core: Improved URN parsing according to RFC8141

- Improved URN parsing to allow consuming URNs that contain 3 or more colons. Previosly URI parser treated some of the colons as separator between host and port causing the URN parsing to fail.
zaleos · Jul 5, 2024 · e73b321 · e73b321
1 parent dc7fc82
commit e73b321
Showing 1 changed file with 33 additions and 24 deletions.
diff --git a/src/core/parser/parse_uri.c b/src/core/parser/parse_uri.c
@@ -185,6 +185,7 @@ int parse_uri(char *buf, int len, struct sip_uri *uri)
 	register char *p;
 	char *end;
 	char *pass;
+	char* f_component;
 	int found_user;
 	int error_headers;
 	uint32_t scheme;
@@ -237,28 +238,35 @@ int parse_uri(char *buf, int len, struct sip_uri *uri)
 	} else                                                 \
 		goto error_bad_char
 
-#define check_host_end         \
-	case ':':                  \
-		/* found the host */   \
-		uri->host.s = s;       \
-		uri->host.len = p - s; \
-		state = URI_PORT;      \
-		s = p + 1;             \
-		break;                 \
-	case ';':                  \
-		uri->host.s = s;       \
-		uri->host.len = p - s; \
-		state = URI_PARAM;     \
-		s = p + 1;             \
-		break;                 \
-	case '?':                  \
-		uri->host.s = s;       \
-		uri->host.len = p - s; \
-		state = URI_HEADERS;   \
-		s = p + 1;             \
-		break;                 \
-	case '&':                  \
-	case '@':                  \
+#define check_host_end                             \
+	case ':':                                      \
+		/* found the host */                       \
+		if(scheme != URN_SCH) {                    \
+			uri->host.s = s;                       \
+			uri->host.len = p - s;                 \
+			state = URI_PORT;                      \
+			s = p + 1;                             \
+		}                                          \
+		break;                                     \
+	case ';':                                      \
+		uri->host.s = s;                           \
+		uri->host.len = p - s;                     \
+		state = URI_PARAM;                         \
+		s = p + 1;                                 \
+		break;                                     \
+	case '?':                                      \
+		uri->host.s = s;                           \
+		uri->host.len = p - s;                     \
+		state = URI_HEADERS;                       \
+		s = p + 1;                                 \
+		break;                                     \
+	case '#':                                      \
+		uri->host.s = s;                           \
+		uri->host.len = p - s;                     \
+		f_component = p;                           \
+		break;                                     \
+	case '&':                                      \
+	case '@':                                      \
 		goto error_bad_char
 
 
@@ -445,6 +453,7 @@ int parse_uri(char *buf, int len, struct sip_uri *uri)
 	end = buf + len;
 	p = buf + 4;
 	found_user = 0;
+	f_component = end;
 	error_headers = 0;
 	b = v = 0;
 	param = param_val = 0;
@@ -493,7 +502,7 @@ int parse_uri(char *buf, int len, struct sip_uri *uri)
 					case '@': /* error no user part, or
 								* be forgiving and accept it ? */
 					default:
-						state = URI_USER;
+						state = (scheme == URN_SCH) ? URI_HOST : URI_USER;
 				}
 				break;
 			case URI_USER:
@@ -1107,7 +1116,7 @@ int parse_uri(char *buf, int len, struct sip_uri *uri)
 		case URI_HOST_P:
 		case URI_HOST6_END:
 			uri->host.s = s;
-			uri->host.len = p - s;
+			uri->host.len = f_component - s;
 			break;
 		case URI_HOST:	  /* error: null host */
 		case URI_HOST6_P: /* error: unterminated ipv6 reference*/