| 1 | #include <Functions/FunctionFactory.h> |
| 2 | #include <Functions/FunctionsStringSearch.h> |
| 3 | #include <common/find_symbols.h> |
| 4 | |
| 5 | namespace DB |
| 6 | { |
| 7 | |
| 8 | struct CutURLParameterImpl |
| 9 | { |
| 10 | static void vector(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()); |
| 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 | size_t prev_offset = 0; |
| 23 | size_t res_offset = 0; |
| 24 | |
| 25 | for (size_t i = 0; i < offsets.size(); ++i) |
| 26 | { |
| 27 | size_t cur_offset = offsets[i]; |
| 28 | |
| 29 | const char * url_begin = reinterpret_cast<const char *>(&data[prev_offset]); |
| 30 | const char * url_end = reinterpret_cast<const char *>(&data[cur_offset]) - 1; |
| 31 | const char * begin_pos = url_begin; |
| 32 | const char * end_pos = begin_pos; |
| 33 | |
| 34 | do |
| 35 | { |
| 36 | const char * query_string_begin = find_first_symbols<'?', '#'>(url_begin, url_end); |
| 37 | if (query_string_begin + 1 >= url_end) |
| 38 | break; |
| 39 | |
| 40 | const char * pos = static_cast<const char *>(memmem(query_string_begin + 1, url_end - query_string_begin - 1, param_str, param_len)); |
| 41 | if (pos == nullptr) |
| 42 | break; |
| 43 | |
| 44 | if (pos[-1] != '?' && pos[-1] != '#' && pos[-1] != '&') |
| 45 | { |
| 46 | pos = nullptr; |
| 47 | break; |
| 48 | } |
| 49 | |
| 50 | begin_pos = pos; |
| 51 | end_pos = begin_pos + param_len; |
| 52 | |
| 53 | /// Skip the value. |
| 54 | while (*end_pos && *end_pos != '&' && *end_pos != '#') |
| 55 | ++end_pos; |
| 56 | |
| 57 | /// Capture '&' before or after the parameter. |
| 58 | if (*end_pos == '&') |
| 59 | ++end_pos; |
| 60 | else if (begin_pos[-1] == '&') |
| 61 | --begin_pos; |
| 62 | } while (false); |
| 63 | |
| 64 | size_t cut_length = (url_end - url_begin) - (end_pos - begin_pos); |
| 65 | res_data.resize(res_offset + cut_length + 1); |
| 66 | memcpySmallAllowReadWriteOverflow15(&res_data[res_offset], url_begin, begin_pos - url_begin); |
| 67 | memcpySmallAllowReadWriteOverflow15(&res_data[res_offset] + (begin_pos - url_begin), end_pos, url_end - end_pos); |
| 68 | res_offset += cut_length + 1; |
| 69 | res_data[res_offset - 1] = 0; |
| 70 | res_offsets[i] = res_offset; |
| 71 | |
| 72 | prev_offset = cur_offset; |
| 73 | } |
| 74 | } |
| 75 | }; |
| 76 | |
| 77 | struct NameCutURLParameter { static constexpr auto name = "cutURLParameter" ; }; |
| 78 | using FunctionCutURLParameter = FunctionsStringSearchToString<CutURLParameterImpl, NameCutURLParameter>; |
| 79 | |
| 80 | void registerFunctionCutURLParameter(FunctionFactory & factory) |
| 81 | { |
| 82 | factory.registerFunction<FunctionCutURLParameter>(); |
| 83 | } |
| 84 | |
| 85 | } |
| 86 | |