Merge pull request #365 from etr/feature/request-path-perf

Optimize per-request critical path performance

Author: etr

src/http_request.cpp

@@ -91,7 +91,7 @@ struct arguments_accumulator {
 };
 
 void http_request::set_method(const std::string& method) {
-    this->method = string_utilities::to_upper_copy(method);
+    this->method = method;
 }
 
 #ifdef HAVE_DAUTH

src/http_utils.cpp

@@ -218,19 +218,15 @@ std::vector<std::string> http_utils::tokenize_url(const std::string& str, const
 }
 
 std::string http_utils::standardize_url(const std::string& url) {
-    std::string n_url = url;
+    if (url.empty()) return url;
 
-    std::string::iterator new_end = std::unique(n_url.begin(), n_url.end(), [](char a, char b) { return (a == b) && (a == '/'); });
-    n_url.erase(new_end, n_url.end());
+    std::string result = url;
 
-    std::string::size_type n_url_length = n_url.length();
+    auto new_end = std::unique(result.begin(), result.end(), [](char a, char b) { return (a == b) && (a == '/'); });
+    result.erase(new_end, result.end());
 
-    std::string result;
-
-    if (n_url_length > 1 && n_url[n_url_length - 1] == '/') {
-        result = n_url.substr(0, n_url_length - 1);
-    } else {
-        result = n_url;
+    if (result.length() > 1 && result.back() == '/') {
+        result.pop_back();
     }
 
     return result;
@@ -302,13 +298,19 @@ uint16_t get_port(const struct sockaddr* sa) {
     }
 }
 
+static inline int hex_digit_value(char c) {
+    if (c >= '0' && c <= '9') return c - '0';
+    if (c >= 'a' && c <= 'f') return c - 'a' + 10;
+    if (c >= 'A' && c <= 'F') return c - 'A' + 10;
+    return -1;
+}
+
 size_t http_unescape(std::string* val) {
     if (val->empty()) return 0;
 
     unsigned int rpos = 0;
     unsigned int wpos = 0;
 
-    unsigned int num;
     unsigned int size = val->size();
 
     while (rpos < size && (*val)[rpos] != '\0') {
@@ -319,11 +321,15 @@ size_t http_unescape(std::string* val) {
                 rpos++;
                 break;
             case '%':
-                if (size > rpos + 2 && ((1 == sscanf(val->substr(rpos + 1, 2).c_str(), "%2x", &num)) || (1 == sscanf(val->substr(rpos + 1, 2).c_str(), "%2X", &num)))) {
-                    (*val)[wpos] = (unsigned char) num;
-                    wpos++;
-                    rpos += 3;
-                    break;
+                if (size > rpos + 2) {
+                    int hi = hex_digit_value((*val)[rpos + 1]);
+                    int lo = hex_digit_value((*val)[rpos + 2]);
+                    if (hi >= 0 && lo >= 0) {
+                        (*val)[wpos] = static_cast<unsigned char>((hi << 4) | lo);
+                        wpos++;
+                        rpos += 3;
+                        break;
+                    }
                 }
             // intentional fall through!
             default:

src/httpserver/details/modded_request.hpp

@@ -37,8 +37,8 @@ namespace details {
 
 struct modded_request {
     struct MHD_PostProcessor *pp = nullptr;
-    std::unique_ptr<std::string> complete_uri;
-    std::unique_ptr<std::string> standardized_url;
+    std::string complete_uri;
+    std::string standardized_url;
     webserver* ws = nullptr;
 
     std::shared_ptr<http_response> (httpserver::http_resource::*callback)(const httpserver::http_request&);

src/httpserver/http_request.hpp

@@ -93,7 +93,8 @@ class http_request {
       * @return a vector of strings containing all pieces
      **/
      const std::vector<std::string> get_path_pieces() const {
-         return http::http_utils::tokenize_url(path);
+         ensure_path_pieces_cached();
+         return cache->path_pieces;
      }
 
      /**
@@ -102,9 +103,9 @@ class http_request {
       * @return the selected piece in form of string
      **/
      const std::string get_path_piece(int index) const {
-         std::vector<std::string> post_path = get_path_pieces();
-         if (static_cast<int>(post_path.size()) > index) {
-             return post_path[index];
+         ensure_path_pieces_cached();
+         if (static_cast<int>(cache->path_pieces.size()) > index) {
+             return cache->path_pieces[index];
          }
          return EMPTY;
      }
@@ -474,7 +475,11 @@ class http_request {
      std::string_view get_connection_value(std::string_view key, enum MHD_ValueKind kind) const;
      const http::header_view_map get_headerlike_values(enum MHD_ValueKind kind) const;
 
-     // Cache certain data items on demand so we can consistently return views
+     // http_request objects are owned by a single connection and are not
+    // shared across threads. Lazy caching (path_pieces, args, etc.) is
+    // safe without synchronization under this invariant.
+
+    // Cache certain data items on demand so we can consistently return views
      // over the data. Some things we transform before returning to the user for
      // simplicity (e.g. query_str, requestor), others out of necessity (arg unescaping).
      // Others (username, password, digested_user) MHD returns as char* that we need
@@ -488,10 +493,19 @@ class http_request {
         std::string digested_user;
 #endif  // HAVE_DAUTH
         std::map<std::string, std::vector<std::string>, http::arg_comparator> unescaped_args;
+        std::vector<std::string> path_pieces;
 
         bool args_populated = false;
+        bool path_pieces_cached = false;
      };
      std::unique_ptr<http_request_data_cache> cache = std::make_unique<http_request_data_cache>();
+     void ensure_path_pieces_cached() const {
+         if (!cache->path_pieces_cached) {
+             cache->path_pieces = http::http_utils::tokenize_url(path);
+             cache->path_pieces_cached = true;
+         }
+     }
+
      // Populate the data cache unescaped_args
      void populate_args() const;
 

src/httpserver/webserver.hpp

@@ -40,11 +40,14 @@
 #include <sys/socket.h>
 #endif
 
+#include <list>
 #include <map>
 #include <memory>
+#include <mutex>
 #include <set>
 #include <shared_mutex>
 #include <string>
+#include <unordered_map>
 #include <vector>
 
 #ifdef HAVE_GNUTLS
@@ -189,6 +192,16 @@ class webserver {
      std::shared_mutex registered_resources_mutex;
      std::map<details::http_endpoint, http_resource*> registered_resources;
      std::map<std::string, http_resource*> registered_resources_str;
+     std::map<details::http_endpoint, http_resource*> registered_resources_regex;
+
+     struct route_cache_entry {
+         details::http_endpoint matched_endpoint;
+         http_resource* resource;
+     };
+     static constexpr size_t ROUTE_CACHE_MAX_SIZE = 256;
+     std::mutex route_cache_mutex;
+     std::list<std::pair<std::string, route_cache_entry>> route_cache_list;
+     std::unordered_map<std::string, std::list<std::pair<std::string, route_cache_entry>>::iterator> route_cache_map;
 
      std::shared_mutex bans_mutex;
      std::set<http::ip_representation> bans;
@@ -227,6 +240,8 @@ class webserver {
 
      MHD_Result complete_request(MHD_Connection* connection, struct details::modded_request* mr, const char* version, const char* method);
 
+     void invalidate_route_cache();
+
 #ifdef HAVE_GNUTLS
      // MHD_PskServerCredentialsCallback signature
      static int psk_cred_handler_func(void* cls,

src/string_utilities.cpp

@@ -22,8 +22,8 @@
 
 #include <algorithm>
 #include <cctype>
-#include <sstream>
 #include <string>
+#include <utility>
 #include <vector>
 
 namespace httpserver {
@@ -45,13 +45,29 @@ const std::string to_lower_copy(const std::string& str) {
 
 const std::vector<std::string> string_split(const std::string& s, char sep, bool collapse) {
     std::vector<std::string> result;
+    if (s.empty()) return result;
 
-    std::istringstream buf(s);
-    for (std::string token; getline(buf, token, sep); ) {
-        if ((collapse && token != "") || !collapse) {
-            result.push_back(token);
+    std::string::size_type start = 0;
+    std::string::size_type end;
+
+    while ((end = s.find(sep, start)) != std::string::npos) {
+        std::string token = s.substr(start, end - start);
+        if (!collapse || !token.empty()) {
+            result.push_back(std::move(token));
         }
+        start = end + 1;
     }
+
+    // Handle the last token (after the final separator)
+    // Only add if there's content or if not collapsing
+    // Note: match istringstream behavior which does not emit trailing empty token
+    if (start < s.size()) {
+        std::string token = s.substr(start);
+        if (!collapse || !token.empty()) {
+            result.push_back(std::move(token));
+        }
+    }
+
     return result;
 }