From afb45aace5d88dda7c6922b57b6e0b899ef7879e Mon Sep 17 00:00:00 2001
From: Jonas Fonseca <fonseca@diku.dk>
Date: Wed, 25 Jan 2006 18:18:01 +0100
Subject: [PATCH] Add support for scanning comment endings such as '--!>'
 correctly

---
 src/dom/sgml/scanner.c              | 20 ++++++++++++++------
 src/dom/test/test-sgml-parser-basic |  9 ++++++++-
 2 files changed, 22 insertions(+), 7 deletions(-)

diff --git a/src/dom/sgml/scanner.c b/src/dom/sgml/scanner.c
index a372c9a59..af90a27b8 100644
--- a/src/dom/sgml/scanner.c
+++ b/src/dom/sgml/scanner.c
@@ -368,12 +368,20 @@ skip_sgml_comment(struct dom_scanner *scanner, unsigned char **string,
 		/* It is always safe to access index -2 and -1 here since we
 		 * are supposed to have '<!--' before this is called. We do
 		 * however need to check that the '-->' are not overlapping any
-		 * preceeding '-'. */
-		if (pos[-2] == '-' && pos[-1] == '-' && &pos[-2] >= *string) {
-			length = pos - *string - 2;
-			*possibly_incomplete = 0;
-			pos++;
-			break;
+		 * preceeding '-'. Additionally also handle the quirky '--!>'
+		 * end sometimes found. */
+		if (pos[-2] == '-') {
+			if (pos[-1] == '-' && &pos[-2] >= *string) {
+				length = pos - *string - 2;
+				*possibly_incomplete = 0;
+				pos++;
+				break;
+			} else if (pos[-1] == '!' && pos[-3] == '-' && &pos[-3] >= *string) {
+				length = pos - *string - 3;
+				*possibly_incomplete = 0;
+				pos++;
+				break;
+			}
 		}
 	}
 
diff --git a/src/dom/test/test-sgml-parser-basic b/src/dom/test/test-sgml-parser-basic
index a203a78ac..1a22b7fed 100755
--- a/src/dom/test/test-sgml-parser-basic
+++ b/src/dom/test/test-sgml-parser-basic
@@ -93,11 +93,18 @@ test_output_equals \
 #text: u'
 
 test_output_equals \
-'Parse bad comment.' \
+'Parse bad comment. (I)' \
 '<!--->s' \
 '
 #comment: ->s'
 
+test_output_equals \
+'Parse bad comment. (II)' \
+'<!--a--!>bad comment' \
+'
+#comment: a
+#text: bad comment'
+
 test_output_equals \
 'Parse empty notation.' \
 '<!>s' \