Feature/http (#16)

* add skeleton and broken http client code. GET returns "Resource temporarily unavailable" errors... * linux compile fix * can GET some pages * Update formatting in README.md * unittest for sending large messages * document bug * Feature/send large message (#14) * introduce send fragment * can pass a fin frame * can send messages which are a perfect multiple of the chunk size * set fin only for last fragment * cleanup * last fragment should be of type CONTINUATION * Add simple send and receive programs * speedups receiving + better way to wait for thing * receive speedup by using linked list of chunks instead of large array * document bug * use chunks to receive data * trailing spaces * Update README.md Add note about message fragmentation. * Feature/ws cli (#15) * New command line tool for transfering files / still very beta. * add readme * use cli11 for argument parsing * json -> msgpack * stop using base64 and use binary which can be stored in message pack * add target for building with homebrew * all CMakeLists are referenced by the top level one * add ws_chat and ws_connect sub commands to ws * cleanup * add echo and broadcast server as ws sub-commands * add gitignore * comments * ping pong added to ws * mv cobra_publisher under ws folder * Update README.md * linux build fix * linux build fix * move http_client to a ws sub-command * simple HTTP post support (urlencode parameters) * can specify extra headers * chunk encoding / simple redirect support / -I option * follow redirects is optional * make README vim markdown plugin friendly * cleanup argument parsing + add socket creation factory * add missing file * http gzip compression * cleanup * doc * Feature/send large message (#14) * introduce send fragment * can pass a fin frame * can send messages which are a perfect multiple of the chunk size * set fin only for last fragment * cleanup * last fragment should be of type CONTINUATION * Add simple send and receive programs * speedups receiving + better way to wait for thing * receive speedup by using linked list of chunks instead of large array * document bug * use chunks to receive data * trailing spaces
2019-02-28 21:54:03 -08:00
parent e881b82511
commit 0cff5065d9
40 changed files with 9871 additions and 206 deletions
--- a/ixwebsocket/IXHttpClient.cpp
+++ b/ixwebsocket/IXHttpClient.cpp
@ -0,0 +1,471 @@
+/*
+ *  IXHttpClient.cpp
+ *  Author: Benjamin Sergeant
+ *  Copyright (c) 2019 Machine Zone, Inc. All rights reserved.
+ */
+
+#include "IXHttpClient.h"
+#include "IXUrlParser.h"
+#include "IXWebSocketHttpHeaders.h"
+#include "IXSocketFactory.h"
+
+#include <sstream>
+#include <iomanip>
+#include <vector>
+#include <cstring>
+
+#include <zlib.h>
+
+namespace ix
+{
+    const std::string HttpClient::kPost = "POST";
+    const std::string HttpClient::kGet = "GET";
+    const std::string HttpClient::kHead = "HEAD";
+
+    HttpClient::HttpClient()
+    {
+
+    }
+
+    HttpClient::~HttpClient()
+    {
+
+    }
+
+    HttpResponse HttpClient::request(
+        const std::string& url,
+        const std::string& verb,
+        const std::string& body,
+        const HttpRequestArgs& args,
+        int redirects)
+    {
+        uint64_t uploadSize = 0;
+        uint64_t downloadSize = 0;
+        int code = 0;
+        WebSocketHttpHeaders headers;
+        std::string payload;
+
+        std::string protocol, host, path, query;
+        int port;
+        bool websocket = false;
+
+        if (!UrlParser::parse(url, protocol, host, path, query, port, websocket))
+        {
+            std::stringstream ss;
+            ss << "Cannot parse url: " << url;
+            return std::make_tuple(code, HttpErrorCode_UrlMalformed,
+                                   headers, payload, ss.str(),
+                                   uploadSize, downloadSize);
+        }
+
+        bool tls = protocol == "https";
+        std::string errorMsg;
+        _socket = createSocket(tls, errorMsg);
+
+        if (!_socket)
+        {
+            return std::make_tuple(code, HttpErrorCode_CannotCreateSocket,
+                                   headers, payload, errorMsg,
+                                   uploadSize, downloadSize);
+        }
+
+        // Build request string
+        std::stringstream ss;
+        ss << verb << " " << path << " HTTP/1.1\r\n";
+        ss << "Host: " << host << "\r\n";
+        ss << "User-Agent: ixwebsocket/1.0.0" << "\r\n";
+        ss << "Accept: */*" << "\r\n";
+
+        if (args.compress)
+        {
+            ss << "Accept-Encoding: gzip" << "\r\n";
+        }
+
+        // Append extra headers
+        for (auto&& it : args.extraHeaders)
+        {
+            ss << it.first << ": " << it.second << "\r\n";
+        }
+
+        if (verb == kPost)
+        {
+            ss << "Content-Length: " << body.size() << "\r\n";
+
+            // Set default Content-Type if unspecified
+            if (args.extraHeaders.find("Content-Type") == args.extraHeaders.end())
+            {
+                ss << "Content-Type: application/x-www-form-urlencoded" << "\r\n";
+            }
+            ss << "\r\n";
+            ss << body;
+        }
+        else
+        {
+            ss << "\r\n";
+        }
+
+        std::string req(ss.str());
+        std::string errMsg;
+        std::atomic<bool> requestInitCancellation(false);
+
+        // Make a cancellation object dealing with connection timeout
+        auto isCancellationRequested =
+            makeCancellationRequestWithTimeout(args.connectTimeout, requestInitCancellation);
+
+        bool success = _socket->connect(host, port, errMsg, isCancellationRequested);
+        if (!success)
+        {
+            std::stringstream ss;
+            ss << "Cannot connect to url: " << url;
+            return std::make_tuple(code, HttpErrorCode_CannotConnect,
+                                   headers, payload, ss.str(),
+                                   uploadSize, downloadSize);
+        }
+
+        // Make a new cancellation object dealing with transfer timeout
+        isCancellationRequested =
+            makeCancellationRequestWithTimeout(args.transferTimeout, requestInitCancellation);
+
+        if (args.verbose)
+        {
+            std::stringstream ss;
+            ss << "Sending " << verb << " request "
+               << "to " << host << ":" << port << std::endl
+               << "request size: " << req.size() << " bytes" << std::endl
+               << "=============" << std::endl
+               << req
+               << "=============" << std::endl
+               << std::endl;
+
+            log(ss.str(), args);
+        }
+
+        if (!_socket->writeBytes(req, isCancellationRequested))
+        {
+            std::string errorMsg("Cannot send request");
+            return std::make_tuple(code, HttpErrorCode_SendError,
+                                   headers, payload, errorMsg,
+                                   uploadSize, downloadSize);
+        }
+
+        uploadSize = req.size();
+
+        auto lineResult = _socket->readLine(isCancellationRequested);
+        auto lineValid = lineResult.first;
+        auto line = lineResult.second;
+
+        if (!lineValid)
+        {
+            std::string errorMsg("Cannot retrieve status line");
+            return std::make_tuple(code, HttpErrorCode_CannotReadStatusLine,
+                                   headers, payload, errorMsg,
+                                   uploadSize, downloadSize);
+        }
+
+        if (args.verbose)
+        {
+            std::stringstream ss;
+            ss << "Status line " << line;
+            log(ss.str(), args);
+        }
+
+        if (sscanf(line.c_str(), "HTTP/1.1 %d", &code) != 1)
+        {
+            std::string errorMsg("Cannot parse response code from status line");
+            return std::make_tuple(code, HttpErrorCode_MissingStatus,
+                                   headers, payload, errorMsg,
+                                   uploadSize, downloadSize);
+        }
+
+        auto result = parseHttpHeaders(_socket, isCancellationRequested);
+        auto headersValid = result.first;
+        headers = result.second;
+
+        if (!headersValid)
+        {
+            std::string errorMsg("Cannot parse http headers");
+            return std::make_tuple(code, HttpErrorCode_HeaderParsingError,
+                                   headers, payload, errorMsg,
+                                   uploadSize, downloadSize);
+        }
+
+        // Redirect ?
+        if ((code >= 301 && code <= 308) && args.followRedirects)
+        {
+            if (headers.find("Location") == headers.end())
+            {
+                std::string errorMsg("Missing location header for redirect");
+                return std::make_tuple(code, HttpErrorCode_MissingLocation,
+                                       headers, payload, errorMsg,
+                                       uploadSize, downloadSize);
+            }
+
+            if (redirects >= args.maxRedirects)
+            {
+                std::stringstream ss;
+                ss << "Too many redirects: " << redirects;
+                return std::make_tuple(code, HttpErrorCode_TooManyRedirects,
+                                       headers, payload, ss.str(),
+                                       uploadSize, downloadSize);
+            }
+
+            // Recurse
+            std::string location = headers["Location"];
+            return request(location, verb, body, args, redirects+1);
+        }
+
+        if (verb == "HEAD")
+        {
+            return std::make_tuple(code, HttpErrorCode_Ok,
+                                   headers, payload, std::string(),
+                                   uploadSize, downloadSize);
+        }
+
+        // Parse response:
+        if (headers.find("Content-Length") != headers.end())
+        {
+            ssize_t contentLength = -1;
+            ss.str("");
+            ss << headers["Content-Length"];
+            ss >> contentLength;
+
+            payload.reserve(contentLength);
+
+            // FIXME: very inefficient way to read bytes, but it works...
+            for (int i = 0; i < contentLength; ++i)
+            {
+                char c;
+                if (!_socket->readByte(&c, isCancellationRequested))
+                {
+                    return std::make_tuple(code, HttpErrorCode_ReadError,
+                                           headers, payload, "Cannot read byte",
+                                           uploadSize, downloadSize);
+                }
+
+                payload += c;
+            }
+        }
+        else if (headers.find("Transfer-Encoding") != headers.end() &&
+                 headers["Transfer-Encoding"] == "chunked")
+        {
+            std::stringstream ss;
+
+            while (true)
+            {
+                lineResult = _socket->readLine(isCancellationRequested);
+                line = lineResult.second;
+
+                if (!lineResult.first)
+                {
+                    return std::make_tuple(code, HttpErrorCode_ChunkReadError,
+                                           headers, payload, errorMsg,
+                                           uploadSize, downloadSize);
+                }
+
+                uint64_t chunkSize;
+                ss.str("");
+                ss << std::hex << line;
+                ss >> chunkSize;
+
+                if (args.verbose)
+                {
+                    std::stringstream oss;
+                    oss << "Reading " << chunkSize << " bytes"
+                        << std::endl;
+                    log(oss.str(), args);
+                }
+
+                payload.reserve(payload.size() + chunkSize);
+
+                // Read another line
+
+                for (uint64_t i = 0; i < chunkSize; ++i)
+                {
+                    char c;
+                    if (!_socket->readByte(&c, isCancellationRequested))
+                    {
+                        errorMsg = "Cannot read byte";
+                        return std::make_tuple(code, HttpErrorCode_ChunkReadError,
+                                               headers, payload, errorMsg,
+                                               uploadSize, downloadSize);
+                    }
+
+                    payload += c;
+                }
+
+                lineResult = _socket->readLine(isCancellationRequested);
+
+                if (!lineResult.first)
+                {
+                    return std::make_tuple(code, HttpErrorCode_ChunkReadError,
+                                           headers, payload, errorMsg,
+                                           uploadSize, downloadSize);
+                }
+
+                if (chunkSize == 0) break;
+            }
+        }
+        else if (code == 204)
+        {
+            ; // 204 is NoContent response code
+        }
+        else
+        {
+            std::string errorMsg("Cannot read http body");
+            return std::make_tuple(code, HttpErrorCode_CannotReadBody,
+                                   headers, payload, errorMsg,
+                                   uploadSize, downloadSize);
+        }
+
+        downloadSize = payload.size();
+
+        // If the content was compressed with gzip, decode it
+        if (headers["Content-Encoding"] == "gzip")
+        {
+            std::string decompressedPayload;
+            if (!gzipInflate(payload, decompressedPayload))
+            {
+                std::string errorMsg("Error decompressing payload");
+                return std::make_tuple(code, HttpErrorCode_Gzip,
+                                       headers, payload, errorMsg,
+                                       uploadSize, downloadSize);
+            }
+            payload = decompressedPayload;
+        }
+
+        return std::make_tuple(code, HttpErrorCode_Ok,
+                               headers, payload, std::string(),
+                               uploadSize, downloadSize);
+    }
+
+    HttpResponse HttpClient::get(const std::string& url,
+                                 const HttpRequestArgs& args)
+    {
+        return request(url, kGet, std::string(), args);
+    }
+
+    HttpResponse HttpClient::head(const std::string& url,
+                                  const HttpRequestArgs& args)
+    {
+        return request(url, kHead, std::string(), args);
+    }
+
+    HttpResponse HttpClient::post(const std::string& url,
+                                  const HttpParameters& httpParameters,
+                                  const HttpRequestArgs& args)
+    {
+        return request(url, kPost, serializeHttpParameters(httpParameters), args);
+    }
+
+    HttpResponse HttpClient::post(const std::string& url,
+                                  const std::string& body,
+                                  const HttpRequestArgs& args)
+    {
+        return request(url, kPost, body, args);
+    }
+
+    std::string HttpClient::urlEncode(const std::string& value)
+    {
+        std::ostringstream escaped;
+        escaped.fill('0');
+        escaped << std::hex;
+
+        for (std::string::const_iterator i = value.begin(), n = value.end();
+             i != n; ++i)
+        {
+            std::string::value_type c = (*i);
+
+            // Keep alphanumeric and other accepted characters intact
+            if (isalnum(c) || c == '-' || c == '_' || c == '.' || c == '~')
+            {
+                escaped << c;
+                continue;
+            }
+
+            // Any other characters are percent-encoded
+            escaped << std::uppercase;
+            escaped << '%' << std::setw(2) << int((unsigned char) c);
+            escaped << std::nouppercase;
+        }
+
+        return escaped.str();
+    }
+
+    std::string HttpClient::serializeHttpParameters(const HttpParameters& httpParameters)
+    {
+        std::stringstream ss;
+        size_t count = httpParameters.size();
+        size_t i = 0;
+
+        for (auto&& it : httpParameters)
+        {
+            ss << urlEncode(it.first)
+               << "="
+               << urlEncode(it.second);
+
+            if (i++ < (count-1))
+            {
+               ss << "&";
+            }
+        }
+        return ss.str();
+    }
+
+    bool HttpClient::gzipInflate(
+        const std::string& in,
+        std::string& out)
+    {
+        z_stream inflateState;
+        std::memset(&inflateState, 0, sizeof(inflateState));
+
+        inflateState.zalloc = Z_NULL;
+        inflateState.zfree = Z_NULL;
+        inflateState.opaque = Z_NULL;
+        inflateState.avail_in = 0;
+        inflateState.next_in = Z_NULL;
+
+        if (inflateInit2(&inflateState, 16+MAX_WBITS) != Z_OK)
+        {
+            return false;
+        }
+
+        inflateState.avail_in = (uInt) in.size();
+        inflateState.next_in = (unsigned char *)(const_cast<char *>(in.data()));
+
+        const int kBufferSize = 1 << 14;
+
+        std::unique_ptr<unsigned char[]> compressBuffer =
+            std::make_unique<unsigned char[]>(kBufferSize);
+
+        do
+        {
+            inflateState.avail_out = (uInt) kBufferSize;
+            inflateState.next_out = compressBuffer.get();
+
+            int ret = inflate(&inflateState, Z_SYNC_FLUSH);
+
+            if (ret == Z_NEED_DICT || ret == Z_DATA_ERROR || ret == Z_MEM_ERROR)
+            {
+                inflateEnd(&inflateState);
+                return false;
+            }
+
+            out.append(
+                reinterpret_cast<char *>(compressBuffer.get()),
+                kBufferSize - inflateState.avail_out
+            );
+        } while (inflateState.avail_out == 0);
+
+        inflateEnd(&inflateState);
+        return true;
+    }
+
+    void HttpClient::log(const std::string& msg,
+                         const HttpRequestArgs& args)
+    {
+        if (args.logger)
+        {
+            args.logger(msg);
+        }
+    }
+}
--- a/ixwebsocket/IXHttpClient.h
+++ b/ixwebsocket/IXHttpClient.h
@ -0,0 +1,106 @@
+/*
+ *  IXHttpClient.h
+ *  Author: Benjamin Sergeant
+ *  Copyright (c) 2019 Machine Zone, Inc. All rights reserved.
+ */
+
+#pragma once
+
+#include <algorithm>
+#include <functional>
+#include <mutex>
+#include <atomic>
+#include <tuple>
+#include <memory>
+#include <map>
+
+#include "IXSocket.h"
+#include "IXWebSocketHttpHeaders.h"
+
+namespace ix
+{
+    enum HttpErrorCode
+    {
+        HttpErrorCode_Ok = 0,
+        HttpErrorCode_CannotConnect = 1,
+        HttpErrorCode_Timeout = 2,
+        HttpErrorCode_Gzip = 3,
+        HttpErrorCode_UrlMalformed = 4,
+        HttpErrorCode_CannotCreateSocket = 5,
+        HttpErrorCode_SendError = 6,
+        HttpErrorCode_ReadError = 7,
+        HttpErrorCode_CannotReadStatusLine = 8,
+        HttpErrorCode_MissingStatus = 9,
+        HttpErrorCode_HeaderParsingError = 10,
+        HttpErrorCode_MissingLocation = 11,
+        HttpErrorCode_TooManyRedirects = 12,
+        HttpErrorCode_ChunkReadError = 13,
+        HttpErrorCode_CannotReadBody = 14
+    };
+
+    using HttpResponse = std::tuple<int, // status
+                                    HttpErrorCode, // error code
+                                    WebSocketHttpHeaders,
+                                    std::string, // payload
+                                    std::string, // error msg
+                                    uint64_t,    // upload size
+                                    uint64_t>;   // download size
+
+    using HttpParameters = std::map<std::string, std::string>;
+    using Logger = std::function<void(const std::string&)>;
+
+    struct HttpRequestArgs
+    {
+        std::string url;
+        WebSocketHttpHeaders extraHeaders;
+        std::string body;
+        int connectTimeout;
+        int transferTimeout;
+        bool followRedirects;
+        int maxRedirects;
+        bool verbose;
+        bool compress;
+        Logger logger;
+    };
+
+    class HttpClient {
+    public:
+        HttpClient();
+        ~HttpClient();
+
+        HttpResponse get(const std::string& url,
+                         const HttpRequestArgs& args);
+        HttpResponse head(const std::string& url,
+                          const HttpRequestArgs& args);
+
+        HttpResponse post(const std::string& url,
+                          const HttpParameters& httpParameters,
+                          const HttpRequestArgs& args);
+        HttpResponse post(const std::string& url,
+                          const std::string& body,
+                          const HttpRequestArgs& args);
+
+    private:
+        HttpResponse request(const std::string& url,
+                             const std::string& verb,
+                             const std::string& body,
+                             const HttpRequestArgs& args,
+                             int redirects = 0);
+
+        std::string serializeHttpParameters(const HttpParameters& httpParameters);
+
+        std::string urlEncode(const std::string& value);
+
+        void log(const std::string& msg, const HttpRequestArgs& args);
+
+        bool gzipInflate(
+            const std::string& in,
+            std::string& out);
+
+        std::shared_ptr<Socket> _socket;
+
+        const static std::string kPost;
+        const static std::string kGet;
+        const static std::string kHead;
+    };
+}
--- a/ixwebsocket/IXSocketFactory.cpp
+++ b/ixwebsocket/IXSocketFactory.cpp
@ -0,0 +1,42 @@
+/*
+ *  IXSocketFactory.cpp
+ *  Author: Benjamin Sergeant
+ *  Copyright (c) 2019 Machine Zone, Inc. All rights reserved.
+ */
+
+#include "IXSocketFactory.h"
+
+#if defined(__APPLE__) or defined(__linux__)
+# ifdef __APPLE__
+#  include <ixwebsocket/IXSocketAppleSSL.h>
+# else
+#  include <ixwebsocket/IXSocketOpenSSL.h>
+# endif
+#endif
+
+namespace ix
+{
+    std::shared_ptr<Socket> createSocket(bool tls,
+                                         std::string& errorMsg)
+    {
+        errorMsg.clear();
+
+        if (!tls)
+        {
+            return std::make_shared<Socket>();
+        }
+        else
+        {
+#ifdef IXWEBSOCKET_USE_TLS
+# ifdef __APPLE__
+            return std::make_shared<SocketAppleSSL>();
+# else
+            return std::make_shared<SocketOpenSSL>();
+# endif
+#else
+            errorMsg = "TLS support is not enabled on this platform.";
+            return nullptr;
+#endif
+        }
+    }
+}
--- a/ixwebsocket/IXSocketFactory.h
+++ b/ixwebsocket/IXSocketFactory.h
@ -0,0 +1,17 @@
+
+/*
+ *  IXSocketFactory.h
+ *  Author: Benjamin Sergeant
+ *  Copyright (c) 2018 Machine Zone, Inc. All rights reserved.
+ */
+
+#pragma once
+
+#include <memory>
+
+namespace ix
+{
+    class Socket;
+    std::shared_ptr<Socket> createSocket(bool tls,
+                                         std::string& errorMsg);
+}
--- a/ixwebsocket/IXUrlParser.cpp
+++ b/ixwebsocket/IXUrlParser.cpp
@ -0,0 +1,104 @@
+/*
+ *  IXUrlParser.cpp
+ *  Author: Benjamin Sergeant
+ *  Copyright (c) 2019 Machine Zone, Inc. All rights reserved.
+ */
+
+#include "IXUrlParser.h"
+
+#include <iostream>
+#include <sstream>
+
+
+namespace ix
+{
+    //
+    // The only difference between those 2 regex is the protocol
+    //
+    std::regex UrlParser::_httpRegex("(http|https)://([^/ :]+):?([^/ ]*)(/?[^ #?]*)\\x3f?([^ #]*)#?([^ ]*)");
+    std::regex UrlParser::_webSocketRegex("(ws|wss)://([^/ :]+):?([^/ ]*)(/?[^ #?]*)\\x3f?([^ #]*)#?([^ ]*)");
+
+    bool UrlParser::parse(const std::string& url,
+                          std::string& protocol,
+                          std::string& host,
+                          std::string& path,
+                          std::string& query,
+                          int& port,
+                          bool websocket)
+    {
+        std::cmatch what;
+        if (!regex_match(url.c_str(), what,
+                         websocket ? _webSocketRegex : _httpRegex))
+        {
+            return false;
+        }
+
+        std::string portStr;
+
+        protocol = std::string(what[1].first, what[1].second);
+        host     = std::string(what[2].first, what[2].second);
+        portStr  = std::string(what[3].first, what[3].second);
+        path     = std::string(what[4].first, what[4].second);
+        query    = std::string(what[5].first, what[5].second);
+
+        if (portStr.empty())
+        {
+            if (protocol == "ws" || protocol == "http")
+            {
+                port = 80;
+            }
+            else if (protocol == "wss" || protocol == "https")
+            {
+                port = 443;
+            }
+            else
+            {
+                // Invalid protocol. Should be caught by regex check
+                // but this missing branch trigger cpplint linter.
+                return false;
+            }
+        }
+        else
+        {
+            std::stringstream ss;
+            ss << portStr;
+            ss >> port;
+        }
+
+        if (path.empty())
+        {
+            path = "/";
+        }
+        else if (path[0] != '/')
+        {
+            path = '/' + path;
+        }
+
+        if (!query.empty())
+        {
+            path += "?";
+            path += query;
+        }
+
+        return true;
+    }
+
+    void UrlParser::printUrl(const std::string& url, bool websocket)
+    {
+        std::string protocol, host, path, query;
+        int port {0};
+
+        if (!parse(url, protocol, host, path, query, port, websocket))
+        {
+            return;
+        }
+
+        std::cout << "[" << url << "]" << std::endl;
+        std::cout << protocol << std::endl;
+        std::cout << host << std::endl;
+        std::cout << port << std::endl;
+        std::cout << path << std::endl;
+        std::cout << query << std::endl;
+        std::cout << "-------------------------------" << std::endl;
+    }
+}
--- a/ixwebsocket/IXUrlParser.h
+++ b/ixwebsocket/IXUrlParser.h
@ -0,0 +1,31 @@
+/*
+ *  IXUrlParser.h
+ *  Author: Benjamin Sergeant
+ *  Copyright (c) 2019 Machine Zone, Inc. All rights reserved.
+ */
+
+#pragma once
+
+#include <string>
+#include <regex>
+
+namespace ix
+{
+    class UrlParser
+    {
+    public:
+        static bool parse(const std::string& url,
+                          std::string& protocol,
+                          std::string& host,
+                          std::string& path,
+                          std::string& query,
+                          int& port,
+                          bool websocket);
+
+        static void printUrl(const std::string& url, bool websocket);
+
+    private:
+        static std::regex _httpRegex;
+        static std::regex _webSocketRegex;
+    };
+}
--- a/ixwebsocket/IXWebSocketHandshake.cpp
+++ b/ixwebsocket/IXWebSocketHandshake.cpp
@ -6,6 +6,7 @@

 #include "IXWebSocketHandshake.h"
 #include "IXSocketConnect.h"
+#include "IXUrlParser.h"

 #include "libwshandshake.hpp"

@ -32,90 +33,6 @@ namespace ix

    }

-    bool WebSocketHandshake::parseUrl(const std::string& url,
-                                      std::string& protocol,
-                                      std::string& host,
-                                      std::string& path,
-                                      std::string& query,
-                                      int& port)
-    {
-        std::regex ex("(ws|wss)://([^/ :]+):?([^/ ]*)(/?[^ #?]*)\\x3f?([^ #]*)#?([^ ]*)");
-        std::cmatch what;
-        if (!regex_match(url.c_str(), what, ex))
-        {
-            return false;
-        }
-
-        std::string portStr;
-
-        protocol = std::string(what[1].first, what[1].second);
-        host     = std::string(what[2].first, what[2].second);
-        portStr  = std::string(what[3].first, what[3].second);
-        path     = std::string(what[4].first, what[4].second);
-        query    = std::string(what[5].first, what[5].second);
-
-        if (portStr.empty())
-        {
-            if (protocol == "ws")
-            {
-                port = 80;
-            }
-            else if (protocol == "wss")
-            {
-                port = 443;
-            }
-            else
-            {
-                // Invalid protocol. Should be caught by regex check
-                // but this missing branch trigger cpplint linter.
-                return false;
-            }
-        }
-        else
-        {
-            std::stringstream ss;
-            ss << portStr;
-            ss >> port;
-        }
-
-        if (path.empty())
-        {
-            path = "/";
-        }
-        else if (path[0] != '/')
-        {
-            path = '/' + path;
-        }
-
-        if (!query.empty())
-        {
-            path += "?";
-            path += query;
-        }
-
-        return true;
-    }
-
-    void WebSocketHandshake::printUrl(const std::string& url)
-    {
-        std::string protocol, host, path, query;
-        int port {0};
-
-        if (!WebSocketHandshake::parseUrl(url, protocol, host,
-                                          path, query, port))
-        {
-            return;
-        }
-
-        std::cout << "[" << url << "]" << std::endl;
-        std::cout << protocol << std::endl;
-        std::cout << host << std::endl;
-        std::cout << port << std::endl;
-        std::cout << path << std::endl;
-        std::cout << query << std::endl;
-        std::cout << "-------------------------------" << std::endl;
-    }
-
    std::string WebSocketHandshake::trim(const std::string& str)
    {
        std::string out(str);
@ -192,61 +109,6 @@ namespace ix
        return s;
    }

-
-    std::pair<bool, WebSocketHttpHeaders> WebSocketHandshake::parseHttpHeaders(
-        const CancellationRequest& isCancellationRequested)
-    {
-        WebSocketHttpHeaders headers;
-
-        char line[256];
-        int i;
-
-        while (true)
-        {
-            int colon = 0;
-
-            for (i = 0;
-                 i < 2 || (i < 255 && line[i-2] != '\r' && line[i-1] != '\n');
-                 ++i)
-            {
-                if (!_socket->readByte(line+i, isCancellationRequested))
-                {
-                    return std::make_pair(false, headers);
-                }
-
-                if (line[i] == ':' && colon == 0)
-                {
-                    colon = i;
-                }
-            }
-            if (line[0] == '\r' && line[1] == '\n')
-            {
-                break;
-            }
-
-            // line is a single header entry. split by ':', and add it to our
-            // header map. ignore lines with no colon.
-            if (colon > 0)
-            {
-                line[i] = '\0';
-                std::string lineStr(line);
-                // colon is ':', colon+1 is ' ', colon+2 is the start of the value.
-                // i is end of string (\0), i-colon is length of string minus key;
-                // subtract 1 for '\0', 1 for '\n', 1 for '\r',
-                // 1 for the ' ' after the ':', and total is -4
-                std::string name(lineStr.substr(0, colon));
-                std::string value(lineStr.substr(colon + 2, i - colon - 4));
-
-                // Make the name lower case.
-                std::transform(name.begin(), name.end(), name.begin(), ::tolower);
-
-                headers[name] = value;
-            }
-        }
-
-        return std::make_pair(true, headers);
-    }
-
    WebSocketInitResult WebSocketHandshake::sendErrorResponse(int code, const std::string& reason)
    {
        std::stringstream ss;
@ -355,7 +217,7 @@ namespace ix
            return WebSocketInitResult(false, status, ss.str());
        }

-        auto result = parseHttpHeaders(isCancellationRequested);
+        auto result = parseHttpHeaders(_socket, isCancellationRequested);
        auto headersValid = result.first;
        auto headers = result.second;

@ -450,7 +312,7 @@ namespace ix
        }

        // Retrieve and validate HTTP headers
-        auto result = parseHttpHeaders(isCancellationRequested);
+        auto result = parseHttpHeaders(_socket, isCancellationRequested);
        auto headersValid = result.first;
        auto headers = result.second;

--- a/ixwebsocket/IXWebSocketHandshake.h
+++ b/ixwebsocket/IXWebSocketHandshake.h
@ -59,19 +59,10 @@ namespace ix
        WebSocketInitResult serverHandshake(int fd,
                                            int timeoutSecs);

-        static bool parseUrl(const std::string& url,
-                             std::string& protocol,
-                             std::string& host,
-                             std::string& path,
-                             std::string& query,
-                             int& port);
-
    private:
-        static void printUrl(const std::string& url);
        std::string genRandomString(const int len);

        // Parse HTTP headers
-        std::pair<bool, WebSocketHttpHeaders> parseHttpHeaders(const CancellationRequest& isCancellationRequested);
        WebSocketInitResult sendErrorResponse(int code, const std::string& reason);

        std::tuple<std::string, std::string, std::string> parseRequestLine(const std::string& line);
--- a/ixwebsocket/IXWebSocketHttpHeaders.cpp
+++ b/ixwebsocket/IXWebSocketHttpHeaders.cpp
@ -0,0 +1,66 @@
+/*
+ *  IXWebSocketHttpHeaders.h
+ *  Author: Benjamin Sergeant
+ *  Copyright (c) 2018 Machine Zone, Inc. All rights reserved.
+ */
+
+#include "IXWebSocketHttpHeaders.h"
+#include "IXSocket.h"
+
+#include <string>
+#include <unordered_map>
+
+namespace ix
+{
+    std::pair<bool, WebSocketHttpHeaders> parseHttpHeaders(
+        std::shared_ptr<Socket> socket,
+        const CancellationRequest& isCancellationRequested)
+    {
+        WebSocketHttpHeaders headers;
+
+        char line[1024];
+        int i;
+
+        while (true)
+        {
+            int colon = 0;
+
+            for (i = 0;
+                 i < 2 || (i < 1023 && line[i-2] != '\r' && line[i-1] != '\n');
+                 ++i)
+            {
+                if (!socket->readByte(line+i, isCancellationRequested))
+                {
+                    return std::make_pair(false, headers);
+                }
+
+                if (line[i] == ':' && colon == 0)
+                {
+                    colon = i;
+                }
+            }
+            if (line[0] == '\r' && line[1] == '\n')
+            {
+                break;
+            }
+
+            // line is a single header entry. split by ':', and add it to our
+            // header map. ignore lines with no colon.
+            if (colon > 0)
+            {
+                line[i] = '\0';
+                std::string lineStr(line);
+                // colon is ':', colon+1 is ' ', colon+2 is the start of the value.
+                // i is end of string (\0), i-colon is length of string minus key;
+                // subtract 1 for '\0', 1 for '\n', 1 for '\r',
+                // 1 for the ' ' after the ':', and total is -4
+                std::string name(lineStr.substr(0, colon));
+                std::string value(lineStr.substr(colon + 2, i - colon - 4));
+
+                headers[name] = value;
+            }
+        }
+
+        return std::make_pair(true, headers);
+    }
+}
--- a/ixwebsocket/IXWebSocketHttpHeaders.h
+++ b/ixwebsocket/IXWebSocketHttpHeaders.h
@ -6,10 +6,40 @@

 #pragma once

+#include "IXCancellationRequest.h"
+
 #include <string>
-#include <unordered_map>
+#include <map>
+#include <memory>
+#include <algorithm>

 namespace ix
 {
-    using WebSocketHttpHeaders = std::unordered_map<std::string, std::string>;
+    class Socket;
+
+    struct CaseInsensitiveLess
+    {
+        // Case Insensitive compare_less binary function
+        struct NocaseCompare
+        {
+            bool operator() (const unsigned char& c1, const unsigned char& c2) const
+            {
+                return std::tolower(c1) < std::tolower(c2);
+            }
+        };
+
+        bool operator() (const std::string & s1, const std::string & s2) const
+        {
+            return std::lexicographical_compare
+                (s1.begin(), s1.end(),   // source range
+                 s2.begin(), s2.end(),   // dest range
+                 NocaseCompare());  // comparison
+        }
+    };
+
+    using WebSocketHttpHeaders = std::map<std::string, std::string, CaseInsensitiveLess>;
+
+    std::pair<bool, WebSocketHttpHeaders> parseHttpHeaders(
+        std::shared_ptr<Socket> socket,
+        const CancellationRequest& isCancellationRequested);
 }
--- a/ixwebsocket/IXWebSocketTransport.cpp
+++ b/ixwebsocket/IXWebSocketTransport.cpp
@ -11,6 +11,8 @@
 #include "IXWebSocketTransport.h"
 #include "IXWebSocketHandshake.h"
 #include "IXWebSocketHttpHeaders.h"
+#include "IXUrlParser.h"
+#include "IXSocketFactory.h"

 #ifdef IXWEBSOCKET_USE_TLS
 # ifdef __APPLE__
@ -70,31 +72,31 @@ namespace ix
    {
        std::string protocol, host, path, query;
        int port;
+        bool websocket = true;

-        if (!WebSocketHandshake::parseUrl(url, protocol, host,
-                                          path, query, port))
+        if (!UrlParser::parse(url, protocol, host, path, query, port, websocket))
        {
            return WebSocketInitResult(false, 0,
                                       std::string("Could not parse URL ") + url);
        }

-        if (protocol == "wss")
+        if (protocol != "ws" && protocol != "wss")
        {
-            _socket.reset();
-#ifdef IXWEBSOCKET_USE_TLS
-# ifdef __APPLE__
-             _socket = std::make_shared<SocketAppleSSL>();
-# else
-             _socket = std::make_shared<SocketOpenSSL>();
-# endif
-#else
-            return WebSocketInitResult(false, 0, "TLS is not supported.");
-#endif
+            std::stringstream ss;
+            ss << "Invalid protocol: " << protocol
+               << " for url " << url
+               << " . Supported protocols are ws and wss";
+
+            return WebSocketInitResult(false, 0, ss.str());
        }
-        else
+
+        bool tls = protocol == "wss";
+        std::string errorMsg;
+        _socket = createSocket(tls, errorMsg);
+
+        if (!_socket)
        {
-            _socket.reset();
-            _socket = std::make_shared<Socket>();
+            return WebSocketInitResult(false, 0, errorMsg);
        }

        WebSocketHandshake webSocketHandshake(_requestInitCancellation,