1 | #include <Common/hex.h> |
2 | #include <Functions/FunctionFactory.h> |
3 | #include <Functions/FunctionStringToString.h> |
4 | #include <common/find_symbols.h> |
5 | |
6 | |
7 | namespace DB |
8 | { |
9 | |
10 | /// We assume that size of the dst buf isn't less than src_size. |
11 | static size_t decodeURL(const char * src, size_t src_size, char * dst) |
12 | { |
13 | const char * src_prev_pos = src; |
14 | const char * src_curr_pos = src; |
15 | const char * src_end = src + src_size; |
16 | char * dst_pos = dst; |
17 | |
18 | while (true) |
19 | { |
20 | src_curr_pos = find_first_symbols<'%'>(src_curr_pos, src_end); |
21 | |
22 | if (src_curr_pos == src_end) |
23 | { |
24 | break; |
25 | } |
26 | else if (src_end - src_curr_pos < 3) |
27 | { |
28 | src_curr_pos = src_end; |
29 | break; |
30 | } |
31 | else |
32 | { |
33 | unsigned char high = unhex(src_curr_pos[1]); |
34 | unsigned char low = unhex(src_curr_pos[2]); |
35 | |
36 | if (high != 0xFF && low != 0xFF) |
37 | { |
38 | unsigned char octet = (high << 4) + low; |
39 | |
40 | size_t bytes_to_copy = src_curr_pos - src_prev_pos; |
41 | memcpySmallAllowReadWriteOverflow15(dst_pos, src_prev_pos, bytes_to_copy); |
42 | dst_pos += bytes_to_copy; |
43 | |
44 | *dst_pos = octet; |
45 | ++dst_pos; |
46 | |
47 | src_prev_pos = src_curr_pos + 3; |
48 | } |
49 | |
50 | src_curr_pos += 3; |
51 | } |
52 | } |
53 | |
54 | if (src_prev_pos < src_curr_pos) |
55 | { |
56 | size_t bytes_to_copy = src_curr_pos - src_prev_pos; |
57 | memcpySmallAllowReadWriteOverflow15(dst_pos, src_prev_pos, bytes_to_copy); |
58 | dst_pos += bytes_to_copy; |
59 | } |
60 | |
61 | return dst_pos - dst; |
62 | } |
63 | |
64 | |
65 | /// Percent decode of URL data. |
66 | struct DecodeURLComponentImpl |
67 | { |
68 | static void vector(const ColumnString::Chars & data, const ColumnString::Offsets & offsets, |
69 | ColumnString::Chars & res_data, ColumnString::Offsets & res_offsets) |
70 | { |
71 | res_data.resize(data.size()); |
72 | size_t size = offsets.size(); |
73 | res_offsets.resize(size); |
74 | |
75 | size_t prev_offset = 0; |
76 | size_t res_offset = 0; |
77 | |
78 | for (size_t i = 0; i < size; ++i) |
79 | { |
80 | const char * src_data = reinterpret_cast<const char *>(&data[prev_offset]); |
81 | size_t src_size = offsets[i] - prev_offset; |
82 | size_t dst_size = decodeURL(src_data, src_size, reinterpret_cast<char *>(res_data.data() + res_offset)); |
83 | |
84 | res_offset += dst_size; |
85 | res_offsets[i] = res_offset; |
86 | prev_offset = offsets[i]; |
87 | } |
88 | |
89 | res_data.resize(res_offset); |
90 | } |
91 | |
92 | [[noreturn]] static void vector_fixed(const ColumnString::Chars &, size_t, ColumnString::Chars &) |
93 | { |
94 | throw Exception("Column of type FixedString is not supported by URL functions" , ErrorCodes::ILLEGAL_COLUMN); |
95 | } |
96 | }; |
97 | |
98 | |
99 | struct NameDecodeURLComponent { static constexpr auto name = "decodeURLComponent" ; }; |
100 | using FunctionDecodeURLComponent = FunctionStringToString<DecodeURLComponentImpl, NameDecodeURLComponent>; |
101 | |
102 | void registerFunctionDecodeURLComponent(FunctionFactory & factory) |
103 | { |
104 | factory.registerFunction<FunctionDecodeURLComponent>(); |
105 | } |
106 | |
107 | } |
108 | |