Optimize field lookups

CHANGELOG.md

@@ -4,6 +4,7 @@ Version 68:
 * Small speed up in fields comparisons
 * Adjust buffer size in fast server
 * Use string_ref in older Boost versions
+* Optimize field lookups
 
 API Changes:
 

include/beast/core/string.hpp

@@ -67,9 +67,24 @@ iequals(
         return false;
     auto p1 = lhs.data();
     auto p2 = rhs.data();
+    char a, b;
     while(n--)
-        if(ascii_tolower(*p1) != ascii_tolower(*p2))
+    {
+        a = *p1++;
+        b = *p2++;
+        if(a != b)
+            goto slow;
+    }
+    return true;
+
+    while(n--)
+    {
+    slow:
+        if(ascii_tolower(a) != ascii_tolower(b))
             return false;
+        a = *p1++;
+        b = *p2++;
+    }
     return true;
 }
 
@@ -92,7 +107,7 @@ iequals(
     return detail::iequals(lhs, rhs);
 }
 
-/** A strictly less predicate for strings, using a case-insensitive comparison.
+/** A case-insensitive less predicate for strings.
 
     The case-comparison operation is defined only for low-ASCII characters.
 */
@@ -100,8 +115,8 @@ struct iless
 {
     bool
     operator()(
-        beast::string_view const& lhs,
+        string_view const& lhs,
-        beast::string_view const& rhs) const
+        string_view const& rhs) const
     {
         using std::begin;
         using std::end;
@@ -115,7 +130,7 @@ struct iless
     }
 };
 
-/** A predicate for string equality, using a case-insensitive comparison.
+/** A case-insensitive equality predicate for strings.
 
     The case-comparison operation is defined only for low-ASCII characters.
 */
@@ -123,8 +138,8 @@ struct iequal
 {
     bool
     operator()(
-        beast::string_view const& lhs,
+        string_view const& lhs,
-        beast::string_view const& rhs) const
+        string_view const& rhs) const
     {
         return iequals(lhs, rhs);
     }

include/beast/http/impl/field.ipp

@@ -8,8 +8,11 @@
 #ifndef BEAST_HTTP_IMPL_FIELD_IPP
 #define BEAST_HTTP_IMPL_FIELD_IPP
 
+#include <beast/core/string.hpp>
 #include <algorithm>
 #include <array>
+#include <unordered_map>
+#include <vector>
 #include <boost/assert.hpp>
 
 namespace beast {
@@ -17,24 +20,70 @@ namespace http {
 
 namespace detail {
 
-class field_strings
+struct field_table
 {
     using array_type =
         std::array<string_view, 352>;
 
-    array_type v_;
+    struct hash
+    {
+        std::size_t
+        operator()(string_view const& s) const
+        {
+            auto const n = s.size();
+            return
+                beast::detail::ascii_tolower(s[0]) *
+                beast::detail::ascii_tolower(s[n/2]) ^
+                beast::detail::ascii_tolower(s[n-1]);   // hist[] = 331, 10, max_load_factor = 0.15f
+        }
+    };
 
-public:
+    struct iequal
-    using const_iterator =
+    {
-        array_type::const_iterator; 
+        // assumes inputs have equal length
+        bool
+        operator()(
+            string_view const& lhs,
+            string_view const& rhs) const
+        {
+            auto p1 = lhs.data();
+            auto p2 = rhs.data();
+            auto pend = lhs.end();
+            char a, b;
+            while(p1 < pend)
+            {
+                a = *p1++;
+                b = *p2++;
+                if(a != b)
+                    goto slow;
+            }
+            return true;
 
+            while(p1 < pend)
+            {
+            slow:
+                if( beast::detail::ascii_tolower(a) !=
+                    beast::detail::ascii_tolower(b))
+                    return false;
+                a = *p1++;
+                b = *p2++;
+            }
+            return true;
+        }
+    };
+
+    using map_type = std::unordered_map<
+        string_view, field, hash, iequal>;
+
+    array_type by_name_;
+    std::vector<map_type> by_size_;
 /*
     From:
     
     https://www.iana.org/assignments/message-headers/message-headers.xhtml
 */
-    field_strings()
+    field_table()
-        : v_({{
+        : by_name_({{
             "<unknown-field>",
             "A-IM",
             "Accept",
@@ -389,61 +438,103 @@ public:
             "Xref"
         }})
     {
+        // find the longest field length
+        std::size_t high = 0;
+        for(auto const& s : by_name_)
+            if(high < s.size())
+                high = s.size();
+        // build by_size map
+        // skip field::unknown
+        by_size_.resize(high + 1);
+        for(auto& map : by_size_)
+            map.max_load_factor(.15f);
+        for(std::size_t i = 1;
+            i < by_name_.size(); ++i)
+        {
+            auto const& s = by_name_[i];
+            by_size_[s.size()].emplace(
+                s, static_cast<field>(i));
+        }
+
+#if 0
+        // This snippet calculates the performance
+        // of the hash function and map settings
+        {
+            std::vector<std::size_t> hist;
+            for(auto const& map : by_size_)
+            {
+                for(std::size_t i = 0; i < map.bucket_count(); ++i)
+                {
+                    auto const n = map.bucket_size(i);
+                    if(n > 0)
+                    {
+                        if(hist.size() < n)
+                            hist.resize(n);
+                        ++hist[n-1];
+                    }
+                }
+            }
+        }
+#endif
     }
 
+    field
+    string_to_field(string_view s) const
+    {
+        if(s.size() >= by_size_.size())
+            return field::unknown;
+        auto const& map = by_size_[s.size()];
+        if(map.empty())
+            return field::unknown;
+        auto it = map.find(s);
+        if(it == map.end())
+            return field::unknown;
+        return it->second;
+    }
+
+    //
+    // Deprecated
+    //
+
+    using const_iterator =
+    array_type::const_iterator; 
+
     std::size_t
     size() const
     {
-        return v_.size();
+        return by_name_.size();
     }
 
     const_iterator
     begin() const
     {
-        return v_.begin();
+        return by_name_.begin();
     }
 
     const_iterator
     end() const
     {
-        return v_.end();
+        return by_name_.end();
     }
 };
 
 inline
-field_strings const&
+field_table const&
-get_field_strings()
+get_field_table()
 {
-    static field_strings const fs;
+    static field_table const tab;
-    return fs;
+    return tab;
 }
 
 template<class = void>
 string_view
 to_string(field f)
 {
-    auto const& v = get_field_strings();
+    auto const& v = get_field_table();
     BOOST_ASSERT(static_cast<unsigned>(f) < v.size());
     return v.begin()[static_cast<unsigned>(f)];
 }
 
-template<class = void>
-field
-string_to_field(string_view s)
-{
-    auto const& v = get_field_strings();
-    auto const it = std::lower_bound(
-        v.begin(), v.end(), s,
-            beast::iless{});
-    if(it == v.end())
-        return field::unknown;
-    if(! iequals(s, *it))
-        return field::unknown;
-    BOOST_ASSERT(iequals(s, to_string(
-        static_cast<field>(it - v.begin()))));
-    return static_cast<field>(it - v.begin());
-}
-
 } // detail
 
 inline
@@ -457,7 +548,7 @@ inline
 field
 string_to_field(string_view s)
 {
-    return detail::string_to_field(s);
+    return detail::get_field_table().string_to_field(s);
 }
 
 } // http

test/benchmarks/nodejs_parser.hpp

@@ -211,6 +211,9 @@ public:
     void
     write_eof(error_code& ec);
 
+    void
+    check_header();
+
 private:
     Derived&
     impl()
@@ -218,231 +221,6 @@ private:
         return *static_cast<Derived*>(this);
     }
 
-    template<class C>
-    class has_on_start_t
-    {
-        template<class T, class R =
-            decltype(std::declval<T>().on_start(), std::true_type{})>
-        static R check(int);
-        template<class>
-        static std::false_type check(...);
-        using type = decltype(check<C>(0));
-    public:
-        static bool const value = type::value;
-    };
-    template<class C>
-    using has_on_start =
-        std::integral_constant<bool, has_on_start_t<C>::value>;
-
-    void
-    call_on_start(std::true_type)
-    {
-        impl().on_start();
-    }
-
-    void
-    call_on_start(std::false_type)
-    {
-    }
-
-    template<class C>
-    class has_on_field_t
-    {
-        template<class T, class R =
-            decltype(std::declval<T>().on_field(
-                std::declval<std::string const&>(),
-                    std::declval<std::string const&>()),
-                        std::true_type{})>
-        static R check(int);
-        template<class>
-        static std::false_type check(...);
-        using type = decltype(check<C>(0));
-    public:
-        static bool const value = type::value;
-    };
-    template<class C>
-    using has_on_field =
-        std::integral_constant<bool, has_on_field_t<C>::value>;
-
-    void
-    call_on_field(std::string const& field,
-        std::string const& value, std::true_type)
-    {
-        impl().on_field(field, value);
-    }
-
-    void
-    call_on_field(std::string const&, std::string const&,
-        std::false_type)
-    {
-    }
-
-    template<class C>
-    class has_on_headers_complete_t
-    {
-        template<class T, class R =
-            decltype(std::declval<T>().on_headers_complete(
-                std::declval<error_code&>()), std::true_type{})>
-        static R check(int);
-        template<class>
-        static std::false_type check(...);
-        using type = decltype(check<C>(0));
-    public:
-        static bool const value = type::value;
-    };
-    template<class C>
-    using has_on_headers_complete =
-        std::integral_constant<bool, has_on_headers_complete_t<C>::value>;
-
-    void
-    call_on_headers_complete(error_code& ec, std::true_type)
-    {
-        impl().on_headers_complete(ec);
-    }
-
-    void
-    call_on_headers_complete(error_code&, std::false_type)
-    {
-    }
-
-    template<class C>
-    class has_on_request_t
-    {
-        template<class T, class R =
-            decltype(std::declval<T>().on_request(
-                std::declval<unsigned>(), std::declval<std::string>(),
-                    std::declval<int>(), std::declval<int>(),
-                        std::declval<bool>(), std::declval<bool>()),
-                            std::true_type{})>
-        static R check(int);
-        template<class>
-        static std::false_type check(...);
-        using type = decltype(check<C>(0));
-    public:
-        static bool const value = type::value;
-    };
-    template<class C>
-    using has_on_request =
-        std::integral_constant<bool, has_on_request_t<C>::value>;
-
-    void
-    call_on_request(unsigned method, std::string url,
-        int major, int minor, bool keep_alive, bool upgrade,
-            std::true_type)
-    {
-        impl().on_request(
-            method, url, major, minor, keep_alive, upgrade);
-    }
-
-    void
-    call_on_request(unsigned, std::string, int, int, bool, bool,
-        std::false_type)
-    {
-    }
-
-    template<class C>
-    class has_on_response_t
-    {
-        template<class T, class R =
-            decltype(std::declval<T>().on_response(
-                std::declval<int>(), std::declval<std::string>,
-                    std::declval<int>(), std::declval<int>(),
-                        std::declval<bool>(), std::declval<bool>()),
-                            std::true_type{})>
-        static R check(int);
-        template<class>
-        static std::false_type check(...);
-#if 0
-        using type = decltype(check<C>(0));
-#else
-        // VFALCO Trait seems broken for http::parser
-        using type = std::true_type;
-#endif
-    public:
-        static bool const value = type::value;
-    };
-    template<class C>
-    using has_on_response =
-        std::integral_constant<bool, has_on_response_t<C>::value>;
-
-    bool
-    call_on_response(int status, std::string text,
-        int major, int minor, bool keep_alive, bool upgrade,
-            std::true_type)
-    {
-        return impl().on_response(
-            status, text, major, minor, keep_alive, upgrade);
-    }
-
-    bool
-    call_on_response(int, std::string, int, int, bool, bool,
-        std::false_type)
-    {
-        // VFALCO Certainly incorrect
-        return true;
-    }
-
-    template<class C>
-    class has_on_body_t
-    {
-        template<class T, class R =
-            decltype(std::declval<T>().on_body(
-                std::declval<void const*>(), std::declval<std::size_t>(),
-                    std::declval<error_code&>()), std::true_type{})>
-        static R check(int);
-        template<class>
-        static std::false_type check(...);
-        using type = decltype(check<C>(0));
-    public:
-        static bool const value = type::value;
-    };
-    template<class C>
-    using has_on_body =
-        std::integral_constant<bool, has_on_body_t<C>::value>;
-
-    void
-    call_on_body(void const* data, std::size_t bytes,
-        error_code& ec, std::true_type)
-    {
-        impl().on_body(data, bytes, ec);
-    }
-
-    void
-    call_on_body(void const*, std::size_t,
-        error_code&, std::false_type)
-    {
-    }
-
-    template<class C>
-    class has_on_complete_t
-    {
-        template<class T, class R =
-            decltype(std::declval<T>().on_complete(), std::true_type{})>
-        static R check(int);
-        template<class>
-        static std::false_type check(...);
-        using type = decltype(check<C>(0));
-    public:
-        static bool const value = type::value;
-    };
-    template<class C>
-    using has_on_complete =
-        std::integral_constant<bool, has_on_complete_t<C>::value>;
-
-    void
-    call_on_complete(std::true_type)
-    {
-        impl().on_complete();
-    }
-
-    void
-    call_on_complete(std::false_type)
-    {
-    }
-
-    void
-    check_header();
-
     static int cb_message_start(http_parser*);
     static int cb_url(http_parser*, char const*, std::size_t);
     static int cb_status(http_parser*, char const*, std::size_t);
@@ -526,7 +304,8 @@ nodejs_basic_parser(nodejs_basic_parser&& other)
 
 template<class Derived>
 auto
-nodejs_basic_parser<Derived>::operator=(nodejs_basic_parser&& other) ->
+nodejs_basic_parser<Derived>::
+operator=(nodejs_basic_parser&& other) ->
     nodejs_basic_parser&
 {
     state_ = other.state_;
@@ -569,7 +348,8 @@ operator=(nodejs_basic_parser const& other) ->
 }
 
 template<class Derived>
-nodejs_basic_parser<Derived>::nodejs_basic_parser(bool request) noexcept
+nodejs_basic_parser<Derived>::
+nodejs_basic_parser(bool request) noexcept
 {
     state_.data = this;
     http_parser_init(&state_, request
@@ -579,7 +359,8 @@ nodejs_basic_parser<Derived>::nodejs_basic_parser(bool request) noexcept
 
 template<class Derived>
 std::size_t
-nodejs_basic_parser<Derived>::write(void const* data,
+nodejs_basic_parser<Derived>::
+write(void const* data,
     std::size_t size, error_code& ec)
 {
     ec_ = &ec;
@@ -596,11 +377,11 @@ nodejs_basic_parser<Derived>::write(void const* data,
 
 template<class Derived>
 void
-nodejs_basic_parser<Derived>::write_eof(error_code& ec)
+nodejs_basic_parser<Derived>::
+write_eof(error_code& ec)
 {
     ec_ = &ec;
-    http_parser_execute(
+    http_parser_execute(&state_, hooks(), nullptr, 0);
-        &state_, hooks(), nullptr, 0);
     if(! ec)
         ec = detail::make_nodejs_error(
             static_cast<int>(state_.http_errno));
@@ -608,13 +389,12 @@ nodejs_basic_parser<Derived>::write_eof(error_code& ec)
 
 template<class Derived>
 void
-nodejs_basic_parser<Derived>::check_header()
+nodejs_basic_parser<Derived>::
+check_header()
 {
     if(! value_.empty())
     {
-        //detail::trim(value_);
+        impl().on_field(field_, value_);
-        call_on_field(field_, value_,
-            has_on_field<Derived>{});
         field_.clear();
         value_.clear();
     }
@@ -622,7 +402,8 @@ nodejs_basic_parser<Derived>::check_header()
 
 template<class Derived>
 int
-nodejs_basic_parser<Derived>::cb_message_start(http_parser* p)
+nodejs_basic_parser<Derived>::
+cb_message_start(http_parser* p)
 {
     auto& t = *reinterpret_cast<nodejs_basic_parser*>(p->data);
     t.complete_ = false;
@@ -630,13 +411,14 @@ nodejs_basic_parser<Derived>::cb_message_start(http_parser* p)
     t.status_.clear();
     t.field_.clear();
     t.value_.clear();
-    t.call_on_start(has_on_start<Derived>{});
+    t.impl().on_start();
     return 0;
 }
 
 template<class Derived>
 int
-nodejs_basic_parser<Derived>::cb_url(http_parser* p,
+nodejs_basic_parser<Derived>::
+cb_url(http_parser* p,
     char const* in, std::size_t bytes)
 {
     auto& t = *reinterpret_cast<nodejs_basic_parser*>(p->data);
@@ -646,7 +428,8 @@ nodejs_basic_parser<Derived>::cb_url(http_parser* p,
 
 template<class Derived>
 int
-nodejs_basic_parser<Derived>::cb_status(http_parser* p,
+nodejs_basic_parser<Derived>::
+cb_status(http_parser* p,
     char const* in, std::size_t bytes)
 {
     auto& t = *reinterpret_cast<nodejs_basic_parser*>(p->data);
@@ -656,7 +439,8 @@ nodejs_basic_parser<Derived>::cb_status(http_parser* p,
 
 template<class Derived>
 int
-nodejs_basic_parser<Derived>::cb_header_field(http_parser* p,
+nodejs_basic_parser<Derived>::
+cb_header_field(http_parser* p,
     char const* in, std::size_t bytes)
 {
     auto& t = *reinterpret_cast<nodejs_basic_parser*>(p->data);
@@ -667,7 +451,8 @@ nodejs_basic_parser<Derived>::cb_header_field(http_parser* p,
 
 template<class Derived>
 int
-nodejs_basic_parser<Derived>::cb_header_value(http_parser* p,
+nodejs_basic_parser<Derived>::
+cb_header_value(http_parser* p,
     char const* in, std::size_t bytes)
 {
     auto& t = *reinterpret_cast<nodejs_basic_parser*>(p->data);
@@ -677,62 +462,68 @@ nodejs_basic_parser<Derived>::cb_header_value(http_parser* p,
 
 template<class Derived>
 int
-nodejs_basic_parser<Derived>::cb_headers_complete(http_parser* p)
+nodejs_basic_parser<Derived>::
+cb_headers_complete(http_parser* p)
 {
     auto& t = *reinterpret_cast<nodejs_basic_parser*>(p->data);
     t.check_header();
-    t.call_on_headers_complete(*t.ec_,
+    t.impl().on_headers_complete(*t.ec_);
-        has_on_headers_complete<Derived>{});
     if(*t.ec_)
         return 1;
     bool const keep_alive =
         http_should_keep_alive(p) != 0;
     if(p->type == http_parser_type::HTTP_REQUEST)
     {
-        t.call_on_request(p->method, t.url_,
+        t.impl().on_request(p->method, t.url_,
             p->http_major, p->http_minor, keep_alive,
-                p->upgrade, has_on_request<Derived>{});
+                p->upgrade);
         return 0;
     }
-    return t.call_on_response(p->status_code, t.status_,
+    return t.impl().on_response(p->status_code, t.status_,
         p->http_major, p->http_minor, keep_alive,
-            p->upgrade, has_on_response<Derived>{}) ? 0 : 1;
+            p->upgrade) ? 0 : 1;
 }
 
 template<class Derived>
 int
-nodejs_basic_parser<Derived>::cb_body(http_parser* p,
+nodejs_basic_parser<Derived>::
+cb_body(http_parser* p,
     char const* in, std::size_t bytes)
 {
     auto& t = *reinterpret_cast<nodejs_basic_parser*>(p->data);
-    t.call_on_body(in, bytes, *t.ec_, has_on_body<Derived>{});
+    t.impl().on_body(in, bytes, *t.ec_);
     return *t.ec_ ? 1 : 0;
 }
 
 template<class Derived>
 int
-nodejs_basic_parser<Derived>::cb_message_complete(http_parser* p)
+nodejs_basic_parser<Derived>::
+cb_message_complete(http_parser* p)
 {
     auto& t = *reinterpret_cast<nodejs_basic_parser*>(p->data);
     t.complete_ = true;
-    t.call_on_complete(has_on_complete<Derived>{});
+    t.impl().on_complete();
     return 0;
 }
 
 template<class Derived>
 int
-nodejs_basic_parser<Derived>::cb_chunk_header(http_parser*)
+nodejs_basic_parser<Derived>::
+cb_chunk_header(http_parser*)
 {
     return 0;
 }
 
 template<class Derived>
 int
-nodejs_basic_parser<Derived>::cb_chunk_complete(http_parser*)
+nodejs_basic_parser<Derived>::
+cb_chunk_complete(http_parser*)
 {
     return 0;
 }
 
+//------------------------------------------------------------------------------
+
 /** A HTTP parser.
 
     The parser may only be used once.

test/benchmarks/parser.cpp

@@ -224,7 +224,7 @@ public:
     void
     testSpeed()
     {
-        static std::size_t constexpr Trials = 3;
+        static std::size_t constexpr Trials = 10;
         static std::size_t constexpr Repeat = 500;
 
         creq_ = build_corpus(N/2, std::true_type{});
@@ -240,6 +240,15 @@ public:
             ((Repeat * size_ + 512) / 1024) << "KB in " <<
                 (Repeat * (creq_.size() + cres_.size())) << " messages";
 
+#if 0
+        timedTest(Trials, "http::parser",
+            [&]
+            {
+                testParser2<request_parser<dynamic_body>>(Repeat, creq_);
+                testParser2<response_parser<dynamic_body>>(Repeat, cres_);
+            });
+#endif
+#if 1
         timedTest(Trials, "http::basic_parser",
             [&]
             {
@@ -250,6 +259,7 @@ public:
                     false, dynamic_body, fields>>(
                         Repeat, cres_);
             });
+#if 1
         timedTest(Trials, "nodejs_parser",
             [&]
             {
@@ -260,6 +270,8 @@ public:
                     false, dynamic_body, fields>>(
                         Repeat, cres_);
             });
+#endif
+#endif
         pass();
     }