Improve TPredicate

* move implementation details into cpp
* reduce sizeof(TPredicate) from 32 bytes to 24 by ordering the fields

6 files changed, 85 insertions, 66 deletions

diff --git a/ydb/core/tx/columnshard/engines/CMakeLists.darwin-x86_64.txt b/ydb/core/tx/columnshard/engines/CMakeLists.darwin-x86_64.txt
index 6e27fb3e510..398cfd3d0cc 100644
--- a/ydb/core/tx/columnshard/engines/CMakeLists.darwin-x86_64.txt
+++ b/ydb/core/tx/columnshard/engines/CMakeLists.darwin-x86_64.txt
@@ -33,5 +33,6 @@ target_sources(tx-columnshard-engines PRIVATE
   ${CMAKE_SOURCE_DIR}/ydb/core/tx/columnshard/engines/indexed_read_data.cpp
   ${CMAKE_SOURCE_DIR}/ydb/core/tx/columnshard/engines/filter.cpp
   ${CMAKE_SOURCE_DIR}/ydb/core/tx/columnshard/engines/portion_info.cpp
+  ${CMAKE_SOURCE_DIR}/ydb/core/tx/columnshard/engines/predicate.cpp
   ${CMAKE_SOURCE_DIR}/ydb/core/tx/columnshard/engines/scalars.cpp
 )
diff --git a/ydb/core/tx/columnshard/engines/CMakeLists.linux-aarch64.txt b/ydb/core/tx/columnshard/engines/CMakeLists.linux-aarch64.txt
index 10c6eaf1a01..ef42c703321 100644
--- a/ydb/core/tx/columnshard/engines/CMakeLists.linux-aarch64.txt
+++ b/ydb/core/tx/columnshard/engines/CMakeLists.linux-aarch64.txt
@@ -34,5 +34,6 @@ target_sources(tx-columnshard-engines PRIVATE
   ${CMAKE_SOURCE_DIR}/ydb/core/tx/columnshard/engines/indexed_read_data.cpp
   ${CMAKE_SOURCE_DIR}/ydb/core/tx/columnshard/engines/filter.cpp
   ${CMAKE_SOURCE_DIR}/ydb/core/tx/columnshard/engines/portion_info.cpp
+  ${CMAKE_SOURCE_DIR}/ydb/core/tx/columnshard/engines/predicate.cpp
   ${CMAKE_SOURCE_DIR}/ydb/core/tx/columnshard/engines/scalars.cpp
 )
diff --git a/ydb/core/tx/columnshard/engines/CMakeLists.linux-x86_64.txt b/ydb/core/tx/columnshard/engines/CMakeLists.linux-x86_64.txt
index 10c6eaf1a01..ef42c703321 100644
--- a/ydb/core/tx/columnshard/engines/CMakeLists.linux-x86_64.txt
+++ b/ydb/core/tx/columnshard/engines/CMakeLists.linux-x86_64.txt
@@ -34,5 +34,6 @@ target_sources(tx-columnshard-engines PRIVATE
   ${CMAKE_SOURCE_DIR}/ydb/core/tx/columnshard/engines/indexed_read_data.cpp
   ${CMAKE_SOURCE_DIR}/ydb/core/tx/columnshard/engines/filter.cpp
   ${CMAKE_SOURCE_DIR}/ydb/core/tx/columnshard/engines/portion_info.cpp
+  ${CMAKE_SOURCE_DIR}/ydb/core/tx/columnshard/engines/predicate.cpp
   ${CMAKE_SOURCE_DIR}/ydb/core/tx/columnshard/engines/scalars.cpp
 )
diff --git a/ydb/core/tx/columnshard/engines/CMakeLists.windows-x86_64.txt b/ydb/core/tx/columnshard/engines/CMakeLists.windows-x86_64.txt
index 6e27fb3e510..398cfd3d0cc 100644
--- a/ydb/core/tx/columnshard/engines/CMakeLists.windows-x86_64.txt
+++ b/ydb/core/tx/columnshard/engines/CMakeLists.windows-x86_64.txt
@@ -33,5 +33,6 @@ target_sources(tx-columnshard-engines PRIVATE
   ${CMAKE_SOURCE_DIR}/ydb/core/tx/columnshard/engines/indexed_read_data.cpp
   ${CMAKE_SOURCE_DIR}/ydb/core/tx/columnshard/engines/filter.cpp
   ${CMAKE_SOURCE_DIR}/ydb/core/tx/columnshard/engines/portion_info.cpp
+  ${CMAKE_SOURCE_DIR}/ydb/core/tx/columnshard/engines/predicate.cpp
   ${CMAKE_SOURCE_DIR}/ydb/core/tx/columnshard/engines/scalars.cpp
 )
diff --git a/ydb/core/tx/columnshard/engines/predicate.cpp b/ydb/core/tx/columnshard/engines/predicate.cpp
new file mode 100644
index 00000000000..493bc06ae2a
--- /dev/null
+++ b/ydb/core/tx/columnshard/engines/predicate.cpp
@@ -0,0 +1,68 @@
+#include "predicate.h"
+
+#include <ydb/core/formats/arrow_helpers.h>
+#include <ydb/core/formats/switch_type.h>
+
+namespace NKikimr::NOlap {
+
+TPredicate::TPredicate(EOperation op, std::shared_ptr<arrow::RecordBatch> batch, bool inclusive) noexcept
+    : Batch(std::move(batch))
+    , Operation(op)
+    , Inclusive(inclusive)
+{}
+
+TPredicate::TPredicate(EOperation op, const TString& serializedBatch, const std::shared_ptr<arrow::Schema>& schema, bool inclusive)
+    : Operation(op)
+    , Inclusive(inclusive)
+{
+    if (!serializedBatch.empty()) {
+        Batch = NArrow::DeserializeBatch(serializedBatch, schema);
+        Y_VERIFY(Batch);
+    }
+}
+
+TVector<TString> TPredicate::ColumnNames() const {
+    TVector<TString> out;
+    out.reserve(Batch->num_columns());
+    for (const auto& field : Batch->schema()->fields()) {
+        out.emplace_back(field->name());
+    }
+    return out;
+}
+
+IOutputStream& operator << (IOutputStream& out, const TPredicate& pred) {
+    out << NSsa::GetFunctionName(pred.Operation);
+    if (pred.Inclusive) {
+        out << "(incl) ";
+    } else {
+        out << "(excl) ";
+    }
+
+    for (i32 i = 0; i < pred.Batch->num_columns(); ++i) {
+        auto array = pred.Batch->column(i);
+        out << pred.Batch->schema()->field(i)->name() << ": ";
+        NArrow::SwitchType(array->type_id(), [&](const auto& type) {
+            using TWrap = std::decay_t<decltype(type)>;
+            using TArray = typename arrow::TypeTraits<typename TWrap::T>::ArrayType;
+
+            auto& typedArray = static_cast<const TArray&>(*array);
+            if (typedArray.IsNull(0)) {
+                out << "NULL";
+            } else {
+                auto value = typedArray.GetView(0);
+                using T = std::decay_t<decltype(value)>;
+                if constexpr (std::is_same_v<T, arrow::util::string_view>) {
+                    out << "'" << std::string_view(value.data(), value.size()) << "'";
+                } else {
+                    out << "'" << value << "'";
+                }
+            }
+            return true;
+        });
+        out << " ";
+    }
+
+    return out;
+}
+
+} // namespace NKikimr::NOlap
diff --git a/ydb/core/tx/columnshard/engines/predicate.h b/ydb/core/tx/columnshard/engines/predicate.h
index 59e5e01eeaa..525009dfbee 100644
--- a/ydb/core/tx/columnshard/engines/predicate.h
+++ b/ydb/core/tx/columnshard/engines/predicate.h
@@ -2,90 +2,37 @@
 
 #include "defs.h"
 
-#include <ydb/core/formats/arrow_helpers.h>
 #include <ydb/core/formats/program.h>
-#include <ydb/core/formats/switch_type.h>
+
+#include <contrib/libs/apache/arrow/cpp/src/arrow/record_batch.h>
 
 namespace NKikimr::NOlap {
 
 struct TPredicate {
     using EOperation = NArrow::EOperation;
 
-    EOperation Operation{EOperation::Unspecified};
     std::shared_ptr<arrow::RecordBatch> Batch;
-    bool Inclusive;
+    EOperation Operation{EOperation::Unspecified};
+    bool Inclusive{false};
 
-    bool Empty() const { return Batch.get() == nullptr; }
+    bool Empty() const noexcept { return Batch.get() == nullptr; }
     bool Good() const { return !Empty() && Batch->num_columns() && Batch->num_rows() == 1; }
-    bool IsFrom() const { return Operation == EOperation::Greater || Operation == EOperation::GreaterEqual; }
-    bool IsTo() const { return Operation == EOperation::Less || Operation == EOperation::LessEqual; }
+    bool IsFrom() const noexcept { return Operation == EOperation::Greater || Operation == EOperation::GreaterEqual; }
+    bool IsTo() const noexcept { return Operation == EOperation::Less || Operation == EOperation::LessEqual; }
 
-    TVector<TString> ColumnNames() const {
-        TVector<TString> out;
-        out.reserve(Batch->num_columns());
-        for (auto& field : Batch->schema()->fields()) {
-            TString name(field->name().data(), field->name().size());
-            out.emplace_back(name);
-        }
-        return out;
-    }
+    TVector<TString> ColumnNames() const;
 
     std::string ToString() const {
         return Empty() ? "()" : Batch->schema()->ToString();
     }
 
-    TPredicate() = default;
+    constexpr TPredicate() noexcept = default;
 
-    TPredicate(EOperation op, std::shared_ptr<arrow::RecordBatch> batch, bool inclusive = false)
-        : Operation(op)
-        , Batch(batch)
-        , Inclusive(inclusive)
-    {}
+    TPredicate(EOperation op, std::shared_ptr<arrow::RecordBatch> batch, bool inclusive = false) noexcept;
 
-    TPredicate(EOperation op, TString serializedBatch, std::shared_ptr<arrow::Schema> schema, bool inclusive)
-        : Operation(op)
-        , Inclusive(inclusive)
-    {
-        if (!serializedBatch.empty()) {
-            Batch = NArrow::DeserializeBatch(serializedBatch, schema);
-            Y_VERIFY(Batch);
-        }
-    }
+    TPredicate(EOperation op, const TString& serializedBatch, const std::shared_ptr<arrow::Schema>& schema, bool inclusive);
 
-    friend IOutputStream& operator << (IOutputStream& out, const TPredicate& pred) {
-        out << NSsa::GetFunctionName(pred.Operation);
-        if (pred.Inclusive) {
-            out << "(incl) ";
-        } else {
-            out << "(excl) ";
-        }
-
-        for (i32 i = 0; i < pred.Batch->num_columns(); ++i) {
-            auto array = pred.Batch->column(i);
-            out << pred.Batch->schema()->field(i)->name() << ": ";
-            NArrow::SwitchType(array->type_id(), [&](const auto& type) {
-                using TWrap = std::decay_t<decltype(type)>;
-                using TArray = typename arrow::TypeTraits<typename TWrap::T>::ArrayType;
-
-                auto& typedArray = static_cast<const TArray&>(*array);
-                if (typedArray.IsNull(0)) {
-                    out << "NULL";
-                } else {
-                    auto value = typedArray.GetView(0);
-                    using T = std::decay_t<decltype(value)>;
-                    if constexpr (std::is_same_v<T, arrow::util::string_view>) {
-                        out << "'" << std::string(value.data(), value.size()) << "'";
-                    } else {
-                        out << "'" << value << "'";
-                    }
-                }
-                return true;
-            });
-            out << " ";
-        }
-
-        return out;
-    }
+    friend IOutputStream& operator << (IOutputStream& out, const TPredicate& pred);
 };
 
-}
+} // namespace NKikimr::NOlap