annotate audioDB.h @ 192:5f3d260ba71d no-big-mmap

Create and status of huge databases works. dump also works, but is as yet untested beyond the 4GB limit.
author mas01cr
date Mon, 19 Nov 2007 18:37:12 +0000
parents c22d53219112
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 "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@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@0 46
mas01cr@108 47 #define O2_OLD_MAGIC ('O'|'2'<<8|'D'<<16|'B'<<24)
mas01cr@108 48 #define O2_MAGIC ('o'|'2'<<8|'d'<<16|'b'<<24)
mas01cr@108 49 #define O2_FORMAT_VERSION (0U)
mas01cr@0 50
mas01cr@0 51 #define O2_DEFAULT_POINTNN (10U)
mas01mc@18 52 #define O2_DEFAULT_TRACKNN (10U)
mas01cr@0 53
mas01mc@7 54 #define O2_DEFAULTDBSIZE (2000000000) // 2GB table size
mas01mc@7 55 //#define O2_DEFAULTDBSIZE (1000000000U) // 1GB table size
mas01cr@0 56
mas01cr@0 57 //#define O2_MAXFILES (1000000)
mas01cr@0 58 #define O2_MAXFILES (10000U) // 10,000 files
mas01cr@0 59 #define O2_MAXFILESTR (256U)
mas01cr@0 60 #define O2_FILETABLESIZE (O2_MAXFILESTR)
mas01mc@18 61 #define O2_TRACKTABLESIZE (sizeof(unsigned))
mas01cr@0 62 #define O2_HEADERSIZE (sizeof(dbTableHeaderT))
mas01cr@0 63 #define O2_MEANNUMVECTORS (1000U)
mas01cr@0 64 #define O2_MAXDIM (1000U)
mas01mc@17 65 #define O2_MAXNN (10000U)
mas01cr@0 66
mas01cr@0 67 // Flags
mas01cr@0 68 #define O2_FLAG_L2NORM (0x1U)
mas01cr@0 69 #define O2_FLAG_MINMAX (0x2U)
mas01cr@0 70 #define O2_FLAG_TIMES (0x20U)
mas01cr@0 71
mas01cr@105 72 // Query types
mas01cr@105 73 #define O2_POINT_QUERY (0x4U)
mas01cr@105 74 #define O2_SEQUENCE_QUERY (0x8U)
mas01cr@105 75 #define O2_TRACK_QUERY (0x10U)
mas01cr@105 76
mas01cr@0 77 // Error Codes
mas01cr@0 78 #define O2_ERR_KEYNOTFOUND (0xFFFFFF00)
mas01cr@0 79
mas01cr@0 80 // Macros
mas01cr@0 81 #define O2_ACTION(a) (strcmp(command,a)==0)
mas01cr@0 82
mas01cr@108 83 #define ALIGN_UP(x,w) ((x) + ((1<<w)-1) & ~((1<<w)-1))
mas01cr@108 84 #define ALIGN_DOWN(x,w) ((x) & ~((1<<w)-1))
mas01cr@108 85
mas01cr@177 86 #define ALIGN_PAGE_UP(x) ((x) + (getpagesize()-1) & ~(getpagesize()-1))
mas01cr@177 87 #define ALIGN_PAGE_DOWN(x) ((x) & ~(getpagesize()-1))
mas01cr@174 88
mas01cr@166 89 #define ENSURE_STRING(x) ((x) ? (x) : "")
mas01cr@166 90
mas01cr@0 91 using namespace std;
mas01cr@0 92
mas01cr@0 93 typedef struct dbTableHeader{
mas01cr@114 94 uint32_t magic;
mas01cr@114 95 uint32_t version;
mas01cr@114 96 uint32_t numFiles;
mas01cr@114 97 uint32_t dim;
mas01cr@114 98 uint32_t flags;
mas01cr@111 99 // FIXME: these lengths and offsets should be size_t or off_t, but
mas01cr@111 100 // that causes this header (and hence audioDB files) to be
mas01cr@111 101 // unportable between 32 and 64-bit architectures. Making them
mas01cr@114 102 // uint32_t isn't the real answer, as it means we won't be able to
mas01cr@114 103 // scale to really large collections easily but it works around the
mas01cr@128 104 // problem. Expanding to 64 bits will of course need a change in
mas01cr@128 105 // file format version. -- CSR, 2007-10-05
mas01cr@192 106 off_t length;
mas01cr@192 107 off_t fileTableOffset;
mas01cr@192 108 off_t trackTableOffset;
mas01cr@192 109 off_t dataOffset;
mas01cr@192 110 off_t l2normTableOffset;
mas01cr@192 111 off_t timesTableOffset;
mas01cr@192 112 off_t dbSize;
mas01cr@0 113 } dbTableHeaderT, *dbTableHeaderPtr;
mas01cr@0 114
mas01cr@0 115
mas01cr@0 116 class audioDB{
mas01cr@0 117
mas01cr@0 118 private:
mas01cr@0 119 gengetopt_args_info args_info;
mas01cr@0 120 unsigned dim;
mas01cr@0 121 const char *dbName;
mas01cr@0 122 const char *inFile;
mas01cr@0 123 const char *hostport;
mas01cr@0 124 const char *key;
mas01mc@18 125 const char* trackFileName;
mas01mc@18 126 ifstream *trackFile;
mas01cr@0 127 const char *command;
mas01cr@131 128 const char *output;
mas01cr@0 129 const char *timesFileName;
mas01cr@0 130 ifstream *timesFile;
mas01cr@0 131
mas01cr@0 132 int dbfid;
mas01cr@177 133 bool forWrite;
mas01cr@0 134 int infid;
mas01cr@0 135 char* db;
mas01cr@0 136 char* indata;
mas01cr@0 137 struct stat statbuf;
mas01cr@0 138 dbTableHeaderPtr dbH;
mas01cr@0 139
mas01cr@0 140 char *fileTable;
mas01mc@18 141 unsigned* trackTable;
mas01cr@0 142 double* dataBuf;
mas01cr@0 143 double* inBuf;
mas01cr@0 144 double* l2normTable;
mas01cr@0 145 double* qNorm;
mas01cr@0 146 double* sNorm;
mas01cr@178 147 double* timesTable;
mas01cr@178 148
mas01cr@178 149 size_t fileTableLength;
mas01cr@178 150 size_t trackTableLength;
mas01cr@192 151 off_t dataBufLength;
mas01cr@178 152 size_t timesTableLength;
mas01cr@178 153 size_t l2normTableLength;
mas01cr@0 154
mas01cr@0 155 // Flags and parameters
mas01cr@0 156 unsigned verbosity; // how much do we want to know?
mas01cr@192 157 off_t 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@0 166 unsigned isClient;
mas01cr@0 167 unsigned isServer;
mas01cr@0 168 unsigned port;
mas01cr@0 169 double timesTol;
mas01mc@17 170 double radius;
mas01mc@17 171
mas01cr@0 172 // Timers
mas01cr@0 173 struct timeval tv1;
mas01cr@0 174 struct timeval tv2;
mas01cr@0 175
mas01cr@0 176 // private methods
mas01cr@32 177 void error(const char* a, const char* b = "", const char *sysFunc = 0);
mas01cr@133 178 void pointQuery(const char* dbName, const char* inFile, adb__queryResponse *adbQueryResponse=0);
mas01cr@133 179 void trackPointQuery(const char* dbName, const char* inFile, adb__queryResponse *adbQueryResponse=0);
mas01cr@133 180 void trackSequenceQueryNN(const char* dbName, const char* inFile, adb__queryResponse *adbQueryResponse=0);
mas01cr@133 181 void trackSequenceQueryRad(const char* dbName, const char* inFile, adb__queryResponse *adbQueryResponse=0);
mas01cr@0 182
mas01cr@177 183 void initDBHeader(const char *dbName);
mas01cr@169 184 void initInputFile(const char *inFile);
mas01cr@177 185 void initTables(const char* dbName, const char* inFile);
mas01cr@0 186 void unitNorm(double* X, unsigned d, unsigned n, double* qNorm);
mas01cr@0 187 void unitNormAndInsertL2(double* X, unsigned dim, unsigned n, unsigned append);
mas01cr@0 188 void insertTimeStamps(unsigned n, ifstream* timesFile, double* timesdata);
mas01cr@0 189 unsigned getKeyPos(char* key);
mas01cr@0 190 public:
mas01cr@0 191
mas01cr@76 192 audioDB(const unsigned argc, char* const argv[]);
mas01cr@133 193 audioDB(const unsigned argc, char* const argv[], adb__queryResponse *adbQueryResponse);
mas01cr@133 194 audioDB(const unsigned argc, char* const argv[], adb__statusResponse *adbStatusResponse);
mas01cr@97 195 void cleanup();
mas01cr@0 196 ~audioDB();
mas01cr@0 197 int processArgs(const unsigned argc, char* const argv[]);
mas01cr@30 198 void get_lock(int fd, bool exclusive);
mas01cr@30 199 void release_lock(int fd);
mas01cr@0 200 void create(const char* dbName);
mas01cr@0 201 void drop();
mas01cr@180 202 bool enough_data_space_free(off_t size);
mas01cr@181 203 void insert_data_vectors(off_t offset, void *buffer, size_t size);
mas01cr@0 204 void insert(const char* dbName, const char* inFile);
mas01cr@0 205 void batchinsert(const char* dbName, const char* inFile);
mas01cr@133 206 void query(const char* dbName, const char* inFile, adb__queryResponse *adbQueryResponse=0);
mas01cr@133 207 void status(const char* dbName, adb__statusResponse *adbStatusResponse=0);
mas01cr@0 208 void ws_status(const char*dbName, char* hostport);
mas01mc@18 209 void ws_query(const char*dbName, const char *trackKey, const char* hostport);
mas01cr@0 210 void l2norm(const char* dbName);
mas01cr@0 211 void dump(const char* dbName);
mas01cr@0 212
mas01cr@0 213 // web services
mas01cr@0 214 void startServer();
mas01cr@0 215
mas01cr@0 216 };
mas01mc@17 217
mas01cr@105 218 #define O2_AUDIODB_INITIALIZERS \
mas01cr@105 219 dim(0), \
mas01cr@105 220 dbName(0), \
mas01cr@105 221 inFile(0), \
mas01cr@105 222 key(0), \
mas01cr@105 223 trackFileName(0), \
mas01cr@105 224 trackFile(0), \
mas01cr@105 225 command(0), \
mas01cr@131 226 output(0), \
mas01cr@105 227 timesFileName(0), \
mas01cr@105 228 timesFile(0), \
mas01cr@105 229 dbfid(0), \
mas01cr@177 230 forWrite(false), \
mas01cr@105 231 infid(0), \
mas01cr@105 232 db(0), \
mas01cr@105 233 indata(0), \
mas01cr@105 234 dbH(0), \
mas01cr@105 235 fileTable(0), \
mas01cr@105 236 trackTable(0), \
mas01cr@105 237 dataBuf(0), \
mas01cr@105 238 l2normTable(0), \
mas01cr@105 239 qNorm(0), \
mas01cr@105 240 timesTable(0), \
mas01cr@178 241 fileTableLength(0), \
mas01cr@178 242 trackTableLength(0), \
mas01cr@178 243 dataBufLength(0), \
mas01cr@178 244 timesTableLength(0), \
mas01cr@178 245 l2normTableLength(0), \
mas01cr@105 246 verbosity(1), \
mas01cr@129 247 size(O2_DEFAULTDBSIZE), \
mas01cr@105 248 queryType(O2_POINT_QUERY), \
mas01cr@105 249 pointNN(O2_DEFAULT_POINTNN), \
mas01cr@105 250 trackNN(O2_DEFAULT_TRACKNN), \
mas01cr@105 251 sequenceLength(16), \
mas01cr@105 252 sequenceHop(1), \
mas01cr@105 253 queryPoint(0), \
mas01cr@105 254 usingQueryPoint(0), \
mas01cr@105 255 usingTimes(0), \
mas01cr@105 256 isClient(0), \
mas01cr@105 257 isServer(0), \
mas01cr@105 258 port(0), \
mas01cr@105 259 timesTol(0.1), \
mas01cr@105 260 radius(0)