Added URL parser.

author: Remko Tronçon <git@el-tramo.be> 2012-09-15 11:10:20 (GMT)
committer: Remko Tronçon <git@el-tramo.be> 2012-09-15 12:05:19 (GMT)
commit: ab402e9e044e6f141a318c9b082671c828b915f3 (patch)
tree: 06194776a0102007b9a5fa5a6650f60b648d813b /Swiften/Base
parent: e94541a7156f4ccceaf1a3f7135b9c89c067883b (diff)
download: swift-contrib-ab402e9e044e6f141a318c9b082671c828b915f3.zip
swift-contrib-ab402e9e044e6f141a318c9b082671c828b915f3.tar.bz2
4 files changed, 266 insertions, 16 deletions
diff --git a/Swiften/Base/SConscript b/Swiften/Base/SConscript
index a5f3592..754164b 100644
--- a/Swiften/Base/SConscript
+++ b/Swiften/Base/SConscript
@@ -13,5 +13,6 @@ objects = swiften_env.SwiftenObject([
 			"RandomGenerator.cpp",
 			"BoostRandomGenerator.cpp",
 			"sleep.cpp",
+			"URL.cpp",
 		])
 swiften_env.Append(SWIFTEN_OBJECTS = [objects])
diff --git a/Swiften/Base/URL.cpp b/Swiften/Base/URL.cpp
new file mode 100644
index 0000000..c36863f
--- /dev/null
+++ b/Swiften/Base/URL.cpp
@@ -0,0 +1,165 @@
+/*
+ * Copyright (c) 2010 Remko Tronçon
+ * Licensed under the GNU General Public License v3.
+ * See Documentation/Licenses/GPLv3.txt for more information.
+ */
+
+#include <Swiften/Base/URL.h>
+
+namespace Swift {
+
+int URL::getPortOrDefaultPort(const URL& url) {
+	if (url.getPort()) {
+		return *url.getPort();
+	}
+	else if (url.getScheme() == "http") {
+		return 80;
+	}
+	else if (url.getScheme() == "https") {
+		return 443;
+	}
+	else {
+		std::cerr << "Unknown scheme: " + url.getScheme() << std::endl;
+		return 80;
+	}
+}
+
+URL URL::fromString(const std::string& urlString) {
+	size_t colonIndex = urlString.find(':');
+	if (colonIndex == std::string::npos) {
+		return URL();
+	}
+	std::string scheme = urlString.substr(0, colonIndex);
+
+	// Authority
+	if (urlString.size() > colonIndex + 2 && urlString[colonIndex+1] == '/' && urlString[colonIndex+2] == '/') {
+		size_t authorityIndex = colonIndex + 3;
+		size_t slashIndex = urlString.find('/', authorityIndex);
+		std::string authority;
+		std::string path;
+		if (slashIndex == std::string::npos) {
+			authority = urlString.substr(authorityIndex);
+			path = "";
+		}
+		else {
+			authority = urlString.substr(authorityIndex, slashIndex - authorityIndex);
+			path = unescape(urlString.substr(slashIndex));
+		}
+
+		size_t atIndex = authority.find('@');
+		std::string userInfo;
+		std::string hostAndPort;
+		if (atIndex != std::string::npos) {
+			userInfo = authority.substr(0, atIndex);
+			hostAndPort = authority.substr(atIndex + 1);
+		}
+		else {
+			userInfo = "";
+			hostAndPort = authority;
+		}
+
+		std::string host;
+		boost::optional<int> port;
+		colonIndex = hostAndPort.find(':');
+		if (colonIndex != std::string::npos) {
+			host = unescape(hostAndPort.substr(0, colonIndex));
+			try {
+				port = boost::lexical_cast<int>(hostAndPort.substr(colonIndex + 1));
+			}
+			catch (const boost::bad_lexical_cast&) {
+				return URL();
+			}
+		}
+		else {
+			host = unescape(hostAndPort);
+		}
+
+		if (port) {
+			return URL(scheme, host, *port, path);
+		}
+		else {
+			return URL(scheme, host, path);
+		}
+	}
+	else {
+		// We don't support URLs without authorities yet
+		return URL();
+	}
+}
+
+// Disabling this code for now, since GCC4.5+boost1.42 (on ubuntu) seems to
+// result in a bug. Replacing it with naive code.
+#if 0
+// Should be in anonymous namespace, but older GCCs complain if we do that
+struct PercentEncodedCharacterFinder {
+template<typename Iterator>
+boost::iterator_range<Iterator> operator()(Iterator begin, Iterator end) {
+	boost::iterator_range<Iterator> r = boost::first_finder("%")(begin, end);
+	if (r.end() == end) {
+		return r;
+	}
+	else {
+		if (r.end() + 1 == end || r.end() + 2 == end) {
+			throw std::runtime_error("Incomplete escape character");
+		}
+		else {
+			r.advance_end(2);
+			return r;
+		}
+	}
+}
+};
+
+struct PercentUnencodeFormatter {
+template<typename FindResult>
+std::string operator()(const FindResult& match) const {
+	std::stringstream s;
+	s << std::hex << std::string(match.begin() + 1, match.end());
+	unsigned int value;
+	s >> value;
+	if (s.fail() || s.bad()) {
+		throw std::runtime_error("Invalid escape character");
+	}
+	unsigned char charValue = static_cast<unsigned char>(value);
+	return std::string(reinterpret_cast<const char*>(&charValue), 1);
+}
+};
+
+std::string unescape(const std::string& s) {
+	try {
+		return boost::find_format_all_copy(s, PercentEncodedCharacterFinder(), PercentUnencodeFormatter());
+	}
+	catch (const std::exception&) {
+		return "";
+	}
+}
+#endif
+
+std::string URL::unescape(const std::string& str) {
+	std::string result;
+	for (size_t i = 0; i < str.size(); ++i) {
+		if (str[i] == '%') {
+			if (i + 3 < str.size()) {
+				std::stringstream s;
+				s << std::hex << str.substr(i+1, 2);
+				unsigned int value;
+				s >> value;
+				if (s.fail() || s.bad()) {
+					return "";
+				}
+				unsigned char charValue = static_cast<unsigned char>(value);
+				result += std::string(reinterpret_cast<const char*>(&charValue), 1);
+				i += 2;
+			}
+			else {
+				return "";
+			}
+		}
+		else {
+			result += str[i];
+		}
+	}
+	return result;
+}
+
+}
diff --git a/Swiften/Base/URL.h b/Swiften/Base/URL.h
index 94dc4cb..9821ed5 100644
--- a/Swiften/Base/URL.h
+++ b/Swiften/Base/URL.h
@@ -8,32 +8,27 @@
 
 #include <string>
 #include <boost/lexical_cast.hpp>
+#include <boost/optional.hpp>
 
 namespace Swift {
 
 class URL {
 	public:
 
-		URL() : scheme(""), user(""), password(""), host(""), port(-1), path(""), isEmpty(true) {
+		URL() : scheme(""), user(""), password(""), host(""), path(""), empty(true) {
 		}
 
-		URL(const std::string& urlString) {
-			host = urlString;
-			port = 80;
-			scheme = "http";
-			isEmpty = false;
-			//FIXME
-		}
-
-		URL(const std::string& scheme, const std::string& host, int port, const std::string& path) : scheme(scheme), user(), password(), host(host), port(port), path(path), isEmpty(false) {
+		URL(const std::string& scheme, const std::string& host, int port, const std::string& path) : scheme(scheme), user(), password(), host(host), port(port), path(path), empty(false) {
+		}	
 
+		URL(const std::string& scheme, const std::string& host, const std::string& path) : scheme(scheme), user(), password(), host(host), path(path), empty(false) {
 		}	
 
 		/**
 		 * Whether the URL is empty.
 		 */
-		bool empty() const {
-			return isEmpty;
+		bool isEmpty() const {
+			return empty;
 		}
 
 		/**
@@ -53,7 +48,7 @@ class URL {
 		/**
 		 * Port number
 		 */
-		int getPort() const {
+		boost::optional<int> getPort() const {
 			return port;
 		}
 
@@ -65,7 +60,7 @@ class URL {
 		}
 
 		const std::string toString() const {
-			if (isEmpty) {
+			if (empty) {
 				return "";
 			}
 			std::string result = scheme + "://";
@@ -86,13 +81,18 @@ class URL {
 			return result;
 		}
 
+		static int getPortOrDefaultPort(const URL& url);
+		static URL fromString(const std::string&);
+		static std::string unescape(const std::string&);
+
+
 	private:
 		std::string scheme;
 		std::string user;
 		std::string password;
 		std::string host;
-		int port;
+		boost::optional<int> port;
 		std::string path;
-		bool isEmpty;
+		bool empty;
 	};
 }
diff --git a/Swiften/Base/UnitTest/URLTest.cpp b/Swiften/Base/UnitTest/URLTest.cpp
new file mode 100644
index 0000000..4de1d33
--- /dev/null
+++ b/Swiften/Base/UnitTest/URLTest.cpp
@@ -0,0 +1,84 @@
+/*
+ * Copyright (c) 2012 Remko Tronçon
+ * Licensed under the GNU General Public License v3.
+ * See Documentation/Licenses/GPLv3.txt for more information.
+ */
+
+#include <cppunit/extensions/HelperMacros.h>
+#include <cppunit/extensions/TestFactoryRegistry.h>
+
+#include <Swiften/Base/URL.h>
+#include <boost/lexical_cast.hpp>
+
+using namespace Swift;
+
+class URLTest : public CppUnit::TestFixture {
+		CPPUNIT_TEST_SUITE(URLTest);
+		CPPUNIT_TEST(testFromString);
+		CPPUNIT_TEST(testFromString_WithoutPath);
+		CPPUNIT_TEST(testFromString_WithPort);
+		CPPUNIT_TEST(testFromString_WithPortWithoutPath);
+		CPPUNIT_TEST(testFromString_WithUserInfo);
+		CPPUNIT_TEST(testFromString_NonASCIIHost);
+		CPPUNIT_TEST(testFromString_NonASCIIPath);
+		CPPUNIT_TEST_SUITE_END();
+
+	public:
+		void testFromString() {
+			URL url = URL::fromString("http://foo.bar/baz/bam");
+			
+			CPPUNIT_ASSERT_EQUAL(std::string("http"), url.getScheme());
+			CPPUNIT_ASSERT_EQUAL(std::string("foo.bar"), url.getHost());
+			CPPUNIT_ASSERT(!url.getPort());
+			CPPUNIT_ASSERT_EQUAL(std::string("/baz/bam"), url.getPath());
+		}
+
+		void testFromString_WithoutPath() {
+			URL url = URL::fromString("http://foo.bar");
+			
+			CPPUNIT_ASSERT_EQUAL(std::string("http"), url.getScheme());
+			CPPUNIT_ASSERT_EQUAL(std::string("foo.bar"), url.getHost());
+			CPPUNIT_ASSERT(!url.getPort());
+			CPPUNIT_ASSERT_EQUAL(std::string(""), url.getPath());
+		}
+
+		void testFromString_WithPort() {
+			URL url = URL::fromString("http://foo.bar:1234/baz/bam");
+			
+			CPPUNIT_ASSERT_EQUAL(std::string("http"), url.getScheme());
+			CPPUNIT_ASSERT_EQUAL(std::string("foo.bar"), url.getHost());
+			CPPUNIT_ASSERT_EQUAL(1234, *url.getPort());
+			CPPUNIT_ASSERT_EQUAL(std::string("/baz/bam"), url.getPath());
+		}
+
+		void testFromString_WithPortWithoutPath() {
+			URL url = URL::fromString("http://foo.bar:1234");
+			
+			CPPUNIT_ASSERT_EQUAL(std::string("http"), url.getScheme());
+			CPPUNIT_ASSERT_EQUAL(std::string("foo.bar"), url.getHost());
+			CPPUNIT_ASSERT_EQUAL(1234, *url.getPort());
+			CPPUNIT_ASSERT_EQUAL(std::string(""), url.getPath());
+		}
+
+		void testFromString_WithUserInfo() {
+			URL url = URL::fromString("http://user:pass@foo.bar/baz/bam");
+			
+			CPPUNIT_ASSERT_EQUAL(std::string("http"), url.getScheme());
+			CPPUNIT_ASSERT_EQUAL(std::string("foo.bar"), url.getHost());
+			CPPUNIT_ASSERT_EQUAL(std::string("/baz/bam"), url.getPath());
+		}
+
+		void testFromString_NonASCIIHost() {
+			URL url = URL::fromString("http://www.tron%C3%A7on.be/baz/bam");
+
+			CPPUNIT_ASSERT_EQUAL(std::string("www.tron\xc3\xa7on.be"), url.getHost());
+		}
+
+		void testFromString_NonASCIIPath() {
+			URL url = URL::fromString("http://foo.bar/baz/tron%C3%A7on/bam");
+			
+			CPPUNIT_ASSERT_EQUAL(std::string("/baz/tron\xc3\xa7on/bam"), url.getPath());
+		}
+};
+
+CPPUNIT_TEST_SUITE_REGISTRATION(URLTest);
author	Remko Tronçon <git@el-tramo.be>	2012-09-15 11:10:20 (GMT)
committer	Remko Tronçon <git@el-tramo.be>	2012-09-15 12:05:19 (GMT)
commit	ab402e9e044e6f141a318c9b082671c828b915f3 (patch)
tree	06194776a0102007b9a5fa5a6650f60b648d813b /Swiften/Base
parent	e94541a7156f4ccceaf1a3f7135b9c89c067883b (diff)
download	swift-contrib-ab402e9e044e6f141a318c9b082671c828b915f3.zip swift-contrib-ab402e9e044e6f141a318c9b082671c828b915f3.tar.bz2