YT: Introduce global config option to convert all protobuf message fields names to snake case

commit_hash:31aeb756f05c80fc7b39c56e8e171dc16f018d13

8 files changed, 151 insertions, 15 deletions

diff --git a/yt/yt/core/yson/config.cpp b/yt/yt/core/yson/config.cpp
index 62632a5a39..eab523b4f8 100644
--- a/yt/yt/core/yson/config.cpp
+++ b/yt/yt/core/yson/config.cpp
@@ -10,6 +10,8 @@ void TProtobufInteropConfig::Register(TRegistrar registrar)
         .Default(EEnumYsonStorageType::String);
     registrar.Parameter("utf8_check", &TThis::Utf8Check)
         .Default(EUtf8Check::Disable);
+    registrar.Parameter("force_snake_case_names", &TThis::ForceSnakeCaseNames)
+        .Default(false);
 }
 
 TProtobufInteropConfigPtr TProtobufInteropConfig::ApplyDynamic(
diff --git a/yt/yt/core/yson/config.h b/yt/yt/core/yson/config.h
index 3c5817c5d2..7238e47605 100644
--- a/yt/yt/core/yson/config.h
+++ b/yt/yt/core/yson/config.h
@@ -21,6 +21,9 @@ public:
     EEnumYsonStorageType DefaultEnumYsonStorageType;
     // Check if string field contains actual UTF-8 string.
     EUtf8Check Utf8Check;
+    // If `true` convert all field names in camel case to snake case.
+    // Note: Applies to each protobuf message only on first seen, changes of flag after do not have effect.
+    bool ForceSnakeCaseNames;
 
     TProtobufInteropConfigPtr ApplyDynamic(const TProtobufInteropDynamicConfigPtr& dynamicConfig) const;
 
diff --git a/yt/yt/core/yson/protobuf_interop.cpp b/yt/yt/core/yson/protobuf_interop.cpp
index c23dcab2a3..eca135a620 100644
--- a/yt/yt/core/yson/protobuf_interop.cpp
+++ b/yt/yt/core/yson/protobuf_interop.cpp
@@ -148,11 +148,12 @@ TString ToUnderscoreCase(const TString& protobufName)
 
 TString DeriveYsonName(const TString& protobufName, const google::protobuf::FileDescriptor* fileDescriptor)
 {
-    if (fileDescriptor->options().GetExtension(NYT::NYson::NProto::derive_underscore_case_names)) {
+    if (fileDescriptor->options().GetExtension(NYT::NYson::NProto::derive_underscore_case_names)
+        || GetProtobufInteropConfig()->ForceSnakeCaseNames)
+    {
         return ToUnderscoreCase(protobufName);
-    } else {
-        return protobufName;
     }
+    return protobufName;
 }
 
 ////////////////////////////////////////////////////////////////////////////////
@@ -167,17 +168,17 @@ TProtobufInteropConfigSingleton* GlobalProtobufInteropConfig()
     return LeakySingleton<TProtobufInteropConfigSingleton>();
 }
 
-TProtobufInteropConfigPtr GetProtobufInteropConfig()
-{
-    return GlobalProtobufInteropConfig()->Config.Acquire();
-}
-
 ////////////////////////////////////////////////////////////////////////////////
 
 } // namespace
 
 ////////////////////////////////////////////////////////////////////////////////
 
+TProtobufInteropConfigPtr GetProtobufInteropConfig()
+{
+    return GlobalProtobufInteropConfig()->Config.Acquire();
+}
+
 void SetProtobufInteropConfig(TProtobufInteropConfigPtr config)
 {
     GlobalProtobufInteropConfig()->Config.Store(std::move(config));
diff --git a/yt/yt/core/yson/protobuf_interop.h b/yt/yt/core/yson/protobuf_interop.h
index bdc06244da..9f9d5be70f 100644
--- a/yt/yt/core/yson/protobuf_interop.h
+++ b/yt/yt/core/yson/protobuf_interop.h
@@ -263,6 +263,7 @@ TString YsonStringToProto(
 
 ////////////////////////////////////////////////////////////////////////////////
 
+TProtobufInteropConfigPtr GetProtobufInteropConfig();
 void SetProtobufInteropConfig(TProtobufInteropConfigPtr config);
 
 ////////////////////////////////////////////////////////////////////////////////
diff --git a/yt/yt/core/yson/unittests/proto/protobuf_yson_casing_ext_ut.proto b/yt/yt/core/yson/unittests/proto/protobuf_yson_casing_ext_ut.proto
index 98ce2c4062..1eb4eeb75b 100644
--- a/yt/yt/core/yson/unittests/proto/protobuf_yson_casing_ext_ut.proto
+++ b/yt/yt/core/yson/unittests/proto/protobuf_yson_casing_ext_ut.proto
@@ -5,3 +5,9 @@ message TExternalProtobuf
     optional int32 SomeField = 1;
     optional int32 AnotherField123 = 2;
 }
+
+message TOtherExternalProtobuf
+{
+    optional int32 SomeField = 1;
+    optional int32 AnotherField123 = 2;
+}
diff --git a/yt/yt/core/yson/unittests/proto/protobuf_yson_schema_ut.proto b/yt/yt/core/yson/unittests/proto/protobuf_yson_schema_ut.proto
index 0e5dbe293c..5ea8fbb320 100644
--- a/yt/yt/core/yson/unittests/proto/protobuf_yson_schema_ut.proto
+++ b/yt/yt/core/yson/unittests/proto/protobuf_yson_schema_ut.proto
@@ -35,3 +35,27 @@ message TTestSchemaMessage
     map<string, int32> string_to_int32_map = 19[(NYT.NYson.NProto.yson_map) = true];
     map<int32, string> int32_to_string_map = 20[(NYT.NYson.NProto.yson_map) = true];
 }
+
+message TCamelCaseStyleTestSchemaMessage1
+{
+    enum EEnum
+    {
+        VALUE_NONE = 0;
+        VALUE_FIRST = 1;
+    }
+
+    optional int32 SomeField = 1;
+    optional EEnum EnumField = 2;
+}
+
+message TCamelCaseStyleTestSchemaMessage2
+{
+    enum EEnum
+    {
+        VALUE_NONE = 0;
+        VALUE_FIRST = 1;
+    }
+
+    optional int32 SomeField = 1;
+    optional EEnum EnumField = 2;
+}
diff --git a/yt/yt/core/yson/unittests/protobuf_yson_schema_ut.cpp b/yt/yt/core/yson/unittests/protobuf_yson_schema_ut.cpp
index 0811eeba85..2076133892 100644
--- a/yt/yt/core/yson/unittests/protobuf_yson_schema_ut.cpp
+++ b/yt/yt/core/yson/unittests/protobuf_yson_schema_ut.cpp
@@ -2,21 +2,24 @@
 
 #include <yt/yt/core/yson/unittests/proto/protobuf_yson_schema_ut.pb.h>
 
+#include <yt/yt/core/yson/config.h>
 #include <yt/yt/core/yson/protobuf_interop.h>
 #include <yt/yt/core/yson/writer.h>
 
 #include <yt/yt/core/ytree/convert.h>
 #include <yt/yt/core/ytree/ypath_client.h>
 
+#include <util/generic/scope.h>
 #include <util/stream/str.h>
 
-
 namespace NYT::NYson {
 namespace {
 
+using namespace NYTree;
+
 ////////////////////////////////////////////////////////////////////////////////
 
-TEST(TProtobufYsonSchemaTest, GetMessageSchema)
+TEST(TProtobufYsonSchemaTest, WriteMessageSchema)
 {
     TStringStream outputStream;
     TYsonWriter ysonWriter(&outputStream, EYsonFormat::Text);
@@ -48,12 +51,65 @@ TEST(TProtobufYsonSchemaTest, GetMessageSchema)
         ];
     })";
 
-    auto expectedNode = NYTree::ConvertToNode(TYsonStringBuf(expected), NYTree::GetEphemeralNodeFactory());
-    auto actualNode = NYTree::ConvertToNode(TYsonStringBuf(outputStream.Str()), NYTree::GetEphemeralNodeFactory());
+    auto expectedNode = ConvertToNode(TYsonStringBuf(expected), GetEphemeralNodeFactory());
+    auto actualNode = ConvertToNode(TYsonStringBuf(outputStream.Str()), GetEphemeralNodeFactory());
+
+    EXPECT_TRUE(AreNodesEqual(actualNode, expectedNode))
+        << "Expected: " << ConvertToYsonString(expectedNode, EYsonFormat::Pretty).AsStringBuf() << "\n\n"
+        << "Actual: " << ConvertToYsonString(actualNode, EYsonFormat::Text).AsStringBuf() << "\n\n";
+}
+
+TEST(TProtobufYsonSchemaTest, CamelCaseNames)
+{
+    TStringStream outputStream;
+    TYsonWriter ysonWriter(&outputStream, EYsonFormat::Text);
+
+    WriteSchema(ReflectProtobufMessageType<NProto::TCamelCaseStyleTestSchemaMessage1>(), &ysonWriter);
+    TStringBuf expected = R"({
+        type_name="struct";
+        members=[
+            {name="SomeField";type="int32";};
+            {name="EnumField";type={type_name="enum";enum_name="EEnum";"values"=["VALUE_NONE";"VALUE_FIRST";];};};
+        ];
+    })";
+
+    auto expectedNode = ConvertToNode(TYsonStringBuf(expected), GetEphemeralNodeFactory());
+    auto actualNode = ConvertToNode(TYsonStringBuf(outputStream.Str()), GetEphemeralNodeFactory());
+
+    EXPECT_TRUE(AreNodesEqual(actualNode, expectedNode))
+        << "Expected: " << ConvertToYsonString(expectedNode, EYsonFormat::Pretty).AsStringBuf() << "\n\n"
+        << "Actual: " << ConvertToYsonString(actualNode, EYsonFormat::Pretty).AsStringBuf() << "\n\n";
+}
+
+TEST(TProtobufYsonSchemaTest, ForceSnakeCaseNames)
+{
+    auto oldConfig = GetProtobufInteropConfig();
+    Y_DEFER {
+        SetProtobufInteropConfig(oldConfig);
+    };
+
+    auto newConfig = CloneYsonStruct(oldConfig);
+    newConfig->ForceSnakeCaseNames = true;
+    SetProtobufInteropConfig(newConfig);
+
+    TStringStream outputStream;
+    TYsonWriter ysonWriter(&outputStream, EYsonFormat::Text);
+
+    WriteSchema(ReflectProtobufMessageType<NProto::TCamelCaseStyleTestSchemaMessage2>(), &ysonWriter);
+    TStringBuf expected = R"({
+        type_name="struct";
+        members=[
+            {name="some_field";type="int32";};
+            {name="enum_field";type={type_name="enum";enum_name="EEnum";"values"=["value_none";"value_first";];};};
+        ];
+    })";
+
+    auto expectedNode = ConvertToNode(TYsonStringBuf(expected), GetEphemeralNodeFactory());
+    auto actualNode = ConvertToNode(TYsonStringBuf(outputStream.Str()), GetEphemeralNodeFactory());
 
-    EXPECT_TRUE(NYTree::AreNodesEqual(actualNode, expectedNode))
-        << "Expected: " << ConvertToYsonString(expectedNode, EYsonFormat::Text, 4).AsStringBuf() << "\n\n"
-        << "Actual: " << ConvertToYsonString(actualNode, EYsonFormat::Text, 4).AsStringBuf() << "\n\n";
+    EXPECT_TRUE(AreNodesEqual(actualNode, expectedNode))
+        << "Expected: " << ConvertToYsonString(expectedNode, EYsonFormat::Pretty).AsStringBuf() << "\n\n"
+        << "Actual: " << ConvertToYsonString(actualNode, EYsonFormat::Pretty).AsStringBuf() << "\n\n";
 }
 
 ////////////////////////////////////////////////////////////////////////////////
diff --git a/yt/yt/core/yson/unittests/protobuf_yson_ut.cpp b/yt/yt/core/yson/unittests/protobuf_yson_ut.cpp
index 7373b8d78c..f369a0a0ff 100644
--- a/yt/yt/core/yson/unittests/protobuf_yson_ut.cpp
+++ b/yt/yt/core/yson/unittests/protobuf_yson_ut.cpp
@@ -22,6 +22,8 @@
 
 #include <google/protobuf/wire_format.h>
 
+#include <util/generic/scope.h>
+
 namespace NYT {
 
 ////////////////////////////////////////////////////////////////////////////////
@@ -2792,5 +2794,46 @@ TEST(TYsonToProtobufTest, Casing)
     EXPECT_EQ(message.anotherfield123(), 2);
 }
 
+TEST(TYsonToProtobufTest, ForceSnakeCaseNames)
+{
+    auto oldConfig = GetProtobufInteropConfig();
+    Y_DEFER {
+        SetProtobufInteropConfig(oldConfig);
+    };
+
+    auto newConfig = CloneYsonStruct(oldConfig);
+    newConfig->ForceSnakeCaseNames = true;
+    SetProtobufInteropConfig(newConfig);
+
+    auto ysonNode = BuildYsonNodeFluently()
+        .BeginMap()
+            .Item("some_field").Value(1)
+            .Item("another_field123").Value(2)
+        .EndMap();
+    auto ysonString = ConvertToYsonString(ysonNode);
+
+    NProto::TOtherExternalProtobuf message;
+    message.ParseFromStringOrThrow(NYson::YsonStringToProto(
+        ysonString,
+        NYson::ReflectProtobufMessageType<NProto::TOtherExternalProtobuf>(),
+        TProtobufWriterOptions()));
+
+    EXPECT_EQ(message.somefield(), 1);
+    EXPECT_EQ(message.anotherfield123(), 2);
+    {
+        TString newYsonString;
+        TStringOutput newYsonOutputStream(newYsonString);
+        TYsonWriter ysonWriter(&newYsonOutputStream, EYsonFormat::Pretty);
+        WriteProtobufMessage(&ysonWriter, message);
+
+        auto originalNode = NYTree::ConvertToNode(ysonString, NYTree::GetEphemeralNodeFactory());
+        auto newNode = NYTree::ConvertToNode(TYsonStringBuf(newYsonString), NYTree::GetEphemeralNodeFactory());
+
+        EXPECT_TRUE(NYTree::AreNodesEqual(originalNode, newNode))
+            << "Expected: " << ysonString.AsStringBuf() << "\n\n"
+            << "Actual: " << newYsonString << "\n\n";
+    }
+}
+
 } // namespace
 } // namespace NYT::NYson