annotate audioDB.h @ 204:2ea1908707c7 refactoring

Filewise refactor. Break apart huge monolithic audioDB.cpp file into seven broadly independent portions: * SOAP * DB creation * insertion * query * dump * common functionality * constructor functions Remove the "using namespace std" from the header file, though that wasn't actually a problem: the problem in question is solved by including adb.nsmap in only soap.cpp. Makefile improvements.
author mas01cr
date Wed, 28 Nov 2007 15:10:28 +0000
parents 72a037f2b1e4
children 3c7c8b84e4f3
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
mas01mc@7 58 //#define O2_DEFAULTDBSIZE (1000000000U) // 1GB table size
mas01cr@0 59
mas01cr@0 60 //#define O2_MAXFILES (1000000)
mas01cr@0 61 #define O2_MAXFILES (10000U) // 10,000 files
mas01cr@0 62 #define O2_MAXFILESTR (256U)
mas01cr@0 63 #define O2_FILETABLESIZE (O2_MAXFILESTR)
mas01mc@18 64 #define O2_TRACKTABLESIZE (sizeof(unsigned))
mas01cr@0 65 #define O2_HEADERSIZE (sizeof(dbTableHeaderT))
mas01cr@0 66 #define O2_MEANNUMVECTORS (1000U)
mas01cr@0 67 #define O2_MAXDIM (1000U)
mas01mc@17 68 #define O2_MAXNN (10000U)
mas01cr@0 69
mas01cr@0 70 // Flags
mas01cr@0 71 #define O2_FLAG_L2NORM (0x1U)
mas01cr@0 72 #define O2_FLAG_MINMAX (0x2U)
mas01cr@193 73 #define O2_FLAG_POWER (0x4U)
mas01cr@0 74 #define O2_FLAG_TIMES (0x20U)
mas01cr@0 75
mas01cr@105 76 // Query types
mas01cr@105 77 #define O2_POINT_QUERY (0x4U)
mas01cr@105 78 #define O2_SEQUENCE_QUERY (0x8U)
mas01cr@105 79 #define O2_TRACK_QUERY (0x10U)
mas01cr@105 80
mas01cr@0 81 // Error Codes
mas01cr@0 82 #define O2_ERR_KEYNOTFOUND (0xFFFFFF00)
mas01cr@0 83
mas01cr@0 84 // Macros
mas01cr@0 85 #define O2_ACTION(a) (strcmp(command,a)==0)
mas01cr@0 86
mas01cr@108 87 #define ALIGN_UP(x,w) ((x) + ((1<<w)-1) & ~((1<<w)-1))
mas01cr@108 88 #define ALIGN_DOWN(x,w) ((x) & ~((1<<w)-1))
mas01cr@108 89
mas01cr@196 90 #define ALIGN_PAGE_UP(x) ((x) + (getpagesize()-1) & ~(getpagesize()-1))
mas01cr@196 91 #define ALIGN_PAGE_DOWN(x) ((x) & ~(getpagesize()-1))
mas01cr@196 92
mas01cr@166 93 #define ENSURE_STRING(x) ((x) ? (x) : "")
mas01cr@166 94
mas01cr@204 95 #define CHECKED_MMAP(type, var, start, length) \
mas01cr@204 96 { void *tmp = mmap(0, length, (PROT_READ | (forWrite ? PROT_WRITE : 0)), MAP_SHARED, dbfid, (start)); \
mas01cr@204 97 if(tmp == (void *) -1) { \
mas01cr@204 98 error("mmap error for db table", #var, "mmap"); \
mas01cr@204 99 } \
mas01cr@204 100 var = (type) tmp; \
mas01cr@204 101 }
mas01cr@0 102
mas01cr@0 103 typedef struct dbTableHeader{
mas01cr@114 104 uint32_t magic;
mas01cr@114 105 uint32_t version;
mas01cr@114 106 uint32_t numFiles;
mas01cr@114 107 uint32_t dim;
mas01cr@114 108 uint32_t flags;
mas01cr@196 109 off_t length;
mas01cr@196 110 off_t fileTableOffset;
mas01cr@196 111 off_t trackTableOffset;
mas01cr@196 112 off_t dataOffset;
mas01cr@196 113 off_t l2normTableOffset;
mas01cr@196 114 off_t timesTableOffset;
mas01cr@196 115 off_t powerTableOffset;
mas01cr@196 116 off_t dbSize;
mas01cr@0 117 } dbTableHeaderT, *dbTableHeaderPtr;
mas01cr@0 118
mas01cr@0 119
mas01cr@0 120 class audioDB{
mas01cr@0 121
mas01cr@0 122 private:
mas01cr@0 123 gengetopt_args_info args_info;
mas01cr@0 124 unsigned dim;
mas01cr@0 125 const char *dbName;
mas01cr@0 126 const char *inFile;
mas01cr@0 127 const char *hostport;
mas01cr@0 128 const char *key;
mas01mc@18 129 const char* trackFileName;
mas01cr@204 130 std::ifstream *trackFile;
mas01cr@0 131 const char *command;
mas01cr@131 132 const char *output;
mas01cr@0 133 const char *timesFileName;
mas01cr@204 134 std::ifstream *timesFile;
mas01cr@193 135 const char *powerFileName;
mas01cr@204 136 std::ifstream *powerFile;
mas01cr@193 137 int powerfd;
mas01cr@0 138
mas01cr@0 139 int dbfid;
mas01cr@196 140 bool forWrite;
mas01cr@0 141 int infid;
mas01cr@0 142 char* db;
mas01cr@0 143 char* indata;
mas01cr@0 144 struct stat statbuf;
mas01cr@0 145 dbTableHeaderPtr dbH;
mas01cr@0 146
mas01cr@0 147 char *fileTable;
mas01mc@18 148 unsigned* trackTable;
mas01cr@0 149 double* dataBuf;
mas01cr@0 150 double* inBuf;
mas01cr@0 151 double* l2normTable;
mas01cr@0 152 double* qNorm;
mas01cr@0 153 double* sNorm;
mas01cr@196 154 double* timesTable;
mas01cr@193 155 double* powerTable;
mas01cr@0 156
mas01cr@196 157 size_t fileTableLength;
mas01cr@196 158 size_t trackTableLength;
mas01cr@196 159 off_t dataBufLength;
mas01cr@196 160 size_t timesTableLength;
mas01cr@196 161 size_t powerTableLength;
mas01cr@196 162 size_t l2normTableLength;
mas01cr@196 163
mas01cr@0 164 // Flags and parameters
mas01cr@0 165 unsigned verbosity; // how much do we want to know?
mas01cr@196 166 off_t size; // given size (for creation)
mas01cr@0 167 unsigned queryType; // point queries default
mas01cr@0 168 unsigned pointNN; // how many point NNs ?
mas01mc@18 169 unsigned trackNN; // how many track NNs ?
mas01cr@0 170 unsigned sequenceLength;
mas01cr@0 171 unsigned sequenceHop;
mas01cr@0 172 unsigned queryPoint;
mas01cr@0 173 unsigned usingQueryPoint;
mas01cr@0 174 unsigned usingTimes;
mas01cr@193 175 unsigned usingPower;
mas01cr@0 176 unsigned isClient;
mas01cr@0 177 unsigned isServer;
mas01cr@0 178 unsigned port;
mas01cr@0 179 double timesTol;
mas01mc@17 180 double radius;
mas01cr@193 181
mas01cr@193 182 bool use_absolute_threshold;
mas01cr@193 183 double absolute_threshold;
mas01cr@193 184 bool use_relative_threshold;
mas01cr@193 185 double relative_threshold;
mas01cr@193 186
mas01mc@17 187
mas01cr@0 188 // Timers
mas01cr@0 189 struct timeval tv1;
mas01cr@0 190 struct timeval tv2;
mas01cr@0 191
mas01cr@0 192 // private methods
mas01cr@32 193 void error(const char* a, const char* b = "", const char *sysFunc = 0);
mas01cr@133 194 void pointQuery(const char* dbName, const char* inFile, adb__queryResponse *adbQueryResponse=0);
mas01cr@133 195 void trackPointQuery(const char* dbName, const char* inFile, adb__queryResponse *adbQueryResponse=0);
mas01cr@193 196 void sequence_sum(double *buffer, int length, int seqlen);
mas01cr@193 197 void sequence_sqrt(double *buffer, int length, int seqlen);
mas01cr@193 198 void sequence_average(double *buffer, int length, int seqlen);
mas01cr@193 199
mas01cr@133 200 void trackSequenceQueryNN(const char* dbName, const char* inFile, adb__queryResponse *adbQueryResponse=0);
mas01cr@133 201 void trackSequenceQueryRad(const char* dbName, const char* inFile, adb__queryResponse *adbQueryResponse=0);
mas01cr@0 202
mas01cr@196 203 void initDBHeader(const char *dbName);
mas01cr@169 204 void initInputFile(const char *inFile);
mas01cr@196 205 void initTables(const char* dbName, const char* inFile);
mas01cr@0 206 void unitNorm(double* X, unsigned d, unsigned n, double* qNorm);
mas01cr@0 207 void unitNormAndInsertL2(double* X, unsigned dim, unsigned n, unsigned append);
mas01cr@204 208 void insertTimeStamps(unsigned n, std::ifstream* timesFile, double* timesdata);
mas01cr@193 209 void insertPowerData(unsigned n, int powerfd, double *powerdata);
mas01cr@0 210 unsigned getKeyPos(char* key);
mas01cr@0 211 public:
mas01cr@0 212
mas01cr@76 213 audioDB(const unsigned argc, char* const argv[]);
mas01cr@133 214 audioDB(const unsigned argc, char* const argv[], adb__queryResponse *adbQueryResponse);
mas01cr@133 215 audioDB(const unsigned argc, char* const argv[], adb__statusResponse *adbStatusResponse);
mas01cr@97 216 void cleanup();
mas01cr@0 217 ~audioDB();
mas01cr@0 218 int processArgs(const unsigned argc, char* const argv[]);
mas01cr@30 219 void get_lock(int fd, bool exclusive);
mas01cr@30 220 void release_lock(int fd);
mas01cr@0 221 void create(const char* dbName);
mas01cr@0 222 void drop();
mas01cr@196 223 bool enough_data_space_free(off_t size);
mas01cr@196 224 void insert_data_vectors(off_t offset, void *buffer, size_t size);
mas01cr@0 225 void insert(const char* dbName, const char* inFile);
mas01cr@0 226 void batchinsert(const char* dbName, const char* inFile);
mas01cr@133 227 void query(const char* dbName, const char* inFile, adb__queryResponse *adbQueryResponse=0);
mas01cr@133 228 void status(const char* dbName, adb__statusResponse *adbStatusResponse=0);
mas01cr@0 229 void ws_status(const char*dbName, char* hostport);
mas01mc@18 230 void ws_query(const char*dbName, const char *trackKey, const char* hostport);
mas01cr@0 231 void l2norm(const char* dbName);
mas01cr@193 232 void power_flag(const char *dbName);
mas01cr@193 233 bool powers_acceptable(double p1, double p2);
mas01cr@0 234 void dump(const char* dbName);
mas01cr@0 235
mas01cr@0 236 // web services
mas01cr@0 237 void startServer();
mas01cr@0 238
mas01cr@0 239 };
mas01mc@17 240
mas01cr@105 241 #define O2_AUDIODB_INITIALIZERS \
mas01cr@105 242 dim(0), \
mas01cr@105 243 dbName(0), \
mas01cr@105 244 inFile(0), \
mas01cr@105 245 key(0), \
mas01cr@105 246 trackFileName(0), \
mas01cr@105 247 trackFile(0), \
mas01cr@105 248 command(0), \
mas01cr@131 249 output(0), \
mas01cr@105 250 timesFileName(0), \
mas01cr@105 251 timesFile(0), \
mas01cr@193 252 powerFileName(0), \
mas01cr@193 253 powerFile(0), \
mas01cr@193 254 powerfd(0), \
mas01cr@105 255 dbfid(0), \
mas01cr@196 256 forWrite(false), \
mas01cr@105 257 infid(0), \
mas01cr@105 258 db(0), \
mas01cr@105 259 indata(0), \
mas01cr@105 260 dbH(0), \
mas01cr@105 261 fileTable(0), \
mas01cr@105 262 trackTable(0), \
mas01cr@105 263 dataBuf(0), \
mas01cr@105 264 l2normTable(0), \
mas01cr@105 265 qNorm(0), \
mas01cr@105 266 timesTable(0), \
mas01cr@196 267 fileTableLength(0), \
mas01cr@196 268 trackTableLength(0), \
mas01cr@196 269 dataBufLength(0), \
mas01cr@196 270 timesTableLength(0), \
mas01cr@196 271 powerTableLength(0), \
mas01cr@196 272 l2normTableLength(0), \
mas01cr@105 273 verbosity(1), \
mas01cr@129 274 size(O2_DEFAULTDBSIZE), \
mas01cr@105 275 queryType(O2_POINT_QUERY), \
mas01cr@105 276 pointNN(O2_DEFAULT_POINTNN), \
mas01cr@105 277 trackNN(O2_DEFAULT_TRACKNN), \
mas01cr@105 278 sequenceLength(16), \
mas01cr@105 279 sequenceHop(1), \
mas01cr@105 280 queryPoint(0), \
mas01cr@105 281 usingQueryPoint(0), \
mas01cr@105 282 usingTimes(0), \
mas01cr@193 283 usingPower(0), \
mas01cr@105 284 isClient(0), \
mas01cr@105 285 isServer(0), \
mas01cr@105 286 port(0), \
mas01cr@105 287 timesTol(0.1), \
mas01cr@193 288 radius(0), \
mas01cr@193 289 use_absolute_threshold(false), \
mas01cr@193 290 absolute_threshold(0.0), \
mas01cr@193 291 use_relative_threshold(false), \
mas01cr@193 292 relative_threshold(0.0)