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