Merge pull request #1936 from libgit2/better-url-parsing

Streamline url-parsing logic.

6 files changed, 135 insertions, 104 deletions

diff --git a/CMakeLists.txt b/CMakeLists.txt
index 2440ed0b8..271af690c 100644
--- a/CMakeLists.txt
+++ b/CMakeLists.txt
@@ -128,6 +128,8 @@ INCLUDE_DIRECTORIES(src include)
 
 IF (WIN32 AND WINHTTP AND NOT MINGW)
 	ADD_DEFINITIONS(-DGIT_WINHTTP)
+	INCLUDE_DIRECTORIES(deps/http-parser)
+	FILE(GLOB SRC_HTTP deps/http-parser/*.c deps/http-parser/*.h)
 ELSE ()
 	IF (NOT AMIGA)
 		FIND_PACKAGE(OpenSSL)
diff --git a/src/netops.c b/src/netops.c
index 7e13f12e7..d7f17b1fc 100644
--- a/src/netops.c
+++ b/src/netops.c
@@ -32,6 +32,7 @@
 #include "netops.h"
 #include "posix.h"
 #include "buffer.h"
+#include "http_parser.h"
 
 #ifdef GIT_WIN32
 static void net_set_error(const char *str)
@@ -582,7 +583,7 @@ int gitno_connection_data_from_url(
 		const char *service_suffix)
 {
 	int error = -1;
-	const char *default_port = NULL;
+	const char *default_port = NULL, *path_search_start = NULL;
 	char *original_host = NULL;
 
 	/* service_suffix is optional */
@@ -594,22 +595,18 @@ int gitno_connection_data_from_url(
 	gitno_connection_data_free_ptrs(data);
 
 	if (!git__prefixcmp(url, prefix_http)) {
-		url = url + strlen(prefix_http);
+		path_search_start = url + strlen(prefix_http);
 		default_port = "80";
 
 		if (data->use_ssl) {
 			giterr_set(GITERR_NET, "Redirect from HTTPS to HTTP is not allowed");
 			goto cleanup;
 		}
-	}
-
-	if (!git__prefixcmp(url, prefix_https)) {
-		url += strlen(prefix_https);
+	} else if (!git__prefixcmp(url, prefix_https)) {
+		path_search_start = url + strlen(prefix_https);
 		default_port = "443";
 		data->use_ssl = true;
-	}
-
-	if (url[0] == '/')
+	} else if (url[0] == '/')
 		default_port = data->use_ssl ? "443" : "80";
 
 	if (!default_port) {
@@ -618,18 +615,19 @@ int gitno_connection_data_from_url(
 	}
 
 	error = gitno_extract_url_parts(
-		&data->host, &data->port, &data->user, &data->pass,
+		&data->host, &data->port, &data->path, &data->user, &data->pass,
 		url, default_port);
 
 	if (url[0] == '/') {
 		/* Relative redirect; reuse original host name and port */
+		path_search_start = url;
 		git__free(data->host);
 		data->host = original_host;
 		original_host = NULL;
 	}
 
 	if (!error) {
-		const char *path = strchr(url, '/');
+		const char *path = strchr(path_search_start, '/');
 		size_t pathlen = strlen(path);
 		size_t suffixlen = service_suffix ? strlen(service_suffix) : 0;
 
@@ -660,55 +658,74 @@ void gitno_connection_data_free_ptrs(gitno_connection_data *d)
 	git__free(d->pass); d->pass = NULL;
 }
 
+#define hex2c(c) ((c | 32) % 39 - 9)
+static char* unescape(char *str)
+{
+	int x, y;
+	int len = strlen(str);
+
+	for (x=y=0; str[y]; ++x, ++y) {
+		if ((str[x] = str[y]) == '%') {
+			if (y < len-2 && isxdigit(str[y+1]) && isxdigit(str[y+2])) {
+				str[x] = (hex2c(str[y+1]) << 4) + hex2c(str[y+2]);
+				y += 2;
+			}
+		}
+	}
+	str[x] = '\0';
+	return str;
+}
+
 int gitno_extract_url_parts(
 		char **host,
 		char **port,
+		char **path,
 		char **username,
 		char **password,
 		const char *url,
 		const char *default_port)
 {
-	char *colon, *slash, *at, *end;
-	const char *start;
-
-	/*
-	 * ==> [user[:pass]@]hostname.tld[:port]/resource
-	 */
-
-	colon = strchr(url, ':');
-	slash = strchr(url, '/');
-	at = strchr(url, '@');
+	struct http_parser_url u = {0};
+	const char *_host, *_port, *_path, *_userinfo;
 
-	if (!slash ||
-	    (colon && (slash < colon))) {
-		giterr_set(GITERR_NET, "Malformed URL");
+	if (http_parser_parse_url(url, strlen(url), false, &u)) {
+		giterr_set(GITERR_NET, "Malformed URL '%s'", url);
 		return GIT_EINVALIDSPEC;
 	}
 
-	start = url;
-	if (at && at < slash) {
-		start = at+1;
-		*username = git__substrdup(url, at - url);
-	}
+	_host = url+u.field_data[UF_HOST].off;
+	_port = url+u.field_data[UF_PORT].off;
+	_path = url+u.field_data[UF_PATH].off;
+	_userinfo = url+u.field_data[UF_USERINFO].off;
 
-	if (colon && colon < at) {
-		git__free(*username);
-		*username = git__substrdup(url, colon-url);
-		*password = git__substrdup(colon+1, at-colon-1);
-		colon = strchr(at, ':');
+	if (u.field_set & (1 << UF_HOST)) {
+		*host = git__substrdup(_host, u.field_data[UF_HOST].len);
+		GITERR_CHECK_ALLOC(*host);
 	}
 
-	if (colon == NULL) {
+	if (u.field_set & (1 << UF_PORT))
+		*port = git__substrdup(_port, u.field_data[UF_PORT].len);
+	else
 		*port = git__strdup(default_port);
-	} else {
-		*port = git__substrdup(colon + 1, slash - colon - 1);
-	}
 	GITERR_CHECK_ALLOC(*port);
 
-	end = colon == NULL ? slash : colon;
+	if (u.field_set & (1 << UF_PATH)) {
+		*path = git__substrdup(_path, u.field_data[UF_PATH].len);
+		GITERR_CHECK_ALLOC(*path);
+	}
 
-	*host = git__substrdup(start, end - start);
-	GITERR_CHECK_ALLOC(*host);
+	if (u.field_set & (1 << UF_USERINFO)) {
+		const char *colon = memchr(_userinfo, ':', u.field_data[UF_USERINFO].len);
+		if (colon) {
+			*username = unescape(git__substrdup(_userinfo, colon - _userinfo));
+			*password = unescape(git__substrdup(colon+1, u.field_data[UF_USERINFO].len - (colon+1-_userinfo)));
+			GITERR_CHECK_ALLOC(*password);
+		} else {
+			*username = git__substrdup(_userinfo, u.field_data[UF_USERINFO].len);
+		}
+		GITERR_CHECK_ALLOC(*username);
+
+	}
 
 	return 0;
 }
diff --git a/src/netops.h b/src/netops.h
index 5c105d6e3..666d66b12 100644
--- a/src/netops.h
+++ b/src/netops.h
@@ -92,6 +92,7 @@ void gitno_connection_data_free_ptrs(gitno_connection_data *data);
 int gitno_extract_url_parts(
 		char **host,
 		char **port,
+		char **path,
 		char **username,
 		char **password,
 		const char *url,
diff --git a/src/transports/git.c b/src/transports/git.c
index 79a9e7dd4..5dcd4eff7 100644
--- a/src/transports/git.c
+++ b/src/transports/git.c
@@ -179,39 +179,33 @@ static int _git_uploadpack_ls(
 	const char *url,
 	git_smart_subtransport_stream **stream)
 {
-	char *host=NULL, *port=NULL, *user=NULL, *pass=NULL;
+	char *host=NULL, *port=NULL, *path=NULL, *user=NULL, *pass=NULL;
+	const char *stream_url = url;
 	git_stream *s;
+	int error = -1;
 
 	*stream = NULL;
-
 	if (!git__prefixcmp(url, prefix_git))
-		url += strlen(prefix_git);
+		stream_url += strlen(prefix_git);
 
-	if (git_stream_alloc(t, url, cmd_uploadpack, stream) < 0)
+	if (git_stream_alloc(t, stream_url, cmd_uploadpack, stream) < 0)
 		return -1;
 
 	s = (git_stream *)*stream;
 
-	if (gitno_extract_url_parts(&host, &port, &user, &pass, url, GIT_DEFAULT_PORT) < 0)
-		goto on_error;
-
-	if (gitno_connect(&s->socket, host, port, 0) < 0)
-		goto on_error;
-
-	t->current_stream = s;
-	git__free(host);
-	git__free(port);
-	git__free(user);
-	git__free(pass);
-	return 0;
+	if (!(error = gitno_extract_url_parts(&host, &port, &path, &user, &pass, url, GIT_DEFAULT_PORT))) {
+		if (!(error = gitno_connect(&s->socket, host, port, 0)))
+			t->current_stream = s;
 
-on_error:
-	if (*stream)
+		git__free(host);
+		git__free(port);
+		git__free(path);
+		git__free(user);
+		git__free(pass);
+	} else if (*stream)
 		git_stream_free(*stream);
 
-	git__free(host);
-	git__free(port);
-	return -1;
+	return error;
 }
 
 static int _git_uploadpack(
@@ -235,39 +229,33 @@ static int _git_receivepack_ls(
 	const char *url,
 	git_smart_subtransport_stream **stream)
 {
-	char *host=NULL, *port=NULL, *user=NULL, *pass=NULL;
+	char *host=NULL, *port=NULL, *path=NULL, *user=NULL, *pass=NULL;
+	const char *stream_url = url;
 	git_stream *s;
+	int error;
 
 	*stream = NULL;
-
 	if (!git__prefixcmp(url, prefix_git))
-		url += strlen(prefix_git);
+		stream_url += strlen(prefix_git);
 
-	if (git_stream_alloc(t, url, cmd_receivepack, stream) < 0)
+	if (git_stream_alloc(t, stream_url, cmd_receivepack, stream) < 0)
 		return -1;
 
 	s = (git_stream *)*stream;
 
-	if (gitno_extract_url_parts(&host, &port, &user, &pass, url, GIT_DEFAULT_PORT) < 0)
-		goto on_error;
-
-	if (gitno_connect(&s->socket, host, port, 0) < 0)
-		goto on_error;
-
-	t->current_stream = s;
-	git__free(host);
-	git__free(port);
-	git__free(user);
-	git__free(pass);
-	return 0;
+	if (!(error = gitno_extract_url_parts(&host, &port, &path, &user, &pass, url, GIT_DEFAULT_PORT))) {
+		if (!(error = gitno_connect(&s->socket, host, port, 0)))
+			t->current_stream = s;
 
-on_error:
-	if (*stream)
+		git__free(host);
+		git__free(port);
+		git__free(path);
+		git__free(user);
+		git__free(pass);
+	} else if (*stream)
 		git_stream_free(*stream);
 
-	git__free(host);
-	git__free(port);
-	return -1;
+	return error;
 }
 
 static int _git_receivepack(
diff --git a/src/transports/ssh.c b/src/transports/ssh.c
index 4e2834b49..4a905e3c9 100644
--- a/src/transports/ssh.c
+++ b/src/transports/ssh.c
@@ -317,7 +317,7 @@ static int _git_ssh_setup_conn(
 	const char *cmd,
 	git_smart_subtransport_stream **stream)
 {
-	char *host=NULL, *port=NULL, *user=NULL, *pass=NULL;
+	char *host=NULL, *port=NULL, *path=NULL, *user=NULL, *pass=NULL;
 	const char *default_port="22";
 	ssh_stream *s;
 	LIBSSH2_SESSION* session=NULL;
@@ -330,8 +330,7 @@ static int _git_ssh_setup_conn(
 	s = (ssh_stream *)*stream;
 
 	if (!git__prefixcmp(url, prefix_ssh)) {
-		url = url + strlen(prefix_ssh);
-		if (gitno_extract_url_parts(&host, &port, &user, &pass, url, default_port) < 0)
+		if (gitno_extract_url_parts(&host, &port, &path, &user, &pass, url, default_port) < 0)
 			goto on_error;
 	} else {
 		if (git_ssh_extract_url_parts(&host, &user, url) < 0)
@@ -390,6 +389,7 @@ static int _git_ssh_setup_conn(
 	t->current_stream = s;
 	git__free(host);
 	git__free(port);
+	git__free(path);
 	git__free(user);
 	git__free(pass);
 
diff --git a/tests-clar/network/urlparse.c b/tests-clar/network/urlparse.c
index 15e841b35..31616275a 100644
--- a/tests-clar/network/urlparse.c
+++ b/tests-clar/network/urlparse.c
@@ -1,12 +1,12 @@
 #include "clar_libgit2.h"
 #include "netops.h"
 
-char *host, *port, *user, *pass;
+char *host, *port, *path, *user, *pass;
 gitno_connection_data conndata;
 
 void test_network_urlparse__initialize(void)
 {
-	host = port = user = pass = NULL;
+	host = port = path = user = pass = NULL;
 	memset(&conndata, 0, sizeof(conndata));
 }
 
@@ -15,6 +15,7 @@ void test_network_urlparse__cleanup(void)
 #define FREE_AND_NULL(x) if (x) { git__free(x); x = NULL; }
 	FREE_AND_NULL(host);
 	FREE_AND_NULL(port);
+	FREE_AND_NULL(path);
 	FREE_AND_NULL(user);
 	FREE_AND_NULL(pass);
 
@@ -23,27 +24,33 @@ void test_network_urlparse__cleanup(void)
 
 void test_network_urlparse__trivial(void)
 {
-	cl_git_pass(gitno_extract_url_parts(&host, &port, &user, &pass,
-				"example.com/resource", "8080"));
+	cl_git_pass(gitno_extract_url_parts(&host, &port, &path, &user, &pass,
+				"http://example.com/resource", "8080"));
 	cl_assert_equal_s(host, "example.com");
 	cl_assert_equal_s(port, "8080");
+	cl_assert_equal_s(path, "/resource");
 	cl_assert_equal_p(user, NULL);
 	cl_assert_equal_p(pass, NULL);
 }
 
-void test_network_urlparse__bad_url(void)
+void test_network_urlparse__encoded_password(void)
 {
-	cl_git_fail_with(gitno_extract_url_parts(&host, &port, &user, &pass,
-				"github.com/git://github.com/foo/bar.git.git", "443"),
-			GIT_EINVALIDSPEC);
+	cl_git_pass(gitno_extract_url_parts(&host, &port, &path, &user, &pass,
+				"https://user:pass%2fis%40bad@hostname.com:1234/", "1"));
+	cl_assert_equal_s(host, "hostname.com");
+	cl_assert_equal_s(port, "1234");
+	cl_assert_equal_s(path, "/");
+	cl_assert_equal_s(user, "user");
+	cl_assert_equal_s(pass, "pass/is@bad");
 }
 
 void test_network_urlparse__user(void)
 {
-	cl_git_pass(gitno_extract_url_parts(&host, &port, &user, &pass,
-				"user@example.com/resource", "8080"));
+	cl_git_pass(gitno_extract_url_parts(&host, &port, &path, &user, &pass,
+				"https://user@example.com/resource", "8080"));
 	cl_assert_equal_s(host, "example.com");
 	cl_assert_equal_s(port, "8080");
+	cl_assert_equal_s(path, "/resource");
 	cl_assert_equal_s(user, "user");
 	cl_assert_equal_p(pass, NULL);
 }
@@ -51,10 +58,11 @@ void test_network_urlparse__user(void)
 void test_network_urlparse__user_pass(void)
 {
 	/* user:pass@hostname.tld/resource */
-	cl_git_pass(gitno_extract_url_parts(&host, &port, &user, &pass,
-				"user:pass@example.com/resource", "8080"));
+	cl_git_pass(gitno_extract_url_parts(&host, &port, &path, &user, &pass,
+				"https://user:pass@example.com/resource", "8080"));
 	cl_assert_equal_s(host, "example.com");
 	cl_assert_equal_s(port, "8080");
+	cl_assert_equal_s(path, "/resource");
 	cl_assert_equal_s(user, "user");
 	cl_assert_equal_s(pass, "pass");
 }
@@ -62,10 +70,11 @@ void test_network_urlparse__user_pass(void)
 void test_network_urlparse__port(void)
 {
 	/* hostname.tld:port/resource */
-	cl_git_pass(gitno_extract_url_parts(&host, &port, &user, &pass,
-				"example.com:9191/resource", "8080"));
+	cl_git_pass(gitno_extract_url_parts(&host, &port, &path, &user, &pass,
+				"https://example.com:9191/resource", "8080"));
 	cl_assert_equal_s(host, "example.com");
 	cl_assert_equal_s(port, "9191");
+	cl_assert_equal_s(path, "/resource");
 	cl_assert_equal_p(user, NULL);
 	cl_assert_equal_p(pass, NULL);
 }
@@ -73,10 +82,11 @@ void test_network_urlparse__port(void)
 void test_network_urlparse__user_port(void)
 {
 	/* user@hostname.tld:port/resource */
-	cl_git_pass(gitno_extract_url_parts(&host, &port, &user, &pass,
-				"user@example.com:9191/resource", "8080"));
+	cl_git_pass(gitno_extract_url_parts(&host, &port, &path, &user, &pass,
+				"https://user@example.com:9191/resource", "8080"));
 	cl_assert_equal_s(host, "example.com");
 	cl_assert_equal_s(port, "9191");
+	cl_assert_equal_s(path, "/resource");
 	cl_assert_equal_s(user, "user");
 	cl_assert_equal_p(pass, NULL);
 }
@@ -84,10 +94,11 @@ void test_network_urlparse__user_port(void)
 void test_network_urlparse__user_pass_port(void)
 {
 	/* user:pass@hostname.tld:port/resource */
-	cl_git_pass(gitno_extract_url_parts(&host, &port, &user, &pass,
-				"user:pass@example.com:9191/resource", "8080"));
+	cl_git_pass(gitno_extract_url_parts(&host, &port, &path, &user, &pass,
+				"https://user:pass@example.com:9191/resource", "8080"));
 	cl_assert_equal_s(host, "example.com");
 	cl_assert_equal_s(port, "9191");
+	cl_assert_equal_s(path, "/resource");
 	cl_assert_equal_s(user, "user");
 	cl_assert_equal_s(pass, "pass");
 }
@@ -116,6 +127,18 @@ void test_network_urlparse__connection_data_ssl(void)
 	cl_assert_equal_i(conndata.use_ssl, true);
 }
 
+void test_network_urlparse__encoded_username_password(void)
+{
+	cl_git_pass(gitno_connection_data_from_url(&conndata,
+				"https://user%2fname:pass%40word%zyx%v@example.com/foo/bar/baz", "bar/baz"));
+	cl_assert_equal_s(conndata.host, "example.com");
+	cl_assert_equal_s(conndata.port, "443");
+	cl_assert_equal_s(conndata.path, "/foo/");
+	cl_assert_equal_s(conndata.user, "user/name");
+	cl_assert_equal_s(conndata.pass, "pass@word%zyx%v");
+	cl_assert_equal_i(conndata.use_ssl, true);
+}
+
 void test_network_urlparse__connection_data_cross_host_redirect(void)
 {
 	conndata.host = git__strdup("bar.com");