URL u_port should be a number not a string.

The idea here is to reduce the dynamic allocations used for
URLs, and also the back and forth with parsing begin strings
and port numbers.  We always resolve to a port number, and
this is easier for everyone.

The real goal in the long term is to eliminate dynamic allocation
of the URL fields altogether, but that requires a little more
work.  This is a step in the right direction.

5 files changed, 104 insertions, 81 deletions

diff --git a/src/core/platform.h b/src/core/platform.h
index 62a321bf..783cc9fc 100644
--- a/src/core/platform.h
+++ b/src/core/platform.h
@@ -343,7 +343,7 @@ extern int nni_tcp_listener_get(
 // Symbolic service names will be looked up assuming SOCK_STREAM, so
 // they may not work with UDP.
 extern void nni_resolv_ip(
-    const char *, const char *, int, bool, nng_sockaddr *sa, nni_aio *);
+    const char *, uint16_t, int, bool, nng_sockaddr *sa, nni_aio *);
 
 // nni_parse_ip parses an IP address, without a port.
 extern int nni_parse_ip(const char *, nng_sockaddr *);
@@ -351,6 +351,10 @@ extern int nni_parse_ip(const char *, nng_sockaddr *);
 // nni_parse_ip_port parses an IP address with an optional port appended.
 extern int nni_parse_ip_port(const char *, nng_sockaddr *);
 
+// nni_get_port_by_name resolves a name (which may be an ASCII representation
+// of a number) to a port number (the value returned is in native byte order.)
+extern int nni_get_port_by_name(const char *, uint16_t *);
+
 //
 // IPC (UNIX Domain Sockets & Named Pipes) Support.
 //
diff --git a/src/core/tcp.c b/src/core/tcp.c
index 7fb67228..5d324a13 100644
--- a/src/core/tcp.c
+++ b/src/core/tcp.c
@@ -1,5 +1,5 @@
 //
-// Copyright 2020 Staysail Systems, Inc. <info@staysail.tech>
+// Copyright 2024 Staysail Systems, Inc. <info@staysail.tech>
 // Copyright 2018 Capitar IT Group BV <info@capitar.com>
 // Copyright 2019 Devolutions <info@devolutions.net>
 //
@@ -10,6 +10,7 @@
 //
 
 #include <stdint.h>
+#include <stdio.h>
 #include <string.h>
 
 #include <nng/nng.h>
@@ -20,8 +21,8 @@
 
 typedef struct {
 	nng_stream_dialer ops;
-	char             *host;
-	char             *port;
+	char              host[256];
+	uint16_t          port;
 	int               af; // address family
 	bool              closed;
 	nng_sockaddr      sa;
@@ -155,8 +156,6 @@ tcp_dialer_free(void *arg)
 		nni_tcp_dialer_fini(d->d);
 	}
 	nni_mtx_fini(&d->mtx);
-	nni_strfree(d->host);
-	nni_strfree(d->port);
 	NNI_FREE_STRUCT(d);
 }
 
@@ -236,17 +235,13 @@ nni_tcp_dialer_alloc(nng_stream_dialer **dp, const nng_url *url)
 {
 	tcp_dialer *d;
 	int         rv;
-	const char *p;
 
 	if ((rv = tcp_dialer_alloc(&d)) != 0) {
 		return (rv);
 	}
 
-	if (((p = url->u_port) == NULL) || (strlen(p) == 0)) {
-		p = nni_url_default_port(url->u_scheme);
-	}
-
-	if ((strlen(p) == 0) || (strlen(url->u_hostname) == 0)) {
+	if ((url->u_port == 0) || strlen(url->u_hostname) == 0 ||
+	    strlen(url->u_hostname) >= sizeof(d->host)) {
 		// Dialer needs both a destination hostname and port.
 		tcp_dialer_free(d);
 		return (NNG_EADDRINVAL);
@@ -260,11 +255,8 @@ nni_tcp_dialer_alloc(nng_stream_dialer **dp, const nng_url *url)
 		d->af = NNG_AF_UNSPEC;
 	}
 
-	if (((d->host = nng_strdup(url->u_hostname)) == NULL) ||
-	    ((d->port = nng_strdup(p)) == NULL)) {
-		tcp_dialer_free(d);
-		return (NNG_ENOMEM);
-	}
+	snprintf(d->host, sizeof(d->host), "%s", url->u_hostname);
+	d->port = url->u_port;
 
 	*dp = (void *) d;
 	return (0);
diff --git a/src/core/url.c b/src/core/url.c
index c1318719..9bdcfd23 100644
--- a/src/core/url.c
+++ b/src/core/url.c
@@ -15,6 +15,7 @@
 #include <stdio.h>
 #include <string.h>
 
+#include "core/platform.h"
 #include "url.h"
 
 static uint8_t
@@ -234,22 +235,26 @@ url_canonify_uri(char **outp, const char *in)
 
 static struct {
 	const char *scheme;
-	const char *port;
+	uint16_t    port;
 } nni_url_default_ports[] = {
 	// This list is not exhaustive, but likely covers the main ones we
 	// care about.  Feel free to add additional ones as use cases arise.
 	// Note also that we don't use "default" ports for SP protocols
 	// that have no "default" port, like tcp:// or tls+tcp://.
 	// clang-format off
-	{ "git", "9418" },
-	{ "gopher", "70" },
-	{ "http", "80" },
-	{ "https", "443" },
-	{ "ssh", "22" },
-	{ "telnet", "23" },
-	{ "ws", "80" },
-	{ "wss", "443" },
-	{ NULL, NULL },
+	{ "git", 9418 },
+	{ "gopher", 70 },
+	{ "http", 80 },
+	{ "https", 443 },
+	{ "ssh", 22 },
+	{ "telnet", 23 },
+	{ "ws", 80 },
+	{ "ws4", 80 },
+	{ "ws6", 80 },
+	{ "wss", 443 },
+	{ "wss4", 443 },
+	{ "wss6", 443 },
+	{ NULL, 0 },
 	// clang-format on
 };
 
@@ -274,6 +279,9 @@ static const char *nni_schemes[] = {
 	"wss",
 	"wss4",
 	"wss6",
+	"udp",
+	"udp4",
+	"udp6",
 	// we don't support these
 	"file",
 	"mailto",
@@ -284,10 +292,11 @@ static const char *nni_schemes[] = {
 	"telnet",
 	"irc",
 	"imap",
+	"imaps",
 	NULL,
 };
 
-const char *
+uint16_t
 nni_url_default_port(const char *scheme)
 {
 	const char *s;
@@ -310,7 +319,7 @@ nni_url_default_port(const char *scheme)
 			break;
 		}
 	}
-	return ("");
+	return (0);
 }
 
 // URLs usually follow the following format:
@@ -503,6 +512,11 @@ nni_url_parse(nni_url **urlp, const char *raw)
 			}
 		}
 	}
+	// hostname length check
+	if (len >= 256) {
+		rv = NNG_EADDRINVAL;
+		goto error;
+	}
 	if ((url->u_hostname = nni_alloc(len + 1)) == NULL) {
 		rv = NNG_ENOMEM;
 		goto error;
@@ -521,13 +535,12 @@ nni_url_parse(nni_url **urlp, const char *raw)
 			rv = NNG_EINVAL;
 			goto error;
 		}
-		url->u_port = nni_strdup(s + 1);
+		rv = nni_get_port_by_name(s + 1, &url->u_port);
+		if (rv != 0) {
+			goto error;
+		}
 	} else {
-		url->u_port = nni_strdup(nni_url_default_port(url->u_scheme));
-	}
-	if (url->u_port == NULL) {
-		rv = NNG_ENOMEM;
-		goto error;
+		url->u_port = nni_url_default_port(url->u_scheme);
 	}
 
 	*urlp = url;
@@ -546,7 +559,6 @@ nni_url_free(nni_url *url)
 		nni_strfree(url->u_userinfo);
 		nni_strfree(url->u_host);
 		nni_strfree(url->u_hostname);
-		nni_strfree(url->u_port);
 		nni_strfree(url->u_path);
 		nni_strfree(url->u_query);
 		nni_strfree(url->u_fragment);
@@ -558,11 +570,11 @@ nni_url_free(nni_url *url)
 int
 nni_url_asprintf(char **str, const nni_url *url)
 {
-	const char *scheme = url->u_scheme;
-	const char *port   = url->u_port;
-	const char *host   = url->u_hostname;
-	const char *hostob = "";
-	const char *hostcb = "";
+	const char *scheme  = url->u_scheme;
+	const char *host    = url->u_hostname;
+	const char *hostob  = "";
+	const char *hostcb  = "";
+	bool        do_port = true;
 
 	if ((strcmp(scheme, "ipc") == 0) || (strcmp(scheme, "inproc") == 0) ||
 	    (strcmp(scheme, "unix") == 0) ||
@@ -570,11 +582,8 @@ nni_url_asprintf(char **str, const nni_url *url)
 		return (nni_asprintf(str, "%s://%s", scheme, url->u_path));
 	}
 
-	if (port != NULL) {
-		if ((strlen(port) == 0) ||
-		    (strcmp(nni_url_default_port(scheme), port) == 0)) {
-			port = NULL;
-		}
+	if (url->u_port == nni_url_default_port(scheme)) {
+		do_port = false;
 	}
 	if (strcmp(host, "*") == 0) {
 		host = "";
@@ -583,9 +592,14 @@ nni_url_asprintf(char **str, const nni_url *url)
 		hostob = "[";
 		hostcb = "]";
 	}
-	return (nni_asprintf(str, "%s://%s%s%s%s%s%s", scheme, hostob, host,
-	    hostcb, port != NULL ? ":" : "", port != NULL ? port : "",
-	    url->u_requri != NULL ? url->u_requri : ""));
+	char portstr[8];
+	if (do_port) {
+		snprintf(portstr, sizeof(portstr), ":%u", url->u_port);
+	} else {
+		portstr[0] = 0;
+	}
+	return (nni_asprintf(str, "%s://%s%s%s%s%s", scheme, hostob, host,
+	    hostcb, portstr, url->u_requri != NULL ? url->u_requri : ""));
 }
 
 // nni_url_asprintf_port is like nni_url_asprintf, but includes a port
@@ -594,12 +608,10 @@ nni_url_asprintf(char **str, const nni_url *url)
 int
 nni_url_asprintf_port(char **str, const nni_url *url, int port)
 {
-	char    portstr[16];
 	nni_url myurl = *url;
 
 	if (port > 0) {
-		(void) snprintf(portstr, sizeof(portstr), "%d", port);
-		myurl.u_port = portstr;
+		myurl.u_port = (uint16_t) port;
 	}
 	return (nni_url_asprintf(str, &myurl));
 }
@@ -618,7 +630,6 @@ nni_url_clone(nni_url **dstp, const nni_url *src)
 	    URL_COPYSTR(dst->u_userinfo, src->u_userinfo) ||
 	    URL_COPYSTR(dst->u_host, src->u_host) ||
 	    URL_COPYSTR(dst->u_hostname, src->u_hostname) ||
-	    URL_COPYSTR(dst->u_port, src->u_port) ||
 	    URL_COPYSTR(dst->u_requri, src->u_requri) ||
 	    URL_COPYSTR(dst->u_path, src->u_path) ||
 	    URL_COPYSTR(dst->u_query, src->u_query) ||
@@ -627,6 +638,7 @@ nni_url_clone(nni_url **dstp, const nni_url *src)
 		return (NNG_ENOMEM);
 	}
 	dst->u_scheme = src->u_scheme;
+	dst->u_port   = src->u_port;
 	*dstp         = dst;
 	return (0);
 }
diff --git a/src/core/url.h b/src/core/url.h
index 1761f6fd..71a093d0 100644
--- a/src/core/url.h
+++ b/src/core/url.h
@@ -13,13 +13,13 @@
 
 #include "core/defs.h"
 
-extern int         nni_url_parse(nni_url **, const char *path);
-extern void        nni_url_free(nni_url *);
-extern int         nni_url_clone(nni_url **, const nni_url *);
-extern const char *nni_url_default_port(const char *);
-extern int         nni_url_asprintf(char **, const nni_url *);
-extern int         nni_url_asprintf_port(char **, const nni_url *, int);
-extern size_t      nni_url_decode(uint8_t *, const char *, size_t);
-extern int         nni_url_to_address(nng_sockaddr *, const nni_url *);
+extern int      nni_url_parse(nni_url **, const char *path);
+extern void     nni_url_free(nni_url *);
+extern int      nni_url_clone(nni_url **, const nni_url *);
+extern uint16_t nni_url_default_port(const char *);
+extern int      nni_url_asprintf(char **, const nni_url *);
+extern int      nni_url_asprintf_port(char **, const nni_url *, int);
+extern size_t   nni_url_decode(uint8_t *, const char *, size_t);
+extern int      nni_url_to_address(nng_sockaddr *, const nni_url *);
 
 #endif // CORE_URL_H
diff --git a/src/core/url_test.c b/src/core/url_test.c
index 232cabc8..b5114551 100644
--- a/src/core/url_test.c
+++ b/src/core/url_test.c
@@ -22,7 +22,7 @@ test_url_host(void)
 	NUTS_TRUE(strcmp(url->u_scheme, "http") == 0);
 	NUTS_TRUE(strcmp(url->u_host, "www.google.com") == 0);
 	NUTS_TRUE(strcmp(url->u_hostname, "www.google.com") == 0);
-	NUTS_TRUE(strcmp(url->u_port, "80") == 0);
+	NUTS_TRUE(url->u_port == 80);
 	NUTS_TRUE(strcmp(url->u_path, "") == 0);
 	NUTS_TRUE(strcmp(url->u_requri, "") == 0);
 	NUTS_TRUE(url->u_query == NULL);
@@ -32,6 +32,20 @@ test_url_host(void)
 }
 
 void
+test_url_host_too_long(void)
+{
+	nng_url *url;
+	char     buffer[512]; //
+
+	memset(buffer, 0, sizeof(buffer));
+	snprintf(buffer, sizeof(buffer), "http://");
+	for (size_t i = strlen(buffer); i < sizeof(buffer) - 1; i++) {
+		buffer[i] = 'a';
+	}
+	NUTS_FAIL(nng_url_parse(&url, buffer), NNG_EADDRINVAL);
+}
+
+void
 test_url_host_port(void)
 {
 	nng_url *url;
@@ -40,7 +54,7 @@ test_url_host_port(void)
 	NUTS_TRUE(strcmp(url->u_scheme, "http") == 0);
 	NUTS_TRUE(strcmp(url->u_host, "www.google.com:1234") == 0);
 	NUTS_TRUE(strcmp(url->u_hostname, "www.google.com") == 0);
-	NUTS_TRUE(strcmp(url->u_port, "1234") == 0);
+	NUTS_TRUE(url->u_port == 1234);
 	NUTS_TRUE(strcmp(url->u_path, "") == 0);
 	NUTS_TRUE(strcmp(url->u_requri, "") == 0);
 	NUTS_TRUE(url->u_query == NULL);
@@ -59,7 +73,7 @@ test_url_host_port_path(void)
 	NUTS_TRUE(strcmp(url->u_scheme, "http") == 0);
 	NUTS_TRUE(strcmp(url->u_host, "www.google.com:1234") == 0);
 	NUTS_TRUE(strcmp(url->u_hostname, "www.google.com") == 0);
-	NUTS_TRUE(strcmp(url->u_port, "1234") == 0);
+	NUTS_TRUE(url->u_port == 1234);
 	NUTS_TRUE(strcmp(url->u_path, "/somewhere") == 0);
 	NUTS_TRUE(strcmp(url->u_requri, "/somewhere") == 0);
 	NUTS_TRUE(url->u_userinfo == NULL);
@@ -79,7 +93,7 @@ test_url_user_info(void)
 	NUTS_MATCH(url->u_userinfo, "garrett");
 	NUTS_MATCH(url->u_host, "www.google.com:1234");
 	NUTS_MATCH(url->u_hostname, "www.google.com");
-	NUTS_MATCH(url->u_port, "1234");
+	NUTS_TRUE(url->u_port == 1234);
 	NUTS_MATCH(url->u_path, "/somewhere");
 	NUTS_MATCH(url->u_requri, "/somewhere");
 	NUTS_NULL(url->u_query);
@@ -97,7 +111,7 @@ test_url_path_query_param(void)
 	NUTS_MATCH(url->u_scheme, "http");
 	NUTS_MATCH(url->u_host, "www.google.com");
 	NUTS_MATCH(url->u_hostname, "www.google.com");
-	NUTS_MATCH(url->u_port, "80");
+	NUTS_TRUE(url->u_port == 80);
 	NUTS_MATCH(url->u_path, "/somewhere");
 	NUTS_MATCH(url->u_query, "result=yes");
 	NUTS_MATCH(url->u_requri, "/somewhere?result=yes");
@@ -117,7 +131,7 @@ test_url_query_param_anchor(void)
 	NUTS_MATCH(url->u_scheme, "http");
 	NUTS_MATCH(url->u_host, "www.google.com");
 	NUTS_MATCH(url->u_hostname, "www.google.com");
-	NUTS_MATCH(url->u_port, "80");
+	NUTS_TRUE(url->u_port == 80);
 	NUTS_MATCH(url->u_path, "/somewhere");
 	NUTS_MATCH(url->u_query, "result=yes");
 	NUTS_MATCH(url->u_fragment, "chapter1");
@@ -136,7 +150,7 @@ test_url_path_anchor(void)
 	NUTS_MATCH(url->u_scheme, "http");
 	NUTS_MATCH(url->u_host, "www.google.com");
 	NUTS_MATCH(url->u_hostname, "www.google.com");
-	NUTS_MATCH(url->u_port, "80");
+	NUTS_TRUE(url->u_port == 80);
 	NUTS_MATCH(url->u_path, "/somewhere");
 	NUTS_MATCH(url->u_fragment, "chapter2");
 	NUTS_MATCH(url->u_requri, "/somewhere#chapter2");
@@ -155,7 +169,7 @@ test_url_anchor(void)
 	NUTS_MATCH(url->u_host, "www.google.com");
 	NUTS_MATCH(url->u_hostname, "www.google.com");
 	NUTS_MATCH(url->u_path, "");
-	NUTS_MATCH(url->u_port, "80");
+	NUTS_TRUE(url->u_port == 80);
 	NUTS_MATCH(url->u_fragment, "chapter3");
 	NUTS_MATCH(url->u_requri, "#chapter3");
 	NUTS_NULL(url->u_query);
@@ -173,7 +187,7 @@ test_url_query_param(void)
 	NUTS_MATCH(url->u_host, "www.google.com");
 	NUTS_MATCH(url->u_hostname, "www.google.com");
 	NUTS_MATCH(url->u_path, "");
-	NUTS_MATCH(url->u_port, "80");
+	NUTS_TRUE(url->u_port == 80);
 	NUTS_MATCH(url->u_query, "color=red");
 	NUTS_MATCH(url->u_requri, "?color=red");
 	NUTS_ASSERT(url != NULL);
@@ -191,7 +205,7 @@ test_url_v6_host(void)
 	NUTS_MATCH(url->u_host, "[::1]");
 	NUTS_MATCH(url->u_hostname, "::1");
 	NUTS_MATCH(url->u_path, "");
-	NUTS_MATCH(url->u_port, "80");
+	NUTS_TRUE(url->u_port == 80);
 	NUTS_NULL(url->u_query);
 	NUTS_NULL(url->u_fragment);
 	NUTS_NULL(url->u_userinfo);
@@ -208,7 +222,7 @@ test_url_v6_host_port(void)
 	NUTS_MATCH(url->u_host, "[::1]:29");
 	NUTS_MATCH(url->u_hostname, "::1");
 	NUTS_MATCH(url->u_path, "");
-	NUTS_MATCH(url->u_port, "29");
+	NUTS_TRUE(url->u_port == 29);
 	NUTS_NULL(url->u_query);
 	NUTS_NULL(url->u_fragment);
 	NUTS_NULL(url->u_userinfo);
@@ -225,7 +239,7 @@ test_url_v6_host_port_path(void)
 	NUTS_MATCH(url->u_host, "[::1]:29");
 	NUTS_MATCH(url->u_hostname, "::1");
 	NUTS_MATCH(url->u_path, "/bottles");
-	NUTS_MATCH(url->u_port, "29");
+	NUTS_TRUE(url->u_port == 29);
 	NUTS_NULL(url->u_query);
 	NUTS_NULL(url->u_fragment);
 	NUTS_NULL(url->u_userinfo);
@@ -242,7 +256,7 @@ test_url_tcp_port(void)
 	NUTS_MATCH(url->u_host, ":9876");
 	NUTS_MATCH(url->u_hostname, "");
 	NUTS_MATCH(url->u_path, "/");
-	NUTS_MATCH(url->u_port, "9876");
+	NUTS_TRUE(url->u_port == 9876);
 	NUTS_NULL(url->u_query);
 	NUTS_NULL(url->u_fragment);
 	NUTS_NULL(url->u_userinfo);
@@ -260,7 +274,7 @@ test_url_bare_ws(void)
 	NUTS_MATCH(url->u_host, "");
 	NUTS_MATCH(url->u_hostname, "");
 	NUTS_MATCH(url->u_path, "");
-	NUTS_MATCH(url->u_port, "80");
+	NUTS_TRUE(url->u_port == 80);
 	NUTS_NULL(url->u_query);
 	NUTS_NULL(url->u_fragment);
 	NUTS_NULL(url->u_userinfo);
@@ -277,7 +291,7 @@ test_url_ws_wildcard(void)
 	NUTS_MATCH(url->u_host, ":12345");
 	NUTS_MATCH(url->u_hostname, "");
 	NUTS_MATCH(url->u_path, "/foobar");
-	NUTS_MATCH(url->u_port, "12345");
+	NUTS_TRUE(url->u_port == 12345);
 	NUTS_NULL(url->u_query);
 	NUTS_NULL(url->u_fragment);
 	NUTS_NULL(url->u_userinfo);
@@ -294,7 +308,7 @@ test_url_ssh(void)
 	NUTS_MATCH(url->u_host, "host.example.com");
 	NUTS_MATCH(url->u_hostname, "host.example.com");
 	NUTS_MATCH(url->u_path, "");
-	NUTS_MATCH(url->u_port, "22");
+	NUTS_TRUE(url->u_port == 22);
 	NUTS_NULL(url->u_query);
 	NUTS_NULL(url->u_fragment);
 	NUTS_MATCH(url->u_userinfo, "user");
@@ -332,7 +346,7 @@ test_url_canonify(void)
 	NUTS_ASSERT(url != NULL);
 	NUTS_MATCH(url->u_scheme, "http");
 	NUTS_MATCH(url->u_hostname, "www.example.com");
-	NUTS_MATCH(url->u_port, "80");
+	NUTS_TRUE(url->u_port == 80);
 	NUTS_MATCH(url->u_path, "/~garrett");
 	nng_url_free(url);
 }
@@ -345,7 +359,7 @@ test_url_path_resolve(void)
 	    nng_url_parse(&url, "http://www.x.com//abc/def/./x/..///./../y"));
 	NUTS_MATCH(url->u_scheme, "http");
 	NUTS_MATCH(url->u_hostname, "www.x.com");
-	NUTS_MATCH(url->u_port, "80");
+	NUTS_TRUE(url->u_port == 80);
 	NUTS_MATCH(url->u_path, "/abc/y");
 	nng_url_free(url);
 }
@@ -359,7 +373,7 @@ test_url_query_info_pass(void)
 	NUTS_ASSERT(url != NULL);
 	NUTS_MATCH(url->u_scheme, "http");
 	NUTS_MATCH(url->u_hostname, "www.x.com");
-	NUTS_MATCH(url->u_port, "80");
+	NUTS_TRUE(url->u_port == 80);
 	NUTS_MATCH(url->u_path, "/");
 	NUTS_MATCH(url->u_query, "/abc/def/./x/.././../y");
 	nng_url_free(url);
@@ -383,7 +397,7 @@ test_url_good_utf8(void)
 	NUTS_ASSERT(url != NULL);
 	NUTS_MATCH(url->u_scheme, "http");
 	NUTS_MATCH(url->u_hostname, "www.x.com");
-	NUTS_MATCH(url->u_port, "80");
+	NUTS_TRUE(url->u_port == 80);
 	NUTS_MATCH(url->u_path, "/\xc2\xa2_cents");
 	nng_url_free(url);
 }
@@ -446,6 +460,7 @@ test_url_decode(void)
 
 NUTS_TESTS = {
 	{ "url host", test_url_host },
+	{ "url host too long", test_url_host_too_long },
 	{ "url host port", test_url_host_port },
 	{ "url host port path", test_url_host_port_path },
 	{ "url user info", test_url_user_info },