From 7489c134f712eaf5d003ce40cc8223c41e05e3fa Mon Sep 17 00:00:00 2001
From: Jonas Fonseca <fonseca@diku.dk>
Date: Sat, 31 Dec 2005 01:47:57 +0100
Subject: [PATCH] Make non-terminated comments and cdata sections have 'the
 rest' as content

---
 src/dom/sgml/scanner.c              | 10 ++++++++--
 src/dom/test/test-sgml-parser-basic | 21 +++++++++++++++++----
 2 files changed, 25 insertions(+), 6 deletions(-)

diff --git a/src/dom/sgml/scanner.c b/src/dom/sgml/scanner.c
index 0da143c64..34fb141d7 100644
--- a/src/dom/sgml/scanner.c
+++ b/src/dom/sgml/scanner.c
@@ -205,7 +205,10 @@ skip_comment(struct dom_scanner *scanner, unsigned char **string)
 		}
 	}
 
-	if (!pos) pos = scanner->end;
+	if (!pos) {
+		pos = scanner->end;
+		length = pos - *string;
+	}
 
 	*string = pos;
 	return length;
@@ -227,7 +230,10 @@ skip_cdata_section(struct dom_scanner *scanner, unsigned char **string)
 		}
 	}
 
-	if (!pos) pos = scanner->end;
+	if (!pos) {
+		pos = scanner->end;
+		length = pos - *string;
+	}
 
 	*string = pos;
 	return length;
diff --git a/src/dom/test/test-sgml-parser-basic b/src/dom/test/test-sgml-parser-basic
index c9ba46775..58635e8eb 100755
--- a/src/dom/test/test-sgml-parser-basic
+++ b/src/dom/test/test-sgml-parser-basic
@@ -64,17 +64,18 @@ element: root
 
 test_output_equals \
 'Parse comment combinations.' \
-'<root><!-- <!-- -- > --><!----></root>' \
+'<root><!-- <!-- -- > --><!--foo--><!----></root>' \
 '
 element: root
   #comment:  <!-- -- > 
+  #comment: foo
   #comment: '
 
 test_output_equals \
-'Parse comment combinations.' \
-'<!--foo-->' \
+'Parse bad comment.' \
+'<!--->s' \
 '
-#comment: foo'
+#comment: ->s'
 
 test_output_equals \
 'Parse an enclosed CDATA section.' \
@@ -83,6 +84,18 @@ test_output_equals \
 element: root
   #cdata-section: ...] ]>...'
 
+test_output_equals \
+'Parse non-enclosed CDATA section.' \
+'<![CDATA[...]]>' \
+'
+#cdata-section: ...'
+
+test_output_equals \
+'Parse a bad CDATA section.' \
+'<![CDATA[...' \
+'
+#cdata-section: ...'
+
 test_output_equals \
 'Parse attributes.' \
 '<root lang="fr" attr name="value with &foo; <stuff"></root>' \