hyperscan-driven snote parser

CMakeLists.txt

@@ -7,6 +7,10 @@ project(xbot
 )
 
 find_package(Boost REQUIRED)
+find_package(PkgConfig REQUIRED)
+
+pkg_check_modules(LIBHS libhs REQUIRED IMPORTED_TARGET)
+
 include(FetchContent)
 
 FetchContent_Declare(
@@ -39,4 +43,4 @@ add_executable(xbot
   ping_thread.cpp irc_parse_thread.cpp registration_thread.cpp
   self_thread.cpp command_thread.cpp)
 target_include_directories(xbot PRIVATE ${CMAKE_CURRENT_BINARY_DIR})
-target_link_libraries(xbot PRIVATE Boost::headers tomlplusplus_tomlplusplus eventpp)
+target_link_libraries(xbot PRIVATE Boost::headers tomlplusplus_tomlplusplus eventpp PkgConfig::LIBHS)

main.cpp

@@ -77,11 +77,16 @@ auto start(boost::asio::io_context & io, Settings const& settings) -> void
     ping_thread(*connection);
     auto const self_thread = SelfThread::start(*connection);
     registration_thread(*connection, settings.password, settings.username, settings.realname, settings.nickname);
-    snote_thread(*connection);
+    SnoteThread::start(*connection);
     CommandThread::start(*connection);
     echo_thread(*connection);
     unhandled_message_thread(*connection);
 
+    connection->add_listener<SnoteEvent>([](SnoteEvent& event)
+    {
+        std::cout << "Snote match " << static_cast<int>(event.tag) << std::endl;
+    });
+
     boost::asio::co_spawn(
         io,
         connection->connect(io, settings.host, settings.service),

snote_thread.cpp

@@ -4,12 +4,93 @@
 #include "connection.hpp"
 
 #include <cstring>
+#include <cstdlib>
+#include <stdexcept>
+#include <utility>
+#include <regex>
 
+namespace {
 
-auto snote_thread(Connection& connection) -> void
+struct SnotePattern
 {
+    SnotePattern(SnoteTag tag, char const* expression, unsigned flags = 0)
+    : tag{tag}
+    , expression{expression}
+    , flags{flags}
+    , regex{expression, std::regex_constants::ECMAScript | std::regex_constants::optimize}
+    {
+    }
+
+    SnoteTag tag;
+    char const* expression;
+    unsigned flags;
+    std::regex regex;
+};
+
+SnotePattern const patterns[] =
+{
+    {SnoteTag::ClientConnecting,
+     R"(^Client connecting: ([^ ]+) \(([^@ ]+)@([^) ]+)\) \[(.*)\] \{([^ ]*)\} <([^ ]*)> \[(.*)\]$)"},
+
+    {SnoteTag::ClientExiting,
+     R"(^Client exiting: ([^ ]+) \(([^@ ]+)@([^) ]+)\) \[(.*)\] \[(.*)\]$)"},
+};
+
+auto setup_database() -> std::unique_ptr<hs_database_t, SnoteThread::DbDeleter>
+{
+    std::vector<char const*> expressions;
+    std::vector<unsigned> flags;
+    std::vector<unsigned> ids;
+
+    expressions.reserve(std::size(patterns));
+    flags.reserve(std::size(patterns));
+    ids.reserve(std::size(patterns));
+
+    unsigned id = 0;
+
+    for (auto const& pattern : patterns)
+    {
+        expressions.push_back(pattern.expression);
+        flags.push_back(pattern.flags);
+        ids.push_back(id++);
+    }
+
+    hs_database_t* db;
+    hs_compile_error *error;
+
+    switch (hs_compile_multi(expressions.data(), flags.data(), ids.data(), expressions.size(), HS_MODE_BLOCK, nullptr, &db, &error))
+    {
+        case HS_COMPILER_ERROR:
+        {
+            std::string msg = error->message;
+            hs_free_compile_error(error);
+            throw std::runtime_error{std::move(msg)};
+        }
+        case HS_SUCCESS:
+            break;
+        default:
+            abort();
+    }
+    return std::unique_ptr<hs_database_t, SnoteThread::DbDeleter>{db};
+}
+
+} // namespace
+
+auto SnoteThread::start(Connection& connection) -> std::shared_ptr<SnoteThread>
+{
+    auto thread = std::make_shared<SnoteThread>();
+
+    thread->db_ = setup_database();
+
+    hs_scratch_t* scratch = nullptr;
+    if (HS_SUCCESS != hs_alloc_scratch(thread->db_.get(), &scratch))
+    {
+        abort();
+    }
+    thread->scratch_ = std::unique_ptr<hs_scratch_t, ScratchDeleter>{scratch};
+
     static char const* const prefix = "*** Notice -- ";
-    connection.add_listener<IrcMsgEvent>([&connection](IrcMsgEvent& event)
+    connection.add_listener<IrcMsgEvent>([&connection, thread](IrcMsgEvent& event)
     {
         auto& args = event.irc.args;
         if (IrcCommand::NOTICE == event.command
@@ -17,7 +98,41 @@ auto snote_thread(Connection& connection) -> void
          && args[1].starts_with(prefix))
         {
             event.handled_ = true;
-            connection.make_event<SnoteEvent>(args[1].substr(strlen(prefix)));
+            auto message = args[1].substr(strlen(prefix));
+            unsigned int match_id = -1;
+            auto const scan_result = hs_scan(thread->db_.get(), message.data(), message.size(), 0, thread->scratch_.get(),
+                [](unsigned int id, unsigned long long from, unsigned long long to, unsigned int flags, void *context) -> int
+                {
+                    int* const match_id = static_cast<int*>(context);
+                    *match_id = id;
+                    return 1; // stop scanning
+                }
+            , &match_id);
+
+            if (scan_result != HS_SUCCESS && scan_result != HS_SCAN_TERMINATED)
+            {
+                abort();
+            }
+
+            if (match_id != -1)
+            {
+                auto& pattern = patterns[match_id];
+                std::match_results<std::string_view::const_iterator> results;                
+                if (not std::regex_match(message.begin(), message.end(), results, pattern.regex))
+                {
+                    // something went wrong - hyperscan disagrees with std::regex
+                    abort();
+                }
+
+                std::vector<std::string_view> parts;
+                for (auto const sub : results)
+                {
+                    parts.push_back(std::string_view{sub.first, sub.second});
+                }
+                connection.make_event<SnoteEvent>(pattern.tag, std::move(parts));
+            }
         }
     });
+
+    return thread;
 }

snote_thread.hpp

@@ -2,13 +2,55 @@
 
 #include "thread.hpp"
 
+#include <hs.h>
+
+#include <memory>
+
 class Connection;
 
-// WIP: Use much finer granularity
-struct SnoteEvent : Event
+enum class SnoteTag
 {
-    SnoteEvent(std::string_view raw) : raw{raw} {}
-    std::string_view raw;
+    ClientConnecting,
+    ClientExiting,
 };
 
-auto snote_thread(Connection& connection) -> void;
+struct SnoteEvent : Event
+{
+    SnoteEvent(SnoteTag tag, std::vector<std::string_view> parts)
+    : tag{tag}
+    , parts{std::move(parts)}
+    {}
+
+    SnoteTag tag;
+    std::vector<std::string_view> parts;
+};
+
+struct SnoteThread
+{
+    struct DbDeleter
+    {
+        auto operator()(hs_database_t * db) const -> void
+        {
+            if (HS_SUCCESS != hs_free_database(db))
+            {
+                abort();
+            }
+        }
+    };
+
+    struct ScratchDeleter
+    {
+        auto operator()(hs_scratch_t * scratch) const -> void
+        {
+            if (HS_SUCCESS != hs_free_scratch(scratch))
+            {
+                abort();
+            }
+        }
+    };
+
+    std::unique_ptr<hs_database_t, DbDeleter> db_;
+    std::unique_ptr<hs_scratch_t, ScratchDeleter> scratch_;
+
+    static auto start(Connection& connection) -> std::shared_ptr<SnoteThread>;
+};