diff --git a/src/Columns/ColumnAggregateFunction.cpp b/src/Columns/ColumnAggregateFunction.cpp index f7e6b1a1ccc..327179f53ab 100644 --- a/src/Columns/ColumnAggregateFunction.cpp +++ b/src/Columns/ColumnAggregateFunction.cpp @@ -362,13 +362,10 @@ void ColumnAggregateFunction::updateHashWithValue(size_t n, SipHash & hash) cons hash.update(wbuf.str().c_str(), wbuf.str().size()); } -void ColumnAggregateFunction::updateWeakHash32(WeakHash32 & hash) const +WeakHash32 ColumnAggregateFunction::getWeakHash32() const { auto s = data.size(); - if (hash.getData().size() != data.size()) - throw Exception(ErrorCodes::LOGICAL_ERROR, "Size of WeakHash32 does not match size of column: " - "column size is {}, hash size is {}", std::to_string(s), hash.getData().size()); - + WeakHash32 hash(s); auto & hash_data = hash.getData(); std::vector v; @@ -379,6 +376,8 @@ void ColumnAggregateFunction::updateWeakHash32(WeakHash32 & hash) const wbuf.finalize(); hash_data[i] = ::updateWeakHash32(v.data(), v.size(), hash_data[i]); } + + return hash; } void ColumnAggregateFunction::updateHashFast(SipHash & hash) const diff --git a/src/Columns/ColumnAggregateFunction.h b/src/Columns/ColumnAggregateFunction.h index a75b27e835c..c13dfbb4587 100644 --- a/src/Columns/ColumnAggregateFunction.h +++ b/src/Columns/ColumnAggregateFunction.h @@ -170,7 +170,7 @@ public: void updateHashWithValue(size_t n, SipHash & hash) const override; - void updateWeakHash32(WeakHash32 & hash) const override; + WeakHash32 getWeakHash32() const override; void updateHashFast(SipHash & hash) const override; diff --git a/src/Columns/ColumnArray.cpp b/src/Columns/ColumnArray.cpp index 0b7e6541560..33d8fec9d1b 100644 --- a/src/Columns/ColumnArray.cpp +++ b/src/Columns/ColumnArray.cpp @@ -271,15 +271,12 @@ void ColumnArray::updateHashWithValue(size_t n, SipHash & hash) const getData().updateHashWithValue(offset + i, hash); } -void ColumnArray::updateWeakHash32(WeakHash32 & hash) const +WeakHash32 ColumnArray::getWeakHash32() const { auto s = offsets->size(); - if (hash.getData().size() != s) - throw Exception(ErrorCodes::LOGICAL_ERROR, "Size of WeakHash32 does not match size of column: " - "column size is {}, hash size is {}", s, hash.getData().size()); + WeakHash32 hash(s); - WeakHash32 internal_hash(data->size()); - data->updateWeakHash32(internal_hash); + WeakHash32 internal_hash = data->getWeakHash32(); Offset prev_offset = 0; const auto & offsets_data = getOffsets(); @@ -300,6 +297,8 @@ void ColumnArray::updateWeakHash32(WeakHash32 & hash) const prev_offset = offsets_data[i]; } + + return hash; } void ColumnArray::updateHashFast(SipHash & hash) const diff --git a/src/Columns/ColumnArray.h b/src/Columns/ColumnArray.h index 53eb5166df8..0e2a94e2aa8 100644 --- a/src/Columns/ColumnArray.h +++ b/src/Columns/ColumnArray.h @@ -82,7 +82,7 @@ public: const char * deserializeAndInsertFromArena(const char * pos) override; const char * skipSerializedInArena(const char * pos) const override; void updateHashWithValue(size_t n, SipHash & hash) const override; - void updateWeakHash32(WeakHash32 & hash) const override; + WeakHash32 getWeakHash32() const override; void updateHashFast(SipHash & hash) const override; void insertRangeFrom(const IColumn & src, size_t start, size_t length) override; void insert(const Field & x) override; diff --git a/src/Columns/ColumnCompressed.h b/src/Columns/ColumnCompressed.h index 934adf07cf4..70233cbc1b7 100644 --- a/src/Columns/ColumnCompressed.h +++ b/src/Columns/ColumnCompressed.h @@ -3,6 +3,7 @@ #include #include #include +#include #include @@ -94,7 +95,7 @@ public: const char * deserializeAndInsertFromArena(const char *) override { throwMustBeDecompressed(); } const char * skipSerializedInArena(const char *) const override { throwMustBeDecompressed(); } void updateHashWithValue(size_t, SipHash &) const override { throwMustBeDecompressed(); } - void updateWeakHash32(WeakHash32 &) const override { throwMustBeDecompressed(); } + WeakHash32 getWeakHash32() const override { throwMustBeDecompressed(); } void updateHashFast(SipHash &) const override { throwMustBeDecompressed(); } ColumnPtr filter(const Filter &, ssize_t) const override { throwMustBeDecompressed(); } void expand(const Filter &, bool) override { throwMustBeDecompressed(); } diff --git a/src/Columns/ColumnConst.cpp b/src/Columns/ColumnConst.cpp index f2cea83db0e..84427e7be2b 100644 --- a/src/Columns/ColumnConst.cpp +++ b/src/Columns/ColumnConst.cpp @@ -137,18 +137,10 @@ void ColumnConst::updatePermutation(PermutationSortDirection /*direction*/, Perm { } -void ColumnConst::updateWeakHash32(WeakHash32 & hash) const +WeakHash32 ColumnConst::getWeakHash32() const { - if (hash.getData().size() != s) - throw Exception(ErrorCodes::LOGICAL_ERROR, "Size of WeakHash32 does not match size of column: " - "column size is {}, hash size is {}", std::to_string(s), std::to_string(hash.getData().size())); - - WeakHash32 element_hash(1); - data->updateWeakHash32(element_hash); - size_t data_hash = element_hash.getData()[0]; - - for (auto & value : hash.getData()) - value = static_cast(intHashCRC32(data_hash, value)); + WeakHash32 element_hash = data->getWeakHash32(); + return WeakHash32(s, element_hash.getData()[0]); } void ColumnConst::compareColumn( diff --git a/src/Columns/ColumnConst.h b/src/Columns/ColumnConst.h index c2c0fa3027c..e087d91dbcb 100644 --- a/src/Columns/ColumnConst.h +++ b/src/Columns/ColumnConst.h @@ -190,7 +190,7 @@ public: data->updateHashWithValue(0, hash); } - void updateWeakHash32(WeakHash32 & hash) const override; + WeakHash32 getWeakHash32() const override; void updateHashFast(SipHash & hash) const override { diff --git a/src/Columns/ColumnDecimal.cpp b/src/Columns/ColumnDecimal.cpp index eb9784c14dd..01e7f1be51a 100644 --- a/src/Columns/ColumnDecimal.cpp +++ b/src/Columns/ColumnDecimal.cpp @@ -28,7 +28,6 @@ namespace ErrorCodes extern const int PARAMETER_OUT_OF_BOUND; extern const int SIZES_OF_COLUMNS_DOESNT_MATCH; extern const int NOT_IMPLEMENTED; - extern const int LOGICAL_ERROR; } template @@ -72,13 +71,10 @@ void ColumnDecimal::updateHashWithValue(size_t n, SipHash & hash) const } template -void ColumnDecimal::updateWeakHash32(WeakHash32 & hash) const +WeakHash32 ColumnDecimal::getWeakHash32() const { auto s = data.size(); - - if (hash.getData().size() != s) - throw Exception(ErrorCodes::LOGICAL_ERROR, "Size of WeakHash32 does not match size of column: " - "column size is {}, hash size is {}", std::to_string(s), std::to_string(hash.getData().size())); + WeakHash32 hash(s); const T * begin = data.data(); const T * end = begin + s; @@ -90,6 +86,8 @@ void ColumnDecimal::updateWeakHash32(WeakHash32 & hash) const ++begin; ++hash_data; } + + return hash; } template diff --git a/src/Columns/ColumnDecimal.h b/src/Columns/ColumnDecimal.h index c4510ba2922..662def45c96 100644 --- a/src/Columns/ColumnDecimal.h +++ b/src/Columns/ColumnDecimal.h @@ -90,7 +90,7 @@ public: const char * deserializeAndInsertFromArena(const char * pos) override; const char * skipSerializedInArena(const char * pos) const override; void updateHashWithValue(size_t n, SipHash & hash) const override; - void updateWeakHash32(WeakHash32 & hash) const override; + WeakHash32 getWeakHash32() const override; void updateHashFast(SipHash & hash) const override; int compareAt(size_t n, size_t m, const IColumn & rhs_, int nan_direction_hint) const override; void getPermutation(IColumn::PermutationSortDirection direction, IColumn::PermutationSortStability stability, diff --git a/src/Columns/ColumnDynamic.h b/src/Columns/ColumnDynamic.h index 27ad0dd583f..9fc85ffcafc 100644 --- a/src/Columns/ColumnDynamic.h +++ b/src/Columns/ColumnDynamic.h @@ -4,6 +4,7 @@ #include #include #include +#include namespace DB @@ -167,9 +168,9 @@ public: void updateHashWithValue(size_t n, SipHash & hash) const override; - void updateWeakHash32(WeakHash32 & hash) const override + WeakHash32 getWeakHash32() const override { - variant_column->updateWeakHash32(hash); + return variant_column->getWeakHash32(); } void updateHashFast(SipHash & hash) const override diff --git a/src/Columns/ColumnFixedString.cpp b/src/Columns/ColumnFixedString.cpp index d7e4eff2727..00ed9b53e4d 100644 --- a/src/Columns/ColumnFixedString.cpp +++ b/src/Columns/ColumnFixedString.cpp @@ -129,14 +129,10 @@ void ColumnFixedString::updateHashWithValue(size_t index, SipHash & hash) const hash.update(reinterpret_cast(&chars[n * index]), n); } -void ColumnFixedString::updateWeakHash32(WeakHash32 & hash) const +WeakHash32 ColumnFixedString::getWeakHash32() const { auto s = size(); - - if (hash.getData().size() != s) - throw Exception(ErrorCodes::LOGICAL_ERROR, "Size of WeakHash32 does not match size of column: " - "column size is {}, " - "hash size is {}", std::to_string(s), std::to_string(hash.getData().size())); + WeakHash32 hash(s); const UInt8 * pos = chars.data(); UInt32 * hash_data = hash.getData().data(); @@ -148,6 +144,8 @@ void ColumnFixedString::updateWeakHash32(WeakHash32 & hash) const pos += n; ++hash_data; } + + return hash; } void ColumnFixedString::updateHashFast(SipHash & hash) const diff --git a/src/Columns/ColumnFixedString.h b/src/Columns/ColumnFixedString.h index 7b46dc11cd6..e8a815a6501 100644 --- a/src/Columns/ColumnFixedString.h +++ b/src/Columns/ColumnFixedString.h @@ -125,7 +125,7 @@ public: void updateHashWithValue(size_t index, SipHash & hash) const override; - void updateWeakHash32(WeakHash32 & hash) const override; + WeakHash32 getWeakHash32() const override; void updateHashFast(SipHash & hash) const override; diff --git a/src/Columns/ColumnFunction.h b/src/Columns/ColumnFunction.h index 6fdc6679d3e..f1c1aaba4f6 100644 --- a/src/Columns/ColumnFunction.h +++ b/src/Columns/ColumnFunction.h @@ -4,6 +4,7 @@ #include #include #include +#include namespace DB @@ -122,9 +123,9 @@ public: throw Exception(ErrorCodes::NOT_IMPLEMENTED, "updateHashWithValue is not implemented for {}", getName()); } - void updateWeakHash32(WeakHash32 &) const override + WeakHash32 getWeakHash32() const override { - throw Exception(ErrorCodes::NOT_IMPLEMENTED, "updateWeakHash32 is not implemented for {}", getName()); + throw Exception(ErrorCodes::NOT_IMPLEMENTED, "getWeakHash32 is not implemented for {}", getName()); } void updateHashFast(SipHash &) const override diff --git a/src/Columns/ColumnLowCardinality.cpp b/src/Columns/ColumnLowCardinality.cpp index 208326fe629..2836ca3f8cd 100644 --- a/src/Columns/ColumnLowCardinality.cpp +++ b/src/Columns/ColumnLowCardinality.cpp @@ -7,8 +7,7 @@ #include #include #include -#include "Storages/IndicesDescription.h" -#include "base/types.h" +#include #include #include @@ -312,19 +311,10 @@ const char * ColumnLowCardinality::skipSerializedInArena(const char * pos) const return getDictionary().skipSerializedInArena(pos); } -void ColumnLowCardinality::updateWeakHash32(WeakHash32 & hash) const +WeakHash32 ColumnLowCardinality::getWeakHash32() const { - auto s = size(); - - if (hash.getData().size() != s) - throw Exception(ErrorCodes::LOGICAL_ERROR, "Size of WeakHash32 does not match size of column: " - "column size is {}, hash size is {}", std::to_string(s), std::to_string(hash.getData().size())); - - const auto & dict = getDictionary().getNestedColumn(); - WeakHash32 dict_hash(dict->size()); - dict->updateWeakHash32(dict_hash); - - idx.updateWeakHash(hash, dict_hash); + WeakHash32 dict_hash = getDictionary().getNestedColumn()->getWeakHash32(); + return idx.getWeakHash(dict_hash); } void ColumnLowCardinality::updateHashFast(SipHash & hash) const @@ -820,10 +810,11 @@ bool ColumnLowCardinality::Index::containsDefault() const return contains; } -void ColumnLowCardinality::Index::updateWeakHash(WeakHash32 & hash, WeakHash32 & dict_hash) const +WeakHash32 ColumnLowCardinality::Index::getWeakHash(const WeakHash32 & dict_hash) const { + WeakHash32 hash(positions->size()); auto & hash_data = hash.getData(); - auto & dict_hash_data = dict_hash.getData(); + const auto & dict_hash_data = dict_hash.getData(); auto update_weak_hash = [&](auto x) { @@ -832,10 +823,11 @@ void ColumnLowCardinality::Index::updateWeakHash(WeakHash32 & hash, WeakHash32 & auto size = data.size(); for (size_t i = 0; i < size; ++i) - hash_data[i] = static_cast(intHashCRC32(dict_hash_data[data[i]], hash_data[i])); + hash_data[i] = dict_hash_data[data[i]]; }; callForType(std::move(update_weak_hash), size_of_type); + return hash; } void ColumnLowCardinality::Index::collectSerializedValueSizes( diff --git a/src/Columns/ColumnLowCardinality.h b/src/Columns/ColumnLowCardinality.h index ac3b725b22f..c193a5043c4 100644 --- a/src/Columns/ColumnLowCardinality.h +++ b/src/Columns/ColumnLowCardinality.h @@ -103,7 +103,7 @@ public: getDictionary().updateHashWithValue(getIndexes().getUInt(n), hash); } - void updateWeakHash32(WeakHash32 & hash) const override; + WeakHash32 getWeakHash32() const override; void updateHashFast(SipHash &) const override; @@ -313,7 +313,7 @@ public: bool containsDefault() const; - void updateWeakHash(WeakHash32 & hash, WeakHash32 & dict_hash) const; + WeakHash32 getWeakHash(const WeakHash32 & dict_hash) const; void collectSerializedValueSizes(PaddedPODArray & sizes, const PaddedPODArray & dict_sizes) const; diff --git a/src/Columns/ColumnMap.cpp b/src/Columns/ColumnMap.cpp index eecea1a273f..95c6aa39943 100644 --- a/src/Columns/ColumnMap.cpp +++ b/src/Columns/ColumnMap.cpp @@ -143,9 +143,9 @@ void ColumnMap::updateHashWithValue(size_t n, SipHash & hash) const nested->updateHashWithValue(n, hash); } -void ColumnMap::updateWeakHash32(WeakHash32 & hash) const +WeakHash32 ColumnMap::getWeakHash32() const { - nested->updateWeakHash32(hash); + return nested->getWeakHash32(); } void ColumnMap::updateHashFast(SipHash & hash) const diff --git a/src/Columns/ColumnMap.h b/src/Columns/ColumnMap.h index 52165d0d74e..11acf68b6f0 100644 --- a/src/Columns/ColumnMap.h +++ b/src/Columns/ColumnMap.h @@ -64,7 +64,7 @@ public: const char * deserializeAndInsertFromArena(const char * pos) override; const char * skipSerializedInArena(const char * pos) const override; void updateHashWithValue(size_t n, SipHash & hash) const override; - void updateWeakHash32(WeakHash32 & hash) const override; + WeakHash32 getWeakHash32() const override; void updateHashFast(SipHash & hash) const override; void insertFrom(const IColumn & src_, size_t n) override; void insertManyFrom(const IColumn & src, size_t position, size_t length) override; diff --git a/src/Columns/ColumnNullable.cpp b/src/Columns/ColumnNullable.cpp index 1d12a59fd59..444676cb7fc 100644 --- a/src/Columns/ColumnNullable.cpp +++ b/src/Columns/ColumnNullable.cpp @@ -56,25 +56,21 @@ void ColumnNullable::updateHashWithValue(size_t n, SipHash & hash) const getNestedColumn().updateHashWithValue(n, hash); } -void ColumnNullable::updateWeakHash32(WeakHash32 & hash) const +WeakHash32 ColumnNullable::getWeakHash32() const { auto s = size(); - if (hash.getData().size() != s) - throw Exception(ErrorCodes::LOGICAL_ERROR, "Size of WeakHash32 does not match size of column: " - "column size is {}, hash size is {}", std::to_string(s), std::to_string(hash.getData().size())); - - WeakHash32 old_hash = hash; - nested_column->updateWeakHash32(hash); + WeakHash32 hash = nested_column->getWeakHash32(); const auto & null_map_data = getNullMapData(); auto & hash_data = hash.getData(); - auto & old_hash_data = old_hash.getData(); - /// Use old data for nulls. + /// Use default for nulls. for (size_t row = 0; row < s; ++row) if (null_map_data[row]) - hash_data[row] = old_hash_data[row]; + hash_data[row] = WeakHash32::kDefaultInitialValue; + + return hash; } void ColumnNullable::updateHashFast(SipHash & hash) const diff --git a/src/Columns/ColumnNullable.h b/src/Columns/ColumnNullable.h index 510a4cacf1e..5ec85c6e7ef 100644 --- a/src/Columns/ColumnNullable.h +++ b/src/Columns/ColumnNullable.h @@ -119,7 +119,7 @@ public: void protect() override; ColumnPtr replicate(const Offsets & replicate_offsets) const override; void updateHashWithValue(size_t n, SipHash & hash) const override; - void updateWeakHash32(WeakHash32 & hash) const override; + WeakHash32 getWeakHash32() const override; void updateHashFast(SipHash & hash) const override; void getExtremes(Field & min, Field & max) const override; // Special function for nullable minmax index diff --git a/src/Columns/ColumnObject.h b/src/Columns/ColumnObject.h index e2936b27994..fac8b7f914b 100644 --- a/src/Columns/ColumnObject.h +++ b/src/Columns/ColumnObject.h @@ -5,6 +5,7 @@ #include #include #include +#include #include @@ -241,7 +242,7 @@ public: const char * deserializeAndInsertFromArena(const char *) override { throwMustBeConcrete(); } const char * skipSerializedInArena(const char *) const override { throwMustBeConcrete(); } void updateHashWithValue(size_t, SipHash &) const override { throwMustBeConcrete(); } - void updateWeakHash32(WeakHash32 &) const override { throwMustBeConcrete(); } + WeakHash32 getWeakHash32() const override { throwMustBeConcrete(); } void updateHashFast(SipHash &) const override { throwMustBeConcrete(); } void expand(const Filter &, bool) override { throwMustBeConcrete(); } bool hasEqualValues() const override { throwMustBeConcrete(); } diff --git a/src/Columns/ColumnSparse.cpp b/src/Columns/ColumnSparse.cpp index 5190ceb49e5..40444757ef1 100644 --- a/src/Columns/ColumnSparse.cpp +++ b/src/Columns/ColumnSparse.cpp @@ -666,20 +666,22 @@ void ColumnSparse::updateHashWithValue(size_t n, SipHash & hash) const values->updateHashWithValue(getValueIndex(n), hash); } -void ColumnSparse::updateWeakHash32(WeakHash32 & hash) const +WeakHash32 ColumnSparse::getWeakHash32() const { - if (hash.getData().size() != _size) - throw Exception(ErrorCodes::LOGICAL_ERROR, "Size of WeakHash32 does not match size of column: " - "column size is {}, hash size is {}", _size, hash.getData().size()); + WeakHash32 values_hash = values->getWeakHash32(); + WeakHash32 hash(size()); + + auto & hash_data = hash.getData(); + auto & values_hash_data = values_hash.getData(); auto offset_it = begin(); - auto & hash_data = hash.getData(); for (size_t i = 0; i < _size; ++i, ++offset_it) { size_t value_index = offset_it.getValueIndex(); - auto data_ref = values->getDataAt(value_index); - hash_data[i] = ::updateWeakHash32(reinterpret_cast(data_ref.data), data_ref.size, hash_data[i]); + hash_data[i] = values_hash_data[value_index]; } + + return hash; } void ColumnSparse::updateHashFast(SipHash & hash) const diff --git a/src/Columns/ColumnSparse.h b/src/Columns/ColumnSparse.h index 12b2def7cf1..80c31c5d5d5 100644 --- a/src/Columns/ColumnSparse.h +++ b/src/Columns/ColumnSparse.h @@ -127,7 +127,7 @@ public: void protect() override; ColumnPtr replicate(const Offsets & replicate_offsets) const override; void updateHashWithValue(size_t n, SipHash & hash) const override; - void updateWeakHash32(WeakHash32 & hash) const override; + WeakHash32 getWeakHash32() const override; void updateHashFast(SipHash & hash) const override; void getExtremes(Field & min, Field & max) const override; diff --git a/src/Columns/ColumnString.cpp b/src/Columns/ColumnString.cpp index a84aea73486..1bab56e6d50 100644 --- a/src/Columns/ColumnString.cpp +++ b/src/Columns/ColumnString.cpp @@ -104,13 +104,10 @@ MutableColumnPtr ColumnString::cloneResized(size_t to_size) const return res; } -void ColumnString::updateWeakHash32(WeakHash32 & hash) const +WeakHash32 ColumnString::getWeakHash32() const { auto s = offsets.size(); - - if (hash.getData().size() != s) - throw Exception(ErrorCodes::LOGICAL_ERROR, "Size of WeakHash32 does not match size of column: " - "column size is {}, hash size is {}", std::to_string(s), std::to_string(hash.getData().size())); + WeakHash32 hash(s); const UInt8 * pos = chars.data(); UInt32 * hash_data = hash.getData().data(); @@ -126,6 +123,8 @@ void ColumnString::updateWeakHash32(WeakHash32 & hash) const prev_offset = offset; ++hash_data; } + + return hash; } diff --git a/src/Columns/ColumnString.h b/src/Columns/ColumnString.h index 39d4684fd89..dec64a23c53 100644 --- a/src/Columns/ColumnString.h +++ b/src/Columns/ColumnString.h @@ -204,7 +204,7 @@ public: hash.update(reinterpret_cast(&chars[offset]), string_size); } - void updateWeakHash32(WeakHash32 & hash) const override; + WeakHash32 getWeakHash32() const override; void updateHashFast(SipHash & hash) const override { diff --git a/src/Columns/ColumnTuple.cpp b/src/Columns/ColumnTuple.cpp index 2159495b68f..c9ad1cd5836 100644 --- a/src/Columns/ColumnTuple.cpp +++ b/src/Columns/ColumnTuple.cpp @@ -300,16 +300,15 @@ void ColumnTuple::updateHashWithValue(size_t n, SipHash & hash) const column->updateHashWithValue(n, hash); } -void ColumnTuple::updateWeakHash32(WeakHash32 & hash) const +WeakHash32 ColumnTuple::getWeakHash32() const { auto s = size(); - - if (hash.getData().size() != s) - throw Exception(ErrorCodes::LOGICAL_ERROR, "Size of WeakHash32 does not match size of column: " - "column size is {}, hash size is {}", std::to_string(s), std::to_string(hash.getData().size())); + WeakHash32 hash(s); for (const auto & column : columns) - column->updateWeakHash32(hash); + hash.update(column->getWeakHash32()); + + return hash; } void ColumnTuple::updateHashFast(SipHash & hash) const diff --git a/src/Columns/ColumnTuple.h b/src/Columns/ColumnTuple.h index 0103f81b242..e553d3dc85b 100644 --- a/src/Columns/ColumnTuple.h +++ b/src/Columns/ColumnTuple.h @@ -74,7 +74,7 @@ public: const char * deserializeAndInsertFromArena(const char * pos) override; const char * skipSerializedInArena(const char * pos) const override; void updateHashWithValue(size_t n, SipHash & hash) const override; - void updateWeakHash32(WeakHash32 & hash) const override; + WeakHash32 getWeakHash32() const override; void updateHashFast(SipHash & hash) const override; void insertRangeFrom(const IColumn & src, size_t start, size_t length) override; ColumnPtr filter(const Filter & filt, ssize_t result_size_hint) const override; diff --git a/src/Columns/ColumnVariant.cpp b/src/Columns/ColumnVariant.cpp index ec47f5dfa74..fe8775e6aac 100644 --- a/src/Columns/ColumnVariant.cpp +++ b/src/Columns/ColumnVariant.cpp @@ -777,36 +777,26 @@ void ColumnVariant::updateHashWithValue(size_t n, SipHash & hash) const variants[localDiscriminatorByGlobal(global_discr)]->updateHashWithValue(offsetAt(n), hash); } -void ColumnVariant::updateWeakHash32(WeakHash32 & hash) const +WeakHash32 ColumnVariant::getWeakHash32() const { auto s = size(); - if (hash.getData().size() != s) - throw Exception(ErrorCodes::LOGICAL_ERROR, "Size of WeakHash32 does not match size of column: " - "column size is {}, hash size is {}", std::to_string(s), std::to_string(hash.getData().size())); - /// If we have only NULLs, keep hash unchanged. if (hasOnlyNulls()) - return; + return WeakHash32(s); /// Optimization for case when there is only 1 non-empty variant and no NULLs. /// In this case we can just calculate weak hash for this variant. if (auto non_empty_local_discr = getLocalDiscriminatorOfOneNoneEmptyVariantNoNulls()) - { - variants[*non_empty_local_discr]->updateWeakHash32(hash); - return; - } + return variants[*non_empty_local_discr]->getWeakHash32(); /// Calculate weak hash for all variants. std::vector nested_hashes; for (const auto & variant : variants) - { - WeakHash32 nested_hash(variant->size()); - variant->updateWeakHash32(nested_hash); - nested_hashes.emplace_back(std::move(nested_hash)); - } + nested_hashes.emplace_back(variant->getWeakHash32()); /// For each row hash is a hash of corresponding row from corresponding variant. + WeakHash32 hash(s); auto & hash_data = hash.getData(); const auto & local_discriminators_data = getLocalDiscriminators(); const auto & offsets_data = getOffsets(); @@ -815,11 +805,10 @@ void ColumnVariant::updateWeakHash32(WeakHash32 & hash) const Discriminator discr = local_discriminators_data[i]; /// Update hash only for non-NULL values if (discr != NULL_DISCRIMINATOR) - { - auto nested_hash = nested_hashes[local_discriminators_data[i]].getData()[offsets_data[i]]; - hash_data[i] = static_cast(hashCRC32(nested_hash, hash_data[i])); - } + hash_data[i] = nested_hashes[discr].getData()[offsets_data[i]]; } + + return hash; } void ColumnVariant::updateHashFast(SipHash & hash) const diff --git a/src/Columns/ColumnVariant.h b/src/Columns/ColumnVariant.h index e5a4498f340..5ba77b54e0e 100644 --- a/src/Columns/ColumnVariant.h +++ b/src/Columns/ColumnVariant.h @@ -203,7 +203,7 @@ public: const char * deserializeVariantAndInsertFromArena(Discriminator global_discr, const char * pos); const char * skipSerializedInArena(const char * pos) const override; void updateHashWithValue(size_t n, SipHash & hash) const override; - void updateWeakHash32(WeakHash32 & hash) const override; + WeakHash32 getWeakHash32() const override; void updateHashFast(SipHash & hash) const override; ColumnPtr filter(const Filter & filt, ssize_t result_size_hint) const override; void expand(const Filter & mask, bool inverted) override; diff --git a/src/Columns/ColumnVector.cpp b/src/Columns/ColumnVector.cpp index 35d9f5386ed..314516f207a 100644 --- a/src/Columns/ColumnVector.cpp +++ b/src/Columns/ColumnVector.cpp @@ -73,13 +73,10 @@ void ColumnVector::updateHashWithValue(size_t n, SipHash & hash) const } template -void ColumnVector::updateWeakHash32(WeakHash32 & hash) const +WeakHash32 ColumnVector::getWeakHash32() const { auto s = data.size(); - - if (hash.getData().size() != s) - throw Exception(ErrorCodes::LOGICAL_ERROR, "Size of WeakHash32 does not match size of column: " - "column size is {}, hash size is {}", std::to_string(s), std::to_string(hash.getData().size())); + WeakHash32 hash(s); const T * begin = data.data(); const T * end = begin + s; @@ -91,6 +88,8 @@ void ColumnVector::updateWeakHash32(WeakHash32 & hash) const ++begin; ++hash_data; } + + return hash; } template diff --git a/src/Columns/ColumnVector.h b/src/Columns/ColumnVector.h index bbd27c91a70..e9903278f36 100644 --- a/src/Columns/ColumnVector.h +++ b/src/Columns/ColumnVector.h @@ -106,7 +106,7 @@ public: void updateHashWithValue(size_t n, SipHash & hash) const override; - void updateWeakHash32(WeakHash32 & hash) const override; + WeakHash32 getWeakHash32() const override; void updateHashFast(SipHash & hash) const override; diff --git a/src/Columns/IColumn.h b/src/Columns/IColumn.h index afa301d5c1c..a00b6c2a510 100644 --- a/src/Columns/IColumn.h +++ b/src/Columns/IColumn.h @@ -277,10 +277,10 @@ public: /// passed bytes to hash must identify sequence of values unambiguously. virtual void updateHashWithValue(size_t n, SipHash & hash) const = 0; - /// Update hash function value. Hash is calculated for each element. + /// Get hash function value. Hash is calculated for each element. /// It's a fast weak hash function. Mainly need to scatter data between threads. /// WeakHash32 must have the same size as column. - virtual void updateWeakHash32(WeakHash32 & hash) const = 0; + virtual WeakHash32 getWeakHash32() const = 0; /// Update state of hash with all column. virtual void updateHashFast(SipHash & hash) const = 0; diff --git a/src/Columns/IColumnDummy.h b/src/Columns/IColumnDummy.h index 27f420fbc71..deddd02c274 100644 --- a/src/Columns/IColumnDummy.h +++ b/src/Columns/IColumnDummy.h @@ -1,6 +1,7 @@ #pragma once #include +#include namespace DB @@ -59,8 +60,9 @@ public: { } - void updateWeakHash32(WeakHash32 & /*hash*/) const override + WeakHash32 getWeakHash32() const override { + return WeakHash32(s); } void updateHashFast(SipHash & /*hash*/) const override diff --git a/src/Columns/IColumnUnique.h b/src/Columns/IColumnUnique.h index f71f19a5da6..ba5be670f87 100644 --- a/src/Columns/IColumnUnique.h +++ b/src/Columns/IColumnUnique.h @@ -1,6 +1,7 @@ #pragma once #include #include +#include namespace DB { @@ -162,9 +163,9 @@ public: throw Exception(ErrorCodes::NOT_IMPLEMENTED, "Method scatter is not supported for ColumnUnique."); } - void updateWeakHash32(WeakHash32 &) const override + WeakHash32 getWeakHash32() const override { - throw Exception(ErrorCodes::NOT_IMPLEMENTED, "Method updateWeakHash32 is not supported for ColumnUnique."); + throw Exception(ErrorCodes::NOT_IMPLEMENTED, "Method getWeakHash32 is not supported for ColumnUnique."); } void updateHashFast(SipHash &) const override diff --git a/src/Columns/tests/gtest_weak_hash_32.cpp b/src/Columns/tests/gtest_weak_hash_32.cpp index 2c95998761b..3143d0ff83c 100644 --- a/src/Columns/tests/gtest_weak_hash_32.cpp +++ b/src/Columns/tests/gtest_weak_hash_32.cpp @@ -60,8 +60,7 @@ TEST(WeakHash32, ColumnVectorU8) data.push_back(i); } - WeakHash32 hash(col->size()); - col->updateWeakHash32(hash); + WeakHash32 hash = col->getWeakHash32(); checkColumn(hash.getData(), col->getData()); } @@ -77,8 +76,7 @@ TEST(WeakHash32, ColumnVectorI8) data.push_back(i); } - WeakHash32 hash(col->size()); - col->updateWeakHash32(hash); + WeakHash32 hash = col->getWeakHash32(); checkColumn(hash.getData(), col->getData()); } @@ -94,8 +92,7 @@ TEST(WeakHash32, ColumnVectorU16) data.push_back(i); } - WeakHash32 hash(col->size()); - col->updateWeakHash32(hash); + WeakHash32 hash = col->getWeakHash32(); checkColumn(hash.getData(), col->getData()); } @@ -111,8 +108,7 @@ TEST(WeakHash32, ColumnVectorI16) data.push_back(i); } - WeakHash32 hash(col->size()); - col->updateWeakHash32(hash); + WeakHash32 hash = col->getWeakHash32(); checkColumn(hash.getData(), col->getData()); } @@ -128,8 +124,7 @@ TEST(WeakHash32, ColumnVectorU32) data.push_back(i << 16u); } - WeakHash32 hash(col->size()); - col->updateWeakHash32(hash); + WeakHash32 hash = col->getWeakHash32(); checkColumn(hash.getData(), col->getData()); } @@ -145,8 +140,7 @@ TEST(WeakHash32, ColumnVectorI32) data.push_back(i << 16); } - WeakHash32 hash(col->size()); - col->updateWeakHash32(hash); + WeakHash32 hash = col->getWeakHash32(); checkColumn(hash.getData(), col->getData()); } @@ -162,8 +156,7 @@ TEST(WeakHash32, ColumnVectorU64) data.push_back(i << 32u); } - WeakHash32 hash(col->size()); - col->updateWeakHash32(hash); + WeakHash32 hash = col->getWeakHash32(); checkColumn(hash.getData(), col->getData()); } @@ -179,8 +172,7 @@ TEST(WeakHash32, ColumnVectorI64) data.push_back(i << 32); } - WeakHash32 hash(col->size()); - col->updateWeakHash32(hash); + WeakHash32 hash = col->getWeakHash32(); checkColumn(hash.getData(), col->getData()); } @@ -204,8 +196,7 @@ TEST(WeakHash32, ColumnVectorU128) } } - WeakHash32 hash(col->size()); - col->updateWeakHash32(hash); + WeakHash32 hash = col->getWeakHash32(); checkColumn(hash.getData(), eq_data); } @@ -221,8 +212,7 @@ TEST(WeakHash32, ColumnVectorI128) data.push_back(i << 32); } - WeakHash32 hash(col->size()); - col->updateWeakHash32(hash); + WeakHash32 hash = col->getWeakHash32(); checkColumn(hash.getData(), col->getData()); } @@ -238,8 +228,7 @@ TEST(WeakHash32, ColumnDecimal32) data.push_back(i << 16); } - WeakHash32 hash(col->size()); - col->updateWeakHash32(hash); + WeakHash32 hash = col->getWeakHash32(); checkColumn(hash.getData(), col->getData()); } @@ -255,8 +244,7 @@ TEST(WeakHash32, ColumnDecimal64) data.push_back(i << 32); } - WeakHash32 hash(col->size()); - col->updateWeakHash32(hash); + WeakHash32 hash = col->getWeakHash32(); checkColumn(hash.getData(), col->getData()); } @@ -272,8 +260,7 @@ TEST(WeakHash32, ColumnDecimal128) data.push_back(i << 32); } - WeakHash32 hash(col->size()); - col->updateWeakHash32(hash); + WeakHash32 hash = col->getWeakHash32(); checkColumn(hash.getData(), col->getData()); } @@ -294,8 +281,7 @@ TEST(WeakHash32, ColumnString1) } } - WeakHash32 hash(col->size()); - col->updateWeakHash32(hash); + WeakHash32 hash = col->getWeakHash32(); checkColumn(hash.getData(), data); } @@ -331,8 +317,7 @@ TEST(WeakHash32, ColumnString2) } } - WeakHash32 hash(col->size()); - col->updateWeakHash32(hash); + WeakHash32 hash = col->getWeakHash32(); checkColumn(hash.getData(), data); } @@ -369,8 +354,7 @@ TEST(WeakHash32, ColumnString3) } } - WeakHash32 hash(col->size()); - col->updateWeakHash32(hash); + WeakHash32 hash = col->getWeakHash32(); checkColumn(hash.getData(), data); } @@ -397,8 +381,7 @@ TEST(WeakHash32, ColumnFixedString) } } - WeakHash32 hash(col->size()); - col->updateWeakHash32(hash); + WeakHash32 hash = col->getWeakHash32(); checkColumn(hash.getData(), data); } @@ -444,8 +427,7 @@ TEST(WeakHash32, ColumnArray) auto col_arr = ColumnArray::create(std::move(val), std::move(off)); - WeakHash32 hash(col_arr->size()); - col_arr->updateWeakHash32(hash); + WeakHash32 hash = col_arr->getWeakHash32(); checkColumn(hash.getData(), eq_data); } @@ -479,8 +461,7 @@ TEST(WeakHash32, ColumnArray2) auto col_arr = ColumnArray::create(std::move(val), std::move(off)); - WeakHash32 hash(col_arr->size()); - col_arr->updateWeakHash32(hash); + WeakHash32 hash = col_arr->getWeakHash32(); checkColumn(hash.getData(), eq_data); } @@ -536,8 +517,7 @@ TEST(WeakHash32, ColumnArrayArray) auto col_arr = ColumnArray::create(std::move(val), std::move(off)); auto col_arr_arr = ColumnArray::create(std::move(col_arr), std::move(off2)); - WeakHash32 hash(col_arr_arr->size()); - col_arr_arr->updateWeakHash32(hash); + WeakHash32 hash = col_arr_arr->getWeakHash32(); checkColumn(hash.getData(), eq_data); } @@ -555,8 +535,7 @@ TEST(WeakHash32, ColumnConst) auto col_const = ColumnConst::create(std::move(inner_col), 256); - WeakHash32 hash(col_const->size()); - col_const->updateWeakHash32(hash); + WeakHash32 hash = col_const->getWeakHash32(); checkColumn(hash.getData(), data); } @@ -576,8 +555,7 @@ TEST(WeakHash32, ColumnLowcardinality) } } - WeakHash32 hash(col->size()); - col->updateWeakHash32(hash); + WeakHash32 hash = col->getWeakHash32(); checkColumn(hash.getData(), data); } @@ -602,8 +580,7 @@ TEST(WeakHash32, ColumnNullable) auto col_null = ColumnNullable::create(std::move(col), std::move(mask)); - WeakHash32 hash(col_null->size()); - col_null->updateWeakHash32(hash); + WeakHash32 hash = col_null->getWeakHash32(); checkColumn(hash.getData(), eq); } @@ -633,8 +610,7 @@ TEST(WeakHash32, ColumnTupleUInt64UInt64) columns.emplace_back(std::move(col2)); auto col_tuple = ColumnTuple::create(std::move(columns)); - WeakHash32 hash(col_tuple->size()); - col_tuple->updateWeakHash32(hash); + WeakHash32 hash = col_tuple->getWeakHash32(); checkColumn(hash.getData(), eq); } @@ -671,8 +647,7 @@ TEST(WeakHash32, ColumnTupleUInt64String) columns.emplace_back(std::move(col2)); auto col_tuple = ColumnTuple::create(std::move(columns)); - WeakHash32 hash(col_tuple->size()); - col_tuple->updateWeakHash32(hash); + WeakHash32 hash = col_tuple->getWeakHash32(); checkColumn(hash.getData(), eq); } @@ -709,8 +684,7 @@ TEST(WeakHash32, ColumnTupleUInt64FixedString) columns.emplace_back(std::move(col2)); auto col_tuple = ColumnTuple::create(std::move(columns)); - WeakHash32 hash(col_tuple->size()); - col_tuple->updateWeakHash32(hash); + WeakHash32 hash = col_tuple->getWeakHash32(); checkColumn(hash.getData(), eq); } @@ -756,8 +730,7 @@ TEST(WeakHash32, ColumnTupleUInt64Array) columns.emplace_back(ColumnArray::create(std::move(val), std::move(off))); auto col_tuple = ColumnTuple::create(std::move(columns)); - WeakHash32 hash(col_tuple->size()); - col_tuple->updateWeakHash32(hash); + WeakHash32 hash = col_tuple->getWeakHash32(); checkColumn(hash.getData(), eq_data); } diff --git a/src/Common/WeakHash.cpp b/src/Common/WeakHash.cpp index 54d973b6296..cb12df84db1 100644 --- a/src/Common/WeakHash.cpp +++ b/src/Common/WeakHash.cpp @@ -1,2 +1,24 @@ #include +#include +#include +namespace DB +{ + +namespace ErrorCodes +{ + extern const int LOGICAL_ERROR; +} + +void WeakHash32::update(const WeakHash32 & other) +{ + size_t size = data.size(); + if (size != other.data.size()) + throw Exception(ErrorCodes::LOGICAL_ERROR, "Size of WeakHash32 does not match:" + "left size is {}, right size is {}", size, other.data.size()); + + for (size_t i = 0; i < size; ++i) + data[i] = static_cast(intHashCRC32(other.data[i], data[i])); +} + +} diff --git a/src/Common/WeakHash.h b/src/Common/WeakHash.h index b59624e64f2..d4a8d63868c 100644 --- a/src/Common/WeakHash.h +++ b/src/Common/WeakHash.h @@ -11,9 +11,8 @@ namespace DB /// The main purpose why this class needed is to support data initialization. Initially, every bit is 1. class WeakHash32 { - static constexpr UInt32 kDefaultInitialValue = ~UInt32(0); - public: + static constexpr UInt32 kDefaultInitialValue = ~UInt32(0); using Container = PaddedPODArray; @@ -22,6 +21,8 @@ public: void reset(size_t size, UInt32 initial_value = kDefaultInitialValue) { data.assign(size, initial_value); } + void update(const WeakHash32 & other); + const Container & getData() const { return data; } Container & getData() { return data; } diff --git a/src/Interpreters/ConcurrentHashJoin.cpp b/src/Interpreters/ConcurrentHashJoin.cpp index 53987694e46..8426fe1a154 100644 --- a/src/Interpreters/ConcurrentHashJoin.cpp +++ b/src/Interpreters/ConcurrentHashJoin.cpp @@ -271,7 +271,7 @@ IColumn::Selector ConcurrentHashJoin::selectDispatchBlock(const Strings & key_co { const auto & key_col = from_block.getByName(key_name).column->convertToFullColumnIfConst(); const auto & key_col_no_lc = recursiveRemoveLowCardinality(recursiveRemoveSparse(key_col)); - key_col_no_lc->updateWeakHash32(hash); + hash.update(key_col_no_lc->getWeakHash32()); } return hashToSelector(hash, num_shards); } diff --git a/src/Interpreters/JoinUtils.cpp b/src/Interpreters/JoinUtils.cpp index 1788c9aca48..180a45d4295 100644 --- a/src/Interpreters/JoinUtils.cpp +++ b/src/Interpreters/JoinUtils.cpp @@ -554,7 +554,7 @@ static Blocks scatterBlockByHashImpl(const Strings & key_columns_names, const Bl for (const auto & key_name : key_columns_names) { ColumnPtr key_col = materializeColumn(block, key_name); - key_col->updateWeakHash32(hash); + hash.update(key_col->getWeakHash32()); } auto selector = hashToSelector(hash, sharder); diff --git a/src/Processors/Transforms/ScatterByPartitionTransform.cpp b/src/Processors/Transforms/ScatterByPartitionTransform.cpp index 6e3cdc0fda1..16d265c9bcb 100644 --- a/src/Processors/Transforms/ScatterByPartitionTransform.cpp +++ b/src/Processors/Transforms/ScatterByPartitionTransform.cpp @@ -109,7 +109,7 @@ void ScatterByPartitionTransform::generateOutputChunks() hash.reset(num_rows); for (const auto & column_number : key_columns) - columns[column_number]->updateWeakHash32(hash); + hash.update(columns[column_number]->getWeakHash32()); const auto & hash_data = hash.getData(); IColumn::Selector selector(num_rows); diff --git a/tests/queries/0_stateless/03205_parallel_window_finctions_and_column_sparse_bug.reference b/tests/queries/0_stateless/03205_parallel_window_finctions_and_column_sparse_bug.reference new file mode 100644 index 00000000000..13e229432ae --- /dev/null +++ b/tests/queries/0_stateless/03205_parallel_window_finctions_and_column_sparse_bug.reference @@ -0,0 +1,10 @@ +false 1 1 +true 1 1 +--- +false 1 1 +false 1 2 +false 1 3 +true 1 1 +true 1 2 +--- +-755809149 0 diff --git a/tests/queries/0_stateless/03205_parallel_window_finctions_and_column_sparse_bug.sql b/tests/queries/0_stateless/03205_parallel_window_finctions_and_column_sparse_bug.sql new file mode 100644 index 00000000000..4cc54bb5ac2 --- /dev/null +++ b/tests/queries/0_stateless/03205_parallel_window_finctions_and_column_sparse_bug.sql @@ -0,0 +1,33 @@ +create table t(c Int32, d Bool) Engine=MergeTree order by c; +system stop merges t; + +insert into t values (1, 0); +insert into t values (1, 0); +insert into t values (1, 1); +insert into t values (1, 0)(1, 1); + +SELECT d, c, row_number() over (partition by d order by c) as c8 FROM t qualify c8=1 order by d settings max_threads=2, allow_experimental_analyzer = 1; +SELECT '---'; +SELECT d, c, row_number() over (partition by d order by c) as c8 FROM t order by d, c8 settings max_threads=2; +SELECT '---'; + +drop table t; + +create table t ( + c Int32 primary key , + s Bool , + w Float64 + ); + +system stop merges t; + +insert into t values(439499072,true,0),(1393290072,true,0); +insert into t values(-1317193174,false,0),(1929066636,false,0); +insert into t values(-2,false,0),(1962246186,true,0),(2054878592,false,0); +insert into t values(-1893563136,true,41.55); +insert into t values(-1338380855,true,-0.7),(-991301833,true,0),(-755809149,false,43.18),(-41,true,0),(3,false,0),(255,false,0),(255,false,0),(189195893,false,0),(195550885,false,9223372036854776000); + +SELECT * FROM ( +SELECT c, min(w) OVER (PARTITION BY s ORDER BY c ASC, s ASC, w ASC) +FROM t limit toUInt64(-1)) +WHERE c = -755809149;