| 1 | // This example shows how to sort structs using complex multiple part keys using |
| 2 | // string_sort. |
| 3 | // |
| 4 | // Copyright Steven Ross 2009-2014. |
| 5 | // |
| 6 | // Distributed under the Boost Software License, Version 1.0. |
| 7 | // (See accompanying file LICENSE_1_0.txt or copy at |
| 8 | // http://www.boost.org/LICENSE_1_0.txt) |
| 9 | |
| 10 | // See http://www.boost.org/libs/sort for library home page. |
| 11 | |
| 12 | #include <boost/sort/spreadsort/string_sort.hpp> |
| 13 | #include <boost/sort/spreadsort/float_sort.hpp> |
| 14 | #include <time.h> |
| 15 | #include <stdio.h> |
| 16 | #include <stdlib.h> |
| 17 | #include <algorithm> |
| 18 | #include <vector> |
| 19 | #include <iostream> |
| 20 | #include <fstream> |
| 21 | #include <string> |
| 22 | using std::string; |
| 23 | using namespace boost::sort::spreadsort; |
| 24 | |
| 25 | //[generalized_functors |
| 26 | struct DATA_TYPE { |
| 27 | time_t birth; |
| 28 | float net_worth; |
| 29 | string first_name; |
| 30 | string last_name; |
| 31 | }; |
| 32 | |
| 33 | static const int birth_size = sizeof(time_t); |
| 34 | static const int first_name_offset = birth_size + sizeof(float); |
| 35 | static const boost::uint64_t base_mask = 0xff; |
| 36 | |
| 37 | struct lessthan { |
| 38 | inline bool operator()(const DATA_TYPE &x, const DATA_TYPE &y) const { |
| 39 | if (x.birth != y.birth) { |
| 40 | return x.birth < y.birth; |
| 41 | } |
| 42 | if (x.net_worth != y.net_worth) { |
| 43 | return x.net_worth < y.net_worth; |
| 44 | } |
| 45 | if (x.first_name != y.first_name) { |
| 46 | return x.first_name < y.first_name; |
| 47 | } |
| 48 | return x.last_name < y.last_name; |
| 49 | } |
| 50 | }; |
| 51 | |
| 52 | struct bracket { |
| 53 | inline unsigned char operator()(const DATA_TYPE &x, size_t offset) const { |
| 54 | // Sort date as a signed int, returning the appropriate byte. |
| 55 | if (offset < birth_size) { |
| 56 | const int bit_shift = 8 * (birth_size - offset - 1); |
| 57 | unsigned char result = (x.birth & (base_mask << bit_shift)) >> bit_shift; |
| 58 | // Handling the sign bit. Unnecessary if the data is always positive. |
| 59 | if (offset == 0) { |
| 60 | return result ^ 128; |
| 61 | } |
| 62 | |
| 63 | return result; |
| 64 | } |
| 65 | |
| 66 | // Sort a signed float. This requires reversing the order of negatives |
| 67 | // because of the way floats are represented in bits. |
| 68 | if (offset < first_name_offset) { |
| 69 | const int bit_shift = 8 * (first_name_offset - offset - 1); |
| 70 | unsigned key = float_mem_cast<float, unsigned>(data: x.net_worth); |
| 71 | unsigned char result = (key & (base_mask << bit_shift)) >> bit_shift; |
| 72 | // Handling the sign. |
| 73 | if (x.net_worth < 0) { |
| 74 | return 255 - result; |
| 75 | } |
| 76 | // Increasing positives so they are higher than negatives. |
| 77 | if (offset == birth_size) { |
| 78 | return 128 + result; |
| 79 | } |
| 80 | |
| 81 | return result; |
| 82 | } |
| 83 | |
| 84 | // Sort a string that is before the end. This approach supports embedded |
| 85 | // nulls. If embedded nulls are not required, then just delete the "* 2" |
| 86 | // and the inside of the following if just becomes: |
| 87 | // return x.first_name[offset - first_name_offset]; |
| 88 | const unsigned first_name_end_offset = |
| 89 | first_name_offset + x.first_name.size() * 2; |
| 90 | if (offset < first_name_end_offset) { |
| 91 | int char_offset = offset - first_name_offset; |
| 92 | // This signals that the string continues. |
| 93 | if (!(char_offset & 1)) { |
| 94 | return 1; |
| 95 | } |
| 96 | return x.first_name[char_offset >> 1]; |
| 97 | } |
| 98 | |
| 99 | // This signals that the string has ended, so that shorter strings come |
| 100 | // before longer ones. |
| 101 | if (offset == first_name_end_offset) { |
| 102 | return 0; |
| 103 | } |
| 104 | |
| 105 | // The final string needs no special consideration. |
| 106 | return x.last_name[offset - first_name_end_offset - 1]; |
| 107 | } |
| 108 | }; |
| 109 | |
| 110 | struct getsize { |
| 111 | inline size_t operator()(const DATA_TYPE &x) const { |
| 112 | return first_name_offset + x.first_name.size() * 2 + 1 + |
| 113 | x.last_name.size(); |
| 114 | } |
| 115 | }; |
| 116 | //] [/generalized_functors] |
| 117 | |
| 118 | //Pass in an argument to test std::sort |
| 119 | int main(int argc, const char ** argv) { |
| 120 | std::ifstream indata; |
| 121 | std::ofstream outfile; |
| 122 | bool stdSort = false; |
| 123 | unsigned loopCount = 1; |
| 124 | for (int u = 1; u < argc; ++u) { |
| 125 | if (std::string(argv[u]) == "-std" ) |
| 126 | stdSort = true; |
| 127 | else |
| 128 | loopCount = atoi(nptr: argv[u]); |
| 129 | } |
| 130 | double total = 0.0; |
| 131 | //Run multiple loops, if requested |
| 132 | std::vector<DATA_TYPE> array; |
| 133 | for (unsigned u = 0; u < loopCount; ++u) { |
| 134 | indata.open(s: "input.txt" , mode: std::ios_base::in | std::ios_base::binary); |
| 135 | if (indata.bad()) { |
| 136 | printf(format: "input.txt could not be opened\n" ); |
| 137 | return 1; |
| 138 | } |
| 139 | |
| 140 | // Read in the data. |
| 141 | DATA_TYPE inval; |
| 142 | while (!indata.eof() ) { |
| 143 | indata >> inval.first_name; |
| 144 | indata >> inval.last_name; |
| 145 | indata.read(s: reinterpret_cast<char *>(&(inval.birth)), n: birth_size); |
| 146 | indata.read(s: reinterpret_cast<char *>(&(inval.net_worth)), n: sizeof(float)); |
| 147 | // Handling nan. |
| 148 | if (inval.net_worth != inval.net_worth) { |
| 149 | inval.net_worth = 0; |
| 150 | } |
| 151 | if (indata.eof()) |
| 152 | break; |
| 153 | array.push_back(x: inval); |
| 154 | } |
| 155 | indata.close(); |
| 156 | |
| 157 | // Sort the data. |
| 158 | clock_t start, end; |
| 159 | double elapsed; |
| 160 | start = clock(); |
| 161 | if (stdSort) { |
| 162 | std::sort(first: array.begin(), last: array.end(), comp: lessthan()); |
| 163 | } else { |
| 164 | //[generalized_functors_call |
| 165 | string_sort(first: array.begin(), last: array.end(), get_character: bracket(), length: getsize(), comp: lessthan()); |
| 166 | //] [/generalized_functors_call] |
| 167 | } |
| 168 | end = clock(); |
| 169 | elapsed = static_cast<double>(end - start); |
| 170 | if (stdSort) { |
| 171 | outfile.open(s: "standard_sort_out.txt" , mode: std::ios_base::out | |
| 172 | std::ios_base::binary | std::ios_base::trunc); |
| 173 | } else { |
| 174 | outfile.open(s: "boost_sort_out.txt" , mode: std::ios_base::out | |
| 175 | std::ios_base::binary | std::ios_base::trunc); |
| 176 | } |
| 177 | if (outfile.good()) { |
| 178 | for (unsigned u = 0; u < array.size(); ++u) |
| 179 | outfile << array[u].birth << " " << array[u].net_worth << " " |
| 180 | << array[u].first_name << " " << array[u].last_name << "\n" ; |
| 181 | outfile.close(); |
| 182 | } |
| 183 | total += elapsed; |
| 184 | array.clear(); |
| 185 | } |
| 186 | if (stdSort) { |
| 187 | printf(format: "std::sort elapsed time %f\n" , total / CLOCKS_PER_SEC); |
| 188 | } else { |
| 189 | printf(format: "spreadsort elapsed time %f\n" , total / CLOCKS_PER_SEC); |
| 190 | } |
| 191 | return 0; |
| 192 | } |
| 193 | |