[00587dc] | 1 | #include "GenericIO.h" |
---|
| 2 | #include <cstring> |
---|
| 3 | #include <vector> |
---|
| 4 | #include <string> |
---|
| 5 | #include <limits> |
---|
| 6 | #include <sstream> |
---|
| 7 | #include <sqlite3ext.h> |
---|
| 8 | SQLITE_EXTENSION_INIT1 |
---|
| 9 | |
---|
| 10 | using namespace std; |
---|
| 11 | using namespace gio; |
---|
| 12 | |
---|
| 13 | class PrinterBase { |
---|
| 14 | public: |
---|
| 15 | virtual ~PrinterBase() {} |
---|
| 16 | virtual void print(sqlite3_context *cxt, size_t i) = 0; |
---|
| 17 | }; |
---|
| 18 | |
---|
| 19 | template <class T> |
---|
| 20 | class Printer : public PrinterBase { |
---|
| 21 | public: |
---|
| 22 | Printer(GenericIO &G, size_t MNE, const string &N) |
---|
| 23 | : Data(MNE + G.requestedExtraSpace()/sizeof(T)) { |
---|
| 24 | G.addVariable(N, Data, true); |
---|
| 25 | } |
---|
| 26 | |
---|
| 27 | virtual void print(sqlite3_context *cxt, size_t i) { |
---|
| 28 | T d = Data[i]; |
---|
| 29 | if (!numeric_limits<T>::is_integer) { |
---|
| 30 | sqlite3_result_double(cxt, double(d)); |
---|
| 31 | } else { |
---|
| 32 | if (sizeof(T) > 4) { |
---|
| 33 | sqlite3_result_int64(cxt, (sqlite3_int64) d); |
---|
| 34 | } else { |
---|
| 35 | sqlite3_result_int(cxt, int(d)); |
---|
| 36 | } |
---|
| 37 | } |
---|
| 38 | } |
---|
| 39 | |
---|
| 40 | protected: |
---|
| 41 | vector<T> Data; |
---|
| 42 | }; |
---|
| 43 | |
---|
| 44 | template <typename T> |
---|
| 45 | PrinterBase *addPrinter(GenericIO::VariableInfo &V, |
---|
| 46 | GenericIO &GIO, size_t MNE) { |
---|
| 47 | if (sizeof(T) != V.Size) |
---|
| 48 | return 0; |
---|
| 49 | |
---|
| 50 | if (V.IsFloat == numeric_limits<T>::is_integer) |
---|
| 51 | return 0; |
---|
| 52 | if (V.IsSigned != numeric_limits<T>::is_signed) |
---|
| 53 | return 0; |
---|
| 54 | |
---|
| 55 | return new Printer<T>(GIO, MNE, V.Name); |
---|
| 56 | } |
---|
| 57 | |
---|
| 58 | // See: Michael Owens. Query Anything with SQLite. |
---|
| 59 | // Dr. Dobb's Journal. Nov. 2007. |
---|
| 60 | // http://www.drdobbs.com/database/query-anything-with-sqlite/202802959 |
---|
| 61 | // See: http://www.sqlite.org/c3ref/load_extension.html |
---|
| 62 | // See: http://www.sqlite.org/vtab.html |
---|
| 63 | |
---|
| 64 | struct gio_vtab { |
---|
| 65 | gio_vtab(const string &FN) : GIO(FN), MaxNumElems(0) { |
---|
| 66 | memset(&vtab, 0, sizeof(sqlite3_vtab)); |
---|
| 67 | |
---|
| 68 | GIO.openAndReadHeader(false); |
---|
| 69 | |
---|
| 70 | int NR = GIO.readNRanks(); |
---|
| 71 | for (int i = 0; i < NR; ++i) { |
---|
| 72 | size_t NElem = GIO.readNumElems(i); |
---|
| 73 | MaxNumElems = max(MaxNumElems, NElem); |
---|
| 74 | } |
---|
| 75 | } |
---|
| 76 | |
---|
| 77 | sqlite3_vtab vtab; |
---|
| 78 | GenericIO GIO; |
---|
| 79 | size_t MaxNumElems; |
---|
| 80 | }; |
---|
| 81 | |
---|
| 82 | struct gio_cursor { |
---|
| 83 | gio_cursor(const GenericIO &G, size_t MaxNElem) : GIO(G) { |
---|
| 84 | memset(&cursor, 0, sizeof(sqlite3_vtab_cursor)); |
---|
| 85 | |
---|
| 86 | vector<GenericIO::VariableInfo> VI; |
---|
| 87 | GIO.getVariableInfo(VI); |
---|
| 88 | |
---|
| 89 | Vars.resize(VI.size()); |
---|
| 90 | for (size_t i = 0; i < VI.size(); ++i) { |
---|
| 91 | PrinterBase *P = 0; |
---|
| 92 | |
---|
| 93 | #define ADD_PRINTER(T) \ |
---|
| 94 | if (!P) P = addPrinter<T>(VI[i], GIO, MaxNElem) |
---|
| 95 | ADD_PRINTER(float); |
---|
| 96 | ADD_PRINTER(double); |
---|
| 97 | ADD_PRINTER(unsigned char); |
---|
| 98 | ADD_PRINTER(signed char); |
---|
| 99 | ADD_PRINTER(int16_t); |
---|
| 100 | ADD_PRINTER(uint16_t); |
---|
| 101 | ADD_PRINTER(int32_t); |
---|
| 102 | ADD_PRINTER(uint32_t); |
---|
| 103 | ADD_PRINTER(int64_t); |
---|
| 104 | ADD_PRINTER(uint64_t); |
---|
| 105 | #undef ADD_PRINTER |
---|
| 106 | |
---|
| 107 | if (P) Printers.push_back(P); |
---|
| 108 | } |
---|
| 109 | } |
---|
| 110 | |
---|
| 111 | sqlite3_vtab_cursor cursor; |
---|
| 112 | int Rank; |
---|
| 113 | size_t Pos; |
---|
| 114 | uint64_t AbsPos; |
---|
| 115 | GenericIO GIO; |
---|
| 116 | vector< vector<char> > Vars; |
---|
| 117 | vector<PrinterBase *> Printers; |
---|
| 118 | vector<bool> RankMask; |
---|
| 119 | }; |
---|
| 120 | |
---|
| 121 | static |
---|
| 122 | int gio_connect(sqlite3* db, void *, int argc, const char *const *argv, |
---|
| 123 | sqlite3_vtab **ppVTab, char **pzErr) { |
---|
| 124 | if (argc < 4) { |
---|
| 125 | *pzErr = sqlite3_mprintf("%s", "No input file specified!"); |
---|
| 126 | return SQLITE_ERROR; |
---|
| 127 | } |
---|
| 128 | |
---|
| 129 | *pzErr = 0; |
---|
| 130 | try { |
---|
| 131 | string FileName(argv[3]); |
---|
| 132 | |
---|
| 133 | gio_vtab *tab = new gio_vtab(FileName); |
---|
| 134 | if (!tab) |
---|
| 135 | return SQLITE_ERROR; |
---|
| 136 | |
---|
| 137 | vector<GenericIO::VariableInfo> VI; |
---|
| 138 | tab->GIO.getVariableInfo(VI); |
---|
| 139 | |
---|
| 140 | string Schema = "CREATE TABLE x(_rank INTEGER"; |
---|
| 141 | if (VI.size()) |
---|
| 142 | Schema += ", "; |
---|
| 143 | |
---|
| 144 | for (size_t i = 0; i < VI.size(); ++i) { |
---|
| 145 | if (VI[i].Size > 8) |
---|
| 146 | continue; |
---|
| 147 | |
---|
| 148 | Schema += VI[i].Name + (VI[i].IsFloat ? " REAL" : " INTEGER"); |
---|
| 149 | if (i != VI.size() - 1) |
---|
| 150 | Schema += ", "; |
---|
| 151 | } |
---|
| 152 | |
---|
| 153 | Schema += ")"; |
---|
| 154 | if (sqlite3_declare_vtab(db, Schema.c_str()) != SQLITE_OK) { |
---|
| 155 | *pzErr = sqlite3_mprintf("Could not declare schema: %s", Schema.c_str()); |
---|
| 156 | delete tab; |
---|
| 157 | return SQLITE_ERROR; |
---|
| 158 | } |
---|
| 159 | |
---|
| 160 | *ppVTab = &tab->vtab; |
---|
| 161 | } catch(exception &e) { |
---|
| 162 | *pzErr = sqlite3_mprintf("%s", e.what()); |
---|
| 163 | return SQLITE_ERROR; |
---|
| 164 | } |
---|
| 165 | |
---|
| 166 | return SQLITE_OK; |
---|
| 167 | } |
---|
| 168 | |
---|
| 169 | static |
---|
| 170 | int gio_create(sqlite3* db, void *pAux, int argc, const char *const *argv, |
---|
| 171 | sqlite3_vtab **ppVTab, char **pzErr) { |
---|
| 172 | return gio_connect(db, pAux, argc, argv, ppVTab, pzErr); |
---|
| 173 | } |
---|
| 174 | |
---|
| 175 | static |
---|
| 176 | int gio_disconnect(sqlite3_vtab *pVTab) { |
---|
| 177 | gio_vtab *tab = (gio_vtab *) pVTab; |
---|
| 178 | |
---|
| 179 | delete tab; |
---|
| 180 | return SQLITE_OK; |
---|
| 181 | } |
---|
| 182 | |
---|
| 183 | static |
---|
| 184 | int gio_destroy(sqlite3_vtab *pVTab) { |
---|
| 185 | return gio_disconnect(pVTab); |
---|
| 186 | } |
---|
| 187 | |
---|
| 188 | static |
---|
| 189 | int gio_best_index(sqlite3_vtab *pVTab, sqlite3_index_info *pInfo) { |
---|
| 190 | gio_vtab *tab = (gio_vtab *) pVTab; |
---|
| 191 | |
---|
| 192 | int IndexIndex = 0; |
---|
| 193 | stringstream IndexName; |
---|
| 194 | for (int i = 0; i < pInfo->nConstraint; ++i) { |
---|
| 195 | if (!pInfo->aConstraint[i].usable) |
---|
| 196 | continue; |
---|
| 197 | |
---|
| 198 | char opcode = 0; |
---|
| 199 | switch (pInfo->aConstraint[i].op) { |
---|
| 200 | case SQLITE_INDEX_CONSTRAINT_EQ: |
---|
| 201 | opcode = 'E'; |
---|
| 202 | break; |
---|
| 203 | case SQLITE_INDEX_CONSTRAINT_GT: |
---|
| 204 | opcode = 'g'; |
---|
| 205 | break; |
---|
| 206 | case SQLITE_INDEX_CONSTRAINT_LT: |
---|
| 207 | opcode = 'l'; |
---|
| 208 | break; |
---|
| 209 | case SQLITE_INDEX_CONSTRAINT_GE: |
---|
| 210 | opcode = 'G'; |
---|
| 211 | break; |
---|
| 212 | case SQLITE_INDEX_CONSTRAINT_LE: |
---|
| 213 | opcode = 'L'; |
---|
| 214 | break; |
---|
| 215 | } |
---|
| 216 | |
---|
| 217 | if (!opcode) |
---|
| 218 | continue; |
---|
| 219 | |
---|
| 220 | // For now, we only support column 0 (_rank) |
---|
| 221 | // TODO: support for x, y, z, etc. |
---|
| 222 | if (pInfo->aConstraint[i].iColumn != 0) |
---|
| 223 | continue; |
---|
| 224 | |
---|
| 225 | if (!IndexName.str().empty()) |
---|
| 226 | IndexName << ";"; |
---|
| 227 | IndexName << pInfo->aConstraint[i].iColumn << ":" << opcode; |
---|
| 228 | |
---|
| 229 | pInfo->aConstraintUsage[i].argvIndex = 1 + IndexIndex++; |
---|
| 230 | pInfo->aConstraintUsage[i].omit = 1; |
---|
| 231 | } |
---|
| 232 | |
---|
| 233 | if (!IndexName.str().empty()) { |
---|
| 234 | pInfo->idxNum = 1; |
---|
| 235 | pInfo->idxStr = sqlite3_mprintf("%s", IndexName.str().c_str()); |
---|
| 236 | if (!pInfo->idxStr) |
---|
| 237 | return SQLITE_NOMEM; |
---|
| 238 | |
---|
| 239 | pInfo->needToFreeIdxStr = 1; |
---|
| 240 | |
---|
| 241 | // This is a crude estimate, without the actual range information, |
---|
| 242 | // this is the best that we can do. |
---|
| 243 | if (tab->GIO.readNRanks() > 0) |
---|
| 244 | pInfo->estimatedCost = (double) tab->GIO.readNumElems(0); |
---|
| 245 | } else { |
---|
| 246 | uint64_t TotalNumElems = tab->GIO.readTotalNumElems(); |
---|
| 247 | if (TotalNumElems == (uint64_t) -1) |
---|
| 248 | TotalNumElems = ((double) tab->GIO.readNumElems(0))* |
---|
| 249 | tab->GIO.readNRanks(); |
---|
| 250 | pInfo->estimatedCost = (double) TotalNumElems; |
---|
| 251 | } |
---|
| 252 | |
---|
| 253 | if (pInfo->nOrderBy == 1 && pInfo->aOrderBy[0].iColumn == 0 && |
---|
| 254 | pInfo->aOrderBy[0].desc == 0 /* ASC */) |
---|
| 255 | pInfo->orderByConsumed = 1; |
---|
| 256 | |
---|
| 257 | return SQLITE_OK; |
---|
| 258 | } |
---|
| 259 | |
---|
| 260 | static |
---|
| 261 | int gio_open(sqlite3_vtab *pVTab, sqlite3_vtab_cursor **ppCursor) { |
---|
| 262 | gio_vtab *tab = (gio_vtab *) pVTab; |
---|
| 263 | gio_cursor *cur = new gio_cursor(tab->GIO, tab->MaxNumElems); |
---|
| 264 | if (!cur) |
---|
| 265 | return SQLITE_ERROR; |
---|
| 266 | |
---|
| 267 | cur->cursor.pVtab = pVTab; |
---|
| 268 | *ppCursor = &cur->cursor; |
---|
| 269 | |
---|
| 270 | return SQLITE_OK; |
---|
| 271 | } |
---|
| 272 | |
---|
| 273 | static |
---|
| 274 | int gio_close(sqlite3_vtab_cursor* pCursor) { |
---|
| 275 | gio_cursor *cur = (gio_cursor *) pCursor; |
---|
| 276 | |
---|
| 277 | delete cur; |
---|
| 278 | return SQLITE_OK; |
---|
| 279 | } |
---|
| 280 | |
---|
| 281 | static |
---|
| 282 | int gio_filter(sqlite3_vtab_cursor* pCursor, int idxNum, const char *idxStr, |
---|
| 283 | int argc, sqlite3_value **argv) { |
---|
| 284 | gio_cursor *cur = (gio_cursor *) pCursor; |
---|
| 285 | gio_vtab *tab = (gio_vtab *) cur->cursor.pVtab; |
---|
| 286 | |
---|
| 287 | cur->RankMask.resize(cur->GIO.readNRanks(), true); |
---|
| 288 | if (idxNum == 1) { |
---|
| 289 | stringstream ss(idxStr); |
---|
| 290 | string IndexSpec; |
---|
| 291 | for (int arg = 0; getline(ss, IndexSpec, ';'); ++arg) { |
---|
| 292 | if (arg >= argc) |
---|
| 293 | return SQLITE_ERROR; |
---|
| 294 | |
---|
| 295 | stringstream tt(IndexSpec); |
---|
| 296 | int Col; char Colon, Op; |
---|
| 297 | tt >> Col >> Colon >> Op; |
---|
| 298 | if (Colon != ':') |
---|
| 299 | return SQLITE_ERROR; |
---|
| 300 | |
---|
| 301 | // TODO: Support for other columns. |
---|
| 302 | if (Col != 0) |
---|
| 303 | return SQLITE_ERROR; |
---|
| 304 | |
---|
| 305 | int Rank = sqlite3_value_int(argv[arg]); |
---|
| 306 | for (int i = 0; i < (int) cur->RankMask.size(); ++i) { |
---|
| 307 | bool Incl = false; |
---|
| 308 | if (Op == 'E') |
---|
| 309 | Incl = (i == Rank); |
---|
| 310 | else if (Op == 'g') |
---|
| 311 | Incl = (i > Rank); |
---|
| 312 | else if (Op == 'G') |
---|
| 313 | Incl = (i >= Rank); |
---|
| 314 | else if (Op == 'l') |
---|
| 315 | Incl = (i < Rank); |
---|
| 316 | else if (Op == 'L') |
---|
| 317 | Incl = (i <= Rank); |
---|
| 318 | |
---|
| 319 | cur->RankMask[i] = cur->RankMask[i] && Incl; |
---|
| 320 | } |
---|
| 321 | } |
---|
| 322 | } |
---|
| 323 | |
---|
| 324 | cur->Pos = 0; |
---|
| 325 | cur->AbsPos = 0; |
---|
| 326 | |
---|
| 327 | cur->Rank = -1; |
---|
| 328 | do { |
---|
| 329 | try { |
---|
| 330 | ++cur->Rank; |
---|
| 331 | if (cur->Rank == cur->GIO.readNRanks()) { |
---|
| 332 | break; |
---|
| 333 | } |
---|
| 334 | |
---|
| 335 | if (cur->RankMask[cur->Rank]) |
---|
| 336 | cur->GIO.readData(cur->Rank, false); |
---|
| 337 | else |
---|
| 338 | cur->AbsPos += cur->GIO.readNumElems(cur->Rank); |
---|
| 339 | } catch (exception &e) { |
---|
| 340 | cerr << e.what() << endl; |
---|
| 341 | return SQLITE_IOERR; |
---|
| 342 | } |
---|
| 343 | } while (!cur->RankMask[cur->Rank] || |
---|
| 344 | cur->GIO.readNumElems(cur->Rank) == 0); |
---|
| 345 | |
---|
| 346 | return SQLITE_OK; |
---|
| 347 | } |
---|
| 348 | |
---|
| 349 | static |
---|
| 350 | int gio_next(sqlite3_vtab_cursor* pCursor) { |
---|
| 351 | gio_cursor *cur = (gio_cursor *) pCursor; |
---|
| 352 | |
---|
| 353 | if (cur->Pos == cur->GIO.readNumElems(cur->Rank) - 1) { |
---|
| 354 | do { |
---|
| 355 | try { |
---|
| 356 | ++cur->Rank; |
---|
| 357 | if (cur->Rank == cur->GIO.readNRanks()) { |
---|
| 358 | break; |
---|
| 359 | } |
---|
| 360 | |
---|
| 361 | if (cur->RankMask[cur->Rank]) |
---|
| 362 | cur->GIO.readData(cur->Rank, false); |
---|
| 363 | else |
---|
| 364 | cur->AbsPos += cur->GIO.readNumElems(cur->Rank); |
---|
| 365 | } catch (exception &e) { |
---|
| 366 | cerr << e.what() << endl; |
---|
| 367 | return SQLITE_IOERR; |
---|
| 368 | } |
---|
| 369 | } while (!cur->RankMask[cur->Rank] || |
---|
| 370 | cur->GIO.readNumElems(cur->Rank) == 0); |
---|
| 371 | |
---|
| 372 | cur->Pos = 0; |
---|
| 373 | } else { |
---|
| 374 | ++cur->Pos; |
---|
| 375 | } |
---|
| 376 | |
---|
| 377 | ++cur->AbsPos; |
---|
| 378 | |
---|
| 379 | return SQLITE_OK; |
---|
| 380 | } |
---|
| 381 | |
---|
| 382 | static |
---|
| 383 | int gio_eof(sqlite3_vtab_cursor* pCursor) { |
---|
| 384 | gio_cursor *cur = (gio_cursor *) pCursor; |
---|
| 385 | |
---|
| 386 | if (cur->Rank == cur->GIO.readNRanks()) |
---|
| 387 | return 1; |
---|
| 388 | |
---|
| 389 | if (cur->Rank == cur->GIO.readNRanks() - 1 && |
---|
| 390 | cur->Pos >= (cur->GIO.readNumElems(cur->Rank) - 1)) |
---|
| 391 | return 1; |
---|
| 392 | |
---|
| 393 | return 0; |
---|
| 394 | } |
---|
| 395 | |
---|
| 396 | static |
---|
| 397 | int gio_column(sqlite3_vtab_cursor* pCursor, sqlite3_context* cxt, int n) { |
---|
| 398 | gio_cursor *cur = (gio_cursor *) pCursor; |
---|
| 399 | |
---|
| 400 | if (n == 0) |
---|
| 401 | sqlite3_result_int(cxt, cur->Rank); |
---|
| 402 | else if ((n-1) >= cur->Printers.size()) |
---|
| 403 | sqlite3_result_null(cxt); |
---|
| 404 | else |
---|
| 405 | cur->Printers[n-1]->print(cxt, cur->Pos); |
---|
| 406 | |
---|
| 407 | return SQLITE_OK; |
---|
| 408 | } |
---|
| 409 | |
---|
| 410 | static |
---|
| 411 | int gio_rowid(sqlite3_vtab_cursor* pCursor, sqlite3_int64 *pRowid) { |
---|
| 412 | gio_cursor *cur = (gio_cursor *) pCursor; |
---|
| 413 | |
---|
| 414 | *pRowid = (sqlite3_int64) cur->AbsPos; |
---|
| 415 | return SQLITE_OK; |
---|
| 416 | } |
---|
| 417 | |
---|
| 418 | extern "C" |
---|
| 419 | int sq3_register_virtual_gio(sqlite3 * db) { |
---|
| 420 | static const struct sqlite3_module gio_module = { |
---|
| 421 | 2, /* iVersion */ |
---|
| 422 | gio_create, |
---|
| 423 | gio_connect, |
---|
| 424 | gio_best_index, |
---|
| 425 | gio_disconnect, |
---|
| 426 | gio_destroy, |
---|
| 427 | gio_open, |
---|
| 428 | gio_close, |
---|
| 429 | gio_filter, |
---|
| 430 | gio_next, |
---|
| 431 | gio_eof, |
---|
| 432 | gio_column, |
---|
| 433 | gio_rowid, |
---|
| 434 | 0, |
---|
| 435 | 0, |
---|
| 436 | 0, |
---|
| 437 | 0, |
---|
| 438 | 0, |
---|
| 439 | 0, |
---|
| 440 | 0, |
---|
| 441 | 0, |
---|
| 442 | 0, |
---|
| 443 | 0 |
---|
| 444 | }; |
---|
| 445 | |
---|
| 446 | return sqlite3_create_module( db, "GenericIO", &gio_module, NULL ); |
---|
| 447 | } |
---|
| 448 | |
---|
| 449 | extern "C" |
---|
| 450 | int sqlite3_extension_init(sqlite3 *db, char **, |
---|
| 451 | const sqlite3_api_routines *pApi) { |
---|
| 452 | SQLITE_EXTENSION_INIT2(pApi); |
---|
| 453 | sq3_register_virtual_gio(db); |
---|
| 454 | return SQLITE_OK; |
---|
| 455 | } |
---|
| 456 | |
---|