annotate audioDB.h @ 193:f9d16137e704

Merge powertable branch -r168:227 to trunk.
author mas01cr
date Wed, 21 Nov 2007 11:35:44 +0000
parents cdd441dcc9a8
children 8c81cacf5aab
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 "adb.nsmap"
mas01cr@0 19 #include "cmdline.h"
mas01cr@0 20
mas01cr@0 21 #define MAXSTR 512
mas01cr@0 22
mas01cr@0 23 // Databse PRIMARY commands
mas01cr@0 24 #define COM_CREATE "--NEW"
mas01cr@0 25 #define COM_INSERT "--INSERT"
mas01cr@0 26 #define COM_BATCHINSERT "--BATCHINSERT"
mas01cr@0 27 #define COM_QUERY "--QUERY"
mas01cr@0 28 #define COM_STATUS "--STATUS"
mas01cr@0 29 #define COM_L2NORM "--L2NORM"
mas01cr@193 30 #define COM_POWER "--POWER"
mas01cr@0 31 #define COM_DUMP "--DUMP"
mas01cr@0 32 #define COM_SERVER "--SERVER"
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@193 53 #define O2_FORMAT_VERSION (1U)
mas01cr@0 54
mas01cr@0 55 #define O2_DEFAULT_POINTNN (10U)
mas01mc@18 56 #define O2_DEFAULT_TRACKNN (10U)
mas01cr@0 57
mas01mc@7 58 #define O2_DEFAULTDBSIZE (2000000000) // 2GB table size
mas01mc@7 59 //#define O2_DEFAULTDBSIZE (1000000000U) // 1GB table size
mas01cr@0 60
mas01cr@0 61 //#define O2_MAXFILES (1000000)
mas01cr@0 62 #define O2_MAXFILES (10000U) // 10,000 files
mas01cr@0 63 #define O2_MAXFILESTR (256U)
mas01cr@0 64 #define O2_FILETABLESIZE (O2_MAXFILESTR)
mas01mc@18 65 #define O2_TRACKTABLESIZE (sizeof(unsigned))
mas01cr@0 66 #define O2_HEADERSIZE (sizeof(dbTableHeaderT))
mas01cr@0 67 #define O2_MEANNUMVECTORS (1000U)
mas01cr@0 68 #define O2_MAXDIM (1000U)
mas01mc@17 69 #define O2_MAXNN (10000U)
mas01cr@0 70
mas01cr@0 71 // Flags
mas01cr@0 72 #define O2_FLAG_L2NORM (0x1U)
mas01cr@0 73 #define O2_FLAG_MINMAX (0x2U)
mas01cr@193 74 #define O2_FLAG_POWER (0x4U)
mas01cr@0 75 #define O2_FLAG_TIMES (0x20U)
mas01cr@0 76
mas01cr@105 77 // Query types
mas01cr@105 78 #define O2_POINT_QUERY (0x4U)
mas01cr@105 79 #define O2_SEQUENCE_QUERY (0x8U)
mas01cr@105 80 #define O2_TRACK_QUERY (0x10U)
mas01cr@105 81
mas01cr@0 82 // Error Codes
mas01cr@0 83 #define O2_ERR_KEYNOTFOUND (0xFFFFFF00)
mas01cr@0 84
mas01cr@0 85 // Macros
mas01cr@0 86 #define O2_ACTION(a) (strcmp(command,a)==0)
mas01cr@0 87
mas01cr@108 88 #define ALIGN_UP(x,w) ((x) + ((1<<w)-1) & ~((1<<w)-1))
mas01cr@108 89 #define ALIGN_DOWN(x,w) ((x) & ~((1<<w)-1))
mas01cr@108 90
mas01cr@166 91 #define ENSURE_STRING(x) ((x) ? (x) : "")
mas01cr@166 92
mas01cr@0 93 using namespace std;
mas01cr@0 94
mas01cr@0 95 typedef struct dbTableHeader{
mas01cr@114 96 uint32_t magic;
mas01cr@114 97 uint32_t version;
mas01cr@114 98 uint32_t numFiles;
mas01cr@114 99 uint32_t dim;
mas01cr@114 100 uint32_t flags;
mas01cr@111 101 // FIXME: these lengths and offsets should be size_t or off_t, but
mas01cr@111 102 // that causes this header (and hence audioDB files) to be
mas01cr@111 103 // unportable between 32 and 64-bit architectures. Making them
mas01cr@114 104 // uint32_t isn't the real answer, as it means we won't be able to
mas01cr@114 105 // scale to really large collections easily but it works around the
mas01cr@128 106 // problem. Expanding to 64 bits will of course need a change in
mas01cr@128 107 // file format version. -- CSR, 2007-10-05
mas01cr@114 108 uint32_t length;
mas01cr@114 109 uint32_t fileTableOffset;
mas01cr@114 110 uint32_t trackTableOffset;
mas01cr@114 111 uint32_t dataOffset;
mas01cr@114 112 uint32_t l2normTableOffset;
mas01cr@114 113 uint32_t timesTableOffset;
mas01cr@193 114 uint32_t powerTableOffset;
mas01cr@128 115 uint32_t dbSize;
mas01cr@0 116 } dbTableHeaderT, *dbTableHeaderPtr;
mas01cr@0 117
mas01cr@0 118
mas01cr@0 119 class audioDB{
mas01cr@0 120
mas01cr@0 121 private:
mas01cr@0 122 gengetopt_args_info args_info;
mas01cr@0 123 unsigned dim;
mas01cr@0 124 const char *dbName;
mas01cr@0 125 const char *inFile;
mas01cr@0 126 const char *hostport;
mas01cr@0 127 const char *key;
mas01mc@18 128 const char* trackFileName;
mas01mc@18 129 ifstream *trackFile;
mas01cr@0 130 const char *command;
mas01cr@131 131 const char *output;
mas01cr@0 132 const char *timesFileName;
mas01cr@0 133 ifstream *timesFile;
mas01cr@193 134 const char *powerFileName;
mas01cr@193 135 ifstream *powerFile;
mas01cr@193 136 int powerfd;
mas01cr@0 137
mas01cr@0 138 int dbfid;
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@0 150 double* qNorm;
mas01cr@0 151 double* sNorm;
mas01cr@0 152 double* timesTable;
mas01cr@193 153 double* powerTable;
mas01cr@0 154
mas01cr@0 155 // Flags and parameters
mas01cr@0 156 unsigned verbosity; // how much do we want to know?
mas01cr@129 157 unsigned size; // given size (for creation)
mas01cr@0 158 unsigned queryType; // point queries default
mas01cr@0 159 unsigned pointNN; // how many point NNs ?
mas01mc@18 160 unsigned trackNN; // how many track NNs ?
mas01cr@0 161 unsigned sequenceLength;
mas01cr@0 162 unsigned sequenceHop;
mas01cr@0 163 unsigned queryPoint;
mas01cr@0 164 unsigned usingQueryPoint;
mas01cr@0 165 unsigned usingTimes;
mas01cr@193 166 unsigned usingPower;
mas01cr@0 167 unsigned isClient;
mas01cr@0 168 unsigned isServer;
mas01cr@0 169 unsigned port;
mas01cr@0 170 double timesTol;
mas01mc@17 171 double radius;
mas01cr@193 172
mas01cr@193 173 bool use_absolute_threshold;
mas01cr@193 174 double absolute_threshold;
mas01cr@193 175 bool use_relative_threshold;
mas01cr@193 176 double relative_threshold;
mas01cr@193 177
mas01mc@17 178
mas01cr@0 179 // Timers
mas01cr@0 180 struct timeval tv1;
mas01cr@0 181 struct timeval tv2;
mas01cr@0 182
mas01cr@0 183 // private methods
mas01cr@32 184 void error(const char* a, const char* b = "", const char *sysFunc = 0);
mas01cr@133 185 void pointQuery(const char* dbName, const char* inFile, adb__queryResponse *adbQueryResponse=0);
mas01cr@133 186 void trackPointQuery(const char* dbName, const char* inFile, adb__queryResponse *adbQueryResponse=0);
mas01cr@193 187 void sequence_sum(double *buffer, int length, int seqlen);
mas01cr@193 188 void sequence_sqrt(double *buffer, int length, int seqlen);
mas01cr@193 189 void sequence_average(double *buffer, int length, int seqlen);
mas01cr@193 190
mas01cr@133 191 void trackSequenceQueryNN(const char* dbName, const char* inFile, adb__queryResponse *adbQueryResponse=0);
mas01cr@133 192 void trackSequenceQueryRad(const char* dbName, const char* inFile, adb__queryResponse *adbQueryResponse=0);
mas01cr@0 193
mas01cr@120 194 void initDBHeader(const char *dbName, bool forWrite);
mas01cr@169 195 void initInputFile(const char *inFile);
mas01cr@27 196 void initTables(const char* dbName, bool forWrite, const char* inFile);
mas01cr@0 197 void unitNorm(double* X, unsigned d, unsigned n, double* qNorm);
mas01cr@0 198 void unitNormAndInsertL2(double* X, unsigned dim, unsigned n, unsigned append);
mas01cr@0 199 void insertTimeStamps(unsigned n, ifstream* timesFile, double* timesdata);
mas01cr@193 200 void insertPowerData(unsigned n, int powerfd, double *powerdata);
mas01cr@0 201 unsigned getKeyPos(char* key);
mas01cr@0 202 public:
mas01cr@0 203
mas01cr@76 204 audioDB(const unsigned argc, char* const argv[]);
mas01cr@133 205 audioDB(const unsigned argc, char* const argv[], adb__queryResponse *adbQueryResponse);
mas01cr@133 206 audioDB(const unsigned argc, char* const argv[], adb__statusResponse *adbStatusResponse);
mas01cr@97 207 void cleanup();
mas01cr@0 208 ~audioDB();
mas01cr@0 209 int processArgs(const unsigned argc, char* const argv[]);
mas01cr@30 210 void get_lock(int fd, bool exclusive);
mas01cr@30 211 void release_lock(int fd);
mas01cr@0 212 void create(const char* dbName);
mas01cr@0 213 void drop();
mas01cr@0 214 void insert(const char* dbName, const char* inFile);
mas01cr@0 215 void batchinsert(const char* dbName, const char* inFile);
mas01cr@133 216 void query(const char* dbName, const char* inFile, adb__queryResponse *adbQueryResponse=0);
mas01cr@133 217 void status(const char* dbName, adb__statusResponse *adbStatusResponse=0);
mas01cr@0 218 void ws_status(const char*dbName, char* hostport);
mas01mc@18 219 void ws_query(const char*dbName, const char *trackKey, const char* hostport);
mas01cr@0 220 void l2norm(const char* dbName);
mas01cr@193 221 void power_flag(const char *dbName);
mas01cr@193 222 bool powers_acceptable(double p1, double p2);
mas01cr@0 223 void dump(const char* dbName);
mas01cr@0 224
mas01cr@0 225 // web services
mas01cr@0 226 void startServer();
mas01cr@0 227
mas01cr@0 228 };
mas01mc@17 229
mas01cr@105 230 #define O2_AUDIODB_INITIALIZERS \
mas01cr@105 231 dim(0), \
mas01cr@105 232 dbName(0), \
mas01cr@105 233 inFile(0), \
mas01cr@105 234 key(0), \
mas01cr@105 235 trackFileName(0), \
mas01cr@105 236 trackFile(0), \
mas01cr@105 237 command(0), \
mas01cr@131 238 output(0), \
mas01cr@105 239 timesFileName(0), \
mas01cr@105 240 timesFile(0), \
mas01cr@193 241 powerFileName(0), \
mas01cr@193 242 powerFile(0), \
mas01cr@193 243 powerfd(0), \
mas01cr@105 244 dbfid(0), \
mas01cr@105 245 infid(0), \
mas01cr@105 246 db(0), \
mas01cr@105 247 indata(0), \
mas01cr@105 248 dbH(0), \
mas01cr@105 249 fileTable(0), \
mas01cr@105 250 trackTable(0), \
mas01cr@105 251 dataBuf(0), \
mas01cr@105 252 l2normTable(0), \
mas01cr@105 253 qNorm(0), \
mas01cr@105 254 timesTable(0), \
mas01cr@105 255 verbosity(1), \
mas01cr@129 256 size(O2_DEFAULTDBSIZE), \
mas01cr@105 257 queryType(O2_POINT_QUERY), \
mas01cr@105 258 pointNN(O2_DEFAULT_POINTNN), \
mas01cr@105 259 trackNN(O2_DEFAULT_TRACKNN), \
mas01cr@105 260 sequenceLength(16), \
mas01cr@105 261 sequenceHop(1), \
mas01cr@105 262 queryPoint(0), \
mas01cr@105 263 usingQueryPoint(0), \
mas01cr@105 264 usingTimes(0), \
mas01cr@193 265 usingPower(0), \
mas01cr@105 266 isClient(0), \
mas01cr@105 267 isServer(0), \
mas01cr@105 268 port(0), \
mas01cr@105 269 timesTol(0.1), \
mas01cr@193 270 radius(0), \
mas01cr@193 271 use_absolute_threshold(false), \
mas01cr@193 272 absolute_threshold(0.0), \
mas01cr@193 273 use_relative_threshold(false), \
mas01cr@193 274 relative_threshold(0.0)