From bead1d53e5025ceb35ac385e49cb1ab143ba3fdd Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?John=20Ankarstr=C3=B6m?= <john@ankarstrom.se>
Date: Fri, 29 Jan 2021 17:29:10 +0000
Subject: Add block-level elements

---
 em | 88 ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 88 insertions(+)
 create mode 100755 em

diff --git a/em b/em
new file mode 100755
index 0000000..13beb29
--- /dev/null
+++ b/em
@@ -0,0 +1,88 @@
+#!/bin/awk -f
+
+# em -- limited hypertext markup language
+
+function newblock(name) {
+	openblock = name
+	openitem = 0
+	expectblock = 0
+	printf "<%s>", name
+}
+
+function breakblock() {
+	if (openitem) {
+		printf "</li>\n"
+		while (itemlevel-- > 1)
+			printf "</%s>\n", openblock
+		itemlevel = 1
+	}
+	if (opendef) printf "</dd>\n"
+	if (openblock) printf "</%s>\n", openblock
+	openitem = 0
+	opendef = 0
+	openblock = 0
+	expectblock = 1
+}
+
+function heading(level, line) {
+	sub("^=* ", "", line)
+	sub(" =*$", "", line)
+	printf "<h%d>%s</h%d>\n", level, line, level
+	# should inline formatting be supported in headings?
+}
+
+function item(level, type, line) {
+	if (openitem) printf "</li>"
+	openitem = 1
+	if (level > itemlevel) printf "<ul>\n"
+	if (level < itemlevel) printf "</ul>\n"
+	itemlevel = level
+	if (type == "ul") sub("^ +- ", "")
+	if (type == "ol") sub("^ +[0-9]+\. ", "")
+	printf "<li>"
+}
+
+function term(line, t) { # t is a local variable
+	if (opendef) printf "</dd>"
+	opendef = 1
+	t = $0; d  = $0
+	sub("^ ", "", t)
+	sub(": .*$", "", t)
+	sub("^ [^:]+: ", "")
+	printf "<dt>%s</dt><dd>", t
+}
+
+BEGIN { expectblock = 1; itemlevel = 1 }
+END { breakblock() }
+
+/^$/						{ breakblock(); getline }
+expectblock && /^  /		{ newblock("table") }
+expectblock && /^ - /		{ newblock("ul") }
+expectblock && /^ [0-9]+\. /	{ newblock("ol") }
+expectblock && /^ .*: /		{ newblock("dl") }
+expectblock && /^	/		{ newblock("pre"); sub("^	", "") }
+expectblock && /^---$/		{ expectblock = 0; printf "<hr/>\n"; next }
+expectblock && /^= .* =$/			{ heading(1, $0); next }
+expectblock && /^== .* ==$/			{ heading(3, $0); next }
+expectblock && /^=== .* ===$/		{ heading(3, $0); next }
+expectblock && /^==== .* ====$/		{ heading(4, $0); next }
+expectblock && /^===== .* =====$/	{ heading(5, $0); next }
+expectblock && /^====== .* ======$/	{ heading(6, $0); next }
+expectblock { newblock("p") }
+
+block = "ul" && /^ -/		{ item(1, "ul", line) }
+block = "ol" && /^ [0-9]+\./	{ item(1, "ol", line) }
+block = "dl" && /^ .*:/		{ term(line) }
+
+block = "ul" && /^  -/		{ item(2, "ul", line) }
+block = "ol" && /^  [0-9]+\./	{ item(2, "ol", line) }
+block = "ul" && /^   -/		{ item(3, "ul", line) }
+block = "ol" && /^   [0-9]+\./	{ item(3, "ol", line) }
+block = "ul" && /^    -/		{ item(4, "ul", line) }
+block = "ol" && /^    [0-9]+\./	{ item(4, "ol", line) }
+block = "ul" && /^     -/		{ item(5, "ul", line) }
+block = "ol" && /^     [0-9]+\./	{ item(5, "ol", line) }
+block = "ul" && /^      -/		{ item(6, "ul", line) }
+block = "ol" && /^      [0-9]+\./	{ item(6, "ol", line) }
+
+{ print $0 }
-- 
cgit v1.2.3