annotate audioDB.h @ 258:d12364d8b9ea adding-emd

added cmdline stubs for distance switch and skeleton methods for EMD
author map01bf
date Fri, 25 Apr 2008 17:40:19 +0000
parents bfd34e8c84fb
children
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@0 32
mas01cr@0 33 // parameters
mas01cr@0 34 #define COM_CLIENT "--client"
mas01cr@0 35 #define COM_DATABASE "--database"
mas01cr@0 36 #define COM_QTYPE "--qtype"
mas01cr@0 37 #define COM_SEQLEN "--sequencelength"
mas01cr@0 38 #define COM_SEQHOP "--sequencehop"
mas01cr@0 39 #define COM_POINTNN "--pointnn"
mas01mc@18 40 #define COM_TRACKNN "--resultlength"
mas01cr@0 41 #define COM_QPOINT "--qpoint"
mas01cr@0 42 #define COM_FEATURES "--features"
mas01cr@0 43 #define COM_QUERYKEY "--key"
mas01cr@0 44 #define COM_KEYLIST "--keyList"
mas01cr@0 45 #define COM_TIMES "--times"
mas01cr@193 46 #define COM_QUERYPOWER "--power"
mas01cr@193 47 #define COM_RELATIVE_THRESH "--relative-threshold"
mas01cr@193 48 #define COM_ABSOLUTE_THRESH "--absolute-threshold"
mas01cr@0 49
mas01cr@108 50 #define O2_OLD_MAGIC ('O'|'2'<<8|'D'<<16|'B'<<24)
mas01cr@108 51 #define O2_MAGIC ('o'|'2'<<8|'d'<<16|'b'<<24)
mas01cr@210 52 #define O2_FORMAT_VERSION (4U)
mas01cr@0 53
mas01cr@0 54 #define O2_DEFAULT_POINTNN (10U)
mas01mc@18 55 #define O2_DEFAULT_TRACKNN (10U)
mas01cr@0 56
map01bf@255 57 //#define O2_DEFAULTDBSIZE (4000000000) // 4GB table size
mas01mc@7 58 #define O2_DEFAULTDBSIZE (2000000000) // 2GB table size
mas01cr@0 59
mas01cr@239 60 #define O2_MAXFILES (20000U)
mas01cr@0 61 #define O2_MAXFILESTR (256U)
mas01cr@0 62 #define O2_FILETABLESIZE (O2_MAXFILESTR)
mas01mc@18 63 #define O2_TRACKTABLESIZE (sizeof(unsigned))
mas01cr@0 64 #define O2_HEADERSIZE (sizeof(dbTableHeaderT))
mas01cr@0 65 #define O2_MEANNUMVECTORS (1000U)
mas01cr@0 66 #define O2_MAXDIM (1000U)
mas01mc@17 67 #define O2_MAXNN (10000U)
mas01cr@0 68
mas01cr@0 69 // Flags
mas01cr@0 70 #define O2_FLAG_L2NORM (0x1U)
mas01cr@0 71 #define O2_FLAG_MINMAX (0x2U)
mas01cr@193 72 #define O2_FLAG_POWER (0x4U)
mas01cr@0 73 #define O2_FLAG_TIMES (0x20U)
mas01cr@0 74
mas01cr@105 75 // Query types
mas01cr@105 76 #define O2_POINT_QUERY (0x4U)
mas01cr@105 77 #define O2_SEQUENCE_QUERY (0x8U)
mas01cr@105 78 #define O2_TRACK_QUERY (0x10U)
map01bf@255 79 #define O2_N_SEQUENCE_QUERY (0x20U)
map01bf@255 80
map01bf@258 81 // Distance types
map01bf@258 82 #define EUCLIDEAN (0x1U)
map01bf@258 83 #define EARTHMOVERS (0x2U)
mas01cr@105 84
mas01cr@0 85 // Error Codes
mas01cr@0 86 #define O2_ERR_KEYNOTFOUND (0xFFFFFF00)
mas01cr@0 87
mas01cr@0 88 // Macros
mas01cr@0 89 #define O2_ACTION(a) (strcmp(command,a)==0)
mas01cr@0 90
mas01cr@108 91 #define ALIGN_UP(x,w) ((x) + ((1<<w)-1) & ~((1<<w)-1))
mas01cr@108 92 #define ALIGN_DOWN(x,w) ((x) & ~((1<<w)-1))
mas01cr@108 93
mas01cr@196 94 #define ALIGN_PAGE_UP(x) ((x) + (getpagesize()-1) & ~(getpagesize()-1))
mas01cr@196 95 #define ALIGN_PAGE_DOWN(x) ((x) & ~(getpagesize()-1))
mas01cr@196 96
mas01cr@166 97 #define ENSURE_STRING(x) ((x) ? (x) : "")
mas01cr@166 98
mas01cr@239 99 #define CHECKED_MMAP(type, var, start, length) \
mas01cr@239 100 { void *tmp = mmap(0, length, (PROT_READ | (forWrite ? PROT_WRITE : 0)), MAP_SHARED, dbfid, (start)); \
mas01cr@239 101 if(tmp == (void *) -1) { \
mas01cr@239 102 error("mmap error for db table", #var, "mmap"); \
mas01cr@239 103 } \
mas01cr@239 104 var = (type) tmp; \
mas01cr@239 105 }
mas01cr@239 106
mas01cr@239 107 #define VERB_LOG(vv, ...) \
mas01cr@239 108 if(verbosity > vv) { \
mas01cr@239 109 fprintf(stderr, __VA_ARGS__); \
mas01cr@239 110 fflush(stderr); \
mas01cr@239 111 }
mas01cr@0 112
mas01cr@210 113 typedef struct dbTableHeader {
mas01cr@114 114 uint32_t magic;
mas01cr@114 115 uint32_t version;
mas01cr@114 116 uint32_t numFiles;
mas01cr@114 117 uint32_t dim;
mas01cr@114 118 uint32_t flags;
mas01cr@210 119 uint32_t headerSize;
mas01cr@196 120 off_t length;
mas01cr@196 121 off_t fileTableOffset;
mas01cr@196 122 off_t trackTableOffset;
mas01cr@196 123 off_t dataOffset;
mas01cr@196 124 off_t l2normTableOffset;
mas01cr@196 125 off_t timesTableOffset;
mas01cr@196 126 off_t powerTableOffset;
mas01cr@196 127 off_t dbSize;
mas01cr@0 128 } dbTableHeaderT, *dbTableHeaderPtr;
mas01cr@0 129
mas01cr@239 130 class Reporter;
mas01cr@0 131
mas01cr@0 132 class audioDB{
mas01cr@0 133
mas01cr@0 134 private:
mas01cr@0 135 gengetopt_args_info args_info;
mas01cr@0 136 unsigned dim;
mas01cr@0 137 const char *dbName;
mas01cr@0 138 const char *inFile;
mas01cr@0 139 const char *hostport;
mas01cr@0 140 const char *key;
mas01mc@18 141 const char* trackFileName;
mas01cr@239 142 std::ifstream *trackFile;
mas01cr@0 143 const char *command;
mas01cr@131 144 const char *output;
mas01cr@0 145 const char *timesFileName;
mas01cr@239 146 std::ifstream *timesFile;
mas01cr@193 147 const char *powerFileName;
mas01cr@239 148 std::ifstream *powerFile;
mas01cr@193 149 int powerfd;
mas01cr@0 150
mas01cr@0 151 int dbfid;
mas01cr@196 152 bool forWrite;
mas01cr@0 153 int infid;
mas01cr@0 154 char* db;
mas01cr@0 155 char* indata;
mas01cr@0 156 struct stat statbuf;
mas01cr@0 157 dbTableHeaderPtr dbH;
mas01cr@0 158
mas01cr@0 159 char *fileTable;
mas01mc@18 160 unsigned* trackTable;
mas01cr@0 161 double* dataBuf;
mas01cr@0 162 double* inBuf;
mas01cr@0 163 double* l2normTable;
mas01cr@196 164 double* timesTable;
mas01cr@193 165 double* powerTable;
mas01cr@0 166
mas01cr@196 167 size_t fileTableLength;
mas01cr@196 168 size_t trackTableLength;
mas01cr@196 169 off_t dataBufLength;
mas01cr@196 170 size_t timesTableLength;
mas01cr@196 171 size_t powerTableLength;
mas01cr@196 172 size_t l2normTableLength;
mas01cr@196 173
mas01cr@0 174 // Flags and parameters
mas01cr@0 175 unsigned verbosity; // how much do we want to know?
mas01cr@196 176 off_t size; // given size (for creation)
mas01cr@0 177 unsigned queryType; // point queries default
map01bf@258 178 unsigned distType; // euclidean default, added to allow EMD switch
mas01cr@0 179 unsigned pointNN; // how many point NNs ?
mas01mc@18 180 unsigned trackNN; // how many track NNs ?
mas01cr@0 181 unsigned sequenceLength;
mas01cr@0 182 unsigned sequenceHop;
mas01cr@239 183 bool normalizedDistance;
mas01cr@0 184 unsigned queryPoint;
mas01cr@0 185 unsigned usingQueryPoint;
mas01cr@0 186 unsigned usingTimes;
mas01cr@193 187 unsigned usingPower;
mas01cr@0 188 unsigned isClient;
mas01cr@0 189 unsigned isServer;
mas01cr@0 190 unsigned port;
mas01cr@0 191 double timesTol;
mas01mc@17 192 double radius;
mas01cr@193 193
mas01cr@193 194 bool use_absolute_threshold;
mas01cr@193 195 double absolute_threshold;
mas01cr@193 196 bool use_relative_threshold;
mas01cr@193 197 double relative_threshold;
mas01cr@193 198
mas01mc@17 199
mas01cr@0 200 // Timers
mas01cr@0 201 struct timeval tv1;
mas01cr@0 202 struct timeval tv2;
mas01cr@0 203
mas01cr@0 204 // private methods
mas01cr@32 205 void error(const char* a, const char* b = "", const char *sysFunc = 0);
mas01cr@193 206 void sequence_sum(double *buffer, int length, int seqlen);
mas01cr@193 207 void sequence_sqrt(double *buffer, int length, int seqlen);
mas01cr@193 208 void sequence_average(double *buffer, int length, int seqlen);
mas01cr@193 209
mas01cr@239 210 void initialize_arrays(int track, unsigned int numVectors, double *query, double *data_buffer, double **D, double **DD);
mas01cr@239 211 void delete_arrays(int track, unsigned int numVectors, double **D, double **DD);
mas01cr@239 212 void read_data(int track, double **data_buffer_p, size_t *data_buffer_size_p);
mas01cr@239 213 void set_up_query(double **qp, double **vqp, double **qnp, double **vqnp, double **qpp, double **vqpp, double *mqdp, unsigned int *nvp);
mas01cr@239 214 void set_up_db(double **snp, double **vsnp, double **spp, double **vspp, double **mddp, unsigned int *dvp);
mas01cr@240 215 void query_loop(const char* dbName, const char* inFile, Reporter *reporter);
mas01cr@0 216
mas01cr@196 217 void initDBHeader(const char *dbName);
mas01cr@169 218 void initInputFile(const char *inFile);
mas01cr@196 219 void initTables(const char* dbName, const char* inFile);
mas01cr@0 220 void unitNorm(double* X, unsigned d, unsigned n, double* qNorm);
mas01cr@0 221 void unitNormAndInsertL2(double* X, unsigned dim, unsigned n, unsigned append);
mas01cr@239 222 void insertTimeStamps(unsigned n, std::ifstream* timesFile, double* timesdata);
mas01cr@193 223 void insertPowerData(unsigned n, int powerfd, double *powerdata);
mas01cr@0 224 unsigned getKeyPos(char* key);
mas01cr@0 225 public:
mas01cr@0 226
mas01cr@76 227 audioDB(const unsigned argc, char* const argv[]);
mas01cr@133 228 audioDB(const unsigned argc, char* const argv[], adb__queryResponse *adbQueryResponse);
mas01cr@133 229 audioDB(const unsigned argc, char* const argv[], adb__statusResponse *adbStatusResponse);
mas01cr@97 230 void cleanup();
mas01cr@0 231 ~audioDB();
mas01cr@0 232 int processArgs(const unsigned argc, char* const argv[]);
mas01cr@30 233 void get_lock(int fd, bool exclusive);
mas01cr@30 234 void release_lock(int fd);
mas01cr@0 235 void create(const char* dbName);
mas01cr@0 236 void drop();
map01bf@255 237 bool enough_per_file_space_free();
mas01cr@196 238 bool enough_data_space_free(off_t size);
mas01cr@196 239 void insert_data_vectors(off_t offset, void *buffer, size_t size);
mas01cr@0 240 void insert(const char* dbName, const char* inFile);
mas01cr@0 241 void batchinsert(const char* dbName, const char* inFile);
mas01cr@133 242 void query(const char* dbName, const char* inFile, adb__queryResponse *adbQueryResponse=0);
mas01cr@133 243 void status(const char* dbName, adb__statusResponse *adbStatusResponse=0);
mas01cr@0 244 void ws_status(const char*dbName, char* hostport);
mas01mc@18 245 void ws_query(const char*dbName, const char *trackKey, const char* hostport);
mas01cr@0 246 void l2norm(const char* dbName);
mas01cr@193 247 void power_flag(const char *dbName);
mas01cr@193 248 bool powers_acceptable(double p1, double p2);
mas01cr@0 249 void dump(const char* dbName);
mas01cr@0 250
mas01cr@0 251 // web services
mas01cr@0 252 void startServer();
mas01cr@0 253
mas01cr@0 254 };
mas01mc@17 255
mas01cr@105 256 #define O2_AUDIODB_INITIALIZERS \
mas01cr@105 257 dim(0), \
mas01cr@105 258 dbName(0), \
mas01cr@105 259 inFile(0), \
mas01cr@105 260 key(0), \
mas01cr@105 261 trackFileName(0), \
mas01cr@105 262 trackFile(0), \
mas01cr@105 263 command(0), \
mas01cr@131 264 output(0), \
mas01cr@105 265 timesFileName(0), \
mas01cr@105 266 timesFile(0), \
mas01cr@193 267 powerFileName(0), \
mas01cr@193 268 powerFile(0), \
mas01cr@193 269 powerfd(0), \
mas01cr@105 270 dbfid(0), \
mas01cr@196 271 forWrite(false), \
mas01cr@105 272 infid(0), \
mas01cr@105 273 db(0), \
mas01cr@105 274 indata(0), \
mas01cr@105 275 dbH(0), \
mas01cr@105 276 fileTable(0), \
mas01cr@105 277 trackTable(0), \
mas01cr@105 278 dataBuf(0), \
mas01cr@105 279 l2normTable(0), \
mas01cr@105 280 timesTable(0), \
mas01cr@196 281 fileTableLength(0), \
mas01cr@196 282 trackTableLength(0), \
mas01cr@196 283 dataBufLength(0), \
mas01cr@196 284 timesTableLength(0), \
mas01cr@196 285 powerTableLength(0), \
mas01cr@196 286 l2normTableLength(0), \
mas01cr@105 287 verbosity(1), \
mas01cr@129 288 size(O2_DEFAULTDBSIZE), \
mas01cr@105 289 queryType(O2_POINT_QUERY), \
map01bf@258 290 distType(EUCLIDEAN), \
mas01cr@105 291 pointNN(O2_DEFAULT_POINTNN), \
mas01cr@105 292 trackNN(O2_DEFAULT_TRACKNN), \
mas01cr@105 293 sequenceLength(16), \
mas01cr@105 294 sequenceHop(1), \
mas01cr@239 295 normalizedDistance(true), \
mas01cr@105 296 queryPoint(0), \
mas01cr@105 297 usingQueryPoint(0), \
mas01cr@105 298 usingTimes(0), \
mas01cr@193 299 usingPower(0), \
mas01cr@105 300 isClient(0), \
mas01cr@105 301 isServer(0), \
mas01cr@105 302 port(0), \
mas01cr@105 303 timesTol(0.1), \
mas01cr@193 304 radius(0), \
mas01cr@193 305 use_absolute_threshold(false), \
mas01cr@193 306 absolute_threshold(0.0), \
mas01cr@193 307 use_relative_threshold(false), \
mas01cr@193 308 relative_threshold(0.0)