Reject IRIs with colons in the first path segment

3 files changed, 20 insertions, 1 deletions

diff --git a/src/uri.c b/src/uri.c
index 7b172e42..7f849d62 100644
--- a/src/uri.c
+++ b/src/uri.c
@@ -141,7 +141,8 @@ serd_uri_parse(const uint8_t* utf8, SerdURI* out)
 {
 	*out = SERD_URI_NULL;
 
-	const uint8_t* ptr = utf8;
+	const uint8_t* ptr            = utf8;
+	bool           path_has_slash = false;
 
 	/* See http://tools.ietf.org/html/rfc3986#section-3
 	   URI = scheme ":" hier-part [ "?" query ] [ "#" fragment ]
@@ -203,6 +204,15 @@ path:
 		switch (c) {
 		case '?': goto query;
 		case '#': goto fragment;
+		case '/':
+			++out->path.len;
+			path_has_slash = true;
+			break;
+		case ':':
+			if (!path_has_slash) {
+				// RFC3987 S2.2: isegment-nz-nc can not contain a colon
+				return SERD_ERR_BAD_SYNTAX;
+			}
 		default:
 			++out->path.len;
 		}
diff --git a/tests/bad/bad-colon-in-first-path-segment.ttl b/tests/bad/bad-colon-in-first-path-segment.ttl
new file mode 100644
index 00000000..ceebdb81
--- /dev/null
+++ b/tests/bad/bad-colon-in-first-path-segment.ttl
@@ -0,0 +1,3 @@
+@prefix eg: <http://example.org/> .
+
+<_:x> <eg:x> <eg:x> .
diff --git a/tests/bad/manifest.ttl b/tests/bad/manifest.ttl
index 1e21397d..9600acee 100644
--- a/tests/bad/manifest.ttl
+++ b/tests/bad/manifest.ttl
@@ -28,6 +28,7 @@
 		<#bad-char-in-local>
 		<#bad-char-in-prefix>
 		<#bad-char-in-uri>
+		<#bad-colon-in-first-path-segment>
 		<#bad-datatype>
 		<#bad-dot-after-subject>
 		<#bad-eof-in-blank>
@@ -174,6 +175,11 @@
 	mf:name "bad-char-in-uri" ;
 	mf:action <bad-char-in-uri.ttl> .
 
+<#bad-colon-in-first-path-segment>
+	rdf:type rdft:TestTurtleNegativeSyntax ;
+	mf:name "bad-colon-in-first-path-segment" ;
+	mf:action <bad-colon-in-first-path-segment.ttl> .
+
 <#bad-datatype>
 	rdf:type rdft:TestTurtleNegativeSyntax ;
 	mf:name "bad-datatype" ;