1 | #include <zlib.h> |
2 | #include <DataTypes/DataTypeString.h> |
3 | #include <Functions/FunctionFactory.h> |
4 | #include <Functions/FunctionStringOrArrayToT.h> |
5 | |
6 | |
7 | namespace |
8 | { |
9 | |
10 | template <class T> |
11 | struct CRCBase |
12 | { |
13 | T tab[256]; |
14 | CRCBase(T polynomial) |
15 | { |
16 | for (size_t i = 0; i < 256; ++i) |
17 | { |
18 | T c = i; |
19 | for (size_t j = 0; j < 8; ++j) |
20 | { |
21 | c = c & 1 ? polynomial ^ (c >> 1) : c >> 1; |
22 | } |
23 | tab[i] = c; |
24 | } |
25 | } |
26 | }; |
27 | |
28 | template <class T, T polynomial> |
29 | struct CRCImpl |
30 | { |
31 | using ReturnType = T; |
32 | |
33 | static T make_crc(const unsigned char *buf, size_t size) |
34 | { |
35 | static CRCBase<ReturnType> base(polynomial); |
36 | |
37 | T i, crc; |
38 | |
39 | crc = 0; |
40 | for (i = 0; i < size; i++) |
41 | { |
42 | crc = base.tab[(crc ^ buf[i]) & 0xff] ^ (crc >> 8); |
43 | } |
44 | return crc; |
45 | } |
46 | }; |
47 | |
48 | static constexpr UInt64 CRC64_ECMA = 0xc96c5795d7870f42ULL; |
49 | struct CRC64ECMAImpl : public CRCImpl<UInt64, CRC64_ECMA> |
50 | { |
51 | static constexpr auto name = "CRC64" ; |
52 | }; |
53 | |
54 | static constexpr UInt32 CRC32_IEEE = 0xedb88320; |
55 | struct CRC32IEEEImpl : public CRCImpl<UInt32, CRC32_IEEE> |
56 | { |
57 | static constexpr auto name = "CRC32IEEE" ; |
58 | }; |
59 | |
60 | struct CRC32ZLIBImpl |
61 | { |
62 | using ReturnType = UInt32; |
63 | static constexpr auto name = "CRC32" ; |
64 | |
65 | static UInt32 make_crc(const unsigned char *buf, size_t size) |
66 | { return crc32_z(0L, buf, size); } |
67 | }; |
68 | |
69 | } // \anonymous |
70 | |
71 | namespace DB |
72 | { |
73 | |
74 | namespace ErrorCodes |
75 | { |
76 | extern const int ILLEGAL_TYPE_OF_ARGUMENT; |
77 | } |
78 | |
79 | template <class Impl> |
80 | struct CRCFunctionWrapper |
81 | { |
82 | static constexpr auto is_fixed_to_constant = true; |
83 | using ReturnType = typename Impl::ReturnType; |
84 | |
85 | static void vector(const ColumnString::Chars & data, const ColumnString::Offsets & offsets, PaddedPODArray<ReturnType> & res) |
86 | { |
87 | size_t size = offsets.size(); |
88 | |
89 | ColumnString::Offset prev_offset = 0; |
90 | for (size_t i = 0; i < size; ++i) |
91 | { |
92 | res[i] = do_crc(data, prev_offset, offsets[i] - prev_offset - 1); |
93 | prev_offset = offsets[i]; |
94 | } |
95 | } |
96 | |
97 | static void vector_fixed_to_constant(const ColumnString::Chars & data, size_t n, ReturnType & res) { res = do_crc(data, 0, n); } |
98 | |
99 | static void vector_fixed_to_vector(const ColumnString::Chars & data, size_t n, PaddedPODArray<ReturnType> & res) |
100 | { |
101 | size_t size = data.size() / n; |
102 | |
103 | for (size_t i = 0; i < size; ++i) |
104 | { |
105 | res[i] = do_crc(data, i * n, n); |
106 | } |
107 | } |
108 | |
109 | [[noreturn]] static void array(const ColumnString::Offsets & /*offsets*/, PaddedPODArray<ReturnType> & /*res*/) |
110 | { |
111 | throw Exception("Cannot apply function " + std::string(Impl::name) + " to Array argument" , ErrorCodes::ILLEGAL_TYPE_OF_ARGUMENT); |
112 | } |
113 | |
114 | private: |
115 | static ReturnType do_crc(const ColumnString::Chars & buf, size_t offset, size_t size) |
116 | { |
117 | const unsigned char * p = reinterpret_cast<const unsigned char *>(&buf[0]) + offset; |
118 | return Impl::make_crc(p, size); |
119 | } |
120 | }; |
121 | |
122 | template <class T> |
123 | using FunctionCRC = FunctionStringOrArrayToT<CRCFunctionWrapper<T>, T, typename T::ReturnType>; |
124 | // The same as IEEE variant, but uses 0xffffffff as initial value |
125 | // This is the default |
126 | // |
127 | // (And zlib is used here, since it has optimized version) |
128 | using FunctionCRC32ZLIB = FunctionCRC<CRC32ZLIBImpl>; |
129 | // Uses CRC-32-IEEE 802.3 polynomial |
130 | using FunctionCRC32IEEE = FunctionCRC<CRC32IEEEImpl>; |
131 | // Uses CRC-64-ECMA polynomial |
132 | using FunctionCRC64ECMA = FunctionCRC<CRC64ECMAImpl>; |
133 | |
134 | template <class T> |
135 | void registerFunctionCRCImpl(FunctionFactory & factory) |
136 | { |
137 | factory.registerFunction<T>(T::name, FunctionFactory::CaseInsensitive); |
138 | } |
139 | |
140 | void registerFunctionCRC(FunctionFactory & factory) |
141 | { |
142 | registerFunctionCRCImpl<FunctionCRC32ZLIB>(factory); |
143 | registerFunctionCRCImpl<FunctionCRC32IEEE>(factory); |
144 | registerFunctionCRCImpl<FunctionCRC64ECMA>(factory); |
145 | } |
146 | |
147 | } |
148 | |