annotate audioDB.h @ 272:5d721f1ead01 sampling

Add utility for quickly computing x_threshold value.
author mas01cr
date Tue, 17 Jun 2008 11:43:40 +0000
parents 4ffa05f25a00
children 1c76d5f41708
rev   line source
mas01cr@0 1 #include <stdio.h>
mas01cr@0 2 #include <stdlib.h>
mas01cr@0 3 #include <sys/types.h>
mas01cr@0 4 #include <sys/stat.h>
mas01cr@0 5 #include <sys/mman.h>
mas01cr@0 6 #include <fcntl.h>
mas01cr@0 7 #include <string.h>
mas01cr@0 8 #include <iostream>
mas01cr@0 9 #include <fstream>
mas01cr@0 10 #include <math.h>
mas01cr@0 11 #include <sys/time.h>
mas01cr@0 12 #include <assert.h>
mas01cr@62 13 #include <float.h>
mas01cr@104 14 #include <signal.h>
mas01cr@0 15
mas01cr@0 16 // includes for web services
mas01cr@0 17 #include "soapH.h"
mas01cr@0 18 #include "cmdline.h"
mas01cr@0 19
mas01cr@0 20 #define MAXSTR 512
mas01cr@0 21
mas01cr@0 22 // Databse PRIMARY commands
mas01cr@0 23 #define COM_CREATE "--NEW"
mas01cr@0 24 #define COM_INSERT "--INSERT"
mas01cr@0 25 #define COM_BATCHINSERT "--BATCHINSERT"
mas01cr@0 26 #define COM_QUERY "--QUERY"
mas01cr@0 27 #define COM_STATUS "--STATUS"
mas01cr@0 28 #define COM_L2NORM "--L2NORM"
mas01cr@193 29 #define COM_POWER "--POWER"
mas01cr@0 30 #define COM_DUMP "--DUMP"
mas01cr@0 31 #define COM_SERVER "--SERVER"
mas01cr@266 32 #define COM_SAMPLE "--SAMPLE"
mas01cr@0 33
mas01cr@0 34 // parameters
mas01cr@0 35 #define COM_CLIENT "--client"
mas01cr@0 36 #define COM_DATABASE "--database"
mas01cr@0 37 #define COM_QTYPE "--qtype"
mas01cr@0 38 #define COM_SEQLEN "--sequencelength"
mas01cr@0 39 #define COM_SEQHOP "--sequencehop"
mas01cr@0 40 #define COM_POINTNN "--pointnn"
mas01mc@18 41 #define COM_TRACKNN "--resultlength"
mas01cr@0 42 #define COM_QPOINT "--qpoint"
mas01cr@0 43 #define COM_FEATURES "--features"
mas01cr@0 44 #define COM_QUERYKEY "--key"
mas01cr@0 45 #define COM_KEYLIST "--keyList"
mas01cr@0 46 #define COM_TIMES "--times"
mas01cr@193 47 #define COM_QUERYPOWER "--power"
mas01cr@193 48 #define COM_RELATIVE_THRESH "--relative-threshold"
mas01cr@193 49 #define COM_ABSOLUTE_THRESH "--absolute-threshold"
mas01cr@0 50
mas01cr@108 51 #define O2_OLD_MAGIC ('O'|'2'<<8|'D'<<16|'B'<<24)
mas01cr@108 52 #define O2_MAGIC ('o'|'2'<<8|'d'<<16|'b'<<24)
mas01cr@210 53 #define O2_FORMAT_VERSION (4U)
mas01cr@0 54
mas01cr@0 55 #define O2_DEFAULT_POINTNN (10U)
mas01mc@18 56 #define O2_DEFAULT_TRACKNN (10U)
mas01cr@0 57
mas01mc@248 58 //#define O2_DEFAULTDBSIZE (4000000000) // 4GB table size
mas01mc@7 59 #define O2_DEFAULTDBSIZE (2000000000) // 2GB table size
mas01cr@0 60
mas01cr@256 61 #define O2_DEFAULT_DATASIZE (1355U) // in MB
mas01cr@256 62 #define O2_DEFAULT_NTRACKS (20000U)
mas01cr@256 63 #define O2_DEFAULT_DATADIM (9U)
mas01cr@256 64
mas01cr@239 65 #define O2_MAXFILES (20000U)
mas01cr@0 66 #define O2_MAXFILESTR (256U)
mas01cr@256 67 #define O2_FILETABLE_ENTRY_SIZE (O2_MAXFILESTR)
mas01cr@256 68 #define O2_TRACKTABLE_ENTRY_SIZE (sizeof(unsigned))
mas01cr@0 69 #define O2_HEADERSIZE (sizeof(dbTableHeaderT))
mas01cr@0 70 #define O2_MEANNUMVECTORS (1000U)
mas01cr@0 71 #define O2_MAXDIM (1000U)
mas01mc@263 72 #define O2_MAXNN (1000000U)
mas01cr@0 73
mas01cr@0 74 // Flags
mas01cr@0 75 #define O2_FLAG_L2NORM (0x1U)
mas01cr@0 76 #define O2_FLAG_MINMAX (0x2U)
mas01cr@193 77 #define O2_FLAG_POWER (0x4U)
mas01cr@0 78 #define O2_FLAG_TIMES (0x20U)
mas01cr@0 79
mas01cr@105 80 // Query types
mas01cr@105 81 #define O2_POINT_QUERY (0x4U)
mas01cr@105 82 #define O2_SEQUENCE_QUERY (0x8U)
mas01cr@105 83 #define O2_TRACK_QUERY (0x10U)
mas01mc@248 84 #define O2_N_SEQUENCE_QUERY (0x20U)
mas01mc@263 85 #define O2_ONE_TO_ONE_N_SEQUENCE_QUERY (0x40U)
mas01mc@248 86
mas01cr@105 87
mas01cr@0 88 // Error Codes
mas01cr@0 89 #define O2_ERR_KEYNOTFOUND (0xFFFFFF00)
mas01cr@0 90
mas01cr@0 91 // Macros
mas01cr@0 92 #define O2_ACTION(a) (strcmp(command,a)==0)
mas01cr@0 93
mas01cr@108 94 #define ALIGN_UP(x,w) ((x) + ((1<<w)-1) & ~((1<<w)-1))
mas01cr@108 95 #define ALIGN_DOWN(x,w) ((x) & ~((1<<w)-1))
mas01cr@108 96
mas01cr@196 97 #define ALIGN_PAGE_UP(x) ((x) + (getpagesize()-1) & ~(getpagesize()-1))
mas01cr@196 98 #define ALIGN_PAGE_DOWN(x) ((x) & ~(getpagesize()-1))
mas01cr@196 99
mas01cr@166 100 #define ENSURE_STRING(x) ((x) ? (x) : "")
mas01cr@166 101
mas01cr@239 102 #define CHECKED_MMAP(type, var, start, length) \
mas01cr@239 103 { void *tmp = mmap(0, length, (PROT_READ | (forWrite ? PROT_WRITE : 0)), MAP_SHARED, dbfid, (start)); \
mas01cr@239 104 if(tmp == (void *) -1) { \
mas01cr@239 105 error("mmap error for db table", #var, "mmap"); \
mas01cr@239 106 } \
mas01cr@239 107 var = (type) tmp; \
mas01cr@239 108 }
mas01cr@239 109
mas01cr@239 110 #define VERB_LOG(vv, ...) \
mas01cr@239 111 if(verbosity > vv) { \
mas01cr@239 112 fprintf(stderr, __VA_ARGS__); \
mas01cr@239 113 fflush(stderr); \
mas01cr@239 114 }
mas01cr@0 115
mas01cr@210 116 typedef struct dbTableHeader {
mas01cr@114 117 uint32_t magic;
mas01cr@114 118 uint32_t version;
mas01cr@114 119 uint32_t numFiles;
mas01cr@114 120 uint32_t dim;
mas01cr@114 121 uint32_t flags;
mas01cr@210 122 uint32_t headerSize;
mas01cr@196 123 off_t length;
mas01cr@196 124 off_t fileTableOffset;
mas01cr@196 125 off_t trackTableOffset;
mas01cr@196 126 off_t dataOffset;
mas01cr@196 127 off_t l2normTableOffset;
mas01cr@196 128 off_t timesTableOffset;
mas01cr@196 129 off_t powerTableOffset;
mas01cr@196 130 off_t dbSize;
mas01cr@0 131 } dbTableHeaderT, *dbTableHeaderPtr;
mas01cr@0 132
mas01cr@239 133 class Reporter;
mas01cr@0 134
mas01cr@0 135 class audioDB{
mas01cr@0 136
mas01cr@0 137 private:
mas01cr@0 138 gengetopt_args_info args_info;
mas01cr@0 139 unsigned dim;
mas01cr@0 140 const char *dbName;
mas01cr@0 141 const char *inFile;
mas01cr@0 142 const char *hostport;
mas01cr@0 143 const char *key;
mas01mc@18 144 const char* trackFileName;
mas01cr@239 145 std::ifstream *trackFile;
mas01cr@0 146 const char *command;
mas01cr@131 147 const char *output;
mas01cr@0 148 const char *timesFileName;
mas01cr@239 149 std::ifstream *timesFile;
mas01cr@193 150 const char *powerFileName;
mas01cr@239 151 std::ifstream *powerFile;
mas01cr@193 152 int powerfd;
mas01cr@0 153
mas01cr@0 154 int dbfid;
mas01cr@196 155 bool forWrite;
mas01cr@0 156 int infid;
mas01cr@0 157 char* db;
mas01cr@0 158 char* indata;
mas01cr@0 159 struct stat statbuf;
mas01cr@0 160 dbTableHeaderPtr dbH;
mas01cr@0 161
mas01cr@0 162 char *fileTable;
mas01mc@18 163 unsigned* trackTable;
mas01cr@0 164 double* dataBuf;
mas01cr@0 165 double* inBuf;
mas01cr@0 166 double* l2normTable;
mas01cr@196 167 double* timesTable;
mas01cr@193 168 double* powerTable;
mas01cr@0 169
mas01cr@196 170 size_t fileTableLength;
mas01cr@196 171 size_t trackTableLength;
mas01cr@196 172 off_t dataBufLength;
mas01cr@196 173 size_t timesTableLength;
mas01cr@196 174 size_t powerTableLength;
mas01cr@196 175 size_t l2normTableLength;
mas01cr@196 176
mas01cr@0 177 // Flags and parameters
mas01cr@0 178 unsigned verbosity; // how much do we want to know?
mas01cr@256 179
mas01cr@256 180 //off_t size; // given size (for creation)
mas01cr@256 181 unsigned datasize; // size in MB
mas01cr@256 182 unsigned ntracks;
mas01cr@256 183 unsigned datadim;
mas01cr@256 184
mas01cr@0 185 unsigned queryType; // point queries default
mas01cr@0 186 unsigned pointNN; // how many point NNs ?
mas01mc@18 187 unsigned trackNN; // how many track NNs ?
mas01cr@0 188 unsigned sequenceLength;
mas01cr@0 189 unsigned sequenceHop;
mas01cr@239 190 bool normalizedDistance;
mas01cr@0 191 unsigned queryPoint;
mas01cr@0 192 unsigned usingQueryPoint;
mas01cr@0 193 unsigned usingTimes;
mas01cr@193 194 unsigned usingPower;
mas01cr@0 195 unsigned isClient;
mas01cr@0 196 unsigned isServer;
mas01cr@0 197 unsigned port;
mas01cr@0 198 double timesTol;
mas01mc@17 199 double radius;
mas01cr@193 200
mas01cr@193 201 bool use_absolute_threshold;
mas01cr@193 202 double absolute_threshold;
mas01cr@193 203 bool use_relative_threshold;
mas01cr@193 204 double relative_threshold;
mas01cr@193 205
mas01mc@17 206
mas01cr@0 207 // Timers
mas01cr@0 208 struct timeval tv1;
mas01cr@0 209 struct timeval tv2;
mas01cr@0 210
mas01cr@0 211 // private methods
mas01cr@32 212 void error(const char* a, const char* b = "", const char *sysFunc = 0);
mas01cr@193 213 void sequence_sum(double *buffer, int length, int seqlen);
mas01cr@193 214 void sequence_sqrt(double *buffer, int length, int seqlen);
mas01cr@193 215 void sequence_average(double *buffer, int length, int seqlen);
mas01cr@193 216
mas01cr@239 217 void initialize_arrays(int track, unsigned int numVectors, double *query, double *data_buffer, double **D, double **DD);
mas01cr@239 218 void delete_arrays(int track, unsigned int numVectors, double **D, double **DD);
mas01cr@239 219 void read_data(int track, double **data_buffer_p, size_t *data_buffer_size_p);
mas01cr@239 220 void set_up_query(double **qp, double **vqp, double **qnp, double **vqnp, double **qpp, double **vqpp, double *mqdp, unsigned int *nvp);
mas01cr@239 221 void set_up_db(double **snp, double **vsnp, double **spp, double **vspp, double **mddp, unsigned int *dvp);
mas01cr@240 222 void query_loop(const char* dbName, const char* inFile, Reporter *reporter);
mas01cr@0 223
mas01cr@196 224 void initDBHeader(const char *dbName);
mas01cr@169 225 void initInputFile(const char *inFile);
mas01cr@196 226 void initTables(const char* dbName, const char* inFile);
mas01cr@0 227 void unitNorm(double* X, unsigned d, unsigned n, double* qNorm);
mas01cr@0 228 void unitNormAndInsertL2(double* X, unsigned dim, unsigned n, unsigned append);
mas01cr@239 229 void insertTimeStamps(unsigned n, std::ifstream* timesFile, double* timesdata);
mas01cr@193 230 void insertPowerData(unsigned n, int powerfd, double *powerdata);
mas01cr@0 231 unsigned getKeyPos(char* key);
mas01cr@0 232 public:
mas01cr@0 233
mas01cr@76 234 audioDB(const unsigned argc, char* const argv[]);
mas01cr@133 235 audioDB(const unsigned argc, char* const argv[], adb__queryResponse *adbQueryResponse);
mas01cr@133 236 audioDB(const unsigned argc, char* const argv[], adb__statusResponse *adbStatusResponse);
mas01cr@97 237 void cleanup();
mas01cr@0 238 ~audioDB();
mas01cr@0 239 int processArgs(const unsigned argc, char* const argv[]);
mas01cr@30 240 void get_lock(int fd, bool exclusive);
mas01cr@30 241 void release_lock(int fd);
mas01cr@0 242 void create(const char* dbName);
mas01cr@0 243 void drop();
mas01cr@251 244 bool enough_per_file_space_free();
mas01cr@196 245 bool enough_data_space_free(off_t size);
mas01cr@196 246 void insert_data_vectors(off_t offset, void *buffer, size_t size);
mas01cr@0 247 void insert(const char* dbName, const char* inFile);
mas01cr@0 248 void batchinsert(const char* dbName, const char* inFile);
mas01cr@133 249 void query(const char* dbName, const char* inFile, adb__queryResponse *adbQueryResponse=0);
mas01cr@133 250 void status(const char* dbName, adb__statusResponse *adbStatusResponse=0);
mas01cr@266 251 unsigned random_track(unsigned *propTable, unsigned total);
mas01cr@266 252 void sample(const char *dbName);
mas01cr@0 253 void ws_status(const char*dbName, char* hostport);
mas01mc@18 254 void ws_query(const char*dbName, const char *trackKey, const char* hostport);
mas01cr@0 255 void l2norm(const char* dbName);
mas01cr@193 256 void power_flag(const char *dbName);
mas01cr@193 257 bool powers_acceptable(double p1, double p2);
mas01cr@0 258 void dump(const char* dbName);
mas01cr@0 259
mas01cr@0 260 // web services
mas01cr@0 261 void startServer();
mas01cr@0 262
mas01cr@0 263 };
mas01mc@17 264
mas01cr@105 265 #define O2_AUDIODB_INITIALIZERS \
mas01cr@105 266 dim(0), \
mas01cr@105 267 dbName(0), \
mas01cr@105 268 inFile(0), \
mas01cr@105 269 key(0), \
mas01cr@105 270 trackFileName(0), \
mas01cr@105 271 trackFile(0), \
mas01cr@105 272 command(0), \
mas01cr@131 273 output(0), \
mas01cr@105 274 timesFileName(0), \
mas01cr@105 275 timesFile(0), \
mas01cr@193 276 powerFileName(0), \
mas01cr@193 277 powerFile(0), \
mas01cr@193 278 powerfd(0), \
mas01cr@105 279 dbfid(0), \
mas01cr@196 280 forWrite(false), \
mas01cr@105 281 infid(0), \
mas01cr@105 282 db(0), \
mas01cr@105 283 indata(0), \
mas01cr@105 284 dbH(0), \
mas01cr@105 285 fileTable(0), \
mas01cr@105 286 trackTable(0), \
mas01cr@105 287 dataBuf(0), \
mas01cr@105 288 l2normTable(0), \
mas01cr@105 289 timesTable(0), \
mas01cr@196 290 fileTableLength(0), \
mas01cr@196 291 trackTableLength(0), \
mas01cr@196 292 dataBufLength(0), \
mas01cr@196 293 timesTableLength(0), \
mas01cr@196 294 powerTableLength(0), \
mas01cr@196 295 l2normTableLength(0), \
mas01cr@105 296 verbosity(1), \
mas01cr@256 297 datasize(O2_DEFAULT_DATASIZE), \
mas01cr@256 298 ntracks(O2_DEFAULT_NTRACKS), \
mas01cr@256 299 datadim(O2_DEFAULT_DATADIM), \
mas01cr@105 300 queryType(O2_POINT_QUERY), \
mas01cr@105 301 pointNN(O2_DEFAULT_POINTNN), \
mas01cr@105 302 trackNN(O2_DEFAULT_TRACKNN), \
mas01cr@105 303 sequenceLength(16), \
mas01cr@105 304 sequenceHop(1), \
mas01cr@239 305 normalizedDistance(true), \
mas01cr@105 306 queryPoint(0), \
mas01cr@105 307 usingQueryPoint(0), \
mas01cr@105 308 usingTimes(0), \
mas01cr@193 309 usingPower(0), \
mas01cr@105 310 isClient(0), \
mas01cr@105 311 isServer(0), \
mas01cr@105 312 port(0), \
mas01cr@105 313 timesTol(0.1), \
mas01cr@193 314 radius(0), \
mas01cr@193 315 use_absolute_threshold(false), \
mas01cr@193 316 absolute_threshold(0.0), \
mas01cr@193 317 use_relative_threshold(false), \
mas01cr@193 318 relative_threshold(0.0)