH[3]}"
-		a="${BASH_REMATCH[2]}"
-
-		printf "$C_SIGIL%-${MARGIN}s" "$s"
-		fold_line "$WIDTH" "$(printf "$C_LINK_TITLE%s $C_LINK_URL%s$C_RESET\n" \
-			"$t" "$a")"
-	else
-		gemini_pre "$1"
-	fi
-}
-
-gemini_header() {
-	local re="^(#+)[[:blank:]]*(.*)"
-	local s t a # sigil, text, annotation(lvl)
-	if ! ${2-false} && [[ "$1" =~ $re ]]; then
-		s="${BASH_REMATCH[1]}"
-		a="${#BASH_REMATCH[1]}"
-		t="${BASH_REMATCH[2]}"
-
-		local hdrfmt
-		hdrfmt="$(eval echo "\$C_HEADER$a")"
-		printf "$C_SIGIL%-${MARGIN}s$hdrfmt%s$C_RESET\n" \
-			"$s" "$(fold_line "$WIDTH" "$t")"
-	else
-		gemini_pre "$1"
-	fi
-}
-
-gemini_list() {
-	local re="^(\*)[[:blank:]]*(.*)"
-	local s t a # sigil, text, annotation(n/a)
-	if ! ${2-false} && [[ "$1" =~ $re ]]; then
-		s="${BASH_REMATCH[1]}"
-		t="${BASH_REMATCH[2]}"
-
-		printf "$C_SIGIL%-${MARGIN}s$C_LIST%s$C_RESET\n" \
-			"$s" "$(fold_line "$WIDTH" "$t")"
-	else
-		gemini_pre "$1"
-	fi
-}
-
-gemini_text() {
-	printf "%${MARGIN}s" ' '
-	if ! ${2-false}; then
-		fold_line "$WIDTH" "$1"
-	else
-		gemini_pre "$1"
-	fi
-}
-
-gemini_pre() {
-	printf "%${MARGIN}s%s" ' ' "$1"
-}
-
-flip() { # flip NAME
-	[[ "${!1}" == true || "${!1}" == false ]] || return 1
-
-	if "${!1}"; then
-		eval "$1=false"
-	else
-		eval "$1=true"
-	fi
-}
-
-fold_line() { # fold_line WIDTH TEXT
-	local width="$1"
-	local ll=0 wl plain
-	# shellcheck disable=2086
-	# TODO: determine if this is the best way to do it
-	set -- $2
-
-	for word; do
-		plain="${word//$'\x1b'\[*([0-9;])m/}"
-		wl=$((${#plain} + 1))
-		if (((ll + wl) >= width)); then
-			printf "\n%${MARGIN}s" ' '
-			ll=$wl
-		else
-			ll=$((ll + wl))
-		fi
-		printf '%s ' "$word"
-	done
-	printf '\n'
-}
-
-# just here for reference
-strip() { # strip control sequences
-	# https://stackoverflow.com/a/55872518
-	shopt -s extglob
-	while IFS='' read -r x; do
-		# remove colors
-		echo "${x//$'\x1b'\[*([0-9;])m/}"
-	done
-}
-
-test() {
-	MARGIN=4
-	WIDTH=60
-	#shopt -s checkwinsize; (:;:)
-	#WIDTH="$((COLUMNS - (MARGIN*2)))"
-	C_LINK_TITLE=$'\e[34m'
-	C_LINK_URL=$'\e[31m'
-	C_RESET=$'\e[0m'
-	typeset_gemini <<-'EOF'
-		# Project Gemini
-		
-		## Overview
-		
-		Gemini is a new internet protocol which:
-		
-		* Is heavier than gopher
-		* Is lighter than the web
-		* Will not replace either
-		* Strives for maximum power to weight ratio
-		* Takes user privacy very seriously
-		
-		## Resources
-		
-		=> docs/	Gemini documentation
-		=> software/	Gemini software
-		=> servers/	Known Gemini servers
-		=> https://lists.orbitalfox.eu/listinfo/gemini	Gemini mailing list
-		=> gemini://gemini.conman.org/test/torture/	Gemini client torture test
-		
-		## Web proxies
-		
-		=> https://portal.mozz.us/?url=gemini%3A%2F%2Fgemini.circumlunar.space%2F&fmt=fixed	Gemini-to-web proxy service
-		=> https://proxy.vulpes.one/gemini/gemini.circumlunar.space	Another Gemini-to-web proxy service
-		
-		## Search engines
-		
-		=> gemini://gus.guru/	Gemini Universal Search engine
-		=> gemini://houston.coder.town	Houston search engine
-		
-		## Geminispace aggregators (experimental!)
-		
-		=> capcom/	CAPCOM
-		=> gemini://rawtext.club:1965/~sloum/spacewalk.gmi	Spacewalk
-		
-		## Free Gemini hosting
-		
-		=> users/	Users with Gemini content on this server
-	EOF
-}
-test
-- 
2.20.1