intern URL scheme

3 files changed, 60 insertions, 24 deletions

diff --git a/include/nng/nng.h b/include/nng/nng.h
index a6f5dc9a..a41deeb4 100644
--- a/include/nng/nng.h
+++ b/include/nng/nng.h
@@ -1083,16 +1083,16 @@ enum nng_errno_enum {
 // give us a convenient way of doing so.
 
 typedef struct nng_url {
-	char *u_rawurl;   // never NULL
-	char *u_scheme;   // never NULL
-	char *u_userinfo; // will be NULL if not specified
-	char *u_host;     // including colon and port
-	char *u_hostname; // name only, will be "" if not specified
-	char *u_port;     // port, will be "" if not specified
-	char *u_path;     // path, will be "" if not specified
-	char *u_query;    // without '?', will be NULL if not specified
-	char *u_fragment; // without '#', will be NULL if not specified
-	char *u_requri;   // includes query and fragment, "" if not specified
+	char       *u_rawurl;   // never NULL
+	const char *u_scheme;   // never NULL
+	char       *u_userinfo; // will be NULL if not specified
+	char       *u_host;     // including colon and port
+	char       *u_hostname; // name only, will be "" if not specified
+	char       *u_port;     // port, will be "" if not specified
+	char       *u_path;     // path, will be "" if not specified
+	char       *u_query;    // without '?', will be NULL if not specified
+	char       *u_fragment; // without '#', will be NULL if not specified
+	char *u_requri; // includes query and fragment, "" if not specified
 } nng_url;
 
 // nng_url_parse parses a URL string into a structured form.
diff --git a/src/core/url.c b/src/core/url.c
index 074be145..c1318719 100644
--- a/src/core/url.c
+++ b/src/core/url.c
@@ -253,6 +253,40 @@ static struct {
 	// clang-format on
 };
 
+// List of schemes that we recognize.  We don't support them all.
+static const char *nni_schemes[] = {
+	"http",
+	"https",
+	"tcp",
+	"tcp4",
+	"tcp6",
+	"tls+tcp",
+	"tls+tcp4",
+	"tls+tcp6",
+	"socket",
+	"inproc",
+	"ipc",
+	"unix",
+	"abstract",
+	"ws",
+	"ws4",
+	"ws6",
+	"wss",
+	"wss4",
+	"wss6",
+	// we don't support these
+	"file",
+	"mailto",
+	"gopher",
+	"ftp",
+	"ssh",
+	"git",
+	"telnet",
+	"irc",
+	"imap",
+	NULL,
+};
+
 const char *
 nni_url_default_port(const char *scheme)
 {
@@ -321,14 +355,16 @@ nni_url_parse(nni_url **urlp, const char *raw)
 		goto error;
 	}
 
-	if ((url->u_scheme = nni_alloc(len + 1)) == NULL) {
-		rv = NNG_ENOMEM;
-		goto error;
+	for (int i = 0; nni_schemes[i] != NULL; i++) {
+		if (strncmp(s, nni_schemes[i], len) == 0) {
+			url->u_scheme = nni_schemes[i];
+			break;
+		}
 	}
-	for (size_t i = 0; i < len; i++) {
-		url->u_scheme[i] = (char) tolower(s[i]);
+	if (url->u_scheme == NULL) {
+		rv = NNG_ENOTSUP;
+		goto error;
 	}
-	url->u_scheme[len] = '\0';
 	s += len + 3; // strlen("://")
 
 	// For compatibility reasons, we treat ipc:// and inproc:// paths
@@ -507,7 +543,6 @@ nni_url_free(nni_url *url)
 {
 	if (url != NULL) {
 		nni_strfree(url->u_rawurl);
-		nni_strfree(url->u_scheme);
 		nni_strfree(url->u_userinfo);
 		nni_strfree(url->u_host);
 		nni_strfree(url->u_hostname);
@@ -580,7 +615,6 @@ nni_url_clone(nni_url **dstp, const nni_url *src)
 		return (NNG_ENOMEM);
 	}
 	if (URL_COPYSTR(dst->u_rawurl, src->u_rawurl) ||
-	    URL_COPYSTR(dst->u_scheme, src->u_scheme) ||
 	    URL_COPYSTR(dst->u_userinfo, src->u_userinfo) ||
 	    URL_COPYSTR(dst->u_host, src->u_host) ||
 	    URL_COPYSTR(dst->u_hostname, src->u_hostname) ||
@@ -592,7 +626,8 @@ nni_url_clone(nni_url **dstp, const nni_url *src)
 		nni_url_free(dst);
 		return (NNG_ENOMEM);
 	}
-	*dstp = dst;
+	dst->u_scheme = src->u_scheme;
+	*dstp         = dst;
 	return (0);
 }
 
diff --git a/src/core/url_test.c b/src/core/url_test.c
index 6504ff6c..232cabc8 100644
--- a/src/core/url_test.c
+++ b/src/core/url_test.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>
 //
 // This software is supplied under the terms of the MIT License, a
@@ -54,8 +54,7 @@ test_url_host_port_path(void)
 {
 	nng_url *url;
 
-	NUTS_PASS(
-	    nng_url_parse(&url, "http://www.google.com:1234/somewhere"));
+	NUTS_PASS(nng_url_parse(&url, "http://www.google.com:1234/somewhere"));
 	NUTS_ASSERT(url != NULL);
 	NUTS_TRUE(strcmp(url->u_scheme, "http") == 0);
 	NUTS_TRUE(strcmp(url->u_host, "www.google.com:1234") == 0);
@@ -310,6 +309,8 @@ test_url_bad_scheme(void)
 	NUTS_NULL(url);
 	NUTS_FAIL(nng_url_parse(&url, "http:www.google.com"), NNG_EINVAL);
 	NUTS_NULL(url);
+	NUTS_FAIL(nng_url_parse(&url, "nosuch://bogus"), NNG_ENOTSUP);
+	NUTS_NULL(url);
 }
 
 void
@@ -327,7 +328,7 @@ test_url_canonify(void)
 {
 	nng_url *url = NULL;
 	NUTS_PASS(nng_url_parse(
-	    &url, "hTTp://www.EXAMPLE.com/bogus/.%2e/%7egarrett"));
+	    &url, "http://www.EXAMPLE.com/bogus/.%2e/%7egarrett"));
 	NUTS_ASSERT(url != NULL);
 	NUTS_MATCH(url->u_scheme, "http");
 	NUTS_MATCH(url->u_hostname, "www.example.com");
@@ -469,4 +470,4 @@ NUTS_TESTS = {
 	{ "url good utf8", test_url_good_utf8 },
 	{ "url decode", test_url_decode },
 	{ NULL, NULL },
-};
\ No newline at end of file
+};