Fix parsing urls containing colon in path element

Author: uralm1

Diff for: include/restc-cpp/Url.h

@@ -3,7 +3,7 @@
 #ifndef RESTC_CPP_URL_H_
 #define RESTC_CPP_URL_H_
 
-#include <boost/utility/string_ref.hpp>
+#include <boost/utility/string_view.hpp>
 
 namespace restc_cpp {
 
@@ -20,19 +20,19 @@ namespace restc_cpp {
 
      Url& operator = (const char *url);
 
-     boost::string_ref GetProtocolName() const { return protocol_name_; }
-     boost::string_ref GetHost() const { return host_; }
-     boost::string_ref GetPort() const { return port_; }
-     boost::string_ref GetPath() const { return path_; }
-     boost::string_ref GetArgs() const { return args_; }
+     boost::string_view GetProtocolName() const { return protocol_name_; }
+     boost::string_view GetHost() const { return host_; }
+     boost::string_view GetPort() const { return port_; }
+     boost::string_view GetPath() const { return path_; }
+     boost::string_view GetArgs() const { return args_; }
      Protocol GetProtocol() const { return protocol_; }
 
  private:
-     boost::string_ref protocol_name_;
-     boost::string_ref host_;
-     boost::string_ref port_;
-     boost::string_ref path_ = "/";
-     boost::string_ref args_;
+     boost::string_view protocol_name_;
+     boost::string_view host_;
+     boost::string_view port_;
+     boost::string_view path_ = "/";
+     boost::string_view args_;
      Protocol protocol_ = Protocol::UNKNOWN;
  };
 

Diff for: include/restc-cpp/url_encode.h

@@ -1,7 +1,16 @@
+#pragma once
+
+#ifndef RESTC_CPP_URL_ENCODE_H_
+#define RESTC_CPP_URL_ENCODE_H_
+
 #include "restc-cpp.h"
 
+#include <boost/utility/string_view.hpp>
+
 namespace restc_cpp {
 
-std::string url_encode(const boost::string_ref& src);
+std::string url_encode(const boost::string_view& src);
 
 } // namespace
+
+#endif // RESTC_CPP_URL_ENCODE_H_

Diff for: src/Url.cpp

@@ -1,7 +1,7 @@
 #include <cassert>
 #include <array>
 
-#include <boost/utility/string_ref.hpp>
+#include <boost/utility/string_view.hpp>
 #include "restc-cpp/restc-cpp.h"
 #include "restc-cpp/Url.h"
 #include "restc-cpp/error.h"
@@ -28,44 +28,48 @@ Url& Url::operator = (const char *url) {
     constexpr auto magic_7 = 7;
 
     assert(url != nullptr && "A valid URL is required");
-    protocol_name_ = boost::string_ref(url);
+    protocol_name_ = boost::string_view(url);
     if (protocol_name_.find("https://") == 0) {
-        protocol_name_ = boost::string_ref(url, magic_8);
+        protocol_name_ = boost::string_view(url, magic_8);
         protocol_ = Protocol::HTTPS;
     } else if (protocol_name_.find("http://") == 0) {
-        protocol_name_ = boost::string_ref(url, magic_7);
+        protocol_name_ = boost::string_view(url, magic_7);
         protocol_ = Protocol::HTTP;
     } else {
         throw ParseException("Invalid protocol in url. Must be 'http[s]://'");
     }
 
-    auto remains = boost::string_ref(protocol_name_.end());
+    auto remains = boost::string_view(protocol_name_.end());
     const auto args_start = remains.find('?');
     if (args_start != string::npos) {
         args_ = {remains.begin() +  args_start + 1,
             remains.size() - (args_start + 1)};
         remains = {remains.begin(), args_start};
     }
+    auto path_start = remains.find('/');
     const auto port_start = remains.find(':');
-    if (port_start != string::npos) {
+    if (port_start != string::npos &&
+        ( path_start == string::npos ||
+          port_start < path_start )
+       ) {
         if (remains.length() <= static_cast<decltype(host_.length())>(port_start + 2)) {
             throw ParseException("Invalid host (no port after column)");
         }
-        //port_ = boost::string_ref(&remains[port_start+1]);
-        //host_ = boost::string_ref(host_.data(), port_start);
+        //port_ = boost::string_view(&remains[port_start+1]);
+        //host_ = boost::string_view(host_.data(), port_start);
         host_ = {remains.begin(), port_start};
         remains = {remains.begin() + port_start + 1, remains.size() - (port_start + 1)};
 
-        const auto path_start = remains.find('/');
         if (path_start != string::npos) {
+            //path_start = remains.find('/');
+            path_start -= port_start + 1;
             path_ = {remains.begin() + path_start, remains.size() - path_start};// &port_[path_start];
             port_ = {remains.begin(), path_start};
             remains = {};
         } else {
             port_ = remains;
         }
     } else {
-        const auto path_start = remains.find('/');
         if (path_start != string::npos) {
             //path_ = &host_[path_start];
             //host_ = boost::string_ref(host_.data(), path_start);

Diff for: src/url_encode.cpp

@@ -35,7 +35,7 @@ allchars_t get_normal_ch() {
 
 } // anonymous namespace
 
-std::string url_encode(const boost::string_ref& src) {
+std::string url_encode(const boost::string_view& src) {
     constexpr auto magic_4 = 4;
     constexpr auto magic_0x0f = 0x0f;
 

Diff for: tests/unit/UrlTests.cpp

@@ -49,6 +49,15 @@ TEST(Url, UrlWithPathAndSlash)
     EXPECT_EQ("/jgaa/"s, url.GetPath());
 }
 
+TEST(Url, UrlWithPathInclColon)
+{
+    Url url("http://github.com/jgaa:test");
+    EXPECT_EQ("github.com"s, url.GetHost());
+    EXPECT_EQ("80"s, url.GetPort());
+    EXPECT_EQ_ENUM(Url::Protocol::HTTP, url.GetProtocol());
+    EXPECT_EQ("/jgaa:test"s, url.GetPath());
+}
+
 TEST(Url, HttpWithPort)
 {
     Url url("http://github.com:56");
@@ -85,6 +94,15 @@ TEST(Url, HttpWithPortAndPath)
     EXPECT_EQ("/jgaa"s, url.GetPath());
 }
 
+TEST(Url, HttpWithPortAndPathInclColon)
+{
+    Url url("http://github.com:12345/jgaa:test");
+    EXPECT_EQ("github.com"s, url.GetHost());
+    EXPECT_EQ("12345"s, url.GetPort());
+    EXPECT_EQ_ENUM(Url::Protocol::HTTP, url.GetProtocol());
+    EXPECT_EQ("/jgaa:test"s, url.GetPath());
+}
+
 TEST(Url, HttpWithPortAndPathPath)
 {
     Url url("http://github.com:12345/jgaa/andmore");