annotate audioDB.cpp @ 337:ccf15a8c3e46

gengetopt does not support multiple dependencies, so added test for database_given for INDEX command. Still cannot figure out why sometimes error(...) throws char* exception and SIG_ABORT
author mas01mc
date Fri, 05 Sep 2008 14:35:44 +0000
parents 100cf66a5825
children cdb31e8b94e3
rev   line source
mas01cr@0 1 #include "audioDB.h"
mas01cr@0 2
mas01mc@308 3 LSH* SERVER_LSH_INDEX_SINGLETON;
mas01mc@324 4 char* SERVER_ADB_ROOT;
mas01mc@324 5 char* SERVER_ADB_FEATURE_ROOT;
mas01mc@308 6
mas01mc@292 7 PointPair::PointPair(Uns32T a, Uns32T b, Uns32T c):trackID(a),qpos(b),spos(c){};
mas01mc@292 8
mas01mc@292 9 bool operator<(const PointPair& a, const PointPair& b){
mas01mc@324 10 return ( (a.trackID<b.trackID) ||
mas01mc@324 11 ( (a.trackID==b.trackID) &&
mas01mc@324 12 ( (a.spos<b.spos) || ( (a.spos==b.spos) && (a.qpos < b.qpos) )) ) );
mas01mc@292 13 }
mas01mc@292 14
mas01mc@292 15 bool operator>(const PointPair& a, const PointPair& b){
mas01mc@324 16 return ( (a.trackID>b.trackID) ||
mas01mc@324 17 ( (a.trackID==b.trackID) &&
mas01mc@324 18 ( (a.spos>b.spos) || ( (a.spos==b.spos) && (a.qpos > b.qpos) )) ) );
mas01mc@292 19 }
mas01mc@292 20
mas01mc@292 21 bool operator==(const PointPair& a, const PointPair& b){
mas01mc@292 22 return ( (a.trackID==b.trackID) && (a.qpos==b.qpos) && (a.spos==b.spos) );
mas01mc@292 23 }
mas01mc@292 24
mas01cr@76 25 audioDB::audioDB(const unsigned argc, char* const argv[]): O2_AUDIODB_INITIALIZERS
mas01cr@76 26 {
mas01cr@0 27 if(processArgs(argc, argv)<0){
mas01cr@0 28 printf("No command found.\n");
mas01cr@0 29 cmdline_parser_print_version ();
mas01cr@0 30 if (strlen(gengetopt_args_info_purpose) > 0)
mas01cr@0 31 printf("%s\n", gengetopt_args_info_purpose);
mas01cr@0 32 printf("%s\n", gengetopt_args_info_usage);
mas01cr@0 33 printf("%s\n", gengetopt_args_info_help[1]);
mas01cr@0 34 printf("%s\n", gengetopt_args_info_help[2]);
mas01cr@0 35 printf("%s\n", gengetopt_args_info_help[0]);
mas01cr@151 36 error("No command found");
mas01cr@0 37 }
mas01cr@77 38
mas01mc@324 39 // Perform database prefix substitution
mas01mc@328 40 if(dbName && adb_root)
mas01mc@324 41 prefix_name((char** const)&dbName, adb_root);
mas01mc@324 42
mas01mc@328 43 if(inFile && adb_feature_root)
mas01mc@328 44 prefix_name((char** const)&inFile, adb_feature_root);
mas01mc@328 45
mas01cr@0 46 if(O2_ACTION(COM_SERVER))
mas01cr@0 47 startServer();
mas01cr@0 48
mas01cr@0 49 else if(O2_ACTION(COM_CREATE))
mas01cr@0 50 create(dbName);
mas01cr@0 51
mas01cr@0 52 else if(O2_ACTION(COM_INSERT))
mas01cr@0 53 insert(dbName, inFile);
mas01cr@0 54
mas01cr@0 55 else if(O2_ACTION(COM_BATCHINSERT))
mas01cr@0 56 batchinsert(dbName, inFile);
mas01cr@0 57
mas01cr@0 58 else if(O2_ACTION(COM_QUERY))
mas01mc@307 59 if(isClient){
mas01mc@329 60 if(query_from_key){
mas01mc@332 61 VERB_LOG(1, "Calling web services query %s on database %s, query=%s\n", radius>0?"(Radius)":"(NN)", dbName, (key&&strlen(key))?key:inFile);
mas01mc@328 62 ws_query_by_key(dbName, key, inFile, (char*)hostport);
mas01mc@329 63 }
mas01mc@329 64 else{
mas01mc@332 65 VERB_LOG(1, "Calling web services query on database %s, query=%s\n", dbName, (key&&strlen(key))?key:inFile);
mas01mc@307 66 ws_query(dbName, inFile, (char*)hostport);
mas01mc@329 67 }
mas01mc@307 68 }
mas01cr@0 69 else
mas01cr@76 70 query(dbName, inFile);
mas01cr@0 71
mas01cr@0 72 else if(O2_ACTION(COM_STATUS))
mas01cr@0 73 if(isClient)
mas01cr@0 74 ws_status(dbName,(char*)hostport);
mas01cr@0 75 else
mas01cr@0 76 status(dbName);
mas01cr@280 77
mas01cr@280 78 else if(O2_ACTION(COM_SAMPLE))
mas01cr@280 79 sample(dbName);
mas01cr@0 80
mas01cr@0 81 else if(O2_ACTION(COM_L2NORM))
mas01cr@0 82 l2norm(dbName);
mas01cr@0 83
mas01cr@193 84 else if(O2_ACTION(COM_POWER))
mas01cr@193 85 power_flag(dbName);
mas01cr@193 86
mas01cr@0 87 else if(O2_ACTION(COM_DUMP))
mas01cr@0 88 dump(dbName);
mas01mc@292 89
mas01mc@334 90 else if(O2_ACTION(COM_LISZT))
mas01mc@334 91 if(isClient)
mas01mc@334 92 ws_liszt(dbName, (char*) hostport);
mas01mc@334 93 else
mas01mc@334 94 liszt(dbName, lisztOffset, lisztLength);
mas01mc@334 95
mas01mc@292 96 else if(O2_ACTION(COM_INDEX))
mas01mc@292 97 index_index_db(dbName);
mas01cr@0 98
mas01cr@0 99 else
mas01cr@0 100 error("Unrecognized command",command);
mas01cr@0 101 }
mas01cr@0 102
mas01cr@133 103 audioDB::audioDB(const unsigned argc, char* const argv[], adb__queryResponse *adbQueryResponse): O2_AUDIODB_INITIALIZERS
mas01cr@76 104 {
mas01cr@97 105 try {
mas01cr@151 106 isServer = 1; // FIXME: Hack
mas01cr@97 107 processArgs(argc, argv);
mas01mc@324 108 // Perform database prefix substitution
mas01mc@328 109 if(dbName && adb_root)
mas01mc@324 110 prefix_name((char** const)&dbName, adb_root);
mas01mc@328 111 if(inFile && adb_feature_root)
mas01mc@328 112 prefix_name((char** const)&inFile, adb_feature_root);
mas01cr@97 113 assert(O2_ACTION(COM_QUERY));
mas01cr@133 114 query(dbName, inFile, adbQueryResponse);
mas01cr@97 115 } catch(char *err) {
mas01cr@97 116 cleanup();
mas01cr@97 117 throw(err);
mas01cr@97 118 }
mas01cr@76 119 }
mas01cr@76 120
mas01cr@133 121 audioDB::audioDB(const unsigned argc, char* const argv[], adb__statusResponse *adbStatusResponse): O2_AUDIODB_INITIALIZERS
mas01cr@76 122 {
mas01cr@97 123 try {
mas01cr@151 124 isServer = 1; // FIXME: Hack
mas01cr@97 125 processArgs(argc, argv);
mas01mc@324 126 // Perform database prefix substitution
mas01mc@328 127 if(dbName && adb_root)
mas01mc@324 128 prefix_name((char** const)&dbName, adb_root);
mas01mc@328 129 if(inFile && adb_feature_root)
mas01mc@328 130 prefix_name((char** const)&inFile, adb_feature_root);
mas01cr@97 131 assert(O2_ACTION(COM_STATUS));
mas01cr@133 132 status(dbName, adbStatusResponse);
mas01cr@97 133 } catch(char *err) {
mas01cr@97 134 cleanup();
mas01cr@97 135 throw(err);
mas01cr@97 136 }
mas01cr@76 137 }
mas01cr@76 138
mas01mc@334 139 audioDB::audioDB(const unsigned argc, char* const argv[], adb__lisztResponse *adbLisztResponse): O2_AUDIODB_INITIALIZERS
mas01mc@334 140 {
mas01mc@334 141 try {
mas01mc@334 142 isServer = 1; // FIXME: Hack
mas01mc@334 143 processArgs(argc, argv);
mas01mc@334 144 // Perform database prefix substitution
mas01mc@334 145 if(dbName && adb_root)
mas01mc@334 146 prefix_name((char** const)&dbName, adb_root);
mas01mc@334 147 assert(O2_ACTION(COM_LISZT));
mas01mc@334 148 liszt(dbName, lisztOffset, lisztLength, adbLisztResponse);
mas01mc@334 149 } catch(char *err) {
mas01mc@334 150 cleanup();
mas01mc@334 151 throw(err);
mas01mc@334 152 }
mas01mc@334 153 }
mas01mc@334 154
mas01cr@97 155 void audioDB::cleanup() {
mas01cr@122 156 cmdline_parser_free(&args_info);
mas01cr@0 157 if(indata)
mas01cr@0 158 munmap(indata,statbuf.st_size);
mas01cr@0 159 if(db)
mas01cr@196 160 munmap(db,getpagesize());
mas01cr@196 161 if(fileTable)
mas01cr@196 162 munmap(fileTable, fileTableLength);
mas01cr@196 163 if(trackTable)
mas01cr@196 164 munmap(trackTable, trackTableLength);
mas01cr@196 165 if(dataBuf)
mas01cr@196 166 munmap(dataBuf, dataBufLength);
mas01cr@196 167 if(timesTable)
mas01cr@196 168 munmap(timesTable, timesTableLength);
mas01mc@314 169 if(powerTable)
mas01mc@314 170 munmap(powerTable, powerTableLength);
mas01cr@196 171 if(l2normTable)
mas01cr@196 172 munmap(l2normTable, l2normTableLength);
mas01mc@324 173 if(featureFileNameTable)
mas01mc@324 174 munmap(featureFileNameTable, fileTableLength);
mas01mc@324 175 if(timesFileNameTable)
mas01mc@324 176 munmap(timesFileNameTable, fileTableLength);
mas01mc@324 177 if(powerFileNameTable)
mas01mc@324 178 munmap(powerFileNameTable, fileTableLength);
mas01mc@292 179 if(trackOffsetTable)
mas01mc@292 180 delete trackOffsetTable;
mas01mc@292 181 if(reporter)
mas01mc@292 182 delete reporter;
mas01mc@292 183 if(exact_evaluation_queue)
mas01mc@292 184 delete exact_evaluation_queue;
mas01cr@284 185 if(rng)
mas01cr@284 186 gsl_rng_free(rng);
mas01mc@292 187 if(vv)
mas01mc@292 188 delete vv;
mas01cr@0 189 if(dbfid>0)
mas01cr@0 190 close(dbfid);
mas01cr@0 191 if(infid>0)
mas01cr@0 192 close(infid);
mas01cr@0 193 if(dbH)
mas01cr@0 194 delete dbH;
mas01mc@308 195 if(lsh!=SERVER_LSH_INDEX_SINGLETON)
mas01mc@308 196 delete lsh;
mas01cr@0 197 }
mas01cr@0 198
mas01cr@97 199 audioDB::~audioDB(){
mas01cr@97 200 cleanup();
mas01cr@97 201 }
mas01cr@97 202
mas01cr@0 203 int audioDB::processArgs(const unsigned argc, char* const argv[]){
mas01cr@0 204
mas01cr@0 205 if(argc<2){
mas01cr@0 206 cmdline_parser_print_version ();
mas01cr@0 207 if (strlen(gengetopt_args_info_purpose) > 0)
mas01cr@0 208 printf("%s\n", gengetopt_args_info_purpose);
mas01cr@0 209 printf("%s\n", gengetopt_args_info_usage);
mas01cr@0 210 printf("%s\n", gengetopt_args_info_help[1]);
mas01cr@0 211 printf("%s\n", gengetopt_args_info_help[2]);
mas01cr@0 212 printf("%s\n", gengetopt_args_info_help[0]);
mas01cr@0 213 exit(0);
mas01cr@0 214 }
mas01cr@0 215
mas01cr@0 216 if (cmdline_parser (argc, argv, &args_info) != 0)
mas01cr@151 217 error("Error parsing command line");
mas01cr@0 218
mas01cr@0 219 if(args_info.help_given){
mas01cr@0 220 cmdline_parser_print_help();
mas01cr@0 221 exit(0);
mas01cr@0 222 }
mas01cr@0 223
mas01cr@0 224 if(args_info.verbosity_given){
mas01cr@239 225 verbosity = args_info.verbosity_arg;
mas01cr@239 226 if(verbosity < 0 || verbosity > 10){
mas01cr@239 227 std::cerr << "Warning: verbosity out of range, setting to 1" << std::endl;
mas01cr@239 228 verbosity = 1;
mas01cr@0 229 }
mas01cr@0 230 }
mas01cr@0 231
mas01cr@129 232 if(args_info.size_given) {
mas01cr@256 233 if(args_info.datasize_given) {
mas01cr@256 234 error("both --size and --datasize given", "");
mas01cr@256 235 }
mas01cr@256 236 if(args_info.ntracks_given) {
mas01cr@256 237 error("both --size and --ntracks given", "");
mas01cr@256 238 }
mas01cr@256 239 if(args_info.datadim_given) {
mas01cr@256 240 error("both --size and --datadim given", "");
mas01cr@256 241 }
mas01cr@196 242 if (args_info.size_arg < 50 || args_info.size_arg > 32000) {
mas01cr@129 243 error("Size out of range", "");
mas01cr@129 244 }
mas01cr@256 245 double ratio = (double) args_info.size_arg * 1000000 / ((double) O2_DEFAULTDBSIZE);
mas01cr@256 246 /* FIXME: what's the safe way of doing this? */
mas01cr@256 247 datasize = (unsigned int) ceil(datasize * ratio);
mas01cr@256 248 ntracks = (unsigned int) ceil(ntracks * ratio);
mas01cr@256 249 } else {
mas01cr@256 250 if(args_info.datasize_given) {
mas01cr@256 251 datasize = args_info.datasize_arg;
mas01cr@256 252 }
mas01cr@256 253 if(args_info.ntracks_given) {
mas01cr@256 254 ntracks = args_info.ntracks_arg;
mas01cr@256 255 }
mas01cr@256 256 if(args_info.datadim_given) {
mas01cr@256 257 datadim = args_info.datadim_arg;
mas01cr@256 258 }
mas01cr@129 259 }
mas01cr@129 260
mas01cr@239 261 if(args_info.radius_given) {
mas01cr@239 262 radius = args_info.radius_arg;
mas01mc@307 263 if(radius < 0 || radius > 1000000000) {
mas01cr@77 264 error("radius out of range");
mas01cr@239 265 } else {
mas01cr@239 266 VERB_LOG(3, "Setting radius to %f\n", radius);
mas01mc@17 267 }
mas01mc@17 268 }
mas01mc@17 269
mas01mc@292 270 sequenceLength = args_info.sequencelength_arg;
mas01mc@292 271 if(sequenceLength < 1 || sequenceLength > 1000) {
mas01mc@292 272 error("seqlen out of range: 1 <= seqlen <= 1000");
mas01mc@292 273 }
mas01mc@292 274 sequenceHop = args_info.sequencehop_arg;
mas01mc@292 275 if(sequenceHop < 1 || sequenceHop > 1000) {
mas01mc@292 276 error("seqhop out of range: 1 <= seqhop <= 1000");
mas01mc@292 277 }
mas01mc@292 278
mas01mc@292 279 if (args_info.absolute_threshold_given) {
mas01mc@292 280 if (args_info.absolute_threshold_arg >= 0) {
mas01mc@292 281 error("absolute threshold out of range: should be negative");
mas01mc@292 282 }
mas01mc@292 283 use_absolute_threshold = true;
mas01mc@292 284 absolute_threshold = args_info.absolute_threshold_arg;
mas01mc@292 285 }
mas01mc@292 286 if (args_info.relative_threshold_given) {
mas01mc@292 287 use_relative_threshold = true;
mas01mc@292 288 relative_threshold = args_info.relative_threshold_arg;
mas01mc@292 289 }
mas01mc@292 290
mas01mc@324 291 if (args_info.adb_root_given){
mas01mc@324 292 adb_root = args_info.adb_root_arg;
mas01mc@324 293 }
mas01mc@324 294
mas01mc@324 295 if (args_info.adb_feature_root_given){
mas01mc@324 296 adb_feature_root = args_info.adb_feature_root_arg;
mas01mc@324 297 }
mas01mc@324 298
mas01mc@324 299 // perform dbName path prefix SERVER-side subsitution
mas01mc@324 300 if(SERVER_ADB_ROOT && !adb_root)
mas01mc@324 301 adb_root = SERVER_ADB_ROOT;
mas01mc@324 302 if(SERVER_ADB_FEATURE_ROOT && !adb_feature_root)
mas01mc@324 303 adb_feature_root = SERVER_ADB_FEATURE_ROOT;
mas01mc@324 304
mas01cr@0 305 if(args_info.SERVER_given){
mas01cr@0 306 command=COM_SERVER;
mas01cr@0 307 port=args_info.SERVER_arg;
mas01cr@0 308 if(port<100 || port > 100000)
mas01cr@0 309 error("port out of range");
mas01cr@151 310 isServer = 1;
mas01cr@105 311 #if defined(O2_DEBUG)
mas01cr@104 312 struct sigaction sa;
mas01cr@104 313 sa.sa_sigaction = sigterm_action;
mas01cr@104 314 sa.sa_flags = SA_SIGINFO | SA_RESTART | SA_NODEFER;
mas01cr@104 315 sigaction(SIGTERM, &sa, NULL);
mas01cr@104 316 sa.sa_sigaction = sighup_action;
mas01cr@104 317 sa.sa_flags = SA_SIGINFO | SA_RESTART | SA_NODEFER;
mas01cr@104 318 sigaction(SIGHUP, &sa, NULL);
mas01cr@105 319 #endif
mas01mc@308 320 if(args_info.load_index_given){
mas01mc@308 321 if(!args_info.database_given)
mas01mc@308 322 error("load_index requires a --database argument");
mas01mc@308 323 else
mas01mc@308 324 dbName=args_info.database_arg;
mas01mc@308 325 if(!args_info.radius_given)
mas01mc@308 326 error("load_index requires a --radius argument");
mas01mc@308 327 if(!args_info.sequencelength_given)
mas01mc@308 328 error("load_index requires a --sequenceLength argument");
mas01mc@308 329 WS_load_index = true;
mas01mc@308 330 }
mas01cr@0 331 return 0;
mas01cr@0 332 }
mas01cr@0 333
mas01cr@0 334 // No return on client command, find database command
mas01cr@105 335 if(args_info.client_given){
mas01cr@105 336 command=COM_CLIENT;
mas01cr@105 337 hostport=args_info.client_arg;
mas01cr@105 338 isClient=1;
mas01cr@105 339 }
mas01cr@0 340
mas01cr@105 341 if(args_info.NEW_given){
mas01cr@105 342 command=COM_CREATE;
mas01cr@105 343 dbName=args_info.database_arg;
mas01cr@105 344 return 0;
mas01cr@105 345 }
mas01cr@0 346
mas01cr@105 347 if(args_info.STATUS_given){
mas01cr@105 348 command=COM_STATUS;
mas01cr@105 349 dbName=args_info.database_arg;
mas01cr@105 350 return 0;
mas01cr@105 351 }
mas01cr@0 352
mas01cr@280 353 if(args_info.SAMPLE_given) {
mas01cr@280 354 command = COM_SAMPLE;
mas01cr@280 355 dbName = args_info.database_arg;
mas01cr@280 356 sequenceLength = args_info.sequencelength_arg;
mas01cr@280 357 if(sequenceLength < 1 || sequenceLength > 1000) {
mas01cr@280 358 error("seqlen out of range: 1 <= seqlen <= 1000");
mas01cr@280 359 }
mas01cr@280 360 nsamples = args_info.nsamples_arg;
mas01cr@280 361 return 0;
mas01cr@280 362 }
mas01cr@280 363
mas01cr@105 364 if(args_info.DUMP_given){
mas01cr@105 365 command=COM_DUMP;
mas01cr@105 366 dbName=args_info.database_arg;
mas01cr@131 367 output = args_info.output_arg;
mas01cr@105 368 return 0;
mas01cr@105 369 }
mas01cr@0 370
mas01cr@105 371 if(args_info.L2NORM_given){
mas01cr@105 372 command=COM_L2NORM;
mas01cr@105 373 dbName=args_info.database_arg;
mas01cr@105 374 return 0;
mas01cr@105 375 }
mas01cr@0 376
mas01cr@193 377 if(args_info.POWER_given){
mas01cr@193 378 command=COM_POWER;
mas01cr@193 379 dbName=args_info.database_arg;
mas01cr@193 380 return 0;
mas01cr@193 381 }
mas01cr@193 382
mas01cr@105 383 if(args_info.INSERT_given){
mas01cr@105 384 command=COM_INSERT;
mas01cr@105 385 dbName=args_info.database_arg;
mas01cr@105 386 inFile=args_info.features_arg;
mas01cr@105 387 if(args_info.key_given)
mas01mc@292 388 if(!args_info.features_given)
mas01mc@292 389 error("INSERT: '-k key' argument depends on '-f features'");
mas01mc@292 390 else
mas01mc@292 391 key=args_info.key_arg;
mas01cr@105 392 if(args_info.times_given){
mas01cr@105 393 timesFileName=args_info.times_arg;
mas01cr@105 394 if(strlen(timesFileName)>0){
mas01cr@239 395 if(!(timesFile = new std::ifstream(timesFileName,std::ios::in)))
mas01cr@105 396 error("Could not open times file for reading", timesFileName);
mas01cr@105 397 usingTimes=1;
mas01cr@105 398 }
mas01cr@105 399 }
mas01cr@193 400 if (args_info.power_given) {
mas01cr@193 401 powerFileName = args_info.power_arg;
mas01cr@193 402 if (strlen(powerFileName) > 0) {
mas01cr@193 403 if (!(powerfd = open(powerFileName, O_RDONLY))) {
mas01cr@193 404 error("Could not open power file for reading", powerFileName, "open");
mas01cr@193 405 }
mas01cr@193 406 usingPower = 1;
mas01cr@193 407 }
mas01cr@193 408 }
mas01cr@105 409 return 0;
mas01cr@105 410 }
mas01cr@105 411
mas01cr@105 412 if(args_info.BATCHINSERT_given){
mas01cr@105 413 command=COM_BATCHINSERT;
mas01cr@105 414 dbName=args_info.database_arg;
mas01cr@105 415 inFile=args_info.featureList_arg;
mas01cr@105 416 if(args_info.keyList_given)
mas01tc@298 417 if(!args_info.featureList_given)
mas01tc@300 418 error("BATCHINSERT: '-K keyList' argument depends on '-F featureList'");
mas01mc@292 419 else
mas01cr@304 420 key=args_info.keyList_arg; // INCONSISTENT NO CHECK
mas01cr@0 421
mas01cr@105 422 /* TO DO: REPLACE WITH
mas01cr@0 423 if(args_info.keyList_given){
mas01mc@18 424 trackFileName=args_info.keyList_arg;
mas01cr@239 425 if(strlen(trackFileName)>0 && !(trackFile = new std::ifstream(trackFileName,std::ios::in)))
mas01mc@18 426 error("Could not open keyList file for reading",trackFileName);
mas01cr@0 427 }
mas01cr@0 428 AND UPDATE BATCHINSERT()
mas01cr@105 429 */
mas01cr@105 430
mas01cr@105 431 if(args_info.timesList_given){
mas01cr@105 432 timesFileName=args_info.timesList_arg;
mas01cr@105 433 if(strlen(timesFileName)>0){
mas01cr@239 434 if(!(timesFile = new std::ifstream(timesFileName,std::ios::in)))
mas01cr@105 435 error("Could not open timesList file for reading", timesFileName);
mas01cr@105 436 usingTimes=1;
mas01cr@105 437 }
mas01cr@105 438 }
mas01cr@193 439 if(args_info.powerList_given){
mas01cr@193 440 powerFileName=args_info.powerList_arg;
mas01cr@193 441 if(strlen(powerFileName)>0){
mas01cr@239 442 if(!(powerFile = new std::ifstream(powerFileName,std::ios::in)))
mas01cr@193 443 error("Could not open powerList file for reading", powerFileName);
mas01cr@193 444 usingPower=1;
mas01cr@193 445 }
mas01cr@193 446 }
mas01cr@105 447 return 0;
mas01cr@105 448 }
mas01mc@292 449
mas01mc@292 450 // Set no_unit_norm flag
mas01mc@292 451 no_unit_norming = args_info.no_unit_norming_flag;
mas01mc@292 452 lsh_use_u_functions = args_info.lsh_use_u_functions_flag;
mas01mc@292 453
mas01mc@292 454 // LSH Index Command
mas01mc@292 455 if(args_info.INDEX_given){
mas01mc@292 456 if(radius <= 0 )
mas01mc@292 457 error("INDEXing requires a Radius argument");
mas01mc@292 458 if(!(sequenceLength>0 && sequenceLength <= O2_MAXSEQLEN))
mas01mc@292 459 error("INDEXing requires 1 <= sequenceLength <= 1000");
mas01mc@292 460 command=COM_INDEX;
mas01mc@337 461 if(!args_info.database_given)
mas01mc@337 462 error("INDEXing requires a database");
mas01mc@292 463 dbName=args_info.database_arg;
mas01mc@292 464
mas01mc@292 465 // Whether to store LSH hash tables for query in core (FORMAT2)
mas01mc@297 466 lsh_in_core = !args_info.lsh_on_disk_flag; // This flag is set to 0 if on_disk requested
mas01mc@292 467
mas01mc@292 468 lsh_param_w = args_info.lsh_w_arg;
mas01mc@292 469 if(!(lsh_param_w>0 && lsh_param_w<=O2_SERIAL_MAX_BINWIDTH))
mas01mc@292 470 error("Indexing parameter w out of range (0.0 < w <= 100.0)");
mas01mc@292 471
mas01mc@292 472 lsh_param_k = args_info.lsh_k_arg;
mas01mc@292 473 if(!(lsh_param_k>0 && lsh_param_k<=O2_SERIAL_MAX_FUNS))
mas01mc@292 474 error("Indexing parameter k out of range (1 <= k <= 100)");
mas01mc@292 475
mas01mc@292 476 lsh_param_m = args_info.lsh_m_arg;
mas01mc@292 477 if(!(lsh_param_m>0 && lsh_param_m<= (1 + (sqrt(1 + O2_SERIAL_MAX_TABLES*8.0)))/2.0))
mas01mc@292 478 error("Indexing parameter m out of range (1 <= m <= 20)");
mas01mc@292 479
mas01mc@292 480 lsh_param_N = args_info.lsh_N_arg;
mas01mc@292 481 if(!(lsh_param_N>0 && lsh_param_N<=O2_SERIAL_MAX_ROWS))
mas01mc@292 482 error("Indexing parameter N out of range (1 <= N <= 1000000)");
mas01mc@292 483
mas01mc@292 484 lsh_param_b = args_info.lsh_b_arg;
mas01mc@292 485 if(!(lsh_param_b>0 && lsh_param_b<=O2_SERIAL_MAX_TRACKBATCH))
mas01mc@292 486 error("Indexing parameter b out of range (1 <= b <= 10000)");
mas01mc@292 487
mas01mc@296 488 lsh_param_ncols = args_info.lsh_ncols_arg;
mas01mc@296 489 if(lsh_in_core) // We don't want to block rows with FORMAT2 indexing
mas01mc@296 490 lsh_param_ncols = O2_SERIAL_MAX_COLS;
mas01mc@292 491 if( !(lsh_param_ncols>0 && lsh_param_ncols<=O2_SERIAL_MAX_COLS))
mas01mc@292 492 error("Indexing parameter ncols out of range (1 <= ncols <= 1000");
mas01mc@292 493
mas01mc@292 494 return 0;
mas01mc@292 495 }
mas01mc@292 496
mas01cr@105 497 // Query command and arguments
mas01cr@105 498 if(args_info.QUERY_given){
mas01cr@105 499 command=COM_QUERY;
mas01cr@105 500 dbName=args_info.database_arg;
mas01mc@292 501 // XOR features and key search
mas01mc@292 502 if(!args_info.features_given && !args_info.key_given || (args_info.features_given && args_info.key_given))
mas01mc@292 503 error("QUERY requires exactly one of either -f features or -k key");
mas01mc@292 504 if(args_info.features_given)
mas01mc@292 505 inFile=args_info.features_arg; // query from file
mas01mc@292 506 else{
mas01mc@292 507 query_from_key = true;
mas01mc@292 508 key=args_info.key_arg; // query from key
mas01mc@292 509 }
mas01mc@292 510
mas01cr@105 511 if(args_info.keyList_given){
mas01cr@105 512 trackFileName=args_info.keyList_arg;
mas01cr@239 513 if(strlen(trackFileName)>0 && !(trackFile = new std::ifstream(trackFileName,std::ios::in)))
mas01cr@105 514 error("Could not open keyList file for reading",trackFileName);
mas01cr@105 515 }
mas01cr@105 516
mas01cr@105 517 if(args_info.times_given){
mas01cr@105 518 timesFileName=args_info.times_arg;
mas01cr@105 519 if(strlen(timesFileName)>0){
mas01cr@239 520 if(!(timesFile = new std::ifstream(timesFileName,std::ios::in)))
mas01cr@105 521 error("Could not open times file for reading", timesFileName);
mas01cr@105 522 usingTimes=1;
mas01cr@105 523 }
mas01cr@105 524 }
mas01cr@193 525
mas01cr@193 526 if(args_info.power_given){
mas01cr@193 527 powerFileName=args_info.power_arg;
mas01cr@193 528 if(strlen(powerFileName)>0){
mas01cr@193 529 if (!(powerfd = open(powerFileName, O_RDONLY))) {
mas01cr@193 530 error("Could not open power file for reading", powerFileName, "open");
mas01cr@193 531 }
mas01cr@193 532 usingPower = 1;
mas01cr@193 533 }
mas01cr@193 534 }
mas01cr@105 535
mas01cr@105 536 // query type
mas01cr@105 537 if(strncmp(args_info.QUERY_arg, "track", MAXSTR)==0)
mas01cr@105 538 queryType=O2_TRACK_QUERY;
mas01cr@105 539 else if(strncmp(args_info.QUERY_arg, "point", MAXSTR)==0)
mas01cr@105 540 queryType=O2_POINT_QUERY;
mas01cr@105 541 else if(strncmp(args_info.QUERY_arg, "sequence", MAXSTR)==0)
mas01cr@105 542 queryType=O2_SEQUENCE_QUERY;
mas01mc@248 543 else if(strncmp(args_info.QUERY_arg, "nsequence", MAXSTR)==0)
mas01mc@248 544 queryType=O2_N_SEQUENCE_QUERY;
mas01mc@263 545 else if(strncmp(args_info.QUERY_arg, "onetoonensequence", MAXSTR)==0)
mas01mc@263 546 queryType=O2_ONE_TO_ONE_N_SEQUENCE_QUERY;
mas01cr@105 547 else
mas01cr@105 548 error("unsupported query type",args_info.QUERY_arg);
mas01cr@105 549
mas01cr@105 550 if(!args_info.exhaustive_flag){
mas01cr@105 551 queryPoint = args_info.qpoint_arg;
mas01cr@105 552 usingQueryPoint=1;
mas01cr@105 553 if(queryPoint<0 || queryPoint >10000)
mas01cr@105 554 error("queryPoint out of range: 0 <= queryPoint <= 10000");
mas01cr@105 555 }
mas01mc@292 556
mas01mc@296 557 // Whether to pre-load LSH hash tables for query (default on, if flag set then off)
mas01mc@297 558 lsh_in_core = !args_info.lsh_on_disk_flag;
mas01mc@292 559
mas01mc@292 560 // Whether to perform exact evaluation of points returned by LSH
mas01mc@292 561 lsh_exact = args_info.lsh_exact_flag;
mas01mc@292 562
mas01cr@105 563 pointNN = args_info.pointnn_arg;
mas01mc@263 564 if(pointNN < 1 || pointNN > O2_MAXNN) {
mas01mc@263 565 error("pointNN out of range: 1 <= pointNN <= 1000000");
mas01cr@105 566 }
mas01cr@105 567 trackNN = args_info.resultlength_arg;
mas01mc@263 568 if(trackNN < 1 || trackNN > O2_MAXNN) {
mas01mc@263 569 error("resultlength out of range: 1 <= resultlength <= 1000000");
mas01cr@105 570 }
mas01cr@105 571 return 0;
mas01cr@105 572 }
mas01mc@334 573
mas01mc@334 574 if(args_info.LISZT_given){
mas01mc@334 575 command = COM_LISZT;
mas01mc@334 576 dbName=args_info.database_arg;
mas01mc@334 577 lisztOffset = args_info.lisztOffset_arg;
mas01mc@334 578 lisztLength = args_info.lisztLength_arg;
mas01mc@334 579 if(args_info.lisztOffset_arg<0) // check upper bound later when database is opened
mas01mc@334 580 error("lisztOffset cannot be negative");
mas01mc@334 581 if(args_info.lisztLength_arg<0)
mas01mc@334 582 error("lisztLength cannot be negative");
mas01mc@334 583 if(lisztLength >1000000)
mas01mc@334 584 error("lisztLength too large (>1000000)");
mas01mc@334 585 return 0;
mas01mc@334 586 }
mas01mc@334 587
mas01cr@105 588 return -1; // no command found
mas01cr@0 589 }
mas01cr@0 590
mas01cr@133 591 void audioDB::status(const char* dbName, adb__statusResponse *adbStatusResponse){
mas01cr@0 592 if(!dbH)
mas01cr@196 593 initTables(dbName, 0);
mas01cr@0 594
mas01cr@0 595 unsigned dudCount=0;
mas01cr@0 596 unsigned nullCount=0;
mas01cr@0 597 for(unsigned k=0; k<dbH->numFiles; k++){
mas01mc@18 598 if(trackTable[k]<sequenceLength){
mas01cr@0 599 dudCount++;
mas01mc@18 600 if(!trackTable[k])
mas01cr@76 601 nullCount++;
mas01cr@0 602 }
mas01cr@0 603 }
mas01cr@76 604
mas01cr@133 605 if(adbStatusResponse == 0) {
mas01cr@76 606
mas01cr@76 607 // Update Header information
mas01cr@239 608 std::cout << "num files:" << dbH->numFiles << std::endl;
mas01cr@239 609 std::cout << "data dim:" << dbH->dim <<std::endl;
mas01cr@76 610 if(dbH->dim>0){
mas01cr@239 611 std::cout << "total vectors:" << dbH->length/(sizeof(double)*dbH->dim)<<std::endl;
mas01mc@324 612 if(dbH->flags & O2_FLAG_LARGE_ADB)
mas01mc@324 613 std::cout << "vectors available:" << O2_MAX_VECTORS - (dbH->length / (sizeof(double)*dbH->dim)) << std::endl;
mas01mc@324 614 else
mas01mc@324 615 std::cout << "vectors available:" << (dbH->timesTableOffset-(dbH->dataOffset+dbH->length))/(sizeof(double)*dbH->dim) << std::endl;
mas01cr@76 616 }
mas01mc@324 617 if( ! (dbH->flags & O2_FLAG_LARGE_ADB) ){
mas01mc@324 618 std::cout << "total bytes:" << dbH->length << " (" << (100.0*dbH->length)/(dbH->timesTableOffset-dbH->dataOffset) << "%)" << std::endl;
mas01mc@324 619 std::cout << "bytes available:" << dbH->timesTableOffset-(dbH->dataOffset+dbH->length) << " (" <<
mas01mc@324 620 (100.0*(dbH->timesTableOffset-(dbH->dataOffset+dbH->length)))/(dbH->timesTableOffset-dbH->dataOffset) << "%)" << std::endl;
mas01mc@324 621 }
mas01mc@301 622 std::cout << "flags:" << " l2norm[" << DISPLAY_FLAG(dbH->flags&O2_FLAG_L2NORM)
mas01mc@301 623 << "] minmax[" << DISPLAY_FLAG(dbH->flags&O2_FLAG_MINMAX)
mas01mc@301 624 << "] power[" << DISPLAY_FLAG(dbH->flags&O2_FLAG_POWER)
mas01mc@324 625 << "] times[" << DISPLAY_FLAG(dbH->flags&O2_FLAG_TIMES)
mas01mc@324 626 << "] largeADB[" << DISPLAY_FLAG(dbH->flags&O2_FLAG_LARGE_ADB)
mas01mc@324 627 << "]" << endl;
mas01mc@324 628
mas01cr@239 629 std::cout << "null count: " << nullCount << " small sequence count " << dudCount-nullCount << std::endl;
mas01cr@76 630 } else {
mas01cr@133 631 adbStatusResponse->result.numFiles = dbH->numFiles;
mas01cr@133 632 adbStatusResponse->result.dim = dbH->dim;
mas01cr@133 633 adbStatusResponse->result.length = dbH->length;
mas01cr@133 634 adbStatusResponse->result.dudCount = dudCount;
mas01cr@133 635 adbStatusResponse->result.nullCount = nullCount;
mas01cr@133 636 adbStatusResponse->result.flags = dbH->flags;
mas01cr@76 637 }
mas01cr@0 638 }
mas01cr@0 639
mas01cr@196 640 void audioDB::l2norm(const char* dbName) {
mas01cr@196 641 forWrite = true;
mas01cr@196 642 initTables(dbName, 0);
mas01mc@324 643 if( !(dbH->flags & O2_FLAG_LARGE_ADB ) && (dbH->length>0) ){
mas01cr@196 644 /* FIXME: should probably be uint64_t */
mas01cr@0 645 unsigned numVectors = dbH->length/(sizeof(double)*dbH->dim);
mas01cr@196 646 CHECKED_MMAP(double *, dataBuf, dbH->dataOffset, dataBufLength);
mas01cr@0 647 unitNormAndInsertL2(dataBuf, dbH->dim, numVectors, 0); // No append
mas01cr@0 648 }
mas01cr@0 649 // Update database flags
mas01cr@0 650 dbH->flags = dbH->flags|O2_FLAG_L2NORM;
mas01cr@0 651 memcpy (db, dbH, O2_HEADERSIZE);
mas01cr@0 652 }
mas01cr@193 653
mas01cr@193 654 void audioDB::power_flag(const char *dbName) {
mas01cr@196 655 forWrite = true;
mas01mc@324 656 initTables(dbName, 0);
mas01mc@324 657 if( !(dbH->flags & O2_FLAG_LARGE_ADB ) && (dbH->length>0) ){
mas01cr@193 658 error("cannot turn on power storage for non-empty database", dbName);
mas01cr@193 659 }
mas01cr@193 660 dbH->flags |= O2_FLAG_POWER;
mas01cr@193 661 memcpy(db, dbH, O2_HEADERSIZE);
mas01cr@193 662 }
mas01cr@193 663
mas01cr@239 664 // Unit norm block of features
mas01cr@0 665
mas01cr@239 666 /* FIXME: in fact this does not unit norm a block of features, it just
mas01cr@239 667 records the L2 norms somewhere. unitNorm() does in fact unit norm
mas01cr@239 668 a block of features. */
mas01cr@0 669 void audioDB::unitNormAndInsertL2(double* X, unsigned dim, unsigned n, unsigned append=0){
mas01cr@0 670 unsigned d;
mas01cr@59 671 double *p;
mas01cr@0 672 unsigned nn = n;
mas01cr@0 673
mas01cr@0 674 assert(l2normTable);
mas01cr@0 675
mas01mc@324 676 if( !(dbH->flags & O2_FLAG_LARGE_ADB) && !append && (dbH->flags & O2_FLAG_L2NORM) )
mas01cr@0 677 error("Database is already L2 normed", "automatic norm on insert is enabled");
mas01cr@0 678
mas01cr@239 679 VERB_LOG(2, "norming %u vectors...", n);
mas01cr@0 680
mas01cr@0 681 double* l2buf = new double[n];
mas01cr@0 682 double* l2ptr = l2buf;
mas01cr@0 683 assert(l2buf);
mas01cr@0 684 assert(X);
mas01cr@0 685
mas01cr@0 686 while(nn--){
mas01cr@0 687 p=X;
mas01cr@0 688 *l2ptr=0.0;
mas01cr@0 689 d=dim;
mas01cr@0 690 while(d--){
mas01cr@0 691 *l2ptr+=*p**p;
mas01cr@0 692 p++;
mas01cr@0 693 }
mas01mc@17 694 l2ptr++;
mas01mc@17 695 X+=dim;
mas01cr@0 696 }
mas01cr@0 697 unsigned offset;
mas01cr@84 698 if(append) {
mas01cr@84 699 // FIXME: a hack, a very palpable hack: the vectors have already
mas01cr@84 700 // been inserted, and dbH->length has already been updated. We
mas01cr@84 701 // need to subtract off again the number of vectors that we've
mas01cr@84 702 // inserted this time...
mas01cr@84 703 offset=(dbH->length/(dbH->dim*sizeof(double)))-n; // number of vectors
mas01cr@84 704 } else {
mas01cr@0 705 offset=0;
mas01cr@84 706 }
mas01cr@0 707 memcpy(l2normTable+offset, l2buf, n*sizeof(double));
mas01cr@0 708 if(l2buf)
mas01mc@17 709 delete[] l2buf;
mas01cr@239 710 VERB_LOG(2, " done.");
mas01cr@193 711 }
mas01cr@193 712
mas01mc@308 713 // This entry point is visited once per instance
mas01mc@308 714 // so it is a good place to set any global state variables
mas01cr@0 715 int main(const unsigned argc, char* const argv[]){
mas01mc@308 716 SERVER_LSH_INDEX_SINGLETON = 0; // Initialize global variables
mas01mc@324 717 SERVER_ADB_ROOT = 0; // Server-side database root prefix
mas01mc@324 718 SERVER_ADB_FEATURE_ROOT = 0; // Server-side features root prefix
mas01cr@0 719 audioDB(argc, argv);
mas01cr@0 720 }