| 1 | #pragma once |
| 2 | |
| 3 | #include <Common/PODArray.h> |
| 4 | #include <Common/memcmpSmall.h> |
| 5 | #include <Common/typeid_cast.h> |
| 6 | #include <Common/assert_cast.h> |
| 7 | #include <Columns/IColumn.h> |
| 8 | #include <Columns/IColumnImpl.h> |
| 9 | #include <Columns/ColumnVectorHelper.h> |
| 10 | #include <Core/Field.h> |
| 11 | |
| 12 | |
| 13 | namespace DB |
| 14 | { |
| 15 | |
| 16 | /** A column of values of "fixed-length string" type. |
| 17 | * If you insert a smaller string, it will be padded with zero bytes. |
| 18 | */ |
| 19 | class ColumnFixedString final : public COWHelper<ColumnVectorHelper, ColumnFixedString> |
| 20 | { |
| 21 | public: |
| 22 | friend class COWHelper<ColumnVectorHelper, ColumnFixedString>; |
| 23 | |
| 24 | using Chars = PaddedPODArray<UInt8>; |
| 25 | |
| 26 | private: |
| 27 | /// Bytes of rows, laid in succession. The strings are stored without a trailing zero byte. |
| 28 | /** NOTE It is required that the offset and type of chars in the object be the same as that of `data in ColumnUInt8`. |
| 29 | * Used in `packFixed` function (AggregationCommon.h) |
| 30 | */ |
| 31 | Chars chars; |
| 32 | /// The size of the rows. |
| 33 | const size_t n; |
| 34 | |
| 35 | template <bool positive> |
| 36 | struct less; |
| 37 | |
| 38 | /** Create an empty column of strings of fixed-length `n` */ |
| 39 | ColumnFixedString(size_t n_) : n(n_) {} |
| 40 | |
| 41 | ColumnFixedString(const ColumnFixedString & src) : chars(src.chars.begin(), src.chars.end()), n(src.n) {} |
| 42 | |
| 43 | public: |
| 44 | std::string getName() const override { return "FixedString(" + std::to_string(n) + ")" ; } |
| 45 | const char * getFamilyName() const override { return "FixedString" ; } |
| 46 | |
| 47 | MutableColumnPtr cloneResized(size_t size) const override; |
| 48 | |
| 49 | size_t size() const override |
| 50 | { |
| 51 | return chars.size() / n; |
| 52 | } |
| 53 | |
| 54 | size_t byteSize() const override |
| 55 | { |
| 56 | return chars.size() + sizeof(n); |
| 57 | } |
| 58 | |
| 59 | size_t allocatedBytes() const override |
| 60 | { |
| 61 | return chars.allocated_bytes() + sizeof(n); |
| 62 | } |
| 63 | |
| 64 | void protect() override |
| 65 | { |
| 66 | chars.protect(); |
| 67 | } |
| 68 | |
| 69 | Field operator[](size_t index) const override |
| 70 | { |
| 71 | return String(reinterpret_cast<const char *>(&chars[n * index]), n); |
| 72 | } |
| 73 | |
| 74 | void get(size_t index, Field & res) const override |
| 75 | { |
| 76 | res.assignString(reinterpret_cast<const char *>(&chars[n * index]), n); |
| 77 | } |
| 78 | |
| 79 | StringRef getDataAt(size_t index) const override |
| 80 | { |
| 81 | return StringRef(&chars[n * index], n); |
| 82 | } |
| 83 | |
| 84 | void insert(const Field & x) override; |
| 85 | |
| 86 | void insertFrom(const IColumn & src_, size_t index) override; |
| 87 | |
| 88 | void insertData(const char * pos, size_t length) override; |
| 89 | |
| 90 | void insertDefault() override |
| 91 | { |
| 92 | chars.resize_fill(chars.size() + n); |
| 93 | } |
| 94 | |
| 95 | virtual void insertManyDefaults(size_t length) override |
| 96 | { |
| 97 | chars.resize_fill(chars.size() + n * length); |
| 98 | } |
| 99 | |
| 100 | void popBack(size_t elems) override |
| 101 | { |
| 102 | chars.resize_assume_reserved(chars.size() - n * elems); |
| 103 | } |
| 104 | |
| 105 | StringRef serializeValueIntoArena(size_t index, Arena & arena, char const *& begin) const override; |
| 106 | |
| 107 | const char * deserializeAndInsertFromArena(const char * pos) override; |
| 108 | |
| 109 | void updateHashWithValue(size_t index, SipHash & hash) const override; |
| 110 | |
| 111 | int compareAt(size_t p1, size_t p2, const IColumn & rhs_, int /*nan_direction_hint*/) const override |
| 112 | { |
| 113 | const ColumnFixedString & rhs = assert_cast<const ColumnFixedString &>(rhs_); |
| 114 | return memcmpSmallAllowOverflow15(chars.data() + p1 * n, rhs.chars.data() + p2 * n, n); |
| 115 | } |
| 116 | |
| 117 | void getPermutation(bool reverse, size_t limit, int nan_direction_hint, Permutation & res) const override; |
| 118 | |
| 119 | void insertRangeFrom(const IColumn & src, size_t start, size_t length) override; |
| 120 | |
| 121 | ColumnPtr filter(const IColumn::Filter & filt, ssize_t result_size_hint) const override; |
| 122 | |
| 123 | ColumnPtr permute(const Permutation & perm, size_t limit) const override; |
| 124 | |
| 125 | ColumnPtr index(const IColumn & indexes, size_t limit) const override; |
| 126 | |
| 127 | template <typename Type> |
| 128 | ColumnPtr indexImpl(const PaddedPODArray<Type> & indexes, size_t limit) const; |
| 129 | |
| 130 | ColumnPtr replicate(const Offsets & offsets) const override; |
| 131 | |
| 132 | MutableColumns scatter(ColumnIndex num_columns, const Selector & selector) const override |
| 133 | { |
| 134 | return scatterImpl<ColumnFixedString>(num_columns, selector); |
| 135 | } |
| 136 | |
| 137 | void gather(ColumnGathererStream & gatherer_stream) override; |
| 138 | |
| 139 | void reserve(size_t size) override |
| 140 | { |
| 141 | chars.reserve(n * size); |
| 142 | } |
| 143 | |
| 144 | void getExtremes(Field & min, Field & max) const override; |
| 145 | |
| 146 | bool structureEquals(const IColumn & rhs) const override |
| 147 | { |
| 148 | if (auto rhs_concrete = typeid_cast<const ColumnFixedString *>(&rhs)) |
| 149 | return n == rhs_concrete->n; |
| 150 | return false; |
| 151 | } |
| 152 | |
| 153 | bool canBeInsideNullable() const override { return true; } |
| 154 | |
| 155 | bool isFixedAndContiguous() const override { return true; } |
| 156 | size_t sizeOfValueIfFixed() const override { return n; } |
| 157 | StringRef getRawData() const override { return StringRef(chars.data(), chars.size()); } |
| 158 | |
| 159 | /// Specialized part of interface, not from IColumn. |
| 160 | void insertString(const String & string) { insertData(string.c_str(), string.size()); } |
| 161 | Chars & getChars() { return chars; } |
| 162 | const Chars & getChars() const { return chars; } |
| 163 | |
| 164 | size_t getN() const { return n; } |
| 165 | }; |
| 166 | |
| 167 | |
| 168 | } |
| 169 | |