Speedup GetPath()

5 files changed, 68 insertions, 51 deletions

diff --git a/TODO b/TODO
index 584b690..76149ec 100644
--- a/TODO
+++ b/TODO
@@ -1,4 +1,3 @@
-Speed up config.GetPath
 weblog: link consistency check (cron?)
 Integrate into Debian: WNPP
 support alternative SSL libs: mbedtls, gnutls, wolfssl, botan, (matrixssl, libressl, cryptlib: not in debian)
diff --git a/config.cpp b/config.cpp
index 2ece1ef..ed28ea9 100644
--- a/config.cpp
+++ b/config.cpp
@@ -56,15 +56,16 @@ void Config::readConfigfile(std::string filename)
     for (const auto& site: element.second) {
      if (site.first != "site"s)
       throw std::runtime_error("<site> expected in <sites>");
+     std::string site_name;
      Site site_struct;
      for (const auto& x: site.second) {
       if (x.first == "name"s) {
-       if (site_struct.name == "")
-        site_struct.name = x.second.data();
+       if (site_name == "")
+        site_name = x.second.data();
        else
         throw std::runtime_error("Found double site name: "s + x.second.data());
       } else if (x.first == "host"s) {
-       if (std::find(site_struct.hosts.begin(), site_struct.hosts.end(), x.second.data()) == site_struct.hosts.end())
+       if (site_struct.hosts.find(x.second.data()) == site_struct.hosts.end())
         site_struct.hosts.insert(x.second.data());
        else
         throw std::runtime_error("Found double site host element: "s + x.second.data());
@@ -110,10 +111,12 @@ void Config::readConfigfile(std::string filename)
       } else
        throw std::runtime_error("Unknown element: "s + x.first);
      }
-     if (std::find_if(m_sites.begin(), m_sites.end(), [&](const Site& site){return site.name == site_struct.name;}) == m_sites.end())
-      m_sites.push_back(site_struct);
+     if (site_name.empty())
+      throw std::runtime_error("Empty site name");
+     if (m_sites.find(site_name) == m_sites.end())
+      m_sites[site_name] = site_struct;
      else
-      throw std::runtime_error("Found double site spec: "s + site_struct.name);
+      throw std::runtime_error("Found double site spec: "s + site_name);
     }
    } else if (element.first == "sockets"s) {
     for (const auto& socket: element.second) {
@@ -140,10 +143,10 @@ void Config::readConfigfile(std::string filename)
         throw std::runtime_error("Unknown protocol: "s + x.second.data());
       } else if (x.first == "site"s) {
        std::string site {x.second.data()};
-       if (std::find(socket_struct.serve_sites.begin(), socket_struct.serve_sites.end(), site) == socket_struct.serve_sites.end()) {
-        socket_struct.serve_sites.push_back(site);
+       if (socket_struct.serve_sites.find(site) == socket_struct.serve_sites.end()) {
+        socket_struct.serve_sites.insert(site);
        } else {
-        throw std::runtime_error("Site "s + site + " already defined for "s + socket_struct.address + " port " + socket_struct.port);
+        throw std::runtime_error("Site "s + site + " already defined for "s + socket_struct.address + ", port " + socket_struct.port);
        }
       } else
        throw std::runtime_error("Unknown element: "s + x.first);
@@ -161,10 +164,26 @@ void Config::readConfigfile(std::string filename)
  for (auto& socket: m_sockets) {
   if (socket.serve_sites.empty()) {
    for (const auto& site: m_sites) {
-    socket.serve_sites.push_back(site.name);
+    socket.serve_sites.insert(site.first);
    }
   }
  }
+
+ validate();
+}
+
+// just throws on inconsistency
+void Config::validate()
+{
+ // make sure all m_sockets.serve_sites are configured in m_sites
+
+ for (auto& socket: m_sockets) {
+  for (auto& serve_site: socket.serve_sites) {
+   if (m_sites.find(serve_site) == m_sites.end())
+    throw std::runtime_error("Found serve_site "s + serve_site + " without configured site"s);
+  }
+ }
+ 
 }
 
 Config::Config(const std::string& filename)
@@ -193,7 +212,7 @@ const std::vector<std::string>& Config::PluginDirectories() const
  return m_plugin_directories;
 }
 
-const std::vector<Site>& Config::Sites() const
+const std::unordered_map<std::string, Site>& Config::Sites() const
 {
  return m_sites;
 }
@@ -217,21 +236,21 @@ void Config::dump() const
  std::cout << std::endl;
 
  for (const auto& site: m_sites) {
-  std::cout << "Site: " << site.name << ":";
-  for (const auto& host: site.hosts)
+  std::cout << "Site: " << site.first << ":";
+  for (const auto& host: site.second.hosts)
    std::cout << " " << host;
   std::cout << std::endl;
-  if (site.paths.size() == 0)
+  if (site.second.paths.size() == 0)
    std::cout << "  Warning: No paths configured." << std::endl;
-  for (const auto& path: site.paths) {
+  for (const auto& path: site.second.paths) {
    std::cout << "  Path: " << path.requested << std::endl;
    for (const auto& param: path.params) {
     std::cout << "    " << param.first << ": " << param.second << std::endl;
    }
   }
-  if (site.key_path != ""s) {
-   std::cout << "  Key: " << site.key_path.generic_string() << std::endl;
-   std::cout << "  Cert: " << site.cert_path.generic_string() << std::endl;
+  if (site.second.key_path != ""s) {
+   std::cout << "  Key: " << site.second.key_path.generic_string() << std::endl;
+   std::cout << "  Cert: " << site.second.cert_path.generic_string() << std::endl;
   }
  }
 
@@ -251,29 +270,26 @@ const Path& Config::GetPath(const Socket& socket, const std::string& requested_h
 {
  //boost::timer::auto_cpu_timer t;
 
- // TODO: speed this up
  std::string host{requested_host};
  const Path* result{nullptr};
  size_t path_len{0}; // find longest matching prefix
 
  RemovePortFromHostname(host);
 
- for (const auto& site: m_sites) {
-  if (std::find(socket.serve_sites.begin(), socket.serve_sites.end(), site.name) != socket.serve_sites.end()) {
-   for (const auto& m_host: site.hosts) {
-    if (m_host == host) {
-     for (const auto& path: site.paths) {
-      if (boost::starts_with(requested_path, path.requested) &&
-          ("/?"s.find(requested_path[path.requested.size()]) != std::string::npos ||
-           requested_path[path.requested.size()] == 0 ||
-           requested_path[path.requested.size() - 1] == '/'
-          ) &&
-          path.requested.size() > path_len)
-      {
-       path_len = path.requested.size();
-       result = &path;
-      }
-     }
+ for (const auto& site_name: socket.serve_sites) {
+  const Site& site {m_sites.at(site_name)}; // omit error check: validation previously made sure this exists
+
+  if (site.hosts.find(host) != site.hosts.end()) {
+   for (const auto& path: site.paths) {
+    if (boost::starts_with(requested_path, path.requested) &&
+        ("/?"s.find(requested_path[path.requested.size()]) != std::string::npos ||
+         requested_path[path.requested.size()] == 0 ||
+         requested_path[path.requested.size() - 1] == '/'
+        ) &&
+        path.requested.size() > path_len)
+    {
+     path_len = path.requested.size();
+     result = &path;
     }
    }
   }
@@ -288,7 +304,7 @@ const Path& Config::GetPath(const Socket& socket, const std::string& requested_h
 bool Config::PluginIsConfigured(const std::string& name) const
 {
  for (const auto& site: m_sites) {
-  for (const auto& path: site.paths) {
+  for (const auto& path: site.second.paths) {
    auto it{path.params.find("plugin")};
    if (it != path.params.end() && it->second == name)
     return true;
diff --git a/config.h b/config.h
index 46cd267..e60bba5 100644
--- a/config.h
+++ b/config.h
@@ -18,7 +18,7 @@ struct Path
 
 struct Site
 {
- std::string name;
+ // std::string name; is the index in the m_sites map
  std::unordered_set<std::string> hosts;
  std::vector<Path> paths;
  fs::path cert_path;
@@ -36,7 +36,7 @@ struct Socket
  std::string address;
  std::string port;
  SocketProtocol protocol;
- std::vector<std::string> serve_sites; // if empty, automatically expand to all configured sites
+ std::unordered_set<std::string> serve_sites; // if empty, automatically expand to all configured sites
 };
 
 class Config
@@ -44,12 +44,13 @@ class Config
  const std::string default_filename{"/etc/webserver.conf"};
 
  void readConfigfile(std::string filename);
+ void validate();
 
  std::string m_user;
  std::string m_group;
  int m_threads;
  std::vector<std::string> m_plugin_directories;
- std::vector<Site> m_sites;
+ std::unordered_map<std::string, Site> m_sites;
  std::vector<Socket> m_sockets;
 
  public:
@@ -62,7 +63,7 @@ class Config
   int Threads() const;
 
   const std::vector<std::string>& PluginDirectories() const;
-  const std::vector<Site>& Sites() const;
+  const std::unordered_map<std::string, Site>& Sites() const;
   const std::vector<Socket>& Sockets() const;
 
   //
diff --git a/https.cpp b/https.cpp
index 508ece9..a5aa118 100644
--- a/https.cpp
+++ b/https.cpp
@@ -660,15 +660,15 @@ void Server::load_certificates()
  // import the real certificates
  for (const auto& serve_site: m_socket.serve_sites) {
   for (const auto& site: m_config.Sites()) {
-   if (site.name == serve_site) {
+   if (site.first == serve_site) {
     std::shared_ptr<ssl::context> ctx {std::make_shared<ssl::context>(tls_method)};
 
     std::cout << "Creating SSL context/cert for site " << serve_site << " on port " << m_socket.port << std::endl;
 
-    load_server_certificate(*ctx, site.cert_path, site.key_path);
+    load_server_certificate(*ctx, site.second.cert_path, site.second.key_path);
     SSL_CTX_set_client_hello_cb(ctx->native_handle(), servername_callback, &m_ctx);
 
-    for (const auto& host: site.hosts) {
+    for (const auto& host: site.second.hosts) {
      std::cout << "  Adding Host " << host << std::endl;
      m_ctx.emplace(host, ctx);
     }
@@ -681,19 +681,19 @@ void Server::reload_certificates()
 {
  for (const auto& serve_site: m_socket.serve_sites) {
   for (const auto& site: m_config.Sites()) {
-   if (site.name == serve_site) {
+   if (site.first == serve_site) {
 
     std::cout << "Updating SSL context/cert for site " << serve_site << " on port " << m_socket.port << std::endl;
 
-    auto it_host {site.hosts.begin()};
-    if (it_host == site.hosts.end()) {
+    auto it_host {site.second.hosts.begin()};
+    if (it_host == site.second.hosts.end()) {
      std::cout << "  Warning: No configured host found." << std::endl;
     } else {
      auto it_ctx {m_ctx.find(*it_host)};
      if (it_ctx == m_ctx.end()) {
       std::cout << "  Warning: No context found for configured host." << std::endl;
      } else {
-      load_server_certificate(*it_ctx->second, site.cert_path, site.key_path);
+      load_server_certificate(*it_ctx->second, site.second.cert_path, site.second.key_path);
      }
     }
    }
diff --git a/plugin.cpp b/plugin.cpp
index 28fdb62..afcbdb3 100644
--- a/plugin.cpp
+++ b/plugin.cpp
@@ -64,22 +64,23 @@ void PluginLoader::load_plugins()
  }
 }
 
+// validate config regarding plugins
 bool PluginLoader::validate_config()
 {
  const auto& sites{m_config.Sites()};
 
  for (const auto& site: sites) {
-  for (const auto& path: site.paths) {
+  for (const auto& path: site.second.paths) {
    // path must contain target and plugin
    auto it {path.params.find("target")};
    if (it == path.params.end()) {
-    std::cout << "Path " << path.requested << " for site " << site.name << " is missing target specification." << std::endl;
+    std::cout << "Path " << path.requested << " for site " << site.first << " is missing target specification." << std::endl;
     return false;
    }
 
    it = path.params.find("plugin");
    if (it == path.params.end()) {
-    std::cout << "Path " << path.requested << " for site " << site.name << " is missing plugin specification." << std::endl;
+    std::cout << "Path " << path.requested << " for site " << site.first << " is missing plugin specification." << std::endl;
     return false;
    }