1 files changed, 108 insertions, 68 deletions
diff --git a/ht.awk b/ht.awk
index a382ae7..9288060 100755
--- a/ht.awk
+++ b/ht.awk

@@ -3,96 +3,136 @@
 # (C) 2022 C. Duckworth
 # ht.awk converts mostly-html (with some conveniences) to actual html
+/^;/ {
-function bufpush(s) {
+        sub(/^;/, "", $0)
-    BUF = BUF (BUF ? "\n" : "") s;
+        print "<!--", esc($0), "-->"
+        next
 }
-function buflush() {
+/^</ {  # Raw HTML
-    if (BUF) print BUF;
+        if (! (tag == "html")) {
-    BUF = "";
+                tag = "html"
-    if (tag && (tag != "html") && (tag != "raw"))
+        }
-        print "</" tag ">";
+        bufpush($0)
+        next
 }
-function esc(t) {
+/^```$/ {       # Raw block
-    # This is of much more limited utility than I initially realized.
+        if (! (tag == "raw")) {
-    gsub(/&/, "\\&amp;", t);
+                tag = "raw"
-    gsub(/</, "\\&lt;", t);
+                getline
-    gsub(/>/, "\\&gt;", t);
+                bufpush("<pre><code>" $0)
-    sub(/^ /, "\\&nbsp;", t);
+        } else {
-    return t;
+                bufpush("</code></pre>")
+                buflush()
+                tag = ""
+        }
+        next
 }
-/^;/ { sub(/^;/,""); print "<!--", esc($0), "-->"; next; }
+/^=>/ { # Links (Gemini-style)
+        if (tag == "raw") {
+                next
+        }
+        link = "<a href=\"" esc($2) "\">" $3
+        for (i = 4; i <= NF; i++) {
+                link = link " " $i
+        }
+        link = link "</a>"
+        bufpush(link)
+        next
+}
-/^</ { # Raw HTML
+/^-/ {  # Unordered lists
-    if (! (tag == "html")) tag = "html";
+        if (tag == "raw") {
-    bufpush($0);
+                next
-    next;
+        }
+        if (! (tag == "ul")) {
+                tag = "ul"
+        }
+        sub(/^-[ \t]*/, "<li>", $0)
 }
-/^```$/ { # Raw block
+/^[0-9]+\./ {   # Ordered lists
-    if (! (tag == "raw")) {
+        if (tag == "raw") {
-        tag = "raw";
+                next
-        getline;
+        }
-        bufpush("<pre><code>" $0);
+        if (! (tag == "ol")) {
-    } else {
+                tag = "ol"
-        bufpush("</code></pre>");
+        }
-        buflush();
+        sub(/^[0-9]+\.[ \t]/, "<li>", $0)
-        tag = "";
-    }
-    next;
 }
-/^=>/ { # Links (Gemini-style)
+/^>/ {  # Blockquotes
-    if (tag == "raw") next;
+        if (tag == "raw") {
-    link = "<a href=\"" esc($2) "\">" $3;
+                next
-    for (i=4;i<=NF;i++) link = link " " $i;
+        }
-    link = link "</a>";
+        if (! (tag == "blockquote")) {
-    bufpush(link);
+                tag = "blockquote"
-    next;
+        }
+        sub(/^>[ \t]*/, "", $0)
 }
-/^-/ { # Unordered lists
+/^#+/ { # Headers
-    if (tag == "raw") next;
+        if (tag == "raw") {
-    if (! (tag == "ul")) tag = "ul";
+                next
-    sub(/^-[ \t]*/, "<li>");
+        }
+        match($0, /^#+/)
+        if (! (tag == "h" RLENGTH)) {
+                buflush()
+                tag = "h" RLENGTH
+        }
+        sub(/^#+[ \t]*/, "", $0)
 }
-/^[0-9]+\./ { # Ordered lists
+/^$/ {
-    if (tag == "raw") next;
+        if (tag == "raw") {
-    if (! (tag == "ol")) tag = "ol";
+                next
-    sub(/^[0-9]+\.[ \t]/, "<li>");
+        }
+        buflush()
+        tag = ""
 }
-/^>/ { # Blockquotes
+/./ {
-    if (tag == "raw") next;
+        if (! tag) {
-    if (! (tag == "blockquote")) tag = "blockquote";
+                tag = "p"
-    sub(/^>[ \t]*/,"");
+        }
+        if (! BUF) {
+                bufpush("<" tag ">")
+        }
+        if (tag == "raw") {
+                $0 = esc($0)
+        }
+        bufpush($0)
 }
-/^#+/ { # Headers
+END {
-    if (tag == "raw") next;
+        buflush()
-    match($0, /^#+/);
-    if (! (tag == "h" RLENGTH)) {
-        buflush();
-        tag = "h" RLENGTH;
-    }
-    sub(/^#+[ \t]*/,"");
 }
-/^$/ {
-    if (tag == "raw") next;
+function buflush()
-    buflush();
+{
-    tag = "";
+        if (BUF) {
+                print BUF
+        }
+        BUF = ""
+        if (tag && (tag != "html") && (tag != "raw")) {
+                print "</" tag ">"
+        }
 }
-/./ {
+function bufpush(s)
-    if (! tag) tag = "p";
+{
-    if (! BUF) bufpush("<" tag ">");
+        BUF = BUF (BUF ? "\n" : "") s
-    if (tag == "raw") $0 = esc($0);
-    bufpush($0);
 }
-END { buflush(); }
+function esc(t)
+{
+        # This is of much more limited utility than I initially realized.
+        gsub(/&/, "\\&amp;", t)
+        gsub(/</, "\\&lt;", t)
+        gsub(/>/, "\\&gt;", t)
+        sub(/^ /, "\\&nbsp;", t)
+        return t
+}

diff --git a/ht.awk b/ht.awk index a382ae7..9288060 100755 --- a/ht.awk +++ b/ht.awk
@@ -3,96 +3,136 @@
3	# (C) 2022 C. Duckworth	3	# (C) 2022 C. Duckworth
4		4
5	# ht.awk converts mostly-html (with some conveniences) to actual html	5	# ht.awk converts mostly-html (with some conveniences) to actual html
6		6	/^;/ {
7	function bufpush(s) {	7	sub(/^;/, "", $0)
8	BUF = BUF (BUF ? "\n" : "") s;	8	print "<!--", esc($0), "-->"
		9	next
9	}	10	}
10		11
11	function buflush() {	12	/^</ { # Raw HTML
12	if (BUF) print BUF;	13	if (! (tag == "html")) {
13	BUF = "";	14	tag = "html"
14	if (tag && (tag != "html") && (tag != "raw"))	15	}
15	print "</" tag ">";	16	bufpush($0)
		17	next
16	}	18	}
17		19
18	function esc(t) {	20	/^```$/ { # Raw block
19	# This is of much more limited utility than I initially realized.	21	if (! (tag == "raw")) {
20	gsub(/&/, "\\&", t);	22	tag = "raw"
21	gsub(/</, "\\<", t);	23	getline
22	gsub(/>/, "\\>", t);	24	bufpush("<pre><code>" $0)
23	sub(/^ /, "\\ ", t);	25	} else {
24	return t;	26	bufpush("</code></pre>")
		27	buflush()
		28	tag = ""
		29	}
		30	next
25	}	31	}
26		32
27	/^;/ { sub(/^;/,""); print "<!--", esc($0), "-->"; next; }	33	/^=>/ { # Links (Gemini-style)
		34	if (tag == "raw") {
		35	next
		36	}
		37	link = "<a href=\"" esc($2) "\">" $3
		38	for (i = 4; i <= NF; i++) {
		39	link = link " " $i
		40	}
		41	link = link "</a>"
		42	bufpush(link)
		43	next
		44	}
28		45
29	/^</ { # Raw HTML	46	/^-/ { # Unordered lists
30	if (! (tag == "html")) tag = "html";	47	if (tag == "raw") {
31	bufpush($0);	48	next
32	next;	49	}
		50	if (! (tag == "ul")) {
		51	tag = "ul"
		52	}
		53	sub(/^-[ \t]*/, "<li>", $0)
33	}	54	}
34		55
35	/^```$/ { # Raw block	56	/^[0-9]+\./ { # Ordered lists
36	if (! (tag == "raw")) {	57	if (tag == "raw") {
37	tag = "raw";	58	next
38	getline;	59	}
39	bufpush("<pre><code>" $0);	60	if (! (tag == "ol")) {
40	} else {	61	tag = "ol"
41	bufpush("</code></pre>");	62	}
42	buflush();	63	sub(/^[0-9]+\.[ \t]/, "<li>", $0)
43	tag = "";
44	}
45	next;
46	}	64	}
47		65
48	/^=>/ { # Links (Gemini-style)	66	/^>/ { # Blockquotes
49	if (tag == "raw") next;	67	if (tag == "raw") {
50	link = "<a href=\"" esc($2) "\">" $3;	68	next
51	for (i=4;i<=NF;i++) link = link " " $i;	69	}
52	link = link "</a>";	70	if (! (tag == "blockquote")) {
53	bufpush(link);	71	tag = "blockquote"
54	next;	72	}
		73	sub(/^>[ \t]*/, "", $0)
55	}	74	}
56		75
57	/^-/ { # Unordered lists	76	/^#+/ { # Headers
58	if (tag == "raw") next;	77	if (tag == "raw") {
59	if (! (tag == "ul")) tag = "ul";	78	next
60	sub(/^-[ \t]*/, "<li>");	79	}
		80	match($0, /^#+/)
		81	if (! (tag == "h" RLENGTH)) {
		82	buflush()
		83	tag = "h" RLENGTH
		84	}
		85	sub(/^#+[ \t]*/, "", $0)
61	}	86	}
62		87
63	/^[0-9]+\./ { # Ordered lists	88	/^$/ {
64	if (tag == "raw") next;	89	if (tag == "raw") {
65	if (! (tag == "ol")) tag = "ol";	90	next
66	sub(/^[0-9]+\.[ \t]/, "<li>");	91	}
		92	buflush()
		93	tag = ""
67	}	94	}
68		95
69	/^>/ { # Blockquotes	96	/./ {
70	if (tag == "raw") next;	97	if (! tag) {
71	if (! (tag == "blockquote")) tag = "blockquote";	98	tag = "p"
72	sub(/^>[ \t]*/,"");	99	}
		100	if (! BUF) {
		101	bufpush("<" tag ">")
		102	}
		103	if (tag == "raw") {
		104	$0 = esc($0)
		105	}
		106	bufpush($0)
73	}	107	}
74		108
75	/^#+/ { # Headers	109	END {
76	if (tag == "raw") next;	110	buflush()
77	match($0, /^#+/);
78	if (! (tag == "h" RLENGTH)) {
79	buflush();
80	tag = "h" RLENGTH;
81	}
82	sub(/^#+[ \t]*/,"");
83	}	111	}
84		112
85	/^$/ {	113
86	if (tag == "raw") next;	114	function buflush()
87	buflush();	115	{
88	tag = "";	116	if (BUF) {
		117	print BUF
		118	}
		119	BUF = ""
		120	if (tag && (tag != "html") && (tag != "raw")) {
		121	print "</" tag ">"
		122	}
89	}	123	}
90		124
91	/./ {	125	function bufpush(s)
92	if (! tag) tag = "p";	126	{
93	if (! BUF) bufpush("<" tag ">");	127	BUF = BUF (BUF ? "\n" : "") s
94	if (tag == "raw") $0 = esc($0);
95	bufpush($0);
96	}	128	}
97		129
98	END { buflush(); }	130	function esc(t)
		131	{
		132	# This is of much more limited utility than I initially realized.
		133	gsub(/&/, "\\&", t)
		134	gsub(/</, "\\<", t)
		135	gsub(/>/, "\\>", t)
		136	sub(/^ /, "\\ ", t)
		137	return t
		138	}