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