| 1 | /* |
| 2 | * Copyright (c) Meta Platforms, Inc. and affiliates. |
| 3 | * All rights reserved. |
| 4 | * |
| 5 | * This source code is licensed under both the BSD-style license (found in the |
| 6 | * LICENSE file in the root directory of this source tree) and the GPLv2 (found |
| 7 | * in the COPYING file in the root directory of this source tree). |
| 8 | * You may select, at your option, one of the above-listed licenses. |
| 9 | */ |
| 10 | |
| 11 | #include <assert.h> |
| 12 | #include <getopt.h> |
| 13 | #include <stdio.h> |
| 14 | #include <string.h> |
| 15 | |
| 16 | #include "config.h" |
| 17 | #include "data.h" |
| 18 | #include "method.h" |
| 19 | |
| 20 | static int g_max_name_len = 0; |
| 21 | |
| 22 | /** Check if a name contains a comma or is too long. */ |
| 23 | static int is_name_bad(char const* name) { |
| 24 | if (name == NULL) |
| 25 | return 1; |
| 26 | int const len = strlen(name); |
| 27 | if (len > g_max_name_len) |
| 28 | g_max_name_len = len; |
| 29 | for (; *name != '\0'; ++name) |
| 30 | if (*name == ',') |
| 31 | return 1; |
| 32 | return 0; |
| 33 | } |
| 34 | |
| 35 | /** Check if any of the names contain a comma. */ |
| 36 | static int are_names_bad() { |
| 37 | for (size_t method = 0; methods[method] != NULL; ++method) |
| 38 | if (is_name_bad(methods[method]->name)) { |
| 39 | fprintf(stderr, "method name %s is bad\n", methods[method]->name); |
| 40 | return 1; |
| 41 | } |
| 42 | for (size_t datum = 0; data[datum] != NULL; ++datum) |
| 43 | if (is_name_bad(data[datum]->name)) { |
| 44 | fprintf(stderr, "data name %s is bad\n", data[datum]->name); |
| 45 | return 1; |
| 46 | } |
| 47 | for (size_t config = 0; configs[config] != NULL; ++config) |
| 48 | if (is_name_bad(configs[config]->name)) { |
| 49 | fprintf(stderr, "config name %s is bad\n", configs[config]->name); |
| 50 | return 1; |
| 51 | } |
| 52 | return 0; |
| 53 | } |
| 54 | |
| 55 | /** |
| 56 | * Option parsing using getopt. |
| 57 | * When you add a new option update: long_options, long_extras, and |
| 58 | * short_options. |
| 59 | */ |
| 60 | |
| 61 | /** Option variables filled by parse_args. */ |
| 62 | static char const* g_output = NULL; |
| 63 | static char const* g_diff = NULL; |
| 64 | static char const* g_cache = NULL; |
| 65 | static char const* g_zstdcli = NULL; |
| 66 | static char const* g_config = NULL; |
| 67 | static char const* g_data = NULL; |
| 68 | static char const* g_method = NULL; |
| 69 | |
| 70 | typedef enum { |
| 71 | required_option, |
| 72 | optional_option, |
| 73 | help_option, |
| 74 | } option_type; |
| 75 | |
| 76 | /** |
| 77 | * Extra state that we need to keep per-option that we can't store in getopt. |
| 78 | */ |
| 79 | struct option_extra { |
| 80 | int id; /**< The short option name, used as an id. */ |
| 81 | char const* help; /**< The help message. */ |
| 82 | option_type opt_type; /**< The option type: required, optional, or help. */ |
| 83 | char const** value; /**< The value to set or NULL if no_argument. */ |
| 84 | }; |
| 85 | |
| 86 | /** The options. */ |
| 87 | static struct option long_options[] = { |
| 88 | {"cache", required_argument, NULL, 'c'}, |
| 89 | {"output", required_argument, NULL, 'o'}, |
| 90 | {"zstd", required_argument, NULL, 'z'}, |
| 91 | {"config", required_argument, NULL, 128}, |
| 92 | {"data", required_argument, NULL, 129}, |
| 93 | {"method", required_argument, NULL, 130}, |
| 94 | {"diff", required_argument, NULL, 'd'}, |
| 95 | {"help", no_argument, NULL, 'h'}, |
| 96 | }; |
| 97 | |
| 98 | static size_t const nargs = sizeof(long_options) / sizeof(long_options[0]); |
| 99 | |
| 100 | /** The extra info for the options. Must be in the same order as the options. */ |
| 101 | static struct option_extra long_extras[] = { |
| 102 | {'c', "the cache directory", required_option, &g_cache}, |
| 103 | {'o', "write the results here", required_option, &g_output}, |
| 104 | {'z', "zstd cli tool", required_option, &g_zstdcli}, |
| 105 | {128, "use this config", optional_option, &g_config}, |
| 106 | {129, "use this data", optional_option, &g_data}, |
| 107 | {130, "use this method", optional_option, &g_method}, |
| 108 | {'d', "compare the results to this file", optional_option, &g_diff}, |
| 109 | {'h', "display this message", help_option, NULL}, |
| 110 | }; |
| 111 | |
| 112 | /** The short options. Must correspond to the options. */ |
| 113 | static char const short_options[] = "c:d:ho:z:"; |
| 114 | |
| 115 | /** Return the help string for the option type. */ |
| 116 | static char const* required_message(option_type opt_type) { |
| 117 | switch (opt_type) { |
| 118 | case required_option: |
| 119 | return "[required]"; |
| 120 | case optional_option: |
| 121 | return "[optional]"; |
| 122 | case help_option: |
| 123 | return ""; |
| 124 | default: |
| 125 | assert(0); |
| 126 | return NULL; |
| 127 | } |
| 128 | } |
| 129 | |
| 130 | /** Print the help for the program. */ |
| 131 | static void print_help(void) { |
| 132 | fprintf(stderr, "regression test runner\n"); |
| 133 | size_t const nargs = sizeof(long_options) / sizeof(long_options[0]); |
| 134 | for (size_t i = 0; i < nargs; ++i) { |
| 135 | if (long_options[i].val < 128) { |
| 136 | /* Long / short - help [option type] */ |
| 137 | fprintf( |
| 138 | stderr, |
| 139 | "--%s / -%c \t- %s %s\n", |
| 140 | long_options[i].name, |
| 141 | long_options[i].val, |
| 142 | long_extras[i].help, |
| 143 | required_message(long_extras[i].opt_type)); |
| 144 | } else { |
| 145 | /* Short / long - help [option type] */ |
| 146 | fprintf( |
| 147 | stderr, |
| 148 | "--%s \t- %s %s\n", |
| 149 | long_options[i].name, |
| 150 | long_extras[i].help, |
| 151 | required_message(long_extras[i].opt_type)); |
| 152 | } |
| 153 | } |
| 154 | } |
| 155 | |
| 156 | /** Parse the arguments. Return 0 on success. Print help on failure. */ |
| 157 | static int parse_args(int argc, char** argv) { |
| 158 | int option_index = 0; |
| 159 | int c; |
| 160 | |
| 161 | while (1) { |
| 162 | c = getopt_long(argc, argv, short_options, long_options, &option_index); |
| 163 | if (c == -1) |
| 164 | break; |
| 165 | |
| 166 | int found = 0; |
| 167 | for (size_t i = 0; i < nargs; ++i) { |
| 168 | if (c == long_extras[i].id && long_extras[i].value != NULL) { |
| 169 | *long_extras[i].value = optarg; |
| 170 | found = 1; |
| 171 | break; |
| 172 | } |
| 173 | } |
| 174 | if (found) |
| 175 | continue; |
| 176 | |
| 177 | switch (c) { |
| 178 | case 'h': |
| 179 | case '?': |
| 180 | default: |
| 181 | print_help(); |
| 182 | return 1; |
| 183 | } |
| 184 | } |
| 185 | |
| 186 | int bad = 0; |
| 187 | for (size_t i = 0; i < nargs; ++i) { |
| 188 | if (long_extras[i].opt_type != required_option) |
| 189 | continue; |
| 190 | if (long_extras[i].value == NULL) |
| 191 | continue; |
| 192 | if (*long_extras[i].value != NULL) |
| 193 | continue; |
| 194 | fprintf( |
| 195 | stderr, |
| 196 | "--%s is a required argument but is not set\n", |
| 197 | long_options[i].name); |
| 198 | bad = 1; |
| 199 | } |
| 200 | if (bad) { |
| 201 | fprintf(stderr, "\n"); |
| 202 | print_help(); |
| 203 | return 1; |
| 204 | } |
| 205 | |
| 206 | return 0; |
| 207 | } |
| 208 | |
| 209 | /** Helper macro to print to stderr and a file. */ |
| 210 | #define tprintf(file, ...) \ |
| 211 | do { \ |
| 212 | fprintf(file, __VA_ARGS__); \ |
| 213 | fprintf(stderr, __VA_ARGS__); \ |
| 214 | } while (0) |
| 215 | /** Helper macro to flush stderr and a file. */ |
| 216 | #define tflush(file) \ |
| 217 | do { \ |
| 218 | fflush(file); \ |
| 219 | fflush(stderr); \ |
| 220 | } while (0) |
| 221 | |
| 222 | void tprint_names( |
| 223 | FILE* results, |
| 224 | char const* data_name, |
| 225 | char const* config_name, |
| 226 | char const* method_name) { |
| 227 | int const data_padding = g_max_name_len - strlen(data_name); |
| 228 | int const config_padding = g_max_name_len - strlen(config_name); |
| 229 | int const method_padding = g_max_name_len - strlen(method_name); |
| 230 | |
| 231 | tprintf( |
| 232 | results, |
| 233 | "%s, %*s%s, %*s%s, %*s", |
| 234 | data_name, |
| 235 | data_padding, |
| 236 | "", |
| 237 | config_name, |
| 238 | config_padding, |
| 239 | "", |
| 240 | method_name, |
| 241 | method_padding, |
| 242 | ""); |
| 243 | } |
| 244 | |
| 245 | /** |
| 246 | * Run all the regression tests and record the results table to results and |
| 247 | * stderr progressively. |
| 248 | */ |
| 249 | static int run_all(FILE* results) { |
| 250 | tprint_names(results, "Data", "Config", "Method"); |
| 251 | tprintf(results, "Total compressed size\n"); |
| 252 | for (size_t method = 0; methods[method] != NULL; ++method) { |
| 253 | if (g_method != NULL && strcmp(methods[method]->name, g_method)) |
| 254 | continue; |
| 255 | for (size_t datum = 0; data[datum] != NULL; ++datum) { |
| 256 | if (g_data != NULL && strcmp(data[datum]->name, g_data)) |
| 257 | continue; |
| 258 | /* Create the state common to all configs */ |
| 259 | method_state_t* state = methods[method]->create(data[datum]); |
| 260 | for (size_t config = 0; configs[config] != NULL; ++config) { |
| 261 | if (g_config != NULL && strcmp(configs[config]->name, g_config)) |
| 262 | continue; |
| 263 | if (config_skip_data(configs[config], data[datum])) |
| 264 | continue; |
| 265 | /* Print the result for the (method, data, config) tuple. */ |
| 266 | result_t const result = |
| 267 | methods[method]->compress(state, configs[config]); |
| 268 | if (result_is_skip(result)) |
| 269 | continue; |
| 270 | tprint_names( |
| 271 | results, |
| 272 | data[datum]->name, |
| 273 | configs[config]->name, |
| 274 | methods[method]->name); |
| 275 | if (result_is_error(result)) { |
| 276 | tprintf(results, "%s\n", result_get_error_string(result)); |
| 277 | } else { |
| 278 | tprintf( |
| 279 | results, |
| 280 | "%llu\n", |
| 281 | (unsigned long long)result_get_data(result).total_size); |
| 282 | } |
| 283 | tflush(results); |
| 284 | } |
| 285 | methods[method]->destroy(state); |
| 286 | } |
| 287 | } |
| 288 | return 0; |
| 289 | } |
| 290 | |
| 291 | /** memcmp() the old results file and the new results file. */ |
| 292 | static int diff_results(char const* actual_file, char const* expected_file) { |
| 293 | data_buffer_t const actual = data_buffer_read(actual_file); |
| 294 | data_buffer_t const expected = data_buffer_read(expected_file); |
| 295 | int ret = 1; |
| 296 | |
| 297 | if (actual.data == NULL) { |
| 298 | fprintf(stderr, "failed to open results '%s' for diff\n", actual_file); |
| 299 | goto out; |
| 300 | } |
| 301 | if (expected.data == NULL) { |
| 302 | fprintf( |
| 303 | stderr, |
| 304 | "failed to open previous results '%s' for diff\n", |
| 305 | expected_file); |
| 306 | goto out; |
| 307 | } |
| 308 | |
| 309 | ret = data_buffer_compare(actual, expected); |
| 310 | if (ret != 0) { |
| 311 | fprintf( |
| 312 | stderr, |
| 313 | "actual results '%s' does not match expected results '%s'\n", |
| 314 | actual_file, |
| 315 | expected_file); |
| 316 | } else { |
| 317 | fprintf(stderr, "actual results match expected results\n"); |
| 318 | } |
| 319 | out: |
| 320 | data_buffer_free(actual); |
| 321 | data_buffer_free(expected); |
| 322 | return ret; |
| 323 | } |
| 324 | |
| 325 | int main(int argc, char** argv) { |
| 326 | /* Parse args and validate modules. */ |
| 327 | int ret = parse_args(argc, argv); |
| 328 | if (ret != 0) |
| 329 | return ret; |
| 330 | |
| 331 | if (are_names_bad()) |
| 332 | return 1; |
| 333 | |
| 334 | /* Initialize modules. */ |
| 335 | method_set_zstdcli(g_zstdcli); |
| 336 | ret = data_init(g_cache); |
| 337 | if (ret != 0) { |
| 338 | fprintf(stderr, "data_init() failed with error=%s\n", strerror(ret)); |
| 339 | return 1; |
| 340 | } |
| 341 | |
| 342 | /* Run the regression tests. */ |
| 343 | ret = 1; |
| 344 | FILE* results = fopen(g_output, "w"); |
| 345 | if (results == NULL) { |
| 346 | fprintf(stderr, "Failed to open the output file\n"); |
| 347 | goto out; |
| 348 | } |
| 349 | ret = run_all(results); |
| 350 | fclose(results); |
| 351 | |
| 352 | if (ret != 0) |
| 353 | goto out; |
| 354 | |
| 355 | if (g_diff) |
| 356 | /* Diff the new results with the previous results. */ |
| 357 | ret = diff_results(g_output, g_diff); |
| 358 | |
| 359 | out: |
| 360 | data_finish(); |
| 361 | return ret; |
| 362 | } |