| 1 | #ifndef __RANDOMFILE_HH__␊ |
| 2 | #define __RANDOMFILE_HH__␊ |
| 3 | ␊ |
| 4 | // copyright (C) 2002, 2003 graydon hoare <graydon@pobox.com>␊ |
| 5 | // all rights reserved.␊ |
| 6 | // licensed to the public under the terms of the GNU GPL (>= 2)␊ |
| 7 | // see the file COPYING for details␊ |
| 8 | ␊ |
| 9 | #include <string>␊ |
| 10 | #include <vector>␊ |
| 11 | #include <stdlib.h>␊ |
| 12 | #include <boost/lexical_cast.hpp>␊ |
| 13 | ␊ |
| 14 | struct file_randomizer␊ |
| 15 | {␊ |
| 16 | ␊ |
| 17 | std::vector<std::string> lines;␊ |
| 18 | std::string prefix;␊ |
| 19 | ␊ |
| 20 | void seed(int seed = 0xf00feed) ␊ |
| 21 | { ␊ |
| 22 | srand (seed); ␊ |
| 23 | }␊ |
| 24 | ␊ |
| 25 | size_t random_index(bool last_line_ok = true)␊ |
| 26 | {␊ |
| 27 | if (last_line_ok)␊ |
| 28 | return static_cast<size_t>(rand() % lines.size());␊ |
| 29 | else␊ |
| 30 | {␊ |
| 31 | ␉if (lines.size() == 0)␊ |
| 32 | ␉ return 0;␊ |
| 33 | ␉else␊ |
| 34 | ␉ return static_cast<size_t>(rand() % (lines.size() - 1));␊ |
| 35 | }␊ |
| 36 | }␊ |
| 37 | ␊ |
| 38 | bool random_bool()␊ |
| 39 | {␊ |
| 40 | return (rand() % 2 == 0);␊ |
| 41 | }␊ |
| 42 | ␊ |
| 43 | void set_prefix(std::string const & p)␊ |
| 44 | {␊ |
| 45 | prefix = p;␊ |
| 46 | }␊ |
| 47 | ␊ |
| 48 | void append_to(std::vector<std::string> & other)␊ |
| 49 | {␊ |
| 50 | for (std::vector<std::string>::const_iterator i = lines.begin();␊ |
| 51 | ␉ i != lines.end(); ++i)␊ |
| 52 | other.push_back(prefix + *i);␊ |
| 53 | }␊ |
| 54 | ␊ |
| 55 | void initial_sequential_lines(int num_lines = 100) {␊ |
| 56 | lines.clear();␊ |
| 57 | for (int i = 0; i < num_lines; ++i)␊ |
| 58 | {␊ |
| 59 | ␉lines.push_back(std::string("initial ") + boost::lexical_cast<std::string>(i));␊ |
| 60 | }␊ |
| 61 | }␊ |
| 62 | ␊ |
| 63 | void append_sequential_lines(int num_lines = 10) {␊ |
| 64 | lines.clear();␊ |
| 65 | for (int i = 0; i < num_lines; ++i)␊ |
| 66 | {␊ |
| 67 | ␉lines.push_back(std::string("append ") + boost::lexical_cast<std::string>(i));␊ |
| 68 | }␊ |
| 69 | }␊ |
| 70 | ␊ |
| 71 | void prepend_sequential_lines(int num_lines = 10) {␊ |
| 72 | lines.clear();␊ |
| 73 | for (int i = 0; i < num_lines; ++i)␊ |
| 74 | {␊ |
| 75 | ␉lines.push_back(std::string("prepend ") + boost::lexical_cast<std::string>(i));␊ |
| 76 | }␊ |
| 77 | }␊ |
| 78 | ␊ |
| 79 | void delete_percent_of_lines_randomly(int percent = 50) {␊ |
| 80 | double scale = static_cast<double>(percent) / 100.0;␊ |
| 81 | double nlines_d = static_cast<double>(lines.size()) * scale;␊ |
| 82 | int nlines = static_cast<int>(nlines_d);␊ |
| 83 | for (int i = 0; i < nlines; ++i)␊ |
| 84 | lines.erase(lines.begin() + random_index(false));␊ |
| 85 | }␊ |
| 86 | ␊ |
| 87 | void insert_sequential_percent_of_lines_randomly(int percent = 50) {␊ |
| 88 | double scale = static_cast<double>(percent) / 100.0;␊ |
| 89 | double nlines_d = static_cast<double>(lines.size()) * scale;␊ |
| 90 | int nlines = static_cast<int>(nlines_d);␊ |
| 91 | for (int i = 0; i < nlines; ++i)␊ |
| 92 | {␊ |
| 93 | ␉lines.insert(lines.begin() + random_index(),␊ |
| 94 | ␉␉ std::string("insert ") + boost::lexical_cast<std::string>(i));␊ |
| 95 | }␊ |
| 96 | }␊ |
| 97 | ␊ |
| 98 | static void build_random_fork(std::vector<std::string> & ancestor,␊ |
| 99 | ␉␉␉␉std::vector<std::string> & left,␊ |
| 100 | ␉␉␉␉std::vector<std::string> & right,␊ |
| 101 | ␉␉␉␉std::vector<std::string> & merged,␊ |
| 102 | ␉␉␉␉int seed,␊ |
| 103 | ␉␉␉␉int n_hunks = 10)␊ |
| 104 | { ␊ |
| 105 | ␊ |
| 106 | file_randomizer fr;␊ |
| 107 | fr.seed(seed);␊ |
| 108 | // maybe prepend something to one side or the other␊ |
| 109 | if (fr.random_bool())␊ |
| 110 | {␊ |
| 111 | ␉fr.prepend_sequential_lines();␊ |
| 112 | ␉if (fr.random_bool())␊ |
| 113 | ␉ fr.append_to(left);␊ |
| 114 | ␉else␊ |
| 115 | ␉ fr.append_to(right);␊ |
| 116 | ␉fr.append_to(merged);␊ |
| 117 | }␊ |
| 118 | fr.lines.clear();␊ |
| 119 | ␊ |
| 120 | for (int h = 0; h < n_hunks; ++h)␊ |
| 121 | {␊ |
| 122 | ␉file_randomizer hr;␊ |
| 123 | ␉hr.seed(seed + h);␊ |
| 124 | ␉hr.set_prefix(std::string("hunk ") + boost::lexical_cast<std::string>(h) + " -- ");␊ |
| 125 | ␉hr.initial_sequential_lines(10);␊ |
| 126 | ␉hr.append_to(ancestor);␊ |
| 127 | ␉if (hr.random_bool())␊ |
| 128 | ␉ {␊ |
| 129 | ␉ // doing an insert␊ |
| 130 | ␉ if (hr.random_bool())␊ |
| 131 | ␉ {␊ |
| 132 | ␉␉// inserting in left␊ |
| 133 | ␉␉hr.append_to(right);␊ |
| 134 | ␉␉hr.insert_sequential_percent_of_lines_randomly();␊ |
| 135 | ␉␉hr.append_to(left);␊ |
| 136 | ␉ }␊ |
| 137 | ␉ else␊ |
| 138 | ␉ {␊ |
| 139 | ␉␉// inserting in right␊ |
| 140 | ␉␉hr.append_to(left);␊ |
| 141 | ␉␉hr.insert_sequential_percent_of_lines_randomly();␊ |
| 142 | ␉␉hr.append_to(right);␊ |
| 143 | ␉ }␊ |
| 144 | ␉ hr.append_to(merged);␊ |
| 145 | ␉ }␊ |
| 146 | ␉else␊ |
| 147 | ␉ {␊ |
| 148 | ␉ // doing a delete␊ |
| 149 | ␉ if (hr.random_bool())␊ |
| 150 | ␉ {␊ |
| 151 | ␉␉// deleting in left␊ |
| 152 | ␉␉hr.append_to(right);␊ |
| 153 | ␉␉hr.delete_percent_of_lines_randomly();␊ |
| 154 | ␉␉hr.append_to(left);␊ |
| 155 | ␉ }␊ |
| 156 | ␉ else␊ |
| 157 | ␉ {␊ |
| 158 | ␉␉// deleting in right␊ |
| 159 | ␉␉hr.append_to(left);␊ |
| 160 | ␉␉hr.delete_percent_of_lines_randomly();␊ |
| 161 | ␉␉hr.append_to(right);␊ |
| 162 | ␉ }␊ |
| 163 | ␉ hr.append_to(merged);␊ |
| 164 | ␉ }␊ |
| 165 | }␊ |
| 166 | ␊ |
| 167 | // maybe append something to one side or the other␊ |
| 168 | if (fr.random_bool())␊ |
| 169 | {␊ |
| 170 | ␉fr.append_sequential_lines();␊ |
| 171 | ␉if (fr.random_bool())␊ |
| 172 | ␉ fr.append_to(left);␊ |
| 173 | ␉else␊ |
| 174 | ␉ fr.append_to(right);␊ |
| 175 | ␉fr.append_to(merged);␊ |
| 176 | }␊ |
| 177 | }␊ |
| 178 | };␊ |
| 179 | ␊ |
| 180 | #endif // __RANDOMFILE_HH__␊ |
