1 | #include <DataTypes/DataTypeString.h> |
2 | #include <Columns/ColumnString.h> |
3 | #include <Functions/FunctionFactory.h> |
4 | #include <Functions/FunctionStringToString.h> |
5 | |
6 | |
7 | namespace DB |
8 | { |
9 | |
10 | namespace ErrorCodes |
11 | { |
12 | extern const int ILLEGAL_COLUMN; |
13 | } |
14 | |
15 | |
16 | /** Reverse the sequence of code points in a UTF-8 encoded string. |
17 | * The result may not match the expected result, because modifying code points (for example, diacritics) may be applied to another symbols. |
18 | * If the string is not encoded in UTF-8, then the behavior is undefined. |
19 | */ |
20 | struct ReverseUTF8Impl |
21 | { |
22 | static void vector(const ColumnString::Chars & data, |
23 | const ColumnString::Offsets & offsets, |
24 | ColumnString::Chars & res_data, |
25 | ColumnString::Offsets & res_offsets) |
26 | { |
27 | res_data.resize(data.size()); |
28 | res_offsets.assign(offsets); |
29 | size_t size = offsets.size(); |
30 | |
31 | ColumnString::Offset prev_offset = 0; |
32 | for (size_t i = 0; i < size; ++i) |
33 | { |
34 | ColumnString::Offset j = prev_offset; |
35 | while (j < offsets[i] - 1) |
36 | { |
37 | if (data[j] < 0xBF) |
38 | { |
39 | res_data[offsets[i] + prev_offset - 2 - j] = data[j]; |
40 | j += 1; |
41 | } |
42 | else if (data[j] < 0xE0) |
43 | { |
44 | memcpy(&res_data[offsets[i] + prev_offset - 2 - j - 1], &data[j], 2); |
45 | j += 2; |
46 | } |
47 | else if (data[j] < 0xF0) |
48 | { |
49 | memcpy(&res_data[offsets[i] + prev_offset - 2 - j - 2], &data[j], 3); |
50 | j += 3; |
51 | } |
52 | else |
53 | { |
54 | res_data[offsets[i] + prev_offset - 2 - j] = data[j]; |
55 | j += 1; |
56 | } |
57 | } |
58 | |
59 | res_data[offsets[i] - 1] = 0; |
60 | prev_offset = offsets[i]; |
61 | } |
62 | } |
63 | |
64 | [[noreturn]] static void vector_fixed(const ColumnString::Chars &, size_t, ColumnString::Chars &) |
65 | { |
66 | throw Exception("Cannot apply function reverseUTF8 to fixed string." , ErrorCodes::ILLEGAL_COLUMN); |
67 | } |
68 | }; |
69 | |
70 | struct NameReverseUTF8 |
71 | { |
72 | static constexpr auto name = "reverseUTF8" ; |
73 | }; |
74 | using FunctionReverseUTF8 = FunctionStringToString<ReverseUTF8Impl, NameReverseUTF8, true>; |
75 | |
76 | void registerFunctionReverseUTF8(FunctionFactory & factory) |
77 | { |
78 | factory.registerFunction<FunctionReverseUTF8>(); |
79 | } |
80 | |
81 | } |
82 | |