annotate audioDB.h @ 245:0d99b008fd6b 0.8.preview

Improve copyright message from audioDB -H Also give audioDB a less optimistic version number, of 0.8.preview. Envisaged progress: * 0.9.beta in the New Year sometime; * 1.0 mid- to end-February.
author mas01cr
date Mon, 17 Dec 2007 16:17:45 +0000
parents 4eb4608e28e1
children 5682c7d7444b bfd34e8c84fb
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
mas01mc@7 57 #define O2_DEFAULTDBSIZE (2000000000) // 2GB table size
mas01cr@0 58
mas01cr@239 59 #define O2_MAXFILES (20000U)
mas01cr@0 60 #define O2_MAXFILESTR (256U)
mas01cr@0 61 #define O2_FILETABLESIZE (O2_MAXFILESTR)
mas01mc@18 62 #define O2_TRACKTABLESIZE (sizeof(unsigned))
mas01cr@0 63 #define O2_HEADERSIZE (sizeof(dbTableHeaderT))
mas01cr@0 64 #define O2_MEANNUMVECTORS (1000U)
mas01cr@0 65 #define O2_MAXDIM (1000U)
mas01mc@17 66 #define O2_MAXNN (10000U)
mas01cr@0 67
mas01cr@0 68 // Flags
mas01cr@0 69 #define O2_FLAG_L2NORM (0x1U)
mas01cr@0 70 #define O2_FLAG_MINMAX (0x2U)
mas01cr@193 71 #define O2_FLAG_POWER (0x4U)
mas01cr@0 72 #define O2_FLAG_TIMES (0x20U)
mas01cr@0 73
mas01cr@105 74 // Query types
mas01cr@105 75 #define O2_POINT_QUERY (0x4U)
mas01cr@105 76 #define O2_SEQUENCE_QUERY (0x8U)
mas01cr@105 77 #define O2_TRACK_QUERY (0x10U)
mas01cr@105 78
mas01cr@0 79 // Error Codes
mas01cr@0 80 #define O2_ERR_KEYNOTFOUND (0xFFFFFF00)
mas01cr@0 81
mas01cr@0 82 // Macros
mas01cr@0 83 #define O2_ACTION(a) (strcmp(command,a)==0)
mas01cr@0 84
mas01cr@108 85 #define ALIGN_UP(x,w) ((x) + ((1<<w)-1) & ~((1<<w)-1))
mas01cr@108 86 #define ALIGN_DOWN(x,w) ((x) & ~((1<<w)-1))
mas01cr@108 87
mas01cr@196 88 #define ALIGN_PAGE_UP(x) ((x) + (getpagesize()-1) & ~(getpagesize()-1))
mas01cr@196 89 #define ALIGN_PAGE_DOWN(x) ((x) & ~(getpagesize()-1))
mas01cr@196 90
mas01cr@166 91 #define ENSURE_STRING(x) ((x) ? (x) : "")
mas01cr@166 92
mas01cr@239 93 #define CHECKED_MMAP(type, var, start, length) \
mas01cr@239 94 { void *tmp = mmap(0, length, (PROT_READ | (forWrite ? PROT_WRITE : 0)), MAP_SHARED, dbfid, (start)); \
mas01cr@239 95 if(tmp == (void *) -1) { \
mas01cr@239 96 error("mmap error for db table", #var, "mmap"); \
mas01cr@239 97 } \
mas01cr@239 98 var = (type) tmp; \
mas01cr@239 99 }
mas01cr@239 100
mas01cr@239 101 #define VERB_LOG(vv, ...) \
mas01cr@239 102 if(verbosity > vv) { \
mas01cr@239 103 fprintf(stderr, __VA_ARGS__); \
mas01cr@239 104 fflush(stderr); \
mas01cr@239 105 }
mas01cr@0 106
mas01cr@210 107 typedef struct dbTableHeader {
mas01cr@114 108 uint32_t magic;
mas01cr@114 109 uint32_t version;
mas01cr@114 110 uint32_t numFiles;
mas01cr@114 111 uint32_t dim;
mas01cr@114 112 uint32_t flags;
mas01cr@210 113 uint32_t headerSize;
mas01cr@196 114 off_t length;
mas01cr@196 115 off_t fileTableOffset;
mas01cr@196 116 off_t trackTableOffset;
mas01cr@196 117 off_t dataOffset;
mas01cr@196 118 off_t l2normTableOffset;
mas01cr@196 119 off_t timesTableOffset;
mas01cr@196 120 off_t powerTableOffset;
mas01cr@196 121 off_t dbSize;
mas01cr@0 122 } dbTableHeaderT, *dbTableHeaderPtr;
mas01cr@0 123
mas01cr@239 124 class Reporter;
mas01cr@0 125
mas01cr@0 126 class audioDB{
mas01cr@0 127
mas01cr@0 128 private:
mas01cr@0 129 gengetopt_args_info args_info;
mas01cr@0 130 unsigned dim;
mas01cr@0 131 const char *dbName;
mas01cr@0 132 const char *inFile;
mas01cr@0 133 const char *hostport;
mas01cr@0 134 const char *key;
mas01mc@18 135 const char* trackFileName;
mas01cr@239 136 std::ifstream *trackFile;
mas01cr@0 137 const char *command;
mas01cr@131 138 const char *output;
mas01cr@0 139 const char *timesFileName;
mas01cr@239 140 std::ifstream *timesFile;
mas01cr@193 141 const char *powerFileName;
mas01cr@239 142 std::ifstream *powerFile;
mas01cr@193 143 int powerfd;
mas01cr@0 144
mas01cr@0 145 int dbfid;
mas01cr@196 146 bool forWrite;
mas01cr@0 147 int infid;
mas01cr@0 148 char* db;
mas01cr@0 149 char* indata;
mas01cr@0 150 struct stat statbuf;
mas01cr@0 151 dbTableHeaderPtr dbH;
mas01cr@0 152
mas01cr@0 153 char *fileTable;
mas01mc@18 154 unsigned* trackTable;
mas01cr@0 155 double* dataBuf;
mas01cr@0 156 double* inBuf;
mas01cr@0 157 double* l2normTable;
mas01cr@196 158 double* timesTable;
mas01cr@193 159 double* powerTable;
mas01cr@0 160
mas01cr@196 161 size_t fileTableLength;
mas01cr@196 162 size_t trackTableLength;
mas01cr@196 163 off_t dataBufLength;
mas01cr@196 164 size_t timesTableLength;
mas01cr@196 165 size_t powerTableLength;
mas01cr@196 166 size_t l2normTableLength;
mas01cr@196 167
mas01cr@0 168 // Flags and parameters
mas01cr@0 169 unsigned verbosity; // how much do we want to know?
mas01cr@196 170 off_t size; // given size (for creation)
mas01cr@0 171 unsigned queryType; // point queries default
mas01cr@0 172 unsigned pointNN; // how many point NNs ?
mas01mc@18 173 unsigned trackNN; // how many track NNs ?
mas01cr@0 174 unsigned sequenceLength;
mas01cr@0 175 unsigned sequenceHop;
mas01cr@239 176 bool normalizedDistance;
mas01cr@0 177 unsigned queryPoint;
mas01cr@0 178 unsigned usingQueryPoint;
mas01cr@0 179 unsigned usingTimes;
mas01cr@193 180 unsigned usingPower;
mas01cr@0 181 unsigned isClient;
mas01cr@0 182 unsigned isServer;
mas01cr@0 183 unsigned port;
mas01cr@0 184 double timesTol;
mas01mc@17 185 double radius;
mas01cr@193 186
mas01cr@193 187 bool use_absolute_threshold;
mas01cr@193 188 double absolute_threshold;
mas01cr@193 189 bool use_relative_threshold;
mas01cr@193 190 double relative_threshold;
mas01cr@193 191
mas01mc@17 192
mas01cr@0 193 // Timers
mas01cr@0 194 struct timeval tv1;
mas01cr@0 195 struct timeval tv2;
mas01cr@0 196
mas01cr@0 197 // private methods
mas01cr@32 198 void error(const char* a, const char* b = "", const char *sysFunc = 0);
mas01cr@193 199 void sequence_sum(double *buffer, int length, int seqlen);
mas01cr@193 200 void sequence_sqrt(double *buffer, int length, int seqlen);
mas01cr@193 201 void sequence_average(double *buffer, int length, int seqlen);
mas01cr@193 202
mas01cr@239 203 void initialize_arrays(int track, unsigned int numVectors, double *query, double *data_buffer, double **D, double **DD);
mas01cr@239 204 void delete_arrays(int track, unsigned int numVectors, double **D, double **DD);
mas01cr@239 205 void read_data(int track, double **data_buffer_p, size_t *data_buffer_size_p);
mas01cr@239 206 void set_up_query(double **qp, double **vqp, double **qnp, double **vqnp, double **qpp, double **vqpp, double *mqdp, unsigned int *nvp);
mas01cr@239 207 void set_up_db(double **snp, double **vsnp, double **spp, double **vspp, double **mddp, unsigned int *dvp);
mas01cr@240 208 void query_loop(const char* dbName, const char* inFile, Reporter *reporter);
mas01cr@0 209
mas01cr@196 210 void initDBHeader(const char *dbName);
mas01cr@169 211 void initInputFile(const char *inFile);
mas01cr@196 212 void initTables(const char* dbName, const char* inFile);
mas01cr@0 213 void unitNorm(double* X, unsigned d, unsigned n, double* qNorm);
mas01cr@0 214 void unitNormAndInsertL2(double* X, unsigned dim, unsigned n, unsigned append);
mas01cr@239 215 void insertTimeStamps(unsigned n, std::ifstream* timesFile, double* timesdata);
mas01cr@193 216 void insertPowerData(unsigned n, int powerfd, double *powerdata);
mas01cr@0 217 unsigned getKeyPos(char* key);
mas01cr@0 218 public:
mas01cr@0 219
mas01cr@76 220 audioDB(const unsigned argc, char* const argv[]);
mas01cr@133 221 audioDB(const unsigned argc, char* const argv[], adb__queryResponse *adbQueryResponse);
mas01cr@133 222 audioDB(const unsigned argc, char* const argv[], adb__statusResponse *adbStatusResponse);
mas01cr@97 223 void cleanup();
mas01cr@0 224 ~audioDB();
mas01cr@0 225 int processArgs(const unsigned argc, char* const argv[]);
mas01cr@30 226 void get_lock(int fd, bool exclusive);
mas01cr@30 227 void release_lock(int fd);
mas01cr@0 228 void create(const char* dbName);
mas01cr@0 229 void drop();
mas01cr@196 230 bool enough_data_space_free(off_t size);
mas01cr@196 231 void insert_data_vectors(off_t offset, void *buffer, size_t size);
mas01cr@0 232 void insert(const char* dbName, const char* inFile);
mas01cr@0 233 void batchinsert(const char* dbName, const char* inFile);
mas01cr@133 234 void query(const char* dbName, const char* inFile, adb__queryResponse *adbQueryResponse=0);
mas01cr@133 235 void status(const char* dbName, adb__statusResponse *adbStatusResponse=0);
mas01cr@0 236 void ws_status(const char*dbName, char* hostport);
mas01mc@18 237 void ws_query(const char*dbName, const char *trackKey, const char* hostport);
mas01cr@0 238 void l2norm(const char* dbName);
mas01cr@193 239 void power_flag(const char *dbName);
mas01cr@193 240 bool powers_acceptable(double p1, double p2);
mas01cr@0 241 void dump(const char* dbName);
mas01cr@0 242
mas01cr@0 243 // web services
mas01cr@0 244 void startServer();
mas01cr@0 245
mas01cr@0 246 };
mas01mc@17 247
mas01cr@105 248 #define O2_AUDIODB_INITIALIZERS \
mas01cr@105 249 dim(0), \
mas01cr@105 250 dbName(0), \
mas01cr@105 251 inFile(0), \
mas01cr@105 252 key(0), \
mas01cr@105 253 trackFileName(0), \
mas01cr@105 254 trackFile(0), \
mas01cr@105 255 command(0), \
mas01cr@131 256 output(0), \
mas01cr@105 257 timesFileName(0), \
mas01cr@105 258 timesFile(0), \
mas01cr@193 259 powerFileName(0), \
mas01cr@193 260 powerFile(0), \
mas01cr@193 261 powerfd(0), \
mas01cr@105 262 dbfid(0), \
mas01cr@196 263 forWrite(false), \
mas01cr@105 264 infid(0), \
mas01cr@105 265 db(0), \
mas01cr@105 266 indata(0), \
mas01cr@105 267 dbH(0), \
mas01cr@105 268 fileTable(0), \
mas01cr@105 269 trackTable(0), \
mas01cr@105 270 dataBuf(0), \
mas01cr@105 271 l2normTable(0), \
mas01cr@105 272 timesTable(0), \
mas01cr@196 273 fileTableLength(0), \
mas01cr@196 274 trackTableLength(0), \
mas01cr@196 275 dataBufLength(0), \
mas01cr@196 276 timesTableLength(0), \
mas01cr@196 277 powerTableLength(0), \
mas01cr@196 278 l2normTableLength(0), \
mas01cr@105 279 verbosity(1), \
mas01cr@129 280 size(O2_DEFAULTDBSIZE), \
mas01cr@105 281 queryType(O2_POINT_QUERY), \
mas01cr@105 282 pointNN(O2_DEFAULT_POINTNN), \
mas01cr@105 283 trackNN(O2_DEFAULT_TRACKNN), \
mas01cr@105 284 sequenceLength(16), \
mas01cr@105 285 sequenceHop(1), \
mas01cr@239 286 normalizedDistance(true), \
mas01cr@105 287 queryPoint(0), \
mas01cr@105 288 usingQueryPoint(0), \
mas01cr@105 289 usingTimes(0), \
mas01cr@193 290 usingPower(0), \
mas01cr@105 291 isClient(0), \
mas01cr@105 292 isServer(0), \
mas01cr@105 293 port(0), \
mas01cr@105 294 timesTol(0.1), \
mas01cr@193 295 radius(0), \
mas01cr@193 296 use_absolute_threshold(false), \
mas01cr@193 297 absolute_threshold(0.0), \
mas01cr@193 298 use_relative_threshold(false), \
mas01cr@193 299 relative_threshold(0.0)