-

bot-motherlib · Feb 10, 2025 · d77d86a · d77d86a
1 parent dd9cca1
commit d77d86a
Show file tree

Hide file tree

Showing 10 changed files with 289 additions and 24 deletions.
diff --git a/benchmarks/fuzzing.hpp b/benchmarks/fuzzing.hpp
@@ -212,7 +212,7 @@ struct Storage {
       file.exceptions(std::fstream::failbit | std::fstream::badbit);
       file.open(path);
       if (!file.is_open()) {
-        throw std::runtime_error("FUCK THIS SHIT");
+        throw std::runtime_error(fmt::format("can't open file on path : {}", path.generic_string()));
       }
       file << *json;
       file.flush();
@@ -243,7 +243,7 @@ struct Storage {
       file.exceptions(std::fstream::failbit | std::fstream::badbit);
       file.open(path);
       if (!file.is_open()) {
-        throw std::runtime_error("FUCK THIS SHIT");
+        throw std::runtime_error(fmt::format("can't open file on path : {}", path.generic_string()));
       }
       file << *json;
       file.flush();

diff --git a/benchmarks/rapid_parse_handler.hpp b/benchmarks/rapid_parse_handler.hpp
@@ -211,7 +211,7 @@ T parse_handler(std::string_view sv) {
   details::rapid_sv_stream ss(sv);
   rapidjson::ParseResult parse_result = reader.Parse(ss, handler);
   if (parse_result.IsError()) {
-    throw std::runtime_error(fmt::format("Fail parse"));
+    throw_json_parse_error();
   }
   return result;
 }

diff --git a/cmake/deps.cmake b/cmake/deps.cmake
@@ -17,7 +17,6 @@ CPMAddPackage(
   OPTIONS "BOOST_ENABLE_CMAKE ON"
 )
 unset(BOOST_INCLUDE_LIBRARIES)
-find_package(Boost 1.87 COMPONENTS system asio pfr json boost_container REQUIRED)
 
 CPMAddPackage("gh:fmtlib/fmt#11.0.2")
 

diff --git a/gtest/parsers/test_oneof.cpp b/gtest/parsers/test_oneof.cpp
@@ -281,4 +281,98 @@ JSON_PARSE_TEST(OneOfHiddenUserTypeMoreData, MessageOrigin) {
   EXPECT_EQ(parsed, expected);
 }
 
+TEST(GeneratorBoost, MissplaceDiscriminator) {
+  MessageOrigin expected{
+      .data =
+          MessageOriginChannel{
+              .date = tgbm::api::Integer{1630454400},
+              .sender_channel =
+                  tgbm::box<Channel>{
+                      Channel{
+                          .id = tgbm::api::Integer{13579},
+                          .name = "News Channel",
+                      },
+                  },
+          },
+  };
+
+  auto json = R"(
+    {
+      "date": 1630454400,
+      "sender_channel": {
+        "id": 13579,
+        "name": "News Channel"
+      },
+      "type": "channel"
+    }
+  )";
+
+  MessageOrigin parsed = GeneratorBoost::parse_json<MessageOrigin>(json);
+  EXPECT_EQ(parsed, expected);
+}
+
+TEST(GeneratorBoost, ArrayMissplaceDiscriminator) {
+  std::vector<MessageOrigin> expected{
+      MessageOrigin{
+          .data =
+              MessageOriginChannel{
+                  .date = tgbm::api::Integer{1630454400},
+                  .sender_channel =
+                      tgbm::box<Channel>{
+                          Channel{
+                              .id = tgbm::api::Integer{13579},
+                              .name = "News Channel",
+                          },
+                      },
+              },
+      },
+      MessageOrigin{
+          .data =
+              MessageOriginChat{
+                  .date = tgbm::api::Integer{1630454400},
+                  .sender_chat =
+                      tgbm::box<Chat>{
+                          Chat{
+                              .id = tgbm::api::Integer{67890},
+                              .title = "Group Chat",
+                          },
+                      },
+              },
+      },
+  };
+
+  auto json = R"(
+[
+   {
+      "date":1630454400,
+      "type":"channel",
+      "sender_channel":{
+         "id":13579,
+         "name":"News Channel"
+      }
+   },
+   {
+      "date":1630454400,
+      "sender_chat":{
+         "id":67890,
+         "title":"Group Chat"
+      },
+      "type":"chat"
+   }
+])";
+
+  auto parsed = GeneratorBoost::parse_json<std::vector<MessageOrigin>>(json);
+  EXPECT_EQ(parsed, expected);
+}
+
+JSON_PARSE_TEST(MissingField, MessageOrigin) {
+  MessageOrigin expected;
+  auto json = R"(
+{
+  "date": 1630454400
+}
+    )";
+  EXPECT_THROW(parse_json(json), tgbm::json::parse_error);
+}
+
 }  // namespace test_oneof
diff --git a/include/tgbm/jsons/boostjson_dom_traits.hpp b/include/tgbm/jsons/boostjson_dom_traits.hpp
@@ -3,6 +3,7 @@
 #include <boost/json.hpp>
 
 #include <tgbm/jsons/dom_traits.hpp>
+#include <tgbm/jsons/errors.hpp>
 #include <tgbm/utils/macro.hpp>
 
 namespace tgbm::json {
@@ -64,7 +65,7 @@ struct dom_traits_for<::boost::json::value> {
   }
 
   static void on_error() {
-    throw std::runtime_error("JSON Error");
+    throw_json_parse_error();
   }
 
   static const type* find_field(const type& json, std::string_view key) {

diff --git a/include/tgbm/jsons/parse_sax/common_api.hpp b/include/tgbm/jsons/parse_sax/common_api.hpp
@@ -7,6 +7,7 @@
 #include <tgbm/jsons/sax.hpp>
 #include <tgbm/utils/pfr_extension.hpp>
 #include <tgbm/utils/traits.hpp>
+#include <tgbm/logger.hpp>
 
 namespace tgbm::json {
 

diff --git a/include/tgbm/jsons/parse_sax/discriminated_api.hpp b/include/tgbm/jsons/parse_sax/discriminated_api.hpp
@@ -20,20 +20,57 @@ struct sax_parser<T> {
     return v.discriminate(key, emplacer);
   }
 
-  static sax_consumer_t parse(T& v, sax_token& tok, dd::with_stack_resource r) {
+  static sax_consumer_t bufferized_parse(T& out, sax_token& tok, dd::with_stack_resource r) {
+    using enum sax_token::kind_e;
+
+    assert(tok.got != key || tok.str_m != T::discriminator);
+
+    std::vector<sax_token_value> buf_toks;
+    buf_toks.reserve(10);
+    {
+      sax_token token;
+      token.got = object_begin;
+      buf_toks.emplace_back(token);
+    }
+
+    do {
+      buf_toks.emplace_back(tok);
+      co_yield {};
+    } while (tok.got != key || tok.str_m != T::discriminator);
+
+    co_yield {};
+    tok.expect(string);
+
+    auto gen_suboneof = get_generator_suboneof(tok.str_m, out, tok, r);
+    auto it = gen_suboneof.cur_iterator();
+    for (auto& buf_tok : buf_toks) {
+      assert(it != gen_suboneof.end());
+      tok = buf_tok.to_view();
+      ++it;
+    }
+
+    if (it != gen_suboneof.end()) {
+      co_yield {};
+      co_yield dd::elements_of(gen_suboneof);
+    }
+  }
+
+  static sax_consumer_t parse(T& out, sax_token& tok, dd::with_stack_resource r) {
     using enum sax_token::kind_e;
     tok.expect(object_begin);
     co_yield {};
     if (tok.got == object_end)
       co_return;
     tok.expect(key);
-    if (tok.got != key || tok.str_m != T::discriminator) [[unlikely]]
-      json::throw_json_parse_error();
+    if (tok.got != key || tok.str_m != T::discriminator) [[unlikely]] {
+      co_yield dd::elements_of(bufferized_parse(out, tok, r));
+      co_return;
+    }
     co_yield {};
     tok.expect(string);
     // change 'got' before generator creation (may be function returning generator)
     tok.got = object_begin;
-    co_yield dd::elements_of(get_generator_suboneof(tok.str_m, v, tok, r));
+    co_yield dd::elements_of(get_generator_suboneof(tok.str_m, out, tok, r));
   }
 };
 

diff --git a/include/tgbm/jsons/rapidjson_dom_traits.hpp b/include/tgbm/jsons/rapidjson_dom_traits.hpp
@@ -9,6 +9,7 @@
 #include <rapidjson/error/en.h>
 
 #include <tgbm/jsons/dom_traits.hpp>
+#include <tgbm/jsons/errors.hpp>
 
 namespace tgbm::json {
 
@@ -75,7 +76,7 @@ struct dom_traits_for<rapidjson::GenericValue<rapidjson::UTF8<>>> {
   }
 
   static void on_error() {
-    throw std::runtime_error("JSON Error");
+    throw_json_parse_error();
   }
 
   static const type* find_field(const type& json, std::string_view key) {