annotate audioDB.h @ 464:f283448a40db

feature DIM upper limit changed. Needed for working with very large data sets
author mas01mc
date Tue, 30 Dec 2008 17:16:02 +0000
parents dc099cd34b5b
children add65705e655
rev   line source
mas01mc@292 1 #ifndef __AUDIODB_H_
mas01mc@292 2 #define __AUDIODB_H_
mas01mc@292 3
mas01cr@0 4 #include <stdio.h>
mas01cr@0 5 #include <stdlib.h>
mas01cr@0 6 #include <sys/types.h>
mas01cr@0 7 #include <sys/stat.h>
mas01cr@0 8 #include <sys/mman.h>
mas01cr@0 9 #include <fcntl.h>
mas01cr@0 10 #include <string.h>
mas01cr@0 11 #include <iostream>
mas01cr@0 12 #include <fstream>
mas01cr@302 13 #include <set>
mas01cr@302 14 #include <string>
mas01cr@0 15 #include <math.h>
mas01cr@0 16 #include <sys/time.h>
mas01cr@0 17 #include <assert.h>
mas01cr@62 18 #include <float.h>
mas01cr@104 19 #include <signal.h>
mas01cr@280 20 #include <gsl/gsl_rng.h>
mas01cr@0 21
mas01mc@292 22 // includes for LSH indexing
mas01mc@292 23 #include "ReporterBase.h"
mas01mc@292 24 #include "lshlib.h"
mas01mc@292 25
mas01cr@0 26 // includes for web services
mas01cr@0 27 #include "soapH.h"
mas01cr@0 28 #include "cmdline.h"
mas01cr@0 29
mas01cr@0 30 #define MAXSTR 512
mas01cr@0 31
mas01cr@0 32 // Databse PRIMARY commands
mas01cr@0 33 #define COM_CREATE "--NEW"
mas01cr@0 34 #define COM_INSERT "--INSERT"
mas01cr@0 35 #define COM_BATCHINSERT "--BATCHINSERT"
mas01cr@0 36 #define COM_QUERY "--QUERY"
mas01cr@0 37 #define COM_STATUS "--STATUS"
mas01cr@0 38 #define COM_L2NORM "--L2NORM"
mas01cr@193 39 #define COM_POWER "--POWER"
mas01cr@0 40 #define COM_DUMP "--DUMP"
mas01cr@0 41 #define COM_SERVER "--SERVER"
mas01mc@292 42 #define COM_INDEX "--INDEX"
mas01cr@280 43 #define COM_SAMPLE "--SAMPLE"
mas01mc@334 44 #define COM_LISZT "--LISZT"
mas01cr@0 45
mas01cr@0 46 // parameters
mas01cr@0 47 #define COM_CLIENT "--client"
mas01cr@0 48 #define COM_DATABASE "--database"
mas01cr@0 49 #define COM_QTYPE "--qtype"
mas01cr@0 50 #define COM_SEQLEN "--sequencelength"
mas01cr@0 51 #define COM_SEQHOP "--sequencehop"
mas01cr@0 52 #define COM_POINTNN "--pointnn"
mas01mc@307 53 #define COM_RADIUS "--radius"
mas01mc@18 54 #define COM_TRACKNN "--resultlength"
mas01cr@0 55 #define COM_QPOINT "--qpoint"
mas01cr@0 56 #define COM_FEATURES "--features"
mas01cr@0 57 #define COM_QUERYKEY "--key"
mas01cr@0 58 #define COM_KEYLIST "--keyList"
mas01cr@0 59 #define COM_TIMES "--times"
mas01cr@193 60 #define COM_QUERYPOWER "--power"
mas01cr@193 61 #define COM_RELATIVE_THRESH "--relative-threshold"
mas01cr@193 62 #define COM_ABSOLUTE_THRESH "--absolute-threshold"
mas01mc@310 63 #define COM_EXHAUSTIVE "--exhaustive"
mas01mc@310 64 #define COM_LSH_EXACT "--lsh_exact"
mas01cr@0 65
mas01mc@314 66 // Because LSH returns NN with P(1)<1 we want to return exact
mas01mc@314 67 // points above this boundary.
mas01mc@314 68 // Because we work in Radius^2 units,
mas01mc@314 69 // The sqrt of this number is the multiplier on the radius
mas01mc@314 70
mas01mc@464 71 #define O2_LSH_EXACT_MULT 1
mas01mc@314 72
mas01cr@108 73 #define O2_OLD_MAGIC ('O'|'2'<<8|'D'<<16|'B'<<24)
mas01cr@108 74 #define O2_MAGIC ('o'|'2'<<8|'d'<<16|'b'<<24)
mas01cr@210 75 #define O2_FORMAT_VERSION (4U)
mas01cr@0 76
mas01cr@0 77 #define O2_DEFAULT_POINTNN (10U)
mas01mc@18 78 #define O2_DEFAULT_TRACKNN (10U)
mas01cr@0 79
mas01mc@248 80 //#define O2_DEFAULTDBSIZE (4000000000) // 4GB table size
mas01mc@7 81 #define O2_DEFAULTDBSIZE (2000000000) // 2GB table size
mas01cr@0 82
mas01mc@295 83 // Bit masks for packing (trackID,pointID) into 32-bit unsigned int
mas01mc@324 84 // This can be controlled at compile time
mas01mc@324 85 #define O2_DEFAULT_LSH_N_POINT_BITS 14
mas01mc@324 86
mas01mc@324 87 // Override the default point bit width for large database support
mas01mc@324 88 #ifndef LSH_N_POINT_BITS
mas01mc@324 89 #define LSH_N_POINT_BITS O2_DEFAULT_LSH_N_POINT_BITS
mas01mc@324 90 #endif
mas01mc@295 91
mas01mc@295 92 // LIMIT PARAMETERS
mas01cr@256 93 #define O2_DEFAULT_DATASIZE (1355U) // in MB
mas01cr@256 94 #define O2_DEFAULT_NTRACKS (20000U)
mas01cr@256 95 #define O2_DEFAULT_DATADIM (9U)
mas01mc@292 96 #define O2_REALTYPE (double)
mas01mc@324 97 #define O2_MAXFILES (1000000U)
mas01cr@0 98 #define O2_MAXFILESTR (256U)
mas01cr@256 99 #define O2_FILETABLE_ENTRY_SIZE (O2_MAXFILESTR)
mas01cr@256 100 #define O2_TRACKTABLE_ENTRY_SIZE (sizeof(unsigned))
mas01cr@0 101 #define O2_HEADERSIZE (sizeof(dbTableHeaderT))
mas01cr@0 102 #define O2_MEANNUMVECTORS (1000U)
mas01mc@464 103 #define O2_MAXDIM (20000U)
mas01mc@263 104 #define O2_MAXNN (1000000U)
mas01mc@292 105 #define O2_MAXSEQLEN (8000U) // maximum feature vectors in a sequence
mas01mc@324 106 #define O2_MAXTRACKS (1000000U) // maximum number of tracks
mas01mc@324 107 #define O2_MAXTRACKLEN (1<<LSH_N_POINT_BITS) // maximum shingles in a track
mas01mc@292 108 #define O2_MAXDOTPRODUCTMEMORY (sizeof(O2_REALTYPE)*O2_MAXSEQLEN*O2_MAXSEQLEN) // 512MB
mas01mc@292 109 #define O2_DISTANCE_TOLERANCE (1e-6)
mas01mc@324 110 #define O2_SERIAL_MAX_TRACKBATCH (1000000)
mas01mc@324 111 #define O2_LARGE_ADB_SIZE (O2_DEFAULT_DATASIZE+1) // datasize at which features are kept externally (in Mbytes)
mas01mc@324 112 #define O2_LARGE_ADB_NTRACKS (O2_DEFAULT_NTRACKS+1) // ntracks at which features are kept externally
mas01mc@324 113 #define O2_MAX_VECTORS ( O2_MEANNUMVECTORS * O2_MAXTRACKS )
mas01cr@0 114
mas01cr@0 115 // Flags
mas01cr@0 116 #define O2_FLAG_L2NORM (0x1U)
mas01cr@0 117 #define O2_FLAG_MINMAX (0x2U)
mas01cr@193 118 #define O2_FLAG_POWER (0x4U)
mas01cr@0 119 #define O2_FLAG_TIMES (0x20U)
mas01mc@324 120 #define O2_FLAG_LARGE_ADB (0x40U)
mas01mc@301 121 #define DISPLAY_FLAG(x) (x?"on":"off")
mas01cr@0 122
mas01cr@105 123 // Query types
mas01cr@105 124 #define O2_POINT_QUERY (0x4U)
mas01cr@105 125 #define O2_SEQUENCE_QUERY (0x8U)
mas01cr@105 126 #define O2_TRACK_QUERY (0x10U)
mas01mc@248 127 #define O2_N_SEQUENCE_QUERY (0x20U)
mas01mc@263 128 #define O2_ONE_TO_ONE_N_SEQUENCE_QUERY (0x40U)
mas01mc@248 129
mas01cr@0 130 // Error Codes
mas01cr@0 131 #define O2_ERR_KEYNOTFOUND (0xFFFFFF00)
mas01cr@0 132
mas01cr@0 133 // Macros
mas01cr@0 134 #define O2_ACTION(a) (strcmp(command,a)==0)
mas01cr@0 135
mas01cr@370 136 #define ALIGN_UP(x,w) (((x) + ((1<<w)-1)) & ~((1<<w)-1))
mas01cr@108 137 #define ALIGN_DOWN(x,w) ((x) & ~((1<<w)-1))
mas01cr@108 138
mas01cr@370 139 #define ALIGN_PAGE_UP(x) (((x) + (getpagesize()-1)) & ~(getpagesize()-1))
mas01cr@196 140 #define ALIGN_PAGE_DOWN(x) ((x) & ~(getpagesize()-1))
mas01cr@196 141
mas01cr@166 142 #define ENSURE_STRING(x) ((x) ? (x) : "")
mas01cr@166 143
mas01cr@239 144 #define CHECKED_MMAP(type, var, start, length) \
mas01cr@239 145 { void *tmp = mmap(0, length, (PROT_READ | (forWrite ? PROT_WRITE : 0)), MAP_SHARED, dbfid, (start)); \
mas01cr@239 146 if(tmp == (void *) -1) { \
mas01cr@239 147 error("mmap error for db table", #var, "mmap"); \
mas01cr@239 148 } \
mas01cr@239 149 var = (type) tmp; \
mas01cr@239 150 }
mas01cr@239 151
mas01cr@370 152 #define CHECKED_READ(fd, buf, count) \
mas01cr@370 153 { size_t tmpcount = count; \
mas01cr@370 154 ssize_t tmp = read(fd, buf, tmpcount); \
mas01cr@370 155 if(tmp == -1) { \
mas01cr@370 156 error("read error", "", "read"); \
mas01cr@370 157 } else if((size_t) tmp != tmpcount) { \
mas01cr@370 158 error("short read", ""); \
mas01cr@370 159 } \
mas01cr@370 160 }
mas01cr@370 161
mas01cr@370 162 #define CHECKED_WRITE(fd, buf, count) \
mas01cr@370 163 { size_t tmpcount = count; \
mas01cr@370 164 ssize_t tmp = write(fd, buf, tmpcount); \
mas01cr@370 165 if(tmp == -1) { \
mas01cr@370 166 error("write error", "", "write"); \
mas01cr@370 167 } else if((size_t) tmp != tmpcount) { \
mas01cr@370 168 error("short write", ""); \
mas01cr@370 169 } \
mas01cr@370 170 }
mas01cr@370 171
mas01cr@239 172 #define VERB_LOG(vv, ...) \
mas01cr@239 173 if(verbosity > vv) { \
mas01cr@239 174 fprintf(stderr, __VA_ARGS__); \
mas01cr@239 175 fflush(stderr); \
mas01cr@239 176 }
mas01cr@0 177
mas01mc@324 178 // We will only use this in a 32-bit address space
mas01mc@324 179 // So map the off_t down to 32-bits first
mas01mc@324 180 #define INSERT_FILETABLE_STRING(TABLE, STR) \
mas01mc@324 181 strncpy(TABLE + dbH->numFiles*O2_FILETABLE_ENTRY_SIZE, STR, strlen(STR));
mas01mc@324 182
mas01mc@324 183 #define SAFE_DELETE(PTR) delete PTR; PTR=0;
mas01mc@324 184 #define SAFE_DELETE_ARRAY(PTR) delete[] PTR; PTR=0;
mas01mc@324 185
mas01mc@308 186 extern LSH* SERVER_LSH_INDEX_SINGLETON;
mas01mc@324 187 extern char* SERVER_ADB_ROOT;
mas01mc@324 188 extern char* SERVER_ADB_FEATURE_ROOT;
mas01mc@308 189
mas01cr@210 190 typedef struct dbTableHeader {
mas01cr@114 191 uint32_t magic;
mas01cr@114 192 uint32_t version;
mas01cr@114 193 uint32_t numFiles;
mas01cr@114 194 uint32_t dim;
mas01cr@114 195 uint32_t flags;
mas01cr@210 196 uint32_t headerSize;
mas01cr@196 197 off_t length;
mas01cr@196 198 off_t fileTableOffset;
mas01cr@196 199 off_t trackTableOffset;
mas01cr@196 200 off_t dataOffset;
mas01cr@196 201 off_t l2normTableOffset;
mas01cr@196 202 off_t timesTableOffset;
mas01cr@196 203 off_t powerTableOffset;
mas01cr@196 204 off_t dbSize;
mas01cr@0 205 } dbTableHeaderT, *dbTableHeaderPtr;
mas01cr@0 206
mas01ik@355 207 typedef struct {
mas01ik@355 208
mas01ik@355 209 unsigned numFiles;
mas01ik@355 210 unsigned dim;
mas01ik@355 211 unsigned length;
mas01ik@355 212 unsigned dudCount;
mas01ik@355 213 unsigned nullCount;
mas01ik@355 214 unsigned flags;
mas01ik@355 215
mas01ik@355 216
mas01ik@355 217 } cppstatus, *cppstatusptr;
mas01mc@292 218
mas01mc@292 219 class PointPair{
mas01mc@292 220 public:
mas01mc@292 221 Uns32T trackID;
mas01mc@292 222 Uns32T qpos;
mas01mc@292 223 Uns32T spos;
mas01mc@292 224 PointPair(Uns32T a, Uns32T b, Uns32T c);
mas01mc@292 225 };
mas01mc@292 226
mas01mc@292 227 bool operator<(const PointPair& a, const PointPair& b);
mas01cr@0 228
mas01mc@308 229 class audioDB{
mas01cr@0 230 private:
mas01cr@0 231 gengetopt_args_info args_info;
mas01cr@0 232 unsigned dim;
mas01cr@0 233 const char *dbName;
mas01cr@0 234 const char *inFile;
mas01cr@0 235 const char *hostport;
mas01cr@0 236 const char *key;
mas01mc@18 237 const char* trackFileName;
mas01cr@239 238 std::ifstream *trackFile;
mas01cr@0 239 const char *command;
mas01cr@131 240 const char *output;
mas01cr@0 241 const char *timesFileName;
mas01cr@239 242 std::ifstream *timesFile;
mas01cr@193 243 const char *powerFileName;
mas01cr@239 244 std::ifstream *powerFile;
mas01mc@324 245 const char* adb_root;
mas01mc@324 246 const char* adb_feature_root;
mas01mc@324 247
mas01cr@193 248 int powerfd;
mas01cr@0 249 int dbfid;
mas01mc@292 250 int lshfid;
mas01cr@196 251 bool forWrite;
mas01cr@0 252 int infid;
mas01cr@0 253 char* db;
mas01cr@0 254 char* indata;
mas01cr@0 255 struct stat statbuf;
mas01cr@0 256 dbTableHeaderPtr dbH;
mas01cr@284 257
mas01cr@284 258 gsl_rng *rng;
mas01cr@0 259
mas01mc@324 260 char* fileTable;
mas01mc@18 261 unsigned* trackTable;
mas01mc@324 262 off_t* trackOffsetTable;
mas01cr@0 263 double* dataBuf;
mas01cr@0 264 double* inBuf;
mas01cr@0 265 double* l2normTable;
mas01cr@196 266 double* timesTable;
mas01cr@193 267 double* powerTable;
mas01cr@0 268
mas01mc@324 269 char* featureFileNameTable;
mas01mc@324 270 char* timesFileNameTable;
mas01mc@324 271 char* powerFileNameTable;
mas01mc@324 272
mas01cr@196 273 size_t fileTableLength;
mas01cr@196 274 size_t trackTableLength;
mas01cr@196 275 off_t dataBufLength;
mas01cr@196 276 size_t timesTableLength;
mas01cr@196 277 size_t powerTableLength;
mas01cr@196 278 size_t l2normTableLength;
mas01cr@196 279
mas01cr@0 280 // Flags and parameters
mas01cr@0 281 unsigned verbosity; // how much do we want to know?
mas01cr@256 282
mas01cr@280 283 unsigned nsamples;
mas01cr@280 284
mas01cr@256 285 //off_t size; // given size (for creation)
mas01cr@256 286 unsigned datasize; // size in MB
mas01cr@256 287 unsigned ntracks;
mas01cr@256 288 unsigned datadim;
mas01cr@256 289
mas01cr@0 290 unsigned queryType; // point queries default
mas01cr@0 291 unsigned pointNN; // how many point NNs ?
mas01mc@18 292 unsigned trackNN; // how many track NNs ?
mas01cr@0 293 unsigned sequenceLength;
mas01cr@0 294 unsigned sequenceHop;
mas01cr@239 295 bool normalizedDistance;
mas01mc@292 296 bool no_unit_norming;
mas01cr@0 297 unsigned queryPoint;
mas01cr@0 298 unsigned usingQueryPoint;
mas01cr@0 299 unsigned usingTimes;
mas01cr@193 300 unsigned usingPower;
mas01cr@0 301 unsigned isClient;
mas01cr@0 302 unsigned isServer;
mas01cr@0 303 unsigned port;
mas01cr@0 304 double timesTol;
mas01mc@17 305 double radius;
mas01mc@292 306 bool query_from_key;
mas01mc@292 307 Uns32T query_from_key_index;
mas01cr@193 308 bool use_absolute_threshold;
mas01cr@193 309 double absolute_threshold;
mas01cr@193 310 bool use_relative_threshold;
mas01cr@193 311 double relative_threshold;
mas01mc@334 312
mas01mc@292 313 ReporterBase* reporter; // track/point reporter
mas01mc@292 314 priority_queue<PointPair, std::vector<PointPair>, std::less<PointPair> >* exact_evaluation_queue;
mas01mc@414 315 set<Uns32T> * allowed_keys; // search restrict list by key
mas01mc@292 316
mas01cr@0 317 // Timers
mas01cr@0 318 struct timeval tv1;
mas01cr@0 319 struct timeval tv2;
mas01mc@334 320
mas01mc@334 321 // LISZT parameters
mas01mc@334 322 unsigned lisztOffset;
mas01mc@334 323 unsigned lisztLength;
mas01mc@334 324
mas01ik@355 325 //for lib / API
mas01cr@370 326 int apierrortemp;
mas01ik@355 327 unsigned UseApiError;
mas01ik@355 328
mas01cr@0 329 // private methods
mas01cr@32 330 void error(const char* a, const char* b = "", const char *sysFunc = 0);
mas01cr@193 331 void sequence_sum(double *buffer, int length, int seqlen);
mas01cr@193 332 void sequence_sqrt(double *buffer, int length, int seqlen);
mas01cr@193 333 void sequence_average(double *buffer, int length, int seqlen);
mas01cr@193 334
mas01ik@355 335
mas01cr@239 336 void initialize_arrays(int track, unsigned int numVectors, double *query, double *data_buffer, double **D, double **DD);
mas01cr@239 337 void delete_arrays(int track, unsigned int numVectors, double **D, double **DD);
mas01mc@324 338 void read_data(int trkfid, int track, double **data_buffer_p, size_t *data_buffer_size_p);
mas01cr@239 339 void set_up_query(double **qp, double **vqp, double **qnp, double **vqnp, double **qpp, double **vqpp, double *mqdp, unsigned int *nvp);
mas01mc@292 340 void set_up_query_from_key(double **qp, double **vqp, double **qnp, double **vqnp, double **qpp, double **vqpp, double *mqdp, unsigned *nvp, Uns32T queryIndex);
mas01cr@239 341 void set_up_db(double **snp, double **vsnp, double **spp, double **vspp, double **mddp, unsigned int *dvp);
mas01mc@292 342 void query_loop(const char* dbName, Uns32T queryIndex);
mas01mc@292 343 void query_loop_points(double* query, double* qnPtr, double* qpPtr, double meanQdur, Uns32T numVectors);
mas01mc@292 344 double dot_product_points(double* q, double* p, Uns32T L);
mas01cr@284 345 void initRNG();
mas01cr@196 346 void initDBHeader(const char *dbName);
mas01mc@324 347 void initInputFile(const char *inFile, bool loadData = true);
mas01mc@292 348 void initTables(const char* dbName, const char* inFile = 0);
mas01mc@292 349 void initTablesFromKey(const char* dbName, const Uns32T queryIndex);
mas01cr@0 350 void unitNorm(double* X, unsigned d, unsigned n, double* qNorm);
mas01cr@0 351 void unitNormAndInsertL2(double* X, unsigned dim, unsigned n, unsigned append);
mas01cr@239 352 void insertTimeStamps(unsigned n, std::ifstream* timesFile, double* timesdata);
mas01cr@193 353 void insertPowerData(unsigned n, int powerfd, double *powerdata);
mas01cr@0 354 unsigned getKeyPos(char* key);
mas01mc@324 355 void prefix_name(char** const name, const char* prefix);
mas01mc@324 356
mas01cr@0 357 public:
mas01cr@370 358 audioDB(const unsigned argc, const char *argv[]);
mas01cr@370 359 audioDB(const unsigned argc, const char *argv[], adb__queryResponse *adbQueryResponse);
mas01cr@370 360 audioDB(const unsigned argc, const char *argv[], adb__statusResponse *adbStatusResponse);
mas01cr@370 361 audioDB(const unsigned argc, const char *argv[], adb__lisztResponse *adbLisztResponse);
mas01cr@370 362 audioDB(const unsigned argc, const char *argv[], int * apierror);
mas01cr@370 363 audioDB(const unsigned argc, const char *argv[], cppstatusptr stat, int * apierror);
mas01cr@370 364 audioDB(const unsigned argc, const char *argv[],adb__queryResponse *adbQueryResponse, int * apierror);
mas01ik@355 365
mas01mc@334 366
mas01cr@97 367 void cleanup();
mas01cr@0 368 ~audioDB();
mas01cr@370 369 int processArgs(const unsigned argc, const char* argv[]);
mas01cr@30 370 void get_lock(int fd, bool exclusive);
mas01cr@30 371 void release_lock(int fd);
mas01cr@0 372 void create(const char* dbName);
mas01cr@251 373 bool enough_per_file_space_free();
mas01cr@196 374 bool enough_data_space_free(off_t size);
mas01cr@196 375 void insert_data_vectors(off_t offset, void *buffer, size_t size);
mas01cr@0 376 void insert(const char* dbName, const char* inFile);
mas01cr@0 377 void batchinsert(const char* dbName, const char* inFile);
mas01mc@324 378 void batchinsert_large_adb(const char* dbName, const char* inFile);
mas01cr@133 379 void query(const char* dbName, const char* inFile, adb__queryResponse *adbQueryResponse=0);
mas01cr@133 380 void status(const char* dbName, adb__statusResponse *adbStatusResponse=0);
mas01ik@355 381 void status(const char* dbName, cppstatusptr status);
mas01ik@355 382
mas01cr@284 383 unsigned random_track(unsigned *propTable, unsigned total);
mas01cr@280 384 void sample(const char *dbName);
mas01cr@0 385 void l2norm(const char* dbName);
mas01cr@193 386 void power_flag(const char *dbName);
mas01cr@193 387 bool powers_acceptable(double p1, double p2);
mas01cr@0 388 void dump(const char* dbName);
mas01mc@334 389 void liszt(const char* dbName, unsigned offset, unsigned numLines, adb__lisztResponse* adbLisztResponse=0);
mas01cr@0 390
mas01mc@292 391 // LSH indexing parameters and data structures
mas01mc@292 392 LSH* lsh;
mas01mc@292 393 bool lsh_in_core; // load LSH tables for query into core (true) or keep on disk (false)
mas01mc@292 394 bool lsh_use_u_functions;
mas01mc@292 395 bool lsh_exact; // flag to indicate use exact evaluation of points returned by LSH
mas01mc@308 396 bool WS_load_index; // flag to indicate that we want to make a Web Services index memory resident
mas01mc@292 397 double lsh_param_w; // Width of LSH hash-function bins
mas01mc@292 398 Uns32T lsh_param_k; // Number of independent hash functions
mas01mc@292 399 Uns32T lsh_param_m; // Combinatorial parameter for m(m-1)/2 hash tables
mas01mc@292 400 Uns32T lsh_param_N; // Number of rows per hash table
mas01mc@292 401 Uns32T lsh_param_b; // Batch size, in number of tracks, per indexing iteration
mas01mc@292 402 Uns32T lsh_param_ncols; // Maximum number of collision in a hash-table row
mas01mc@324 403 Uns32T lsh_n_point_bits; // How many bits to use to encode point ID within a track
mas01mc@324 404
mas01mc@292 405
mas01mc@292 406 // LSH vector<> containers for one in-core copy of a set of feature vectors
mas01mc@292 407 vector<float>::iterator vi; // feature vector iterator
mas01mc@292 408 vector<vector<float> > *vv; // one-track's worth data
mas01mc@292 409
mas01mc@292 410 // LSH indexing and retrieval methods
mas01mc@292 411 void index_index_db(const char* dbName);
mas01mc@292 412 void index_initialize(double**,double**,double**,double**,unsigned int*);
mas01mc@292 413 void index_insert_tracks(Uns32T start_track, Uns32T end_track, double** fvpp, double** sNormpp,double** snPtrp, double** sPowerp, double** spPtrp);
mas01mc@292 414 int index_insert_track(Uns32T trackID, double** fvpp, double** snpp, double** sppp);
mas01mc@292 415 Uns32T index_insert_shingles(vector<vector<float> >*, Uns32T trackID, double* spp);
mas01mc@292 416 void index_make_shingle(vector<vector<float> >*, Uns32T idx, double* fvp, Uns32T dim, Uns32T seqLen);
mas01mc@292 417 int index_norm_shingles(vector<vector<float> >*, double* snp, double* spp);
mas01mc@292 418 int index_query_loop(const char* dbName, Uns32T queryIndex);
mas01mc@292 419 vector<vector<float> >* index_initialize_shingles(Uns32T sz);
mas01mc@292 420 int index_init_query(const char* dbName);
mas01mc@292 421 int index_exists(const char* dbName, double radius, Uns32T sequenceLength);
mas01mc@292 422 char* index_get_name(const char*dbName, double radius, Uns32T sequenceLength);
mas01mc@414 423 static void index_add_point(void* instance, Uns32T pointID, Uns32T qpos, float dist); // static point reporter callback method
mas01mc@324 424 static Uns32T index_to_trackID(Uns32T lshID, Uns32T nPntBits); // Convert lsh point index to audioDB trackID
mas01mc@324 425 static Uns32T index_to_trackPos(Uns32T lshID, Uns32T nPntBits); // Convert lsh point index to audioDB trackPos (spos)
mas01mc@324 426 static Uns32T index_from_trackInfo(Uns32T trackID, Uns32T pntID, Uns32T nPntBits); // Convert audioDB trackID and trackPos to an lsh point index
mas01mc@292 427 void initialize_exact_evalutation_queue();
mas01mc@292 428 void index_insert_exact_evaluation_queue(Uns32T trackID, Uns32T qpos, Uns32T spos);
mas01mc@308 429 LSH* index_allocate(char* indexName, bool load_hashTables);
mas01mc@324 430 void init_track_aux_data(Uns32T trackID, double* fvp, double** sNormpp,double** snPtrp, double** sPowerp, double** spPtrp);
mas01mc@414 431 void initialize_allowed_keys(std::ifstream*); // implementation of restrict lists using STL "set" class
mas01mc@414 432 int is_in_allowed_keys(Uns32T trackID); // test method for allowed_keys used during search
mas01mc@324 433
mas01mc@292 434 // Web Services
mas01cr@0 435 void startServer();
mas01ik@355 436
mas01mc@308 437 void ws_status(const char*dbName, char* hostport);
mas01mc@308 438 void ws_query(const char*dbName, const char *featureFileName, const char* hostport);
mas01mc@328 439 void ws_query_by_key(const char*dbName, const char *trackKey, const char* featureFileName, const char* hostport);
mas01mc@334 440 void ws_liszt(const char* dbName, char* hostport);
mas01mc@334 441
mas01cr@0 442 };
mas01mc@17 443
mas01mc@292 444 #define O2_AUDIODB_INITIALIZERS \
mas01mc@292 445 dim(0), \
mas01mc@292 446 dbName(0), \
mas01mc@292 447 inFile(0), \
mas01mc@292 448 key(0), \
mas01mc@292 449 trackFileName(0), \
mas01mc@292 450 trackFile(0), \
mas01mc@292 451 command(0), \
mas01mc@292 452 output(0), \
mas01mc@292 453 timesFileName(0), \
mas01mc@292 454 timesFile(0), \
mas01mc@292 455 powerFileName(0), \
mas01mc@292 456 powerFile(0), \
mas01mc@324 457 adb_root(0), \
mas01mc@324 458 adb_feature_root(0), \
mas01mc@324 459 powerfd(0), \
mas01mc@292 460 dbfid(0), \
mas01mc@292 461 lshfid(0), \
mas01mc@292 462 forWrite(false), \
mas01mc@292 463 infid(0), \
mas01mc@292 464 db(0), \
mas01mc@292 465 indata(0), \
mas01mc@292 466 dbH(0), \
mas01mc@292 467 rng(0), \
mas01mc@292 468 fileTable(0), \
mas01mc@292 469 trackTable(0), \
mas01mc@292 470 trackOffsetTable(0), \
mas01mc@292 471 dataBuf(0), \
mas01mc@292 472 l2normTable(0), \
mas01mc@292 473 timesTable(0), \
mas01mc@314 474 powerTable(0), \
mas01mc@324 475 featureFileNameTable(0), \
mas01mc@324 476 timesFileNameTable(0), \
mas01mc@324 477 powerFileNameTable(0), \
mas01mc@292 478 fileTableLength(0), \
mas01mc@292 479 trackTableLength(0), \
mas01mc@292 480 dataBufLength(0), \
mas01mc@292 481 timesTableLength(0), \
mas01mc@292 482 powerTableLength(0), \
mas01mc@292 483 l2normTableLength(0), \
mas01mc@292 484 verbosity(1), \
mas01mc@292 485 nsamples(2000), \
mas01mc@292 486 datasize(O2_DEFAULT_DATASIZE), \
mas01mc@292 487 ntracks(O2_DEFAULT_NTRACKS), \
mas01mc@292 488 datadim(O2_DEFAULT_DATADIM), \
mas01mc@292 489 queryType(O2_POINT_QUERY), \
mas01mc@292 490 pointNN(O2_DEFAULT_POINTNN), \
mas01mc@292 491 trackNN(O2_DEFAULT_TRACKNN), \
mas01mc@292 492 sequenceLength(16), \
mas01mc@292 493 sequenceHop(1), \
mas01mc@292 494 normalizedDistance(true), \
mas01mc@292 495 no_unit_norming(false), \
mas01mc@292 496 queryPoint(0), \
mas01mc@292 497 usingQueryPoint(0), \
mas01mc@292 498 usingTimes(0), \
mas01mc@292 499 usingPower(0), \
mas01mc@292 500 isClient(0), \
mas01mc@292 501 isServer(0), \
mas01mc@292 502 port(0), \
mas01mc@292 503 timesTol(0.1), \
mas01mc@292 504 radius(0), \
mas01mc@292 505 query_from_key(false), \
mas01mc@292 506 query_from_key_index(O2_ERR_KEYNOTFOUND), \
mas01mc@292 507 use_absolute_threshold(false), \
mas01mc@292 508 absolute_threshold(0.0), \
mas01mc@292 509 use_relative_threshold(false), \
mas01mc@292 510 relative_threshold(0.0), \
mas01mc@292 511 reporter(0), \
mas01mc@292 512 exact_evaluation_queue(0), \
mas01mc@414 513 allowed_keys(0), \
mas01mc@334 514 lisztOffset(0), \
mas01mc@334 515 lisztLength(0), \
mas01cr@370 516 apierrortemp(0), \
mas01cr@370 517 UseApiError(0), \
mas01mc@292 518 lsh(0), \
mas01mc@292 519 lsh_in_core(false), \
mas01mc@292 520 lsh_use_u_functions(false), \
mas01mc@292 521 lsh_exact(false), \
mas01mc@308 522 WS_load_index(false), \
mas01mc@292 523 lsh_param_k(0), \
mas01mc@292 524 lsh_param_m(0), \
mas01mc@292 525 lsh_param_N(0), \
mas01mc@292 526 lsh_param_b(0), \
mas01mc@292 527 lsh_param_ncols(0), \
mas01mc@324 528 lsh_n_point_bits(0), \
mas01cr@370 529 vv(0)
mas01mc@292 530 #endif