YTORM-275: Fix TInstant deserialization from microsecond values

47845f7aede6898d43680bd7c6f1d48c05607e34

3 files changed, 65 insertions, 11 deletions

diff --git a/library/cpp/yt/misc/cast-inl.h b/library/cpp/yt/misc/cast-inl.h
index ceacda91bd..59be41ac1c 100644
--- a/library/cpp/yt/misc/cast-inl.h
+++ b/library/cpp/yt/misc/cast-inl.h
@@ -9,7 +9,6 @@
 #include <util/string/cast.h>
 #include <util/string/printf.h>
 
-#include <concepts>
 #include <type_traits>
 
 namespace NYT {
@@ -78,6 +77,8 @@ inline TString FormatInvalidCastValue(char8_t value)
 
 } // namespace NDetail
 
+////////////////////////////////////////////////////////////////////////////////
+
 template <class T, class S>
 bool TryIntegralCast(S value, T* result)
 {
@@ -93,8 +94,12 @@ T CheckedIntegralCast(S value)
 {
     T result;
     if (!TryIntegralCast<T>(value, &result)) {
-        throw TSimpleException(Sprintf("Argument value %s is out of expected range",
-            NYT::NDetail::FormatInvalidCastValue(value).c_str()));
+        throw TSimpleException(Sprintf("Error casting %s value \"%s\" to %s: value is out of expected range [%s; %s]",
+            TypeName<S>().c_str(),
+            NYT::NDetail::FormatInvalidCastValue(value).c_str(),
+            TypeName<T>().c_str(),
+            ::ToString(std::numeric_limits<T>::min()).c_str(),
+            ::ToString(std::numeric_limits<T>::max()).c_str()));
     }
     return result;
 }
@@ -119,7 +124,8 @@ T CheckedEnumCast(S value)
 {
     T result;
     if (!TryEnumCast<T>(value, &result)) {
-        throw TSimpleException(Sprintf("Invalid value %d of enum type %s",
+        throw TSimpleException(Sprintf("Error casting %s value \"%d\" to enum %s",
+            TypeName<S>().c_str(),
             static_cast<int>(value),
             TEnumTraits<T>::GetTypeName().data()));
     }
diff --git a/yt/yt/client/table_client/unittests/serialization_ut.cpp b/yt/yt/client/table_client/unittests/serialization_ut.cpp
index 9280dd1f76..e48de97ab8 100644
--- a/yt/yt/client/table_client/unittests/serialization_ut.cpp
+++ b/yt/yt/client/table_client/unittests/serialization_ut.cpp
@@ -1,3 +1,4 @@
+#include <yt/yt/client/table_client/helpers.h>
 #include <yt/yt/client/table_client/schema.h>
 
 #include <yt/yt/library/formats/format.h>
@@ -66,6 +67,27 @@ TEST(TSchemaSerialization, Deleted)
         "Stable name should be set for a deleted column");
 }
 
+TEST(TInstantSerialization, YsonCompatibility)
+{
+    auto convert = [] (auto value) {
+        TUnversionedValue unversioned;
+        ToUnversionedValue(&unversioned, value, /*rowBuffer*/ nullptr);
+        auto node = NYTree::ConvertToNode(unversioned);
+        return NYTree::ConvertTo<TInstant>(node);
+    };
+
+    TInstant now = TInstant::Now();
+    TInstant lower = TInstant::TInstant::ParseIso8601("1970-03-01");
+    TInstant upper = TInstant::ParseIso8601("2100-01-01");
+
+    EXPECT_EQ(now, convert(now));
+    EXPECT_EQ(TInstant::MilliSeconds(now.MilliSeconds()), convert(now.MilliSeconds()));
+    EXPECT_EQ(lower, convert(lower));
+    EXPECT_EQ(TInstant::MilliSeconds(lower.MilliSeconds()), convert(lower.MilliSeconds()));
+    EXPECT_EQ(upper, convert(upper));
+    EXPECT_EQ(TInstant::MilliSeconds(upper.MilliSeconds()), convert(upper.MilliSeconds()));
+}
+
 ////////////////////////////////////////////////////////////////////////////////
 
 } // namespace
diff --git a/yt/yt/core/ytree/serialize.cpp b/yt/yt/core/ytree/serialize.cpp
index 553ab3b03f..897439d31c 100644
--- a/yt/yt/core/ytree/serialize.cpp
+++ b/yt/yt/core/ytree/serialize.cpp
@@ -16,6 +16,33 @@ using namespace google::protobuf::io;
 
 ////////////////////////////////////////////////////////////////////////////////
 
+// Unversioned values use microsecond precision for TInstant values,
+// while YSON deserializes values with millisecond precision.
+// It can lead to unexpected results when converting TInstant -> TUnversionedValue -> INodePtr -> TInstant.
+// These boundaries allow to correctly deserialize microsecond values back to TInstant.
+//
+// log2(timeEpoch("2100-01-01") * 10**3) < 42.
+// log2(timeEpoch("1970-03-01") * 10**6) > 42.
+// log2(timeEpoch("2100-01-01") * 10**6) < 52.
+static constexpr ui64 MicrosecondLowerWidthBoundary = 42;
+static constexpr ui64 MicrosecondUpperWidthBoundary = 52;
+static constexpr ui64 UnixTimeMicrosecondLowerBoundary = 1ull << MicrosecondLowerWidthBoundary;
+static constexpr ui64 UnixTimeMicrosecondUpperBoundary = 1ull << MicrosecondUpperWidthBoundary;
+
+TInstant ConvertRawValueToUnixTime(ui64 value)
+{
+    if (value < UnixTimeMicrosecondLowerBoundary) {
+        return TInstant::MilliSeconds(value);
+    } else if (value < UnixTimeMicrosecondUpperBoundary) {
+        return TInstant::MicroSeconds(value);
+    } else {
+        THROW_ERROR_EXCEPTION("Value %Qv does not represent valid UNIX time",
+            value);
+    }
+}
+
+////////////////////////////////////////////////////////////////////////////////
+
 EYsonType GetYsonType(const TYsonString& yson)
 {
     return yson.GetType();
@@ -272,17 +299,16 @@ void Deserialize(TInstant& value, INodePtr node)
 {
     switch (node->GetType()) {
         case ENodeType::Int64: {
-            auto ms = node->AsInt64()->GetValue();
-            if (ms < 0) {
-                THROW_ERROR_EXCEPTION("Instant cannot be negative");
-            }
-            value = TInstant::MilliSeconds(ms);
+            auto ms = CheckedIntegralCast<ui64>(node->AsInt64()->GetValue());
+            value = ConvertRawValueToUnixTime(ms);
             break;
         }
 
-        case ENodeType::Uint64:
-            value = TInstant::MilliSeconds(node->AsUint64()->GetValue());
+        case ENodeType::Uint64: {
+            auto ms = node->AsUint64()->GetValue();
+            value = ConvertRawValueToUnixTime(ms);
             break;
+        }
 
         case ENodeType::Double: {
             auto ms = node->AsDouble()->GetValue();