From 59e5a214a8be015b00004b1961a83f48dd6d65f4 Mon Sep 17 00:00:00 2001
From: Case Duckworth
Date: Mon, 29 Jan 2024 00:02:27 -0600
Subject: Initial commit

---
 subtext.awk | 165 ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
 test.st     |  33 ++++++++++++
 2 files changed, 198 insertions(+)
 create mode 100644 subtext.awk
 create mode 100644 test.st

diff --git a/subtext.awk b/subtext.awk
new file mode 100644
index 0000000..c3e135b
--- /dev/null
+++ b/subtext.awk
@@ -0,0 +1,165 @@
+# subtext -- text substitutor -*- awk -*-
+# (C) C Duckworth <acdw@acdw.net>
+
+BEGIN {
+	## Tuneables
+	postproc = get_value(postproc, "ST_POSTPROC", "shexpand")
+	bodyfunc = get_value(bodyfunc, "ST_BODYFUNC", "sub_text")
+	sopath = get_value(sopath, "ST_SOPATH", ".:"ENVIRON["HOME"]"/.subtext")
+	split(sopath, asopath, ":")
+	# postproc = postproc ? postproc : "shexpand"
+	# bodyfunc = bodyfunc ? bodyfunc : "sub_text"
+	## Globals
+	# Booleans
+	true = 1
+	false = 0
+	## Wrap the text in a function
+	pretext = "### begin text\n"bodyfunc"(){\n"
+	posttext = "}\n### end text"
+	## Prelude function strings
+	# Ask sed to do these b/c awk has no capture groups ;_;
+	shellfix = "sed -E" \
+		" -e 's/`/\\\\`/g'" \
+		" -e 's/(^|[^\\$])\\$([^\\$]|$)/\\1\\\\$\\2/g'" \
+		" -e 's/(^|[^\\$])\\$(\\$+)([^\\$]|$)/\\1\\2\\3/g'"
+	shxwrap = " -e 's/^/:/'"
+	htmlfix = "sed -E" \
+		" -e 's#([^\\\\]|^)&#\\1\\&amp;#g'" \
+		" -e 's#([^\\\\]|^)<#\\1\\&lt;#g'" \
+		" -e 's#([^\\\\]|^)>#\\1\\&gt;#g'" \
+		" -e 's#\\\\([&<>])#\\1#g'";
+	## Prelude
+	par = "#!/bin/sh\n### kernel\n" \
+		"preface()(sed \"s/^/$1/\")\n" \
+		"unpreface()(sed \"s/^$1//\")\n" \
+		"shexpand()(eval \"$( (echo 'cat<<.';preface +;echo .)" \
+		" | unpreface + )\")\n" \
+		"### library\n" \
+		"shellfix()(" shellfix ")\n" \
+		"htmlfix()(" htmlfix ")\n" \
+		"### variables\n" \
+		"ST_POSTPROC=" postproc "\n" \
+		"ST_BODYFUNC=" bodyfunc "\n" \
+		"ST_SOPATH=" sopath "\n" \
+		"### header\n"
+}
+
+### End a block
+
+end[endn] && $0 == end[endn] {
+	par = par "\n" end[endn--] "\n)"
+	printpar()
+	subdocp = false
+	next
+}
+
+### Line continuation
+
+/\\$/ {
+	getline nl
+	sub(/\\$/,"")
+	$0 = $0 " " nl
+}
+
+### Special commands
+## These call subtext-internal functions
+
+/^\.so/ {			# Insert $2 verbatim (if in sopath), else error
+	source($2)
+	next
+}
+
+### Escape sequences
+
+/^\.\.\./ {			# Delimit document (end is optional)
+	printpar()
+	docp = !docp
+	if (docp) print pretext "unpreface ':'<<'_'|eval \"$ST_POSTPROC\""
+	else end_text()
+	next
+}
+
+/^\.\./ && docp {		# Begin a heredoc
+	# ..[<command>] [<options>] [<<delim]
+	## ends with DELIM or '..'
+	subdocp = true
+	end[++endn] = (match($0,"<<") ? substr($0,RSTART+RLENGTH) : "..")
+	command = substr($0, 3, RSTART ? RSTART - 2 : length)
+	$0 = "$$(" (command ? command : "cat") "<<" end[endn]
+}
+
+/^\./ && docp {			# One-line command
+	# .<command> [<parameters>]
+	## wraps the line in $( ... ), basically (also quotes)
+	specialp = 2
+	gsub(/"/,"\\\\&")
+	ln = "$$(" substr($1, 2)
+	for (f=2;f<=NF;f++) ln = ln " \"" $f "\""
+	ln = ln ")"
+	$0 = ln
+}
+
+/^\\/ && docp { # \ at the beginning of a line escapes the next character
+	$0 = substr($0, 2)
+}
+
+### Book-keeping
+
+/^$/ {
+	if (!par) next
+	printpar()
+}
+
+{ par = par (par?"\n":"") $0 }
+{ if (--specialp < 0) specialp = 0 }
+
+END {
+	if (dead) exit dead
+	if (par) printpar()
+	while (endn > 0)
+		print "\n" end[endn--] "\n)"
+	if (docp) end_text()
+}
+
+function end_text() {
+	print "_\n" posttext
+}
+
+function printpar() {
+	specialp = specialp || (match(par, /^[ 	]*</))
+	if (docp) {
+		if (!subdocp && !specialp) par = "<p>" par "</p>"
+		shx = shellfix shxwrap
+		print par | shx
+		close(shx)
+	} else print par
+	par = ""
+}
+
+function get_value(var, env_var, default) {
+	if (var) return var
+	else if (ENVIRON[env_var]) return ENVIRON[env_var]
+	else return default
+}
+
+function source(name) {
+	found = false
+	sp = ""
+	for (dir in asopath) {
+		fn = asopath[dir] "/" name
+		sp = asopath[dir] "\n\t" sp
+		while ((getline ln < fn) > 0) {
+			found = true
+			par = par (par?"\n":"") ln
+		}
+		if (found) break
+	}
+	if (!found)
+		die("Couldn't source " name "; looked in:\n\t" sp, 9)
+}
+
+function die(message, code) {
+	dead = code
+	print "!!" FILENAME ":" NR ": " message > "/dev/stderr"
+	exit
+}
diff --git a/test.st b/test.st
new file mode 100644
index 0000000..b11bd87
--- /dev/null
+++ b/test.st
@@ -0,0 +1,33 @@
+.so etlib.sh
+alias verse=cat
+title="test file"
+...
+here's a test file
+.a href="https://example.com" a link or something!!
+
+.h1 it has "headers" or whatever
+
+..blockquote title=foo
+it has paragraphs of quotes
+..
+
+$$(echo it has shell $scripty-stuff)
+
+..
+escaped par!
+omg wow
+..
+
+..p
+another one :O
+
+holy crap
+this is great
+..
+
+i should have more paragraphs
+
+they are wonderful
+i love them
+
+very much
-- 
cgit 1.4.1-21-gabe81