annotate audioDB.h @ 223:e2e561eef3d6 refactoring

Bump up O2_MAXFILES. It turns out that the given value is a little bit niggardly for e.g. KSA_CHARM with 1s frames. (This does mean that we waste more space for tables; the calculation of table sizes could probably be revisited.)
author mas01cr
date Wed, 05 Dec 2007 13:51:34 +0000
parents 4a66c326bbf1
children 083956accbce
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@199 52 #define O2_FORMAT_VERSION (3U)
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@223 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@204 93 #define CHECKED_MMAP(type, var, start, length) \
mas01cr@204 94 { void *tmp = mmap(0, length, (PROT_READ | (forWrite ? PROT_WRITE : 0)), MAP_SHARED, dbfid, (start)); \
mas01cr@204 95 if(tmp == (void *) -1) { \
mas01cr@204 96 error("mmap error for db table", #var, "mmap"); \
mas01cr@204 97 } \
mas01cr@204 98 var = (type) tmp; \
mas01cr@204 99 }
mas01cr@0 100
mas01cr@0 101 typedef struct dbTableHeader{
mas01cr@114 102 uint32_t magic;
mas01cr@114 103 uint32_t version;
mas01cr@114 104 uint32_t numFiles;
mas01cr@114 105 uint32_t dim;
mas01cr@114 106 uint32_t flags;
mas01cr@196 107 off_t length;
mas01cr@196 108 off_t fileTableOffset;
mas01cr@196 109 off_t trackTableOffset;
mas01cr@196 110 off_t dataOffset;
mas01cr@196 111 off_t l2normTableOffset;
mas01cr@196 112 off_t timesTableOffset;
mas01cr@196 113 off_t powerTableOffset;
mas01cr@196 114 off_t dbSize;
mas01cr@0 115 } dbTableHeaderT, *dbTableHeaderPtr;
mas01cr@0 116
mas01cr@0 117
mas01cr@0 118 class audioDB{
mas01cr@0 119
mas01cr@0 120 private:
mas01cr@0 121 gengetopt_args_info args_info;
mas01cr@0 122 unsigned dim;
mas01cr@0 123 const char *dbName;
mas01cr@0 124 const char *inFile;
mas01cr@0 125 const char *hostport;
mas01cr@0 126 const char *key;
mas01mc@18 127 const char* trackFileName;
mas01cr@204 128 std::ifstream *trackFile;
mas01cr@0 129 const char *command;
mas01cr@131 130 const char *output;
mas01cr@0 131 const char *timesFileName;
mas01cr@204 132 std::ifstream *timesFile;
mas01cr@193 133 const char *powerFileName;
mas01cr@204 134 std::ifstream *powerFile;
mas01cr@193 135 int powerfd;
mas01cr@0 136
mas01cr@0 137 int dbfid;
mas01cr@196 138 bool forWrite;
mas01cr@0 139 int infid;
mas01cr@0 140 char* db;
mas01cr@0 141 char* indata;
mas01cr@0 142 struct stat statbuf;
mas01cr@0 143 dbTableHeaderPtr dbH;
mas01cr@0 144
mas01cr@0 145 char *fileTable;
mas01mc@18 146 unsigned* trackTable;
mas01cr@0 147 double* dataBuf;
mas01cr@0 148 double* inBuf;
mas01cr@0 149 double* l2normTable;
mas01cr@196 150 double* timesTable;
mas01cr@193 151 double* powerTable;
mas01cr@0 152
mas01cr@196 153 size_t fileTableLength;
mas01cr@196 154 size_t trackTableLength;
mas01cr@196 155 off_t dataBufLength;
mas01cr@196 156 size_t timesTableLength;
mas01cr@196 157 size_t powerTableLength;
mas01cr@196 158 size_t l2normTableLength;
mas01cr@196 159
mas01cr@0 160 // Flags and parameters
mas01cr@0 161 unsigned verbosity; // how much do we want to know?
mas01cr@196 162 off_t size; // given size (for creation)
mas01cr@0 163 unsigned queryType; // point queries default
mas01cr@0 164 unsigned pointNN; // how many point NNs ?
mas01mc@18 165 unsigned trackNN; // how many track NNs ?
mas01cr@0 166 unsigned sequenceLength;
mas01cr@0 167 unsigned sequenceHop;
mas01cr@0 168 unsigned queryPoint;
mas01cr@0 169 unsigned usingQueryPoint;
mas01cr@0 170 unsigned usingTimes;
mas01cr@193 171 unsigned usingPower;
mas01cr@0 172 unsigned isClient;
mas01cr@0 173 unsigned isServer;
mas01cr@0 174 unsigned port;
mas01cr@0 175 double timesTol;
mas01mc@17 176 double radius;
mas01cr@193 177
mas01cr@193 178 bool use_absolute_threshold;
mas01cr@193 179 double absolute_threshold;
mas01cr@193 180 bool use_relative_threshold;
mas01cr@193 181 double relative_threshold;
mas01cr@193 182
mas01mc@17 183
mas01cr@0 184 // Timers
mas01cr@0 185 struct timeval tv1;
mas01cr@0 186 struct timeval tv2;
mas01cr@0 187
mas01cr@0 188 // private methods
mas01cr@32 189 void error(const char* a, const char* b = "", const char *sysFunc = 0);
mas01cr@193 190 void sequence_sum(double *buffer, int length, int seqlen);
mas01cr@193 191 void sequence_sqrt(double *buffer, int length, int seqlen);
mas01cr@193 192 void sequence_average(double *buffer, int length, int seqlen);
mas01cr@193 193
mas01cr@208 194 void initialize_arrays(int track, unsigned int numVectors, double *query, double *data_buffer, double **D, double **DD);
mas01cr@211 195 void delete_arrays(int track, unsigned int numVectors, double **D, double **DD);
mas01cr@209 196 void read_data(int track, double **data_buffer_p, size_t *data_buffer_size_p);
mas01cr@222 197 void set_up_query(double **qp, double **vqp, double **qnp, double **vqnp, double **qpp, double **vqpp, double *mqdp, unsigned int *nvp);
mas01cr@222 198 void set_up_db(double **snp, double **vsnp, double **spp, double **vspp, double **mddp, unsigned int *dvp);
mas01cr@133 199 void trackSequenceQueryNN(const char* dbName, const char* inFile, adb__queryResponse *adbQueryResponse=0);
mas01cr@133 200 void trackSequenceQueryRad(const char* dbName, const char* inFile, adb__queryResponse *adbQueryResponse=0);
mas01cr@0 201
mas01cr@196 202 void initDBHeader(const char *dbName);
mas01cr@169 203 void initInputFile(const char *inFile);
mas01cr@196 204 void initTables(const char* dbName, const char* inFile);
mas01cr@0 205 void unitNorm(double* X, unsigned d, unsigned n, double* qNorm);
mas01cr@0 206 void unitNormAndInsertL2(double* X, unsigned dim, unsigned n, unsigned append);
mas01cr@204 207 void insertTimeStamps(unsigned n, std::ifstream* timesFile, double* timesdata);
mas01cr@193 208 void insertPowerData(unsigned n, int powerfd, double *powerdata);
mas01cr@0 209 unsigned getKeyPos(char* key);
mas01cr@0 210 public:
mas01cr@0 211
mas01cr@76 212 audioDB(const unsigned argc, char* const argv[]);
mas01cr@133 213 audioDB(const unsigned argc, char* const argv[], adb__queryResponse *adbQueryResponse);
mas01cr@133 214 audioDB(const unsigned argc, char* const argv[], adb__statusResponse *adbStatusResponse);
mas01cr@97 215 void cleanup();
mas01cr@0 216 ~audioDB();
mas01cr@0 217 int processArgs(const unsigned argc, char* const argv[]);
mas01cr@30 218 void get_lock(int fd, bool exclusive);
mas01cr@30 219 void release_lock(int fd);
mas01cr@0 220 void create(const char* dbName);
mas01cr@0 221 void drop();
mas01cr@196 222 bool enough_data_space_free(off_t size);
mas01cr@196 223 void insert_data_vectors(off_t offset, void *buffer, size_t size);
mas01cr@0 224 void insert(const char* dbName, const char* inFile);
mas01cr@0 225 void batchinsert(const char* dbName, const char* inFile);
mas01cr@133 226 void query(const char* dbName, const char* inFile, adb__queryResponse *adbQueryResponse=0);
mas01cr@133 227 void status(const char* dbName, adb__statusResponse *adbStatusResponse=0);
mas01cr@0 228 void ws_status(const char*dbName, char* hostport);
mas01mc@18 229 void ws_query(const char*dbName, const char *trackKey, const char* hostport);
mas01cr@0 230 void l2norm(const char* dbName);
mas01cr@193 231 void power_flag(const char *dbName);
mas01cr@193 232 bool powers_acceptable(double p1, double p2);
mas01cr@0 233 void dump(const char* dbName);
mas01cr@0 234
mas01cr@0 235 // web services
mas01cr@0 236 void startServer();
mas01cr@0 237
mas01cr@0 238 };
mas01mc@17 239
mas01cr@105 240 #define O2_AUDIODB_INITIALIZERS \
mas01cr@105 241 dim(0), \
mas01cr@105 242 dbName(0), \
mas01cr@105 243 inFile(0), \
mas01cr@105 244 key(0), \
mas01cr@105 245 trackFileName(0), \
mas01cr@105 246 trackFile(0), \
mas01cr@105 247 command(0), \
mas01cr@131 248 output(0), \
mas01cr@105 249 timesFileName(0), \
mas01cr@105 250 timesFile(0), \
mas01cr@193 251 powerFileName(0), \
mas01cr@193 252 powerFile(0), \
mas01cr@193 253 powerfd(0), \
mas01cr@105 254 dbfid(0), \
mas01cr@196 255 forWrite(false), \
mas01cr@105 256 infid(0), \
mas01cr@105 257 db(0), \
mas01cr@105 258 indata(0), \
mas01cr@105 259 dbH(0), \
mas01cr@105 260 fileTable(0), \
mas01cr@105 261 trackTable(0), \
mas01cr@105 262 dataBuf(0), \
mas01cr@105 263 l2normTable(0), \
mas01cr@105 264 timesTable(0), \
mas01cr@196 265 fileTableLength(0), \
mas01cr@196 266 trackTableLength(0), \
mas01cr@196 267 dataBufLength(0), \
mas01cr@196 268 timesTableLength(0), \
mas01cr@196 269 powerTableLength(0), \
mas01cr@196 270 l2normTableLength(0), \
mas01cr@105 271 verbosity(1), \
mas01cr@129 272 size(O2_DEFAULTDBSIZE), \
mas01cr@105 273 queryType(O2_POINT_QUERY), \
mas01cr@105 274 pointNN(O2_DEFAULT_POINTNN), \
mas01cr@105 275 trackNN(O2_DEFAULT_TRACKNN), \
mas01cr@105 276 sequenceLength(16), \
mas01cr@105 277 sequenceHop(1), \
mas01cr@105 278 queryPoint(0), \
mas01cr@105 279 usingQueryPoint(0), \
mas01cr@105 280 usingTimes(0), \
mas01cr@193 281 usingPower(0), \
mas01cr@105 282 isClient(0), \
mas01cr@105 283 isServer(0), \
mas01cr@105 284 port(0), \
mas01cr@105 285 timesTol(0.1), \
mas01cr@193 286 radius(0), \
mas01cr@193 287 use_absolute_threshold(false), \
mas01cr@193 288 absolute_threshold(0.0), \
mas01cr@193 289 use_relative_threshold(false), \
mas01cr@193 290 relative_threshold(0.0)