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