sip,uri,test: Escape SIP URIs (#740)

* Escape SIP URIs

According to RFC 3261, SIP URIs need to escape UTF-8 characters.

* uric & print: use %W for printing escaped characters

* uri: remove ipv6 ifdef checks

* uric: cast isalnum argument

include/re_uri.h

@@ -43,3 +43,6 @@ int uri_param_escape(struct re_printf *pf, const struct pl *pl);
 int uri_param_unescape(struct re_printf *pf, const struct pl *pl);
 int uri_header_escape(struct re_printf *pf, const struct pl *pl);
 int uri_header_unescape(struct re_printf *pf, const struct pl *pl);
+int uri_display_name_escape(struct re_printf *pf, const struct pl *pl);
+int uri_escape(struct re_printf *pf, const char *uri);
+int uri_escape_pl(struct re_printf *pf, const struct pl *pl);

src/sip/ctrans.c

@@ -59,7 +59,8 @@ static bool route_handler(const struct sip_hdr *hdr, const struct sip_msg *msg,
 			  void *arg)
 {
 	(void)msg;
-	return 0 != mbuf_printf(arg, "Route: %r\r\n", &hdr->val);
+	return 0 != mbuf_printf(arg, "Route: %H\r\n", uri_escape_pl,
+				&hdr->val);
 }
 
 

src/sip/dialog.c

@@ -113,16 +113,21 @@ int sip_dialog_alloc(struct sip_dialog **dlgp,
 	}
 
 	for (i=0; i<routec; i++) {
-		err |= mbuf_printf(dlg->mb, "Route: <%s;lr>\r\n", routev[i]);
+		err |= mbuf_printf(dlg->mb, "Route: <%H;lr>\r\n", uri_escape,
+				   routev[i]);
 		if (i == 0)
 			rend = dlg->mb->pos - 2;
 	}
-	err |= mbuf_printf(dlg->mb, "To: <%s>\r\n", to_uri);
+	err |= mbuf_printf(dlg->mb, "To: <%H>\r\n", uri_escape, to_uri);
 	dlg->cpos = dlg->mb->pos;
-	err |= mbuf_printf(dlg->mb, "From: %s%s%s<%s>;tag=%016llx\r\n",
-			   from_name ? "\"" : "", from_name,
-			   from_name ? "\" " : "",
-			   from_uri, ltag);
+	err |= mbuf_printf(dlg->mb, "From: ");
+	if (from_name) {
+		pl_set_str(&pl, from_name);
+		err |= mbuf_printf(dlg->mb, "\"%H\" ", uri_display_name_escape,
+				   &pl);
+	}
+	err |= mbuf_printf(dlg->mb, "<%H>", uri_escape, from_uri);
+	err |= mbuf_printf(dlg->mb, ";tag=%016llx\r\n", ltag);
 	if (err)
 		goto out;
 
@@ -156,7 +161,7 @@ static bool record_route_handler(const struct sip_hdr *hdr,
 	struct route_enc *renc = arg;
 	(void)msg;
 
-	if (mbuf_printf(renc->mb, "Route: %r\r\n", &hdr->val))
+	if (mbuf_printf(renc->mb, "Route: %H\r\n", uri_escape_pl, &hdr->val))
 		return true;
 
 	if (!renc->end)

src/sip/request.c

@@ -189,7 +189,8 @@ static int request(struct sip_request *req, enum sip_transp tp,
 		goto out;
 
 	mbuf_set_pos(mbs, 0);
-	err  = mbuf_printf(mb, "%s %s SIP/2.0\r\n", req->met, req->uri);
+	err  = mbuf_printf(mb, "%s ", req->met);
+	err |= mbuf_printf(mb, "%H SIP/2.0\r\n", uri_escape, req->uri);
 	err |= mbuf_printf(mb, "Via: SIP/2.0/%s %J;branch=%s;rport\r\n",
 			   sip_transp_name(tp), &laddr, branch);
 	err |= mbuf_write_mem(mb, mbuf_buf(mbs), mbuf_get_left(mbs));

src/uri/uri.c

@@ -49,11 +49,9 @@ int uri_encode(struct re_printf *pf, const struct uri *uri)
 	/* The IPv6 address is delimited by '[' and ']' */
 	switch (uri->af) {
 
-#ifdef HAVE_INET6
 	case AF_INET6:
 		err = re_hprintf(pf, "[%r]", &uri->host);
 		break;
-#endif
 
 	default:
 		err = re_hprintf(pf, "%r", &uri->host);
@@ -271,3 +269,82 @@ int uri_headers_apply(const struct pl *pl, uri_apply_h *ah, void *arg)
 
 	return err;
 }
+
+
+/**
+ * Escape SIP URI
+ *
+ * @param pf  Print function to encode into
+ * @param uri URI
+ *
+ * @return 0 if success, otherwise errorcode
+ */
+int uri_escape(struct re_printf *pf, const char *uri)
+{
+	struct pl pl;
+	pl_set_str(&pl, uri);
+	return uri_escape_pl(pf, &pl);
+}
+
+
+/**
+ * Escape SIP URI
+ *
+ * @param pf Print function to encode into
+ * @param pl URI
+ *
+ * @return 0 if success, otherwise errorcode
+ */
+int uri_escape_pl(struct re_printf *pf, const struct pl *pl)
+{
+	int err;
+	struct uri uri;
+
+	if (!pl)
+		return 0;
+
+	err = uri_decode(&uri, pl);
+	if (err)
+		return err;
+
+	if (!pl_isset(&uri.scheme) || !pl_isset(&uri.host))
+		return EINVAL;
+
+	err = re_hprintf(pf, "%r:", &uri.scheme);
+	if (err)
+		return err;
+
+	if (pl_isset(&uri.user)) {
+		err = re_hprintf(pf, "%H", uri_user_escape, &uri.user);
+
+		if (pl_isset(&uri.password))
+			err |= re_hprintf(pf, ":%H", uri_password_escape,
+					  &uri.password);
+
+		err |= pf->vph("@", 1, pf->arg);
+		if (err)
+			return err;
+	}
+
+	/* The IPv6 address is delimited by '[' and ']' */
+	switch (uri.af) {
+
+	case AF_INET6:
+		err = re_hprintf(pf, "[%r]", &uri.host);
+		break;
+
+	default:
+		err = re_hprintf(pf, "%r", &uri.host);
+		break;
+	}
+	if (err)
+		return err;
+
+	if (uri.port)
+		err = re_hprintf(pf, ":%u", uri.port);
+
+	err |= re_hprintf(pf, "%r%r%r", &uri.path, &uri.params,
+			  &uri.headers);
+
+	return err;
+}

src/uri/uric.c

@@ -40,7 +40,7 @@ static bool is_mark(int c)
 
 static bool is_unreserved(char c)
 {
-	return isalnum(c) || is_mark(c);
+	return isalnum((unsigned char)c) || is_mark(c);
 }
 
 
@@ -120,6 +120,33 @@ static bool is_param_unreserved(char c)
 }
 
 
+static bool is_token_non_alphanum(int c)
+{
+	switch (c) {
+
+	case '-':
+	case '.':
+	case '!':
+	case '%':
+	case '*':
+	case '_':
+	case '+':
+	case '`':
+	case '\'':
+	case '~':
+		return true;
+	default:
+		return false;
+	}
+}
+
+
+static bool is_token(char c)
+{
+	return isalnum((unsigned char)c) || is_token_non_alphanum(c);
+}
+
+
 static bool is_paramchar(char c)
 {
 	return is_param_unreserved(c) || is_unreserved(c);
@@ -147,7 +174,7 @@ static int comp_escape(struct re_printf *pf, const struct pl *pl, esc_h *eh)
 			err = pf->vph(&c, 1, pf->arg);
 		}
 		else {
-			err = re_hprintf(pf, "%%%02X", c);
+			err = re_hprintf(pf, "%%%W", &c, 1);
 		}
 	}
 
@@ -304,3 +331,17 @@ int uri_header_unescape(struct re_printf *pf, const struct pl *pl)
 {
 	return comp_unescape(pf, pl, is_hvalue);
 }
+
+
+/**
+ * Escape display name
+ *
+ * @param pf Print function
+ * @param pl String to escape
+ *
+ * @return 0 if success, otherwise errorcode
+ */
+int uri_display_name_escape(struct re_printf *pf, const struct pl *pl)
+{
+	return comp_escape(pf, pl, is_token);
+}

test/uri.c

@@ -178,7 +178,8 @@ int test_uri_user(void)
 	} uriv[] = {
 		{"alice%40atlanta.com", "[email protected]"},
 		{"project%20x", "project x"},
-		{"*21%23", "*21#"}
+		{"*21%23", "*21#"},
+		{"*21%23%C3%A4%E2%82%AC%40%20", "*21#ä\xE2\x82\xAC@ "}
 	};
 	struct mbuf mbe, mbd;
 	int err = EINVAL;