From f02261b71dc27edc6cf9bd241262dbd0923a1a3e Mon Sep 17 00:00:00 2001
From: anmartan <a.martin@zaleos.net>
Date: Thu, 4 Jul 2024 09:30:02 +0200
Subject: [PATCH] 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.
---
 src/core/parser/parse_uri.c | 53 +++++++++++++++++++++----------------
 1 file changed, 30 insertions(+), 23 deletions(-)

diff --git a/src/core/parser/parse_uri.c b/src/core/parser/parse_uri.c
index 9ccb62ee51e2..026bb0c624d8 100644
--- a/src/core/parser/parse_uri.c
+++ b/src/core/parser/parse_uri.c
@@ -186,6 +186,7 @@ int parse_uri(char *buf, int len, struct sip_uri *uri)
 	char *end;
 	char *pass;
 	int found_user;
+	int found_host;
 	int error_headers;
 	uint32_t scheme;
 	uri_type backup_urit;
@@ -237,28 +238,33 @@ 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 || found_host == 0) { \
+			uri->host.s = s;                       \
+			found_host = 1;                        \
+		}                                          \
+		uri->host.len = p - uri->host.s;           \
+		if(scheme != URN_SCH) {                    \
+			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 '@':                                      \
 		goto error_bad_char
 
 
@@ -445,6 +451,7 @@ int parse_uri(char *buf, int len, struct sip_uri *uri)
 	end = buf + len;
 	p = buf + 4;
 	found_user = 0;
+	found_host = 0;
 	error_headers = 0;
 	b = v = 0;
 	param = param_val = 0;
@@ -493,7 +500,7 @@ 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: