Use Boost.Beast to parse HTTP request (#6294)

This commit is contained in:
Siarhei Fedartsou
2022-08-02 22:12:05 +02:00
committed by GitHub
parent 00816722dd
commit e7185b4bcb
7 changed files with 73 additions and 440 deletions
+64 -30
View File
@@ -1,12 +1,10 @@
#include "server/connection.hpp"
#include "server/request_handler.hpp"
#include "server/request_parser.hpp"
#include <boost/algorithm/string/predicate.hpp>
#include <boost/bind.hpp>
#include <boost/iostreams/filter/gzip.hpp>
#include <boost/iostreams/filtering_stream.hpp>
#include <vector>
namespace osrm
@@ -16,12 +14,32 @@ namespace server
Connection::Connection(boost::asio::io_context &io_context, RequestHandler &handler)
: strand(boost::asio::make_strand(io_context)), TCP_socket(strand), timer(strand),
request_handler(handler)
request_handler(handler), http_request_parser(std::make_optional<RequestParser>())
{
}
boost::asio::ip::tcp::socket &Connection::socket() { return TCP_socket; }
namespace
{
http::compression_type select_compression(const boost::beast::http::fields &fields)
{
const auto header_value = fields[boost::beast::http::field::accept_encoding];
/* giving gzip precedence over deflate */
if (boost::icontains(header_value, "deflate"))
{
return http::deflate_rfc1951;
}
if (boost::icontains(header_value, "gzip"))
{
return http::gzip_rfc1952;
}
return http::no_compression;
}
} // namespace
/// Start the first asynchronous operation for the connection.
void Connection::start()
{
@@ -60,20 +78,45 @@ void Connection::handle_read(const boost::system::error_code &error, std::size_t
timer.expires_from_now(boost::posix_time::seconds(0));
}
boost::beast::error_code ec;
http_request_parser->put(boost::asio::buffer(incoming_data_buffer, bytes_transferred), ec);
// no error detected, let's parse the request
http::compression_type compression_type(http::no_compression);
RequestParser::RequestStatus result;
std::tie(result, compression_type) =
request_parser.parse(current_request,
incoming_data_buffer.data(),
incoming_data_buffer.data() + bytes_transferred);
// the request has been parsed
if (result == RequestParser::RequestStatus::valid)
if (ec)
{
if (ec == boost::beast::http::error::need_more)
{
// we don't have a result yet, so continue reading
TCP_socket.async_read_some(boost::asio::buffer(incoming_data_buffer),
boost::bind(&Connection::handle_read,
this->shared_from_this(),
boost::asio::placeholders::error,
boost::asio::placeholders::bytes_transferred));
}
else
{
// request is not parseable
current_reply = http::reply::stock_reply(http::reply::bad_request);
boost::asio::async_write(TCP_socket,
current_reply.to_buffers(),
boost::bind(&Connection::handle_write,
this->shared_from_this(),
boost::asio::placeholders::error));
}
}
else
{
// the request has been parsed
const auto &message = http_request_parser->get();
compression_type = select_compression(message);
fill_request(message, current_request);
boost::system::error_code ec;
current_request.endpoint = TCP_socket.remote_endpoint(ec).address();
if (ec)
{
util::Log(logDEBUG) << "Socket remote endpoint error: " << ec.message();
@@ -127,25 +170,6 @@ void Connection::handle_read(const boost::system::error_code &error, std::size_t
this->shared_from_this(),
boost::asio::placeholders::error));
}
else if (result == RequestParser::RequestStatus::invalid)
{ // request is not parseable
current_reply = http::reply::stock_reply(http::reply::bad_request);
boost::asio::async_write(TCP_socket,
current_reply.to_buffers(),
boost::bind(&Connection::handle_write,
this->shared_from_this(),
boost::asio::placeholders::error));
}
else
{
// we don't have a result yet, so continue reading
TCP_socket.async_read_some(boost::asio::buffer(incoming_data_buffer),
boost::bind(&Connection::handle_read,
this->shared_from_this(),
boost::asio::placeholders::error,
boost::asio::placeholders::bytes_transferred));
}
}
/// Handle completion of a write operation.
@@ -158,7 +182,7 @@ void Connection::handle_write(const boost::system::error_code &error)
--processed_requests;
current_request = http::request();
current_reply = http::reply();
request_parser = RequestParser();
http_request_parser.emplace();
incoming_data_buffer = boost::array<char, 8192>();
output_buffer.clear();
this->start();
@@ -220,5 +244,15 @@ std::vector<char> Connection::compress_buffers(const std::vector<char> &uncompre
return compressed_data;
}
void Connection::fill_request(const RequestParser::value_type &http_message,
http::request &current_request)
{
current_request.uri = http_message.target().to_string();
current_request.agent = http_message[boost::beast::http::field::user_agent].to_string();
current_request.referrer = http_message[boost::beast::http::field::referer].to_string();
current_request.connection = http_message[boost::beast::http::field::connection].to_string();
}
} // namespace server
} // namespace osrm
-302
View File
@@ -1,302 +0,0 @@
#include "server/request_parser.hpp"
#include "server/http/compression_type.hpp"
#include "server/http/header.hpp"
#include "server/http/request.hpp"
#include <boost/algorithm/string/predicate.hpp>
#include <string>
namespace osrm
{
namespace server
{
RequestParser::RequestParser()
: state(internal_state::method_start), current_header({"", ""}),
selected_compression(http::no_compression)
{
}
std::tuple<RequestParser::RequestStatus, http::compression_type>
RequestParser::parse(http::request &current_request, char *begin, char *end)
{
while (begin != end)
{
RequestStatus result = consume(current_request, *begin++);
if (result != RequestStatus::indeterminate)
{
return std::make_tuple(result, selected_compression);
}
}
RequestStatus result = RequestStatus::indeterminate;
return std::make_tuple(result, selected_compression);
}
RequestParser::RequestStatus RequestParser::consume(http::request &current_request,
const char input)
{
switch (state)
{
case internal_state::method_start:
if (!is_char(input) || is_CTL(input) || is_special(input))
{
return RequestStatus::invalid;
}
state = internal_state::method;
return RequestStatus::indeterminate;
case internal_state::method:
if (input == ' ')
{
state = internal_state::uri;
return RequestStatus::indeterminate;
}
if (!is_char(input) || is_CTL(input) || is_special(input))
{
return RequestStatus::invalid;
}
return RequestStatus::indeterminate;
case internal_state::uri_start:
if (is_CTL(input))
{
return RequestStatus::invalid;
}
state = internal_state::uri;
current_request.uri.push_back(input);
return RequestStatus::indeterminate;
case internal_state::uri:
if (input == ' ')
{
state = internal_state::http_version_h;
return RequestStatus::indeterminate;
}
if (is_CTL(input))
{
return RequestStatus::invalid;
}
current_request.uri.push_back(input);
return RequestStatus::indeterminate;
case internal_state::http_version_h:
if (input == 'H')
{
state = internal_state::http_version_t_1;
return RequestStatus::indeterminate;
}
return RequestStatus::invalid;
case internal_state::http_version_t_1:
if (input == 'T')
{
state = internal_state::http_version_t_2;
return RequestStatus::indeterminate;
}
return RequestStatus::invalid;
case internal_state::http_version_t_2:
if (input == 'T')
{
state = internal_state::http_version_p;
return RequestStatus::indeterminate;
}
return RequestStatus::invalid;
case internal_state::http_version_p:
if (input == 'P')
{
state = internal_state::http_version_slash;
return RequestStatus::indeterminate;
}
return RequestStatus::invalid;
case internal_state::http_version_slash:
if (input == '/')
{
state = internal_state::http_version_major_start;
return RequestStatus::indeterminate;
}
return RequestStatus::invalid;
case internal_state::http_version_major_start:
if (is_digit(input))
{
state = internal_state::http_version_major;
return RequestStatus::indeterminate;
}
return RequestStatus::invalid;
case internal_state::http_version_major:
if (input == '.')
{
state = internal_state::http_version_minor_start;
return RequestStatus::indeterminate;
}
if (is_digit(input))
{
return RequestStatus::indeterminate;
}
return RequestStatus::invalid;
case internal_state::http_version_minor_start:
if (is_digit(input))
{
state = internal_state::http_version_minor;
return RequestStatus::indeterminate;
}
return RequestStatus::invalid;
case internal_state::http_version_minor:
if (input == '\r')
{
state = internal_state::expecting_newline_1;
return RequestStatus::indeterminate;
}
if (is_digit(input))
{
return RequestStatus::indeterminate;
}
return RequestStatus::invalid;
case internal_state::expecting_newline_1:
if (input == '\n')
{
state = internal_state::header_line_start;
return RequestStatus::indeterminate;
}
return RequestStatus::invalid;
case internal_state::header_line_start:
if (boost::iequals(current_header.name, "Accept-Encoding"))
{
/* giving gzip precedence over deflate */
if (boost::icontains(current_header.value, "deflate"))
{
selected_compression = http::deflate_rfc1951;
}
if (boost::icontains(current_header.value, "gzip"))
{
selected_compression = http::gzip_rfc1952;
}
}
if (boost::iequals(current_header.name, "Referer"))
{
current_request.referrer = current_header.value;
}
if (boost::iequals(current_header.name, "User-Agent"))
{
current_request.agent = current_header.value;
}
if (boost::iequals(current_header.name, "Connection"))
{
current_request.connection = current_header.value;
}
if (input == '\r')
{
state = internal_state::expecting_newline_3;
return RequestStatus::indeterminate;
}
if (!is_char(input) || is_CTL(input) || is_special(input))
{
return RequestStatus::invalid;
}
state = internal_state::header_name;
current_header.clear();
current_header.name.push_back(input);
return RequestStatus::indeterminate;
case internal_state::header_lws:
if (input == '\r')
{
state = internal_state::expecting_newline_2;
return RequestStatus::indeterminate;
}
if (input == ' ' || input == '\t')
{
return RequestStatus::indeterminate;
}
if (is_CTL(input))
{
return RequestStatus::invalid;
}
state = internal_state::header_value;
return RequestStatus::indeterminate;
case internal_state::header_name:
if (input == ':')
{
state = internal_state::header_value;
return RequestStatus::indeterminate;
}
if (!is_char(input) || is_CTL(input) || is_special(input))
{
return RequestStatus::invalid;
}
current_header.name.push_back(input);
return RequestStatus::indeterminate;
case internal_state::header_value:
if (input == ' ')
{
state = internal_state::header_value;
return RequestStatus::indeterminate;
}
if (input == '\r')
{
state = internal_state::expecting_newline_2;
return RequestStatus::indeterminate;
}
if (is_CTL(input))
{
return RequestStatus::invalid;
}
current_header.value.push_back(input);
return RequestStatus::indeterminate;
case internal_state::expecting_newline_2:
if (input == '\n')
{
state = internal_state::header_line_start;
return RequestStatus::indeterminate;
}
return RequestStatus::invalid;
default: // expecting_newline_3
return input == '\n' ? RequestStatus::valid : RequestStatus::invalid;
}
}
bool RequestParser::is_char(const int character) const
{
return character >= 0 && character <= 127;
}
bool RequestParser::is_CTL(const int character) const
{
return (character >= 0 && character <= 31) || (character == 127);
}
bool RequestParser::is_special(const int character) const
{
switch (character)
{
case '(':
case ')':
case '<':
case '>':
case '@':
case ',':
case ';':
case ':':
case '\\':
case '"':
case '/':
case '[':
case ']':
case '?':
case '=':
case '{':
case '}':
case ' ':
case '\t':
return true;
default:
return false;
}
}
bool RequestParser::is_digit(const int character) const
{
return character >= '0' && character <= '9';
}
} // namespace server
} // namespace osrm