ext/fts3/fts3_aux.c

   1 /*
   2 ** 2011 Jan 27
   3 **
   4 ** The author disclaims copyright to this source code.  In place of
   5 ** a legal notice, here is a blessing:
   6 **
   7 **    May you do good and not evil.
   8 **    May you find forgiveness for yourself and forgive others.
   9 **    May you share freely, never taking more than you give.
  10 **
  11 ******************************************************************************
  12 **
  13 */
  14 #include "fts3Int.h"
  15 #if !defined(SQLITE_CORE) || defined(SQLITE_ENABLE_FTS3)
  16
  17 #include <string.h>
  18 #include <assert.h>
  19
  20 typedef struct Fts3auxTable Fts3auxTable;
  21 typedef struct Fts3auxCursor Fts3auxCursor;
  22
  23 struct Fts3auxTable {
  24   sqlite3_vtab base;              /* Base class used by SQLite core */
  25   Fts3Table *pFts3Tab;
  26 };
  27
  28 struct Fts3auxCursor {
  29   sqlite3_vtab_cursor base;       /* Base class used by SQLite core */
  30   Fts3MultiSegReader csr;        /* Must be right after "base" */
  31   Fts3SegFilter filter;
  32   char *zStop;
  33   int nStop;                      /* Byte-length of string zStop */
  34   int iLangid;                    /* Language id to query */
  35   int isEof;                      /* True if cursor is at EOF */
  36   sqlite3_int64 iRowid;           /* Current rowid */
  37
  38   int iCol;                       /* Current value of 'col' column */
  39   int nStat;                      /* Size of aStat[] array */
  40   struct Fts3auxColstats {
  41     sqlite3_int64 nDoc;           /* 'documents' values for current csr row */
  42     sqlite3_int64 nOcc;           /* 'occurrences' values for current csr row */
  43   } *aStat;
  44 };
  45
  46 /*
  47 ** Schema of the terms table.
  48 */
  49 #define FTS3_AUX_SCHEMA \
  50   "CREATE TABLE x(term, col, documents, occurrences, languageid HIDDEN)"
  51
  52 /*
  53 ** This function does all the work for both the xConnect and xCreate methods.
  54 ** These tables have no persistent representation of their own, so xConnect
  55 ** and xCreate are identical operations.
  56 */
  57 static int fts3auxConnectMethod(
  58   sqlite3 *db,                    /* Database connection */
  59   void *pUnused,                  /* Unused */
  60   int argc,                       /* Number of elements in argv array */
  61   const char * const *argv,       /* xCreate/xConnect argument array */
  62   sqlite3_vtab **ppVtab,          /* OUT: New sqlite3_vtab object */
  63   char **pzErr                    /* OUT: sqlite3_malloc'd error message */
  64 ){
  65   char const *zDb;                /* Name of database (e.g. "main") */
  66   char const *zFts3;              /* Name of fts3 table */
  67   int nDb;                        /* Result of strlen(zDb) */
  68   int nFts3;                      /* Result of strlen(zFts3) */
  69   int nByte;                      /* Bytes of space to allocate here */
  70   int rc;                         /* value returned by declare_vtab() */
  71   Fts3auxTable *p;                /* Virtual table object to return */
  72
  73   UNUSED_PARAMETER(pUnused);
  74
  75   /* The user should invoke this in one of two forms:
  76   **
  77   **     CREATE VIRTUAL TABLE xxx USING fts4aux(fts4-table);
  78   **     CREATE VIRTUAL TABLE xxx USING fts4aux(fts4-table-db, fts4-table);
  79   */
  80   if( argc!=4 && argc!=5 ) goto bad_args;
  81
  82   zDb = argv[1];
  83   nDb = (int)strlen(zDb);
  84   if( argc==5 ){
  85     if( nDb==4 && 0==sqlite3_strnicmp("temp", zDb, 4) ){
  86       zDb = argv[3];
  87       nDb = (int)strlen(zDb);
  88       zFts3 = argv[4];
  89     }else{
  90       goto bad_args;
  91     }
  92   }else{
  93     zFts3 = argv[3];
  94   }
  95   nFts3 = (int)strlen(zFts3);
  96
  97   rc = sqlite3_declare_vtab(db, FTS3_AUX_SCHEMA);
  98   if( rc!=SQLITE_OK ) return rc;
  99
 100   nByte = sizeof(Fts3auxTable) + sizeof(Fts3Table) + nDb + nFts3 + 2;
 101   p = (Fts3auxTable *)sqlite3_malloc(nByte);
 102   if( !p ) return SQLITE_NOMEM;
 103   memset(p, 0, nByte);
 104
 105   p->pFts3Tab = (Fts3Table *)&p[1];
 106   p->pFts3Tab->zDb = (char *)&p->pFts3Tab[1];
 107   p->pFts3Tab->zName = &p->pFts3Tab->zDb[nDb+1];
 108   p->pFts3Tab->db = db;
 109   p->pFts3Tab->nIndex = 1;
 110
 111   memcpy((char *)p->pFts3Tab->zDb, zDb, nDb);
 112   memcpy((char *)p->pFts3Tab->zName, zFts3, nFts3);
 113   sqlite3Fts3Dequote((char *)p->pFts3Tab->zName);
 114
 115   *ppVtab = (sqlite3_vtab *)p;
 116   return SQLITE_OK;
 117
 118  bad_args:
 119   sqlite3Fts3ErrMsg(pzErr, "invalid arguments to fts4aux constructor");
 120   return SQLITE_ERROR;
 121 }
 122
 123 /*
 124 ** This function does the work for both the xDisconnect and xDestroy methods.
 125 ** These tables have no persistent representation of their own, so xDisconnect
 126 ** and xDestroy are identical operations.
 127 */
 128 static int fts3auxDisconnectMethod(sqlite3_vtab *pVtab){
 129   Fts3auxTable *p = (Fts3auxTable *)pVtab;
 130   Fts3Table *pFts3 = p->pFts3Tab;
 131   int i;
 132
 133   /* Free any prepared statements held */
 134   for(i=0; i<SizeofArray(pFts3->aStmt); i++){
 135     sqlite3_finalize(pFts3->aStmt[i]);
 136   }
 137   sqlite3_free(pFts3->zSegmentsTbl);
 138   sqlite3_free(p);
 139   return SQLITE_OK;
 140 }
 141
 142 #define FTS4AUX_EQ_CONSTRAINT 1
 143 #define FTS4AUX_GE_CONSTRAINT 2
 144 #define FTS4AUX_LE_CONSTRAINT 4
 145
 146 /*
 147 ** xBestIndex - Analyze a WHERE and ORDER BY clause.
 148 */
 149 static int fts3auxBestIndexMethod(
 150   sqlite3_vtab *pVTab,
 151   sqlite3_index_info *pInfo
 152 ){
 153   int i;
 154   int iEq = -1;
 155   int iGe = -1;
 156   int iLe = -1;
 157   int iLangid = -1;
 158   int iNext = 1;                  /* Next free argvIndex value */
 159
 160   UNUSED_PARAMETER(pVTab);
 161
 162   /* This vtab delivers always results in "ORDER BY term ASC" order. */
 163   if( pInfo->nOrderBy==1
 164    && pInfo->aOrderBy[0].iColumn==0
 165    && pInfo->aOrderBy[0].desc==0
 166   ){
 167     pInfo->orderByConsumed = 1;
 168   }
 169
 170   /* Search for equality and range constraints on the "term" column.
 171   ** And equality constraints on the hidden "languageid" column. */
 172   for(i=0; i<pInfo->nConstraint; i++){
 173     if( pInfo->aConstraint[i].usable ){
 174       int op = pInfo->aConstraint[i].op;
 175       int iCol = pInfo->aConstraint[i].iColumn;
 176
 177       if( iCol==0 ){
 178         if( op==SQLITE_INDEX_CONSTRAINT_EQ ) iEq = i;
 179         if( op==SQLITE_INDEX_CONSTRAINT_LT ) iLe = i;
 180         if( op==SQLITE_INDEX_CONSTRAINT_LE ) iLe = i;
 181         if( op==SQLITE_INDEX_CONSTRAINT_GT ) iGe = i;
 182         if( op==SQLITE_INDEX_CONSTRAINT_GE ) iGe = i;
 183       }
 184       if( iCol==4 ){
 185         if( op==SQLITE_INDEX_CONSTRAINT_EQ ) iLangid = i;
 186       }
 187     }
 188   }
 189
 190   if( iEq>=0 ){
 191     pInfo->idxNum = FTS4AUX_EQ_CONSTRAINT;
 192     pInfo->aConstraintUsage[iEq].argvIndex = iNext++;
 193     pInfo->estimatedCost = 5;
 194   }else{
 195     pInfo->idxNum = 0;
 196     pInfo->estimatedCost = 20000;
 197     if( iGe>=0 ){
 198       pInfo->idxNum += FTS4AUX_GE_CONSTRAINT;
 199       pInfo->aConstraintUsage[iGe].argvIndex = iNext++;
 200       pInfo->estimatedCost /= 2;
 201     }
 202     if( iLe>=0 ){
 203       pInfo->idxNum += FTS4AUX_LE_CONSTRAINT;
 204       pInfo->aConstraintUsage[iLe].argvIndex = iNext++;
 205       pInfo->estimatedCost /= 2;
 206     }
 207   }
 208   if( iLangid>=0 ){
 209     pInfo->aConstraintUsage[iLangid].argvIndex = iNext++;
 210     pInfo->estimatedCost--;
 211   }
 212
 213   return SQLITE_OK;
 214 }
 215
 216 /*
 217 ** xOpen - Open a cursor.
 218 */
 219 static int fts3auxOpenMethod(sqlite3_vtab *pVTab, sqlite3_vtab_cursor **ppCsr){
 220   Fts3auxCursor *pCsr;            /* Pointer to cursor object to return */
 221
 222   UNUSED_PARAMETER(pVTab);
 223
 224   pCsr = (Fts3auxCursor *)sqlite3_malloc(sizeof(Fts3auxCursor));
 225   if( !pCsr ) return SQLITE_NOMEM;
 226   memset(pCsr, 0, sizeof(Fts3auxCursor));
 227
 228   *ppCsr = (sqlite3_vtab_cursor *)pCsr;
 229   return SQLITE_OK;
 230 }
 231
 232 /*
 233 ** xClose - Close a cursor.
 234 */
 235 static int fts3auxCloseMethod(sqlite3_vtab_cursor *pCursor){
 236   Fts3Table *pFts3 = ((Fts3auxTable *)pCursor->pVtab)->pFts3Tab;
 237   Fts3auxCursor *pCsr = (Fts3auxCursor *)pCursor;
 238
 239   sqlite3Fts3SegmentsClose(pFts3);
 240   sqlite3Fts3SegReaderFinish(&pCsr->csr);
 241   sqlite3_free((void *)pCsr->filter.zTerm);
 242   sqlite3_free(pCsr->zStop);
 243   sqlite3_free(pCsr->aStat);
 244   sqlite3_free(pCsr);
 245   return SQLITE_OK;
 246 }
 247
 248 static int fts3auxGrowStatArray(Fts3auxCursor *pCsr, int nSize){
 249   if( nSize>pCsr->nStat ){
 250     struct Fts3auxColstats *aNew;
 251     aNew = (struct Fts3auxColstats *)sqlite3_realloc(pCsr->aStat,
 252         sizeof(struct Fts3auxColstats) * nSize
 253     );
 254     if( aNew==0 ) return SQLITE_NOMEM;
 255     memset(&aNew[pCsr->nStat], 0,
 256         sizeof(struct Fts3auxColstats) * (nSize - pCsr->nStat)
 257     );
 258     pCsr->aStat = aNew;
 259     pCsr->nStat = nSize;
 260   }
 261   return SQLITE_OK;
 262 }
 263
 264 /*
 265 ** xNext - Advance the cursor to the next row, if any.
 266 */
 267 static int fts3auxNextMethod(sqlite3_vtab_cursor *pCursor){
 268   Fts3auxCursor *pCsr = (Fts3auxCursor *)pCursor;
 269   Fts3Table *pFts3 = ((Fts3auxTable *)pCursor->pVtab)->pFts3Tab;
 270   int rc;
 271
 272   /* Increment our pretend rowid value. */
 273   pCsr->iRowid++;
 274
 275   for(pCsr->iCol++; pCsr->iCol<pCsr->nStat; pCsr->iCol++){
 276     if( pCsr->aStat[pCsr->iCol].nDoc>0 ) return SQLITE_OK;
 277   }
 278
 279   rc = sqlite3Fts3SegReaderStep(pFts3, &pCsr->csr);
 280   if( rc==SQLITE_ROW ){
 281     int i = 0;
 282     int nDoclist = pCsr->csr.nDoclist;
 283     char *aDoclist = pCsr->csr.aDoclist;
 284     int iCol;
 285
 286     int eState = 0;
 287
 288     if( pCsr->zStop ){
 289       int n = (pCsr->nStop<pCsr->csr.nTerm) ? pCsr->nStop : pCsr->csr.nTerm;
 290       int mc = memcmp(pCsr->zStop, pCsr->csr.zTerm, n);
 291       if( mc<0 || (mc==0 && pCsr->csr.nTerm>pCsr->nStop) ){
 292         pCsr->isEof = 1;
 293         return SQLITE_OK;
 294       }
 295     }
 296
 297     if( fts3auxGrowStatArray(pCsr, 2) ) return SQLITE_NOMEM;
 298     memset(pCsr->aStat, 0, sizeof(struct Fts3auxColstats) * pCsr->nStat);
 299     iCol = 0;
 300
 301     while( i<nDoclist ){
 302       sqlite3_int64 v = 0;
 303
 304       i += sqlite3Fts3GetVarint(&aDoclist[i], &v);
 305       switch( eState ){
 306         /* State 0. In this state the integer just read was a docid. */
 307         case 0:
 308           pCsr->aStat[0].nDoc++;
 309           eState = 1;
 310           iCol = 0;
 311           break;
 312
 313         /* State 1. In this state we are expecting either a 1, indicating
 314         ** that the following integer will be a column number, or the
 315         ** start of a position list for column 0.
 316         **
 317         ** The only difference between state 1 and state 2 is that if the
 318         ** integer encountered in state 1 is not 0 or 1, then we need to
 319         ** increment the column 0 "nDoc" count for this term.
 320         */
 321         case 1:
 322           assert( iCol==0 );
 323           if( v>1 ){
 324             pCsr->aStat[1].nDoc++;
 325           }
 326           eState = 2;
 327           /* fall through */
 328
 329         case 2:
 330           if( v==0 ){       /* 0x00. Next integer will be a docid. */
 331             eState = 0;
 332           }else if( v==1 ){ /* 0x01. Next integer will be a column number. */
 333             eState = 3;
 334           }else{            /* 2 or greater. A position. */
 335             pCsr->aStat[iCol+1].nOcc++;
 336             pCsr->aStat[0].nOcc++;
 337           }
 338           break;
 339
 340         /* State 3. The integer just read is a column number. */
 341         default: assert( eState==3 );
 342           iCol = (int)v;
 343           if( fts3auxGrowStatArray(pCsr, iCol+2) ) return SQLITE_NOMEM;
 344           pCsr->aStat[iCol+1].nDoc++;
 345           eState = 2;
 346           break;
 347       }
 348     }
 349
 350     pCsr->iCol = 0;
 351     rc = SQLITE_OK;
 352   }else{
 353     pCsr->isEof = 1;
 354   }
 355   return rc;
 356 }
 357
 358 /*
 359 ** xFilter - Initialize a cursor to point at the start of its data.
 360 */
 361 static int fts3auxFilterMethod(
 362   sqlite3_vtab_cursor *pCursor,   /* The cursor used for this query */
 363   int idxNum,                     /* Strategy index */
 364   const char *idxStr,             /* Unused */
 365   int nVal,                       /* Number of elements in apVal */
 366   sqlite3_value **apVal           /* Arguments for the indexing scheme */
 367 ){
 368   Fts3auxCursor *pCsr = (Fts3auxCursor *)pCursor;
 369   Fts3Table *pFts3 = ((Fts3auxTable *)pCursor->pVtab)->pFts3Tab;
 370   int rc;
 371   int isScan = 0;
 372   int iLangVal = 0;               /* Language id to query */
 373
 374   int iEq = -1;                   /* Index of term=? value in apVal */
 375   int iGe = -1;                   /* Index of term>=? value in apVal */
 376   int iLe = -1;                   /* Index of term<=? value in apVal */
 377   int iLangid = -1;               /* Index of languageid=? value in apVal */
 378   int iNext = 0;
 379
 380   UNUSED_PARAMETER(nVal);
 381   UNUSED_PARAMETER(idxStr);
 382
 383   assert( idxStr==0 );
 384   assert( idxNum==FTS4AUX_EQ_CONSTRAINT || idxNum==0
 385        || idxNum==FTS4AUX_LE_CONSTRAINT || idxNum==FTS4AUX_GE_CONSTRAINT
 386        || idxNum==(FTS4AUX_LE_CONSTRAINT|FTS4AUX_GE_CONSTRAINT)
 387   );
 388
 389   if( idxNum==FTS4AUX_EQ_CONSTRAINT ){
 390     iEq = iNext++;
 391   }else{
 392     isScan = 1;
 393     if( idxNum & FTS4AUX_GE_CONSTRAINT ){
 394       iGe = iNext++;
 395     }
 396     if( idxNum & FTS4AUX_LE_CONSTRAINT ){
 397       iLe = iNext++;
 398     }
 399   }
 400   if( iNext<nVal ){
 401     iLangid = iNext++;
 402   }
 403
 404   /* In case this cursor is being reused, close and zero it. */
 405   testcase(pCsr->filter.zTerm);
 406   sqlite3Fts3SegReaderFinish(&pCsr->csr);
 407   sqlite3_free((void *)pCsr->filter.zTerm);
 408   sqlite3_free(pCsr->aStat);
 409   memset(&pCsr->csr, 0, ((u8*)&pCsr[1]) - (u8*)&pCsr->csr);
 410
 411   pCsr->filter.flags = FTS3_SEGMENT_REQUIRE_POS|FTS3_SEGMENT_IGNORE_EMPTY;
 412   if( isScan ) pCsr->filter.flags |= FTS3_SEGMENT_SCAN;
 413
 414   if( iEq>=0 || iGe>=0 ){
 415     const unsigned char *zStr = sqlite3_value_text(apVal[0]);
 416     assert( (iEq==0 && iGe==-1) || (iEq==-1 && iGe==0) );
 417     if( zStr ){
 418       pCsr->filter.zTerm = sqlite3_mprintf("%s", zStr);
 419       pCsr->filter.nTerm = sqlite3_value_bytes(apVal[0]);
 420       if( pCsr->filter.zTerm==0 ) return SQLITE_NOMEM;
 421     }
 422   }
 423
 424   if( iLe>=0 ){
 425     pCsr->zStop = sqlite3_mprintf("%s", sqlite3_value_text(apVal[iLe]));
 426     pCsr->nStop = sqlite3_value_bytes(apVal[iLe]);
 427     if( pCsr->zStop==0 ) return SQLITE_NOMEM;
 428   }
 429
 430   if( iLangid>=0 ){
 431     iLangVal = sqlite3_value_int(apVal[iLangid]);
 432
 433     /* If the user specified a negative value for the languageid, use zero
 434     ** instead. This works, as the "languageid=?" constraint will also
 435     ** be tested by the VDBE layer. The test will always be false (since
 436     ** this module will not return a row with a negative languageid), and
 437     ** so the overall query will return zero rows.  */
 438     if( iLangVal<0 ) iLangVal = 0;
 439   }
 440   pCsr->iLangid = iLangVal;
 441
 442   rc = sqlite3Fts3SegReaderCursor(pFts3, iLangVal, 0, FTS3_SEGCURSOR_ALL,
 443       pCsr->filter.zTerm, pCsr->filter.nTerm, 0, isScan, &pCsr->csr
 444   );
 445   if( rc==SQLITE_OK ){
 446     rc = sqlite3Fts3SegReaderStart(pFts3, &pCsr->csr, &pCsr->filter);
 447   }
 448
 449   if( rc==SQLITE_OK ) rc = fts3auxNextMethod(pCursor);
 450   return rc;
 451 }
 452
 453 /*
 454 ** xEof - Return true if the cursor is at EOF, or false otherwise.
 455 */
 456 static int fts3auxEofMethod(sqlite3_vtab_cursor *pCursor){
 457   Fts3auxCursor *pCsr = (Fts3auxCursor *)pCursor;
 458   return pCsr->isEof;
 459 }
 460
 461 /*
 462 ** xColumn - Return a column value.
 463 */
 464 static int fts3auxColumnMethod(
 465   sqlite3_vtab_cursor *pCursor,   /* Cursor to retrieve value from */
 466   sqlite3_context *pCtx,          /* Context for sqlite3_result_xxx() calls */
 467   int iCol                        /* Index of column to read value from */
 468 ){
 469   Fts3auxCursor *p = (Fts3auxCursor *)pCursor;
 470
 471   assert( p->isEof==0 );
 472   switch( iCol ){
 473     case 0: /* term */
 474       sqlite3_result_text(pCtx, p->csr.zTerm, p->csr.nTerm, SQLITE_TRANSIENT);
 475       break;
 476
 477     case 1: /* col */
 478       if( p->iCol ){
 479         sqlite3_result_int(pCtx, p->iCol-1);
 480       }else{
 481         sqlite3_result_text(pCtx, "*", -1, SQLITE_STATIC);
 482       }
 483       break;
 484
 485     case 2: /* documents */
 486       sqlite3_result_int64(pCtx, p->aStat[p->iCol].nDoc);
 487       break;
 488
 489     case 3: /* occurrences */
 490       sqlite3_result_int64(pCtx, p->aStat[p->iCol].nOcc);
 491       break;
 492
 493     default: /* languageid */
 494       assert( iCol==4 );
 495       sqlite3_result_int(pCtx, p->iLangid);
 496       break;
 497   }
 498
 499   return SQLITE_OK;
 500 }
 501
 502 /*
 503 ** xRowid - Return the current rowid for the cursor.
 504 */
 505 static int fts3auxRowidMethod(
 506   sqlite3_vtab_cursor *pCursor,   /* Cursor to retrieve value from */
 507   sqlite_int64 *pRowid            /* OUT: Rowid value */
 508 ){
 509   Fts3auxCursor *pCsr = (Fts3auxCursor *)pCursor;
 510   *pRowid = pCsr->iRowid;
 511   return SQLITE_OK;
 512 }
 513
 514 /*
 515 ** Register the fts3aux module with database connection db. Return SQLITE_OK
 516 ** if successful or an error code if sqlite3_create_module() fails.
 517 */
 518 int sqlite3Fts3InitAux(sqlite3 *db){
 519   static const sqlite3_module fts3aux_module = {
 520      0,                           /* iVersion      */
 521      fts3auxConnectMethod,        /* xCreate       */
 522      fts3auxConnectMethod,        /* xConnect      */
 523      fts3auxBestIndexMethod,      /* xBestIndex    */
 524      fts3auxDisconnectMethod,     /* xDisconnect   */
 525      fts3auxDisconnectMethod,     /* xDestroy      */
 526      fts3auxOpenMethod,           /* xOpen         */
 527      fts3auxCloseMethod,          /* xClose        */
 528      fts3auxFilterMethod,         /* xFilter       */
 529      fts3auxNextMethod,           /* xNext         */
 530      fts3auxEofMethod,            /* xEof          */
 531      fts3auxColumnMethod,         /* xColumn       */
 532      fts3auxRowidMethod,          /* xRowid        */
 533      0,                           /* xUpdate       */
 534      0,                           /* xBegin        */
 535      0,                           /* xSync         */
 536      0,                           /* xCommit       */
 537      0,                           /* xRollback     */
 538      0,                           /* xFindFunction */
 539      0,                           /* xRename       */
 540      0,                           /* xSavepoint    */
 541      0,                           /* xRelease      */
 542      0                            /* xRollbackTo   */
 543   };
 544   int rc;                         /* Return code */
 545
 546   rc = sqlite3_create_module(db, "fts4aux", &fts3aux_module, 0);
 547   return rc;
 548 }
 549
 550 #endif /* !defined(SQLITE_CORE) || defined(SQLITE_ENABLE_FTS3) */