Added URL parsing class

Allows the parsing of full URLs into easy chunks. Added as a helper class for users of the library dealing with URLs, and to potentially replace some of the HTTP parsing code in the future.

CMakeLists.txt

@@ -17,9 +17,10 @@ endif()
 set( INCLUDE_PATH "${PROJECT_SOURCE_DIR}/include" )
 set( SOURCE_PATH "${PROJECT_SOURCE_DIR}/src" )
 
-set(SOURCE_FILES main.cpp src/TcpSocket.cpp include/frnetlib/TcpSocket.h src/TcpListener.cpp include/frnetlib/TcpListener.h src/Socket.cpp include/frnetlib/Socket.h src/Packet.cpp include/frnetlib/Packet.h include/frnetlib/NetworkEncoding.h src/SocketSelector.cpp include/frnetlib/SocketSelector.h src/HttpSocket.cpp include/frnetlib/HttpSocket.h src/HttpRequest.cpp include/frnetlib/HttpRequest.h src/HttpResponse.cpp include/frnetlib/HttpResponse.h src/Http.cpp include/frnetlib/Http.h src/SSLSocket.cpp include/frnetlib/SSLSocket.h src/SSLListener.cpp include/frnetlib/SSLListener.h include/frnetlib/SSLContext.h src/SocketReactor.cpp include/frnetlib/SocketReactor.h include/frnetlib/Packetable.h include/frnetlib/Listener.h)
+set(SOURCE_FILES main.cpp src/TcpSocket.cpp include/frnetlib/TcpSocket.h src/TcpListener.cpp include/frnetlib/TcpListener.h src/Socket.cpp include/frnetlib/Socket.h src/Packet.cpp include/frnetlib/Packet.h include/frnetlib/NetworkEncoding.h src/SocketSelector.cpp include/frnetlib/SocketSelector.h src/HttpSocket.cpp include/frnetlib/HttpSocket.h src/HttpRequest.cpp include/frnetlib/HttpRequest.h src/HttpResponse.cpp include/frnetlib/HttpResponse.h src/Http.cpp include/frnetlib/Http.h src/SSLSocket.cpp include/frnetlib/SSLSocket.h src/SSLListener.cpp include/frnetlib/SSLListener.h include/frnetlib/SSLContext.h src/SocketReactor.cpp include/frnetlib/SocketReactor.h include/frnetlib/Packetable.h include/frnetlib/Listener.h src/URL.cpp include/frnetlib/URL.h)
 
 include_directories(include)
+set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -std=c++11 -m64 -fPIC -std=c++14 -pthread -lmbedtls -lmbedx509 -lmbedcrypto")
 
 # Set the library output directory
 set( LIBRARY_OUTPUT_PATH "${PROJECT_BINARY_DIR}/lib" )

include/frnetlib/URL.h

@@ -0,0 +1,123 @@
+//
+// Created by fred.nicolson on 23/05/17.
+//
+
+#ifndef FRNETLIB_URLPARSER_H
+#define FRNETLIB_URLPARSER_H
+#include <string>
+#include <unordered_map>
+
+//Note, a URL looks like this:  scheme:[//host[:port]][/path][?query][#fragment]
+
+class URL
+{
+public:
+    enum Scheme
+    {
+        HTTP = 0,
+        HTTPS = 1,
+        FTP = 2,
+        MAILTO = 3,
+        IRC = 4,
+        SFTP = 5,
+        Unknown = 6,
+    };
+
+    /*!
+     * Constructors
+     */
+    URL() = default;
+    URL(const std::string &url);
+
+    /*!
+     * Parses a given URL, extracting its various components
+     *
+     * @param url The URL to parse
+     */
+    void parse(std::string url);
+
+    /*!
+     * Get the URL scheme
+     */
+    inline Scheme get_scheme()
+    {
+        return scheme;
+    }
+
+    /*
+     * Get the URL host
+     */
+    inline const std::string &get_host()
+    {
+        return host;
+    }
+
+    /*
+     * Get the URL port
+     */
+    inline const std::string &get_port()
+    {
+        return port;
+    }
+
+    /*
+     * Get the URL path
+     */
+    inline const std::string &get_path()
+    {
+        return path;
+    }
+
+    /*
+     * Get the URL query
+     */
+    inline const std::string &get_query()
+    {
+        return query;
+    }
+
+    /*
+     * Get the URL fragment
+     */
+    inline const std::string &get_fragment()
+    {
+        return fragment;
+    }
+
+    /*!
+     * Converts a string to a scheme enum.
+     *
+     * @param scheme The string scheme to convert
+     * @return The associated scheme enum value. Scheme::Unknown on failure.
+     */
+    static Scheme string_to_scheme(const std::string &scheme);
+
+    /*!
+     * Converts a scheme enum value to a string
+     *
+     * @param scheme The scheme value to convert
+     * @return The string version
+     */
+    static const std::string &scheme_to_string(Scheme scheme);
+
+private:
+    /*!
+     * Converts a string to lower case
+     *
+     * @param str The string to convert
+     * @return The converted string
+     */
+    static std::string to_lower(const std::string &str);
+
+    //State
+    Scheme scheme;
+    std::string host;
+    std::string port;
+    std::string path;
+    std::string query;
+    std::string fragment;
+    static std::unordered_map<std::string, URL::Scheme> scheme_string_map;
+};
+
+
+#endif //FRNETLIB_URLPARSER_H

src/URL.cpp

@@ -0,0 +1,127 @@
+//
+// Created by fred.nicolson on 23/05/17.
+//
+
+#include <stdexcept>
+#include <algorithm>
+#include <iostream>
+#include "frnetlib/URL.h"
+
+std::unordered_map<std::string, URL::Scheme> URL::scheme_string_map = {
+        {"http", URL::HTTP},
+        {"https", URL::HTTPS},
+        {"sftp", URL::FTP},
+        {"mailto", URL::MAILTO},
+        {"irc", URL::IRC},
+        {"sftp", URL::SFTP},
+        {"unknown", URL::Unknown}
+};
+
+URL::URL(const std::string &url)
+{
+    parse(url);
+}
+
+
+void URL::parse(std::string url)
+{
+    size_t parse_offset = 0;
+    size_t pos = 0;
+
+    //Check to see if a scheme exists
+    pos = url.find("://");
+    if(pos != std::string::npos)
+    {
+        auto scheme_pos = scheme_string_map.find(to_lower(url.substr(0, pos)));
+        scheme = (scheme_pos == scheme_string_map.end()) ? URL::Unknown : scheme_pos->second;
+        parse_offset = pos + 3;
+    }
+
+    //Check to see if there's a port
+    pos = url.find(":", parse_offset);
+    if(pos != std::string::npos)
+    {
+        //Store host
+        host = url.substr(parse_offset, pos - parse_offset);
+        parse_offset += host.size();
+
+        //Find end of port
+        size_t port_end = url.find("/", parse_offset);
+        port_end = (port_end == std::string::npos) ? url.size() : port_end;
+        port = url.substr(pos + 1, port_end - pos - 1);
+        parse_offset = port_end + 1;
+    }
+    else
+    {
+        //Store host
+        pos = url.find("/", parse_offset);
+        pos = (pos != std::string::npos) ? pos : url.find("?", parse_offset);
+        pos = (pos != std::string::npos) ? pos : url.find("#", parse_offset);
+        pos = (pos != std::string::npos) ? pos : url.size();
+        host = url.substr(parse_offset, pos - parse_offset);
+        parse_offset = pos + 1;
+    }
+
+    //Exit if done
+    if(parse_offset >= url.size())
+        return;
+
+    //Extract the path
+    pos = url.find("?", parse_offset);
+    if(pos != std::string::npos)
+    {
+        path = url.substr(parse_offset, pos - parse_offset);
+        parse_offset = pos + 1;
+    }
+    else
+    {
+        pos = url.find("#", parse_offset);
+        pos = (pos != std::string::npos) ? pos : url.find("?", parse_offset);
+        pos = (pos != std::string::npos) ? pos : url.size();
+        path = url.substr(parse_offset, pos - parse_offset);
+        parse_offset = pos + 1;
+    }
+
+    //Extract the query
+    pos = url.find("#", parse_offset - 1);
+    if(pos != std::string::npos)
+    {
+        if(pos + 1 != parse_offset)
+            query = url.substr(parse_offset, pos - parse_offset);
+        fragment = url.substr(pos + 1, url.size() - pos - 1);
+    }
+    else
+    {
+        if(parse_offset >= url.size())
+            return;
+        query = url.substr(parse_offset, url.size() - parse_offset);
+    }
+
+    return;
+}
+
+URL::Scheme URL::string_to_scheme(const std::string &scheme)
+{
+    auto iter = scheme_string_map.find(to_lower(scheme));
+    if(iter == scheme_string_map.end())
+        return URL::Unknown;
+    return iter->second;
+}
+
+std::string URL::to_lower(const std::string &str)
+{
+    std::string out = str;
+    std::transform(out.begin(), out.end(), out.begin(), ::tolower);
+    return out;
+}
+
+const std::string &URL::scheme_to_string(URL::Scheme scheme)
+{
+    auto iter = std::find_if(scheme_string_map.begin(), scheme_string_map.end(), [](const auto &i){
+        return i->second == scheme;
+    });
+
+    if(iter == scheme_string_map.end())
+        throw std::logic_error("Unknown URL::Scheme value " + std::to_string(scheme));
+    return iter->first;
+}