1 | #include <Functions/FunctionFactory.h> |
2 | #include <Functions/FunctionsStringSearch.h> |
3 | #include <common/find_symbols.h> |
4 | |
5 | namespace DB |
6 | { |
7 | |
8 | struct |
9 | { |
10 | static void (const ColumnString::Chars & data, |
11 | const ColumnString::Offsets & offsets, |
12 | std::string pattern, |
13 | ColumnString::Chars & res_data, ColumnString::Offsets & res_offsets) |
14 | { |
15 | res_data.reserve(data.size() / 5); |
16 | res_offsets.resize(offsets.size()); |
17 | |
18 | pattern += '='; |
19 | const char * param_str = pattern.c_str(); |
20 | size_t param_len = pattern.size(); |
21 | |
22 | ColumnString::Offset prev_offset = 0; |
23 | ColumnString::Offset res_offset = 0; |
24 | |
25 | for (size_t i = 0; i < offsets.size(); ++i) |
26 | { |
27 | ColumnString::Offset cur_offset = offsets[i]; |
28 | |
29 | const char * str = reinterpret_cast<const char *>(&data[prev_offset]); |
30 | const char * end = reinterpret_cast<const char *>(&data[cur_offset]); |
31 | |
32 | /// Find query string or fragment identifier. |
33 | /// Note that we support parameters in fragment identifier in the same way as in query string. |
34 | |
35 | const char * const query_string_begin = find_first_symbols<'?', '#'>(str, end); |
36 | |
37 | /// Will point to the beginning of "name=value" pair. Then it will be reassigned to the beginning of "value". |
38 | const char * param_begin = nullptr; |
39 | |
40 | if (query_string_begin + 1 < end) |
41 | { |
42 | param_begin = query_string_begin + 1; |
43 | |
44 | while (true) |
45 | { |
46 | param_begin = static_cast<const char *>(memmem(param_begin, end - param_begin, param_str, param_len)); |
47 | |
48 | if (!param_begin) |
49 | break; |
50 | |
51 | if (param_begin[-1] != '?' && param_begin[-1] != '#' && param_begin[-1] != '&') |
52 | { |
53 | /// Parameter name is different but has the same suffix. |
54 | param_begin += param_len; |
55 | continue; |
56 | } |
57 | else |
58 | { |
59 | param_begin += param_len; |
60 | break; |
61 | } |
62 | } |
63 | } |
64 | |
65 | if (param_begin) |
66 | { |
67 | const char * param_end = find_first_symbols<'&', '#'>(param_begin, end); |
68 | if (param_end == end) |
69 | param_end = param_begin + strlen(param_begin); |
70 | |
71 | size_t param_size = param_end - param_begin; |
72 | |
73 | res_data.resize(res_offset + param_size + 1); |
74 | memcpySmallAllowReadWriteOverflow15(&res_data[res_offset], param_begin, param_size); |
75 | res_offset += param_size; |
76 | } |
77 | else |
78 | { |
79 | /// No parameter found, put empty string in result. |
80 | res_data.resize(res_offset + 1); |
81 | } |
82 | |
83 | res_data[res_offset] = 0; |
84 | ++res_offset; |
85 | res_offsets[i] = res_offset; |
86 | |
87 | prev_offset = cur_offset; |
88 | } |
89 | } |
90 | }; |
91 | |
92 | struct { static constexpr auto = "extractURLParameter" ; }; |
93 | using = FunctionsStringSearchToString<ExtractURLParameterImpl, NameExtractURLParameter>; |
94 | |
95 | void (FunctionFactory & factory) |
96 | { |
97 | factory.registerFunction<FunctionExtractURLParameter>(); |
98 | } |
99 | |
100 | } |
101 | |