From 3ab42a5fb02a2652cae089bd3f936729fd49382d Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?John=20Ankarstr=C3=B6m?= <john@ankarstrom.se>
Date: Sun, 31 Jan 2021 17:53:13 +0000
Subject: Support [A-Za-z] in list item values

HTML/CSS makes this a bit awkward; all the better if em
can handle it for you.
---
 emparse | 75 ++++++++++++++++++++++++++++++++++++++++++++---------------------
 1 file changed, 51 insertions(+), 24 deletions(-)

(limited to 'emparse')

diff --git a/emparse b/emparse
index 3985a86..ed3a4c7 100755
--- a/emparse
+++ b/emparse
@@ -2,15 +2,20 @@
 
 # aux/emparse -- parse em source
 
-BEGIN { expectblock = 1; itemlevel = 1 }
+BEGIN {
+	expectblock = 1
+	itemlevel = 1
+	for (n=0; n < 256; n++)
+		ord[sprintf("%c", n)] = n
+}
 END { breakblock() }
 
 /^$/							{ breakblock(); getline }
 expectblock && /^  /			{ newblock("table") }
 expectblock && /^ > /			{ newblock("blockquote") }
 expectblock && /^ - /			{ newblock("ul") }
-expectblock && /^ [0-9]+\. /		{ newblock("ol") }
-expectblock && /^ \[[0-9]+\] /	{ newblock("nl") }
+expectblock && /^ [0-9a-z]+\. /	{ newblock("ol") }
+expectblock && /^ \[[0-9a-z]+\] /	{ newblock("nl") }
 expectblock && /^ .*: /			{ newblock("dl") }
 expectblock && /^	/			{ newblock("pre") }
 expectblock && /^---$/			{ expectblock = 0; printf "<hr/>\n"; next }
@@ -26,20 +31,20 @@ openblock == "pre" { sub("^	", ""); $0 = escape($0); printf "%s\n", $0; next }
 
 openblock == "blockquote" && /^ > /		{ item(1, "blockquote", line) }
 openblock == "ul" && /^ - /				{ item(1, "ul", line) }
-openblock == "ol" && /^ [0-9]+\. /			{ item(1, "ol", line) }
-openblock == "nl" && /^ \[[0-9]+\] /		{ item(1, "nl", line) }
+openblock == "ol" && /^ [0-9a-z]+\. /		{ item(1, "ol", line) }
+openblock == "nl" && /^ \[[0-9a-z]+\] /		{ item(1, "nl", line) }
 openblock == "dl" && /^ .*: /				{ term(line) }
 
-(openblock == "ul" || openblock == "ol") && /^  -/			{ item(2, "ul", line) }
-(openblock == "ul" || openblock == "ol") && /^  [0-9]+\./		{ item(2, "ol", line) }
-(openblock == "ul" || openblock == "ol") && /^   -/			{ item(3, "ul", line) }
-(openblock == "ul" || openblock == "ol") && /^   [0-9]+\./	{ item(3, "ol", line) }
-(openblock == "ul" || openblock == "ol") && /^    -/			{ item(4, "ul", line) }
-(openblock == "ul" || openblock == "ol") && /^    [0-9]+\./	{ item(4, "ol", line) }
-(openblock == "ul" || openblock == "ol") && /^     -/			{ item(5, "ul", line) }
-(openblock == "ul" || openblock == "ol") && /^     [0-9]+\./	{ item(5, "ol", line) }
-(openblock == "ul" || openblock == "ol") && /^      -/		{ item(6, "ul", line) }
-(openblock == "ul" || openblock == "ol") && /^      [0-9]+\./	{ item(6, "ol", line) }
+(openblock == "ul" || openblock == "ol") && /^  -/				{ item(2, "ul", line) }
+(openblock == "ul" || openblock == "ol") && /^  [0-9a-z]+\./		{ item(2, "ol", line) }
+(openblock == "ul" || openblock == "ol") && /^   -/				{ item(3, "ul", line) }
+(openblock == "ul" || openblock == "ol") && /^   [0-9a-z]+\./		{ item(3, "ol", line) }
+(openblock == "ul" || openblock == "ol") && /^    -/				{ item(4, "ul", line) }
+(openblock == "ul" || openblock == "ol") && /^    [0-9a-z]+\./		{ item(4, "ol", line) }
+(openblock == "ul" || openblock == "ol") && /^     -/				{ item(5, "ul", line) }
+(openblock == "ul" || openblock == "ol") && /^     [0-9a-z]+\./		{ item(5, "ol", line) }
+(openblock == "ul" || openblock == "ol") && /^      -/			{ item(6, "ul", line) }
+(openblock == "ul" || openblock == "ol") && /^      [0-9a-z]+\./	{ item(6, "ol", line) }
 
 { format($0) } # inline formatting
 
@@ -56,11 +61,11 @@ function format(line) {
 		else if (match(w[i], "^&lt;(\\./.*|\\.\\./.*|#.*|.*/.*)&gt;\\)[.,:;?!]$"))
 			w[i] = link(substr(w[i], 5, RLENGTH-10)) substr(w[i], RLENGTH-1)
 
-		if (match(w[i], "^\\[[0-9]+\\]$"))
+		if (match(w[i], "^\\[[0-9a-z]+\\]$"))
 			w[i] = ref(substr(w[i], 2, RLENGTH-2))
-		else if (match(w[i], "^\\[[0-9]+\\][.,:;?!)]$"))
+		else if (match(w[i], "^\\[[0-9a-z]+\\][.,:;?!)]$"))
 			w[i] = ref(substr(w[i], 2, RLENGTH-3)) substr(w[i], RLENGTH)
-		else if (match(w[i], "^\\[[0-9]+\\]\\)[.,:;?!]$"))
+		else if (match(w[i], "^\\[[0-9a-z]+\\]\\)[.,:;?!]$"))
 			w[i] = ref(substr(w[i], 2, RLENGTH-4)) substr(w[i], RLENGTH-1)
 
 		if (!openformat) {
@@ -186,16 +191,18 @@ function item(level, type, line) {
 		printf "<li>"
 	}
 	if (type == "ol") {
-		match($0, "[0-9]+\.")
+		match($0, "[0-9a-z]+\.")
 		v = substr($0, RSTART, RLENGTH-1)
-		sub("^ +[0-9]+\. ", "")
-		printf "<li value=\"%s\">", v
+		sub("^ +[0-9a-z]+\. ", "")
+		listvalid(v)
+		printf "<li value=\"%s\" style=\"list-style-type: %s\">", listnum(v), listtype(v)
 	}
 	if (type == "nl") {
-		match($0, "\\[[0-9]+\\]")
+		match($0, "\\[[0-9a-z]+\\]")
 		v = substr($0, RSTART+1, RLENGTH-2)
-		sub("^ \\[[0-9]+\\] ", "")
-		printf "<li value=\"%s\" id=\"ref%s\">", v, v
+		sub("^ \\[[0-9a-z]+\\] ", "")
+		listvalid(v)
+		printf "<li value=\"%s\" style=\"list-style-type: %s\" id=\"ref%s\">", listnum(v), listtype(v), v
 	}
 }
 
@@ -220,6 +227,26 @@ function link(h) {
 	return "<a href=\"" h "\">" h "</a>"
 }
 
+function listvalid(v) {
+	if (match(v, "^[0-9]+$")) return
+	if (match(v, "^[a-z]$")) return
+	if (match(v, "^[A-Z]$")) return
+	printf "error: '%s' is not a valid list item value\n", v > "/dev/stderr"
+	exit 1
+}
+
+function listnum(v) {
+	if (match(v, "^[0-9]+$")) return v
+	if (match(v, "^[a-z]$")) return ord[v]-96
+	if (match(v, "^[A-Z]$")) return ord[v]-64
+}
+
+function listtype(v) {
+	if (match(v, "^[0-9]+$")) return "decimal"
+	if (match(v, "^[a-z]$")) return "lower-alpha"
+	if (match(v, "^[A-Z]$")) return "upper-alpha"
+}
+
 function escape(s) {
 	gsub("&", "\\&amp;", s)
 	gsub("<", "\\&lt;", s)
-- 
cgit v1.2.3