YT-23439: Fix block size estimation in scan writer

commit_hash:6bc02d261d376710e0d08f9372def30d3acfa949

6 files changed, 18 insertions, 17 deletions

diff --git a/yt/yt/client/table_client/config.cpp b/yt/yt/client/table_client/config.cpp
index 49898440e2..6bfd4b410c 100644
--- a/yt/yt/client/table_client/config.cpp
+++ b/yt/yt/client/table_client/config.cpp
@@ -464,6 +464,8 @@ void TChunkWriterOptions::Register(TRegistrar registrar)
         .Default(false);
     registrar.Parameter("enable_column_meta_in_chunk_meta", &TThis::EnableColumnMetaInChunkMeta)
         .Default(true);
+    registrar.Parameter("consider_min_row_range_data_weight", &TThis::ConsiderMinRowRangeDataWeight)
+        .Default(true);
 
     registrar.Parameter("schema_modification", &TThis::SchemaModification)
         .Default(ETableSchemaModification::None);
diff --git a/yt/yt/client/table_client/config.h b/yt/yt/client/table_client/config.h
index a8accf974b..2f91bcd032 100644
--- a/yt/yt/client/table_client/config.h
+++ b/yt/yt/client/table_client/config.h
@@ -429,6 +429,7 @@ public:
     bool EnableRowCountInColumnarStatistics;
     bool EnableSegmentMetaInBlocks;
     bool EnableColumnMetaInChunkMeta;
+    bool ConsiderMinRowRangeDataWeight;
 
     NYTree::INodePtr CastAnyToCompositeNode;
 
diff --git a/yt/yt/core/misc/bit_packed_unsigned_vector-inl.h b/yt/yt/core/misc/bit_packed_unsigned_vector-inl.h
index d1abbbbd0a..3dbb3ce3e5 100644
--- a/yt/yt/core/misc/bit_packed_unsigned_vector-inl.h
+++ b/yt/yt/core/misc/bit_packed_unsigned_vector-inl.h
@@ -25,7 +25,7 @@ inline size_t CompressedUnsignedVectorSizeInWords(ui64 maxValue, size_t count)
 
 inline size_t CompressedUnsignedVectorSizeInBytes(ui64 maxValue, size_t count)
 {
-    static size_t wordSize = sizeof(ui64);
+    static constexpr size_t wordSize = sizeof(ui64);
     return CompressedUnsignedVectorSizeInWords(maxValue, count) * wordSize;
 }
 
diff --git a/yt/yt/core/misc/bit_packed_unsigned_vector.cpp b/yt/yt/core/misc/bit_packed_unsigned_vector.cpp
index 05ff7ed88f..bec799a9f1 100644
--- a/yt/yt/core/misc/bit_packed_unsigned_vector.cpp
+++ b/yt/yt/core/misc/bit_packed_unsigned_vector.cpp
@@ -8,24 +8,26 @@ namespace NYT {
 
 ////////////////////////////////////////////////////////////////////////////////
 
-void PrepareDiffFromExpected(std::vector<ui32>* values, ui32* expected, ui32* maxDiff)
+std::pair<ui32, ui32> PrepareDiffFromExpected(std::vector<ui32>* values)
 {
+    ui32 expected = 0;
+    ui32 maxDiff = 0;
+
     if (values->empty()) {
-        *expected = 0;
-        *maxDiff = 0;
-        return;
+        return {expected, maxDiff};
     }
 
-    *expected = DivRound<int>(values->back(), values->size());
+    expected = DivRound<int>(values->back(), values->size());
 
-    *maxDiff = 0;
     i64 expectedValue = 0;
     for (int i = 0; i < std::ssize(*values); ++i) {
-        expectedValue += *expected;
-        i32 diff = values->at(i) - expectedValue;
+        expectedValue += expected;
+        i32 diff = (*values)[i] - expectedValue;
         (*values)[i] = ZigZagEncode32(diff);
-        *maxDiff = std::max(*maxDiff, (*values)[i]);
+        maxDiff = std::max(maxDiff, (*values)[i]);
     }
+
+    return {expected, maxDiff};
 }
 
 ////////////////////////////////////////////////////////////////////////////////
diff --git a/yt/yt/core/misc/bit_packed_unsigned_vector.h b/yt/yt/core/misc/bit_packed_unsigned_vector.h
index 5a12aaba56..4f95561267 100644
--- a/yt/yt/core/misc/bit_packed_unsigned_vector.h
+++ b/yt/yt/core/misc/bit_packed_unsigned_vector.h
@@ -59,7 +59,7 @@ private:
 
 ////////////////////////////////////////////////////////////////////////////////
 
-void PrepareDiffFromExpected(std::vector<ui32>* values, ui32* expected, ui32* maxDiff);
+std::pair<ui32, ui32> PrepareDiffFromExpected(std::vector<ui32>* values);
 
 ////////////////////////////////////////////////////////////////////////////////
 
diff --git a/yt/yt/library/column_converters/string_column_converter.cpp b/yt/yt/library/column_converters/string_column_converter.cpp
index 60b4e28337..797f64f4ee 100644
--- a/yt/yt/library/column_converters/string_column_converter.cpp
+++ b/yt/yt/library/column_converters/string_column_converter.cpp
@@ -128,9 +128,7 @@ private:
         auto offsets = GetDirectDenseOffsets();
 
         // Save offsets as diff from expected.
-        ui32 expectedLength;
-        ui32 maxDiff;
-        PrepareDiffFromExpected(&offsets, &expectedLength, &maxDiff);
+        auto [expectedLength, maxDiff] = PrepareDiffFromExpected(&offsets);
 
         auto directDataSize = DirectBuffer_->GetSize();
         auto directData = DirectBuffer_->Finish();
@@ -213,9 +211,7 @@ private:
         auto idsRef = TSharedRef::MakeCopy<TConverterTag>(TRef(ids.data(), sizeof(ui32) * ids.size()));
 
         // 2. Dictionary offsets.
-        ui32 expectedLength;
-        ui32 maxDiff;
-        PrepareDiffFromExpected(&dictionaryOffsets, &expectedLength, &maxDiff);
+        auto [expectedLength, maxDiff] = PrepareDiffFromExpected(&dictionaryOffsets);
         auto dictionaryOffsetsRef = TSharedRef::MakeCopy<TConverterTag>(TRef(dictionaryOffsets.data(), sizeof(ui32) * dictionaryOffsets.size()));
 
         auto primaryColumn = std::make_shared<TBatchColumn>();