From b7e14e7b79a73f69044d91c6af6cf8b62e82437c Mon Sep 17 00:00:00 2001
From: Hiltjo Posthuma <hiltjo@codemadness.org>
Date: Sun, 11 Mar 2018 16:32:37 +0100
Subject: xml: improve comment parsing

note that ---> is officially invalid XML, but we allow it anyway.
---
 xml.c | 39 ++++++++++++++++++++++-----------------
 1 file changed, 22 insertions(+), 17 deletions(-)

diff --git a/xml.c b/xml.c
index ec5d05c..bff793d 100644
--- a/xml.c
+++ b/xml.c
@@ -107,36 +107,41 @@ xml_parseattrs(XMLParser *x)
 static void
 xml_parsecomment(XMLParser *x)
 {
-	static const char *end = "-->";
 	size_t datalen = 0, i = 0;
-	char tmp[4];
 	int c;
 
 	if (x->xmlcommentstart)
 		x->xmlcommentstart(x);
 	while ((c = x->getnext()) != EOF) {
-		if (c == end[i]) {
-			if (end[++i] == '\0') { /* end */
+		if (c == '-' || c == '>') {
+			if (x->xmlcomment) {
 				x->data[datalen] = '\0';
+				x->xmlcomment(x, x->data, datalen);
+				datalen = 0;
+			}
+		}
+
+		if (c == '-') {
+			if (++i > 2) {
 				if (x->xmlcomment)
-					x->xmlcomment(x, x->data, datalen);
-				if (x->xmlcommentend)
-					x->xmlcommentend(x);
-				return;
+					for (; i > 2; i--)
+						x->xmlcomment(x, "-", 1);
+				i = 2;
 			}
+			continue;
+		} else if (c == '>' && i == 2) {
+			if (x->xmlcommentend)
+				x->xmlcommentend(x);
+			return;
 		} else if (i) {
 			if (x->xmlcomment) {
-				x->data[datalen] = '\0';
-				if (datalen)
-					x->xmlcomment(x, x->data, datalen);
-				memcpy(tmp, end, i);
-				tmp[i] = '\0';
-				x->xmlcomment(x, tmp, i);
+				for (; i > 0; i--)
+					x->xmlcomment(x, "-", 1);
 			}
 			i = 0;
-			x->data[0] = c;
-			datalen = 1;
-		} else if (datalen < sizeof(x->data) - 1) {
+		}
+
+		if (datalen < sizeof(x->data) - 1) {
 			x->data[datalen++] = c;
 		} else {
 			x->data[datalen] = '\0';
-- 
cgit v1.2.3