optimize header parse by zero-copy (#396)

include/cinatra/cinatra_log_wrapper.hpp

@@ -1,3 +1,4 @@
+#pragma once
 #include <iostream>
 namespace cinatra {
 struct null_logger_t {

include/cinatra/coro_http_client.hpp

@@ -22,6 +22,7 @@
 #include "async_simple/coro/Lazy.h"
 #include "cinatra_log_wrapper.hpp"
 #include "http_parser.hpp"
+#include "picohttpparser.h"
 #include "response_cv.hpp"
 #include "uri.hpp"
 #include "websocket.hpp"
@@ -50,11 +51,13 @@ inline ClientInjectAction inject_write_failed = ClientInjectAction::none;
 inline ClientInjectAction inject_read_failed = ClientInjectAction::none;
 #endif
 
+struct http_header;
+
 struct resp_data {
   std::error_code net_err;
   int status;
   std::string_view resp_body;
-  std::vector<std::pair<std::string, std::string>> resp_headers;
+  std::span<http_header> resp_headers;
   bool eof;
 #ifdef BENCHMARK_TEST
   uint64_t total;
@@ -1216,7 +1219,7 @@ class coro_http_client {
       return std::make_error_code(std::errc::protocol_error);
     }
     read_buf_.consume(header_size);  // header size
-    data.resp_headers = get_headers(parser);
+    data.resp_headers = parser.get_headers();
     data.status = parser.status();
     return {};
   }
@@ -1234,8 +1237,7 @@ class coro_http_client {
         break;
       }
 
-      http_parser parser;
-      ec = handle_header(data, parser, size);
+      ec = handle_header(data, parser_, size);
 #ifdef INJECT_FOR_HTTP_CLIENT_TEST
       if (inject_header_valid == ClientInjectAction::header_error) {
         ec = std::make_error_code(std::errc::protocol_error);
@@ -1248,32 +1250,32 @@ class coro_http_client {
         break;
       }
 
-      is_keep_alive = parser.keep_alive();
+      is_keep_alive = parser_.keep_alive();
       if (method == http_method::HEAD) {
         co_return data;
       }
 
-      bool is_ranges = parser.is_ranges();
+      bool is_ranges = parser_.is_ranges();
       if (is_ranges) {
         is_keep_alive = true;
       }
-      if (parser.is_chunked()) {
+      if (parser_.is_chunked()) {
         is_keep_alive = true;
         ec = co_await handle_chunked(data, std::move(ctx));
         break;
       }
 
       redirect_uri_.clear();
-      bool is_redirect = parser.is_location();
+      bool is_redirect = parser_.is_location();
       if (is_redirect)
-        redirect_uri_ = parser.get_header_value("Location");
+        redirect_uri_ = parser_.get_header_value("Location");
 
-      size_t content_len = (size_t)parser.body_len();
+      size_t content_len = (size_t)parser_.body_len();
 #ifdef BENCHMARK_TEST
-      total_len_ = parser.total_len();
+      total_len_ = parser_.total_len();
 #endif
 
-      if ((size_t)parser.body_len() <= read_buf_.size()) {
+      if ((size_t)parser_.body_len() <= read_buf_.size()) {
         // Now get entire content, additional data will discard.
         // copy body.
         if (content_len > 0) {
@@ -1548,20 +1550,6 @@ class coro_http_client {
     co_return resp_data{{}, 200};
   }
 
-  std::vector<std::pair<std::string, std::string>> get_headers(
-      http_parser &parser) {
-    std::vector<std::pair<std::string, std::string>> resp_headers;
-
-    auto [headers, num_headers] = parser.get_headers();
-    for (size_t i = 0; i < num_headers; i++) {
-      resp_headers.emplace_back(
-          std::string(headers[i].name, headers[i].name_len),
-          std::string(headers[i].value, headers[i].value_len));
-    }
-
-    return resp_headers;
-  }
-
   // this function must be called before async_ws_connect.
   async_simple::coro::Lazy<void> async_read_ws() {
     resp_data data{};
@@ -1722,6 +1710,7 @@ class coro_http_client {
     return has_http_scheme;
   }
 
+  http_parser parser_;
   coro_io::ExecutorWrapper<> executor_wrapper_;
   coro_io::period_timer timer_;
   std::shared_ptr<socket_t> socket_;

include/cinatra/http_parser.hpp

@@ -1,48 +1,62 @@
 #pragma once
+#include <algorithm>
+#include <array>
 #include <cctype>
+#include <span>
 #include <string>
 #include <string_view>
-#include <vector>
 
+#include "cinatra_log_wrapper.hpp"
 #include "picohttpparser.h"
 
 using namespace std::string_view_literals;
 
+#ifndef CINATRA_MAX_HTTP_HEADER_FIELD_SIZE
+#define CINATRA_MAX_HTTP_HEADER_FIELD_SIZE 100
+#endif
+
 namespace cinatra {
 class http_parser {
  public:
   int parse_response(const char *data, size_t size, int last_len) {
     int minor_version;
 
-    num_headers_ = sizeof(headers_) / sizeof(headers_[0]);
+    num_headers_ = CINATRA_MAX_HTTP_HEADER_FIELD_SIZE;
     const char *msg;
     size_t msg_len;
-    header_len_ =
-        phr_parse_response(data, size, &minor_version, &status_, &msg, &msg_len,
-                           headers_, &num_headers_, last_len);
+    header_len_ = cinatra::detail::phr_parse_response(
+        data, size, &minor_version, &status_, &msg, &msg_len, headers_.data(),
+        &num_headers_, last_len);
     msg_ = {msg, msg_len};
-    auto header_value = this->get_header_value("content-length");
+    auto header_value = this->get_header_value("content-length"sv);
     if (header_value.empty()) {
       body_len_ = 0;
     }
     else {
       body_len_ = atoi(header_value.data());
     }
-
+    if (header_len_ < 0) [[unlikely]] {
+      CINATRA_LOG_WARNING << "parse http head failed";
+      if (size == CINATRA_MAX_HTTP_HEADER_FIELD_SIZE) {
+        CINATRA_LOG_ERROR << "the field of http head is out of max limit "
+                          << CINATRA_MAX_HTTP_HEADER_FIELD_SIZE
+                          << ", you can define macro "
+                             "CINATRA_MAX_HTTP_HEADER_FIELD_SIZE to expand it.";
+      }
+    }
     return header_len_;
   }
 
   std::string_view get_header_value(std::string_view key) const {
     for (size_t i = 0; i < num_headers_; i++) {
-      if (iequal(headers_[i].name, headers_[i].name_len, key.data()))
-        return std::string_view(headers_[i].value, headers_[i].value_len);
+      if (iequal(headers_[i].name, key))
+        return headers_[i].value;
     }
-
     return {};
   }
 
   bool is_chunked() const {
-    auto transfer_encoding = this->get_header_value("transfer-encoding");
+    auto transfer_encoding = this->get_header_value("transfer-encoding"sv);
     if (transfer_encoding == "chunked"sv) {
       return true;
     }
@@ -51,21 +65,21 @@ class http_parser {
   }
 
   bool is_ranges() const {
-    auto transfer_encoding = this->get_header_value("Accept-Ranges");
+    auto transfer_encoding = this->get_header_value("Accept-Ranges"sv);
     return !transfer_encoding.empty();
   }
 
   bool is_websocket() const {
-    auto upgrade = this->get_header_value("Upgrade");
+    auto upgrade = this->get_header_value("Upgrade"sv);
     return upgrade == "WebSocket"sv || upgrade == "websocket"sv;
   }
 
   bool keep_alive() const {
     if (is_websocket()) {
       return true;
     }
-    auto val = this->get_header_value("connection");
-    if (val.empty() || iequal(val.data(), val.length(), "keep-alive")) {
+    auto val = this->get_header_value("connection"sv);
+    if (val.empty() || iequal(val, "keep-alive"sv)) {
       return true;
     }
 
@@ -81,55 +95,29 @@ class http_parser {
   int total_len() const { return header_len_ + body_len_; }
 
   bool is_location() {
-    auto location = this->get_header_value("Location");
+    auto location = this->get_header_value("Location"sv);
     return !location.empty();
   }
 
   std::string_view msg() const { return msg_; }
 
-  std::pair<phr_header *, size_t> get_headers() {
-    return {headers_, num_headers_};
-  }
-
-  void set_headers(
-      const std::vector<std::pair<std::string, std::string>> &headers) {
-    num_headers_ = headers.size();
-    for (size_t i = 0; i < num_headers_; i++) {
-      headers_[i].name = headers[i].first.data();
-      headers_[i].name_len = headers[i].first.size();
-      headers_[i].value = headers[i].second.data();
-      headers_[i].value_len = headers[i].second.size();
-    }
+  std::span<http_header> get_headers() {
+    return {headers_.data(), num_headers_};
   }
 
  private:
-  std::string_view get_header_value(phr_header *headers, size_t num_headers,
-                                    std::string_view key) {
-    for (size_t i = 0; i < num_headers; i++) {
-      if (iequal(headers[i].name, headers[i].name_len, key.data()))
-        return std::string_view(headers[i].value, headers[i].value_len);
-    }
-
-    return {};
-  }
-
-  bool iequal(const char *s, size_t l, const char *t) const {
-    if (strlen(t) != l)
-      return false;
-
-    for (size_t i = 0; i < l; i++) {
-      if (std::tolower(s[i]) != std::tolower(t[i]))
-        return false;
-    }
-
-    return true;
+  bool iequal(std::string_view a, std::string_view b) const {
+    return std::equal(a.begin(), a.end(), b.begin(), b.end(),
+                      [](char a, char b) {
+                        return tolower(a) == tolower(b);
+                      });
   }
 
   int status_ = 0;
   std::string_view msg_;
   size_t num_headers_ = 0;
   int header_len_ = 0;
   int body_len_ = 0;
-  struct phr_header headers_[100];
+  std::array<http_header, CINATRA_MAX_HTTP_HEADER_FIELD_SIZE> headers_;
 };
 }  // namespace cinatra