1 /* 2 ** 2001 September 15 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 ** This module contains C code that generates VDBE code used to process 13 ** the WHERE clause of SQL statements. This module is responsible for 14 ** generating the code that loops through a table looking for applicable 15 ** rows. Indices are selected and used to speed the search when doing 16 ** so is applicable. Because this module is responsible for selecting 17 ** indices, you might also think of this module as the "query optimizer". 18 */ 19 #include "sqliteInt.h" 20 21 22 /* 23 ** Trace output macros 24 */ 25 #if defined(SQLITE_TEST) || defined(SQLITE_DEBUG) 26 int sqlite3WhereTrace = 0; 27 #endif 28 #if defined(SQLITE_TEST) && defined(SQLITE_DEBUG) 29 # define WHERETRACE(X) if(sqlite3WhereTrace) sqlite3DebugPrintf X 30 #else 31 # define WHERETRACE(X) 32 #endif 33 34 /* Forward reference 35 */ 36 typedef struct WhereClause WhereClause; 37 typedef struct WhereMaskSet WhereMaskSet; 38 typedef struct WhereOrInfo WhereOrInfo; 39 typedef struct WhereAndInfo WhereAndInfo; 40 typedef struct WhereCost WhereCost; 41 42 /* 43 ** The query generator uses an array of instances of this structure to 44 ** help it analyze the subexpressions of the WHERE clause. Each WHERE 45 ** clause subexpression is separated from the others by AND operators, 46 ** usually, or sometimes subexpressions separated by OR. 47 ** 48 ** All WhereTerms are collected into a single WhereClause structure. 49 ** The following identity holds: 50 ** 51 ** WhereTerm.pWC->a[WhereTerm.idx] == WhereTerm 52 ** 53 ** When a term is of the form: 54 ** 55 ** X <op> <expr> 56 ** 57 ** where X is a column name and <op> is one of certain operators, 58 ** then WhereTerm.leftCursor and WhereTerm.u.leftColumn record the 59 ** cursor number and column number for X. WhereTerm.eOperator records 60 ** the <op> using a bitmask encoding defined by WO_xxx below. The 61 ** use of a bitmask encoding for the operator allows us to search 62 ** quickly for terms that match any of several different operators. 63 ** 64 ** A WhereTerm might also be two or more subterms connected by OR: 65 ** 66 ** (t1.X <op> <expr>) OR (t1.Y <op> <expr>) OR .... 67 ** 68 ** In this second case, wtFlag as the TERM_ORINFO set and eOperator==WO_OR 69 ** and the WhereTerm.u.pOrInfo field points to auxiliary information that 70 ** is collected about the 71 ** 72 ** If a term in the WHERE clause does not match either of the two previous 73 ** categories, then eOperator==0. The WhereTerm.pExpr field is still set 74 ** to the original subexpression content and wtFlags is set up appropriately 75 ** but no other fields in the WhereTerm object are meaningful. 76 ** 77 ** When eOperator!=0, prereqRight and prereqAll record sets of cursor numbers, 78 ** but they do so indirectly. A single WhereMaskSet structure translates 79 ** cursor number into bits and the translated bit is stored in the prereq 80 ** fields. The translation is used in order to maximize the number of 81 ** bits that will fit in a Bitmask. The VDBE cursor numbers might be 82 ** spread out over the non-negative integers. For example, the cursor 83 ** numbers might be 3, 8, 9, 10, 20, 23, 41, and 45. The WhereMaskSet 84 ** translates these sparse cursor numbers into consecutive integers 85 ** beginning with 0 in order to make the best possible use of the available 86 ** bits in the Bitmask. So, in the example above, the cursor numbers 87 ** would be mapped into integers 0 through 7. 88 ** 89 ** The number of terms in a join is limited by the number of bits 90 ** in prereqRight and prereqAll. The default is 64 bits, hence SQLite 91 ** is only able to process joins with 64 or fewer tables. 92 */ 93 typedef struct WhereTerm WhereTerm; 94 struct WhereTerm { 95 Expr *pExpr; /* Pointer to the subexpression that is this term */ 96 int iParent; /* Disable pWC->a[iParent] when this term disabled */ 97 int leftCursor; /* Cursor number of X in "X <op> <expr>" */ 98 union { 99 int leftColumn; /* Column number of X in "X <op> <expr>" */ 100 WhereOrInfo *pOrInfo; /* Extra information if eOperator==WO_OR */ 101 WhereAndInfo *pAndInfo; /* Extra information if eOperator==WO_AND */ 102 } u; 103 u16 eOperator; /* A WO_xx value describing <op> */ 104 u8 wtFlags; /* TERM_xxx bit flags. See below */ 105 u8 nChild; /* Number of children that must disable us */ 106 WhereClause *pWC; /* The clause this term is part of */ 107 Bitmask prereqRight; /* Bitmask of tables used by pExpr->pRight */ 108 Bitmask prereqAll; /* Bitmask of tables referenced by pExpr */ 109 }; 110 111 /* 112 ** Allowed values of WhereTerm.wtFlags 113 */ 114 #define TERM_DYNAMIC 0x01 /* Need to call sqlite3ExprDelete(db, pExpr) */ 115 #define TERM_VIRTUAL 0x02 /* Added by the optimizer. Do not code */ 116 #define TERM_CODED 0x04 /* This term is already coded */ 117 #define TERM_COPIED 0x08 /* Has a child */ 118 #define TERM_ORINFO 0x10 /* Need to free the WhereTerm.u.pOrInfo object */ 119 #define TERM_ANDINFO 0x20 /* Need to free the WhereTerm.u.pAndInfo obj */ 120 #define TERM_OR_OK 0x40 /* Used during OR-clause processing */ 121 #ifdef SQLITE_ENABLE_STAT3 122 # define TERM_VNULL 0x80 /* Manufactured x>NULL or x<=NULL term */ 123 #else 124 # define TERM_VNULL 0x00 /* Disabled if not using stat3 */ 125 #endif 126 127 /* 128 ** An instance of the following structure holds all information about a 129 ** WHERE clause. Mostly this is a container for one or more WhereTerms. 130 ** 131 ** Explanation of pOuter: For a WHERE clause of the form 132 ** 133 ** a AND ((b AND c) OR (d AND e)) AND f 134 ** 135 ** There are separate WhereClause objects for the whole clause and for 136 ** the subclauses "(b AND c)" and "(d AND e)". The pOuter field of the 137 ** subclauses points to the WhereClause object for the whole clause. 138 */ 139 struct WhereClause { 140 Parse *pParse; /* The parser context */ 141 WhereMaskSet *pMaskSet; /* Mapping of table cursor numbers to bitmasks */ 142 Bitmask vmask; /* Bitmask identifying virtual table cursors */ 143 WhereClause *pOuter; /* Outer conjunction */ 144 u8 op; /* Split operator. TK_AND or TK_OR */ 145 u16 wctrlFlags; /* Might include WHERE_AND_ONLY */ 146 int nTerm; /* Number of terms */ 147 int nSlot; /* Number of entries in a[] */ 148 WhereTerm *a; /* Each a[] describes a term of the WHERE cluase */ 149 #if defined(SQLITE_SMALL_STACK) 150 WhereTerm aStatic[1]; /* Initial static space for a[] */ 151 #else 152 WhereTerm aStatic[8]; /* Initial static space for a[] */ 153 #endif 154 }; 155 156 /* 157 ** A WhereTerm with eOperator==WO_OR has its u.pOrInfo pointer set to 158 ** a dynamically allocated instance of the following structure. 159 */ 160 struct WhereOrInfo { 161 WhereClause wc; /* Decomposition into subterms */ 162 Bitmask indexable; /* Bitmask of all indexable tables in the clause */ 163 }; 164 165 /* 166 ** A WhereTerm with eOperator==WO_AND has its u.pAndInfo pointer set to 167 ** a dynamically allocated instance of the following structure. 168 */ 169 struct WhereAndInfo { 170 WhereClause wc; /* The subexpression broken out */ 171 }; 172 173 /* 174 ** An instance of the following structure keeps track of a mapping 175 ** between VDBE cursor numbers and bits of the bitmasks in WhereTerm. 176 ** 177 ** The VDBE cursor numbers are small integers contained in 178 ** SrcList_item.iCursor and Expr.iTable fields. For any given WHERE 179 ** clause, the cursor numbers might not begin with 0 and they might 180 ** contain gaps in the numbering sequence. But we want to make maximum 181 ** use of the bits in our bitmasks. This structure provides a mapping 182 ** from the sparse cursor numbers into consecutive integers beginning 183 ** with 0. 184 ** 185 ** If WhereMaskSet.ix[A]==B it means that The A-th bit of a Bitmask 186 ** corresponds VDBE cursor number B. The A-th bit of a bitmask is 1<<A. 187 ** 188 ** For example, if the WHERE clause expression used these VDBE 189 ** cursors: 4, 5, 8, 29, 57, 73. Then the WhereMaskSet structure 190 ** would map those cursor numbers into bits 0 through 5. 191 ** 192 ** Note that the mapping is not necessarily ordered. In the example 193 ** above, the mapping might go like this: 4->3, 5->1, 8->2, 29->0, 194 ** 57->5, 73->4. Or one of 719 other combinations might be used. It 195 ** does not really matter. What is important is that sparse cursor 196 ** numbers all get mapped into bit numbers that begin with 0 and contain 197 ** no gaps. 198 */ 199 struct WhereMaskSet { 200 int n; /* Number of assigned cursor values */ 201 int ix[BMS]; /* Cursor assigned to each bit */ 202 }; 203 204 /* 205 ** A WhereCost object records a lookup strategy and the estimated 206 ** cost of pursuing that strategy. 207 */ 208 struct WhereCost { 209 WherePlan plan; /* The lookup strategy */ 210 double rCost; /* Overall cost of pursuing this search strategy */ 211 Bitmask used; /* Bitmask of cursors used by this plan */ 212 }; 213 214 /* 215 ** Bitmasks for the operators that indices are able to exploit. An 216 ** OR-ed combination of these values can be used when searching for 217 ** terms in the where clause. 218 */ 219 #define WO_IN 0x001 220 #define WO_EQ 0x002 221 #define WO_LT (WO_EQ<<(TK_LT-TK_EQ)) 222 #define WO_LE (WO_EQ<<(TK_LE-TK_EQ)) 223 #define WO_GT (WO_EQ<<(TK_GT-TK_EQ)) 224 #define WO_GE (WO_EQ<<(TK_GE-TK_EQ)) 225 #define WO_MATCH 0x040 226 #define WO_ISNULL 0x080 227 #define WO_OR 0x100 /* Two or more OR-connected terms */ 228 #define WO_AND 0x200 /* Two or more AND-connected terms */ 229 #define WO_NOOP 0x800 /* This term does not restrict search space */ 230 231 #define WO_ALL 0xfff /* Mask of all possible WO_* values */ 232 #define WO_SINGLE 0x0ff /* Mask of all non-compound WO_* values */ 233 234 /* 235 ** Value for wsFlags returned by bestIndex() and stored in 236 ** WhereLevel.wsFlags. These flags determine which search 237 ** strategies are appropriate. 238 ** 239 ** The least significant 12 bits is reserved as a mask for WO_ values above. 240 ** The WhereLevel.wsFlags field is usually set to WO_IN|WO_EQ|WO_ISNULL. 241 ** But if the table is the right table of a left join, WhereLevel.wsFlags 242 ** is set to WO_IN|WO_EQ. The WhereLevel.wsFlags field can then be used as 243 ** the "op" parameter to findTerm when we are resolving equality constraints. 244 ** ISNULL constraints will then not be used on the right table of a left 245 ** join. Tickets #2177 and #2189. 246 */ 247 #define WHERE_ROWID_EQ 0x00001000 /* rowid=EXPR or rowid IN (...) */ 248 #define WHERE_ROWID_RANGE 0x00002000 /* rowid<EXPR and/or rowid>EXPR */ 249 #define WHERE_COLUMN_EQ 0x00010000 /* x=EXPR or x IN (...) or x IS NULL */ 250 #define WHERE_COLUMN_RANGE 0x00020000 /* x<EXPR and/or x>EXPR */ 251 #define WHERE_COLUMN_IN 0x00040000 /* x IN (...) */ 252 #define WHERE_COLUMN_NULL 0x00080000 /* x IS NULL */ 253 #define WHERE_INDEXED 0x000f0000 /* Anything that uses an index */ 254 #define WHERE_NOT_FULLSCAN 0x100f3000 /* Does not do a full table scan */ 255 #define WHERE_IN_ABLE 0x000f1000 /* Able to support an IN operator */ 256 #define WHERE_TOP_LIMIT 0x00100000 /* x<EXPR or x<=EXPR constraint */ 257 #define WHERE_BTM_LIMIT 0x00200000 /* x>EXPR or x>=EXPR constraint */ 258 #define WHERE_BOTH_LIMIT 0x00300000 /* Both x>EXPR and x<EXPR */ 259 #define WHERE_IDX_ONLY 0x00800000 /* Use index only - omit table */ 260 #define WHERE_ORDERBY 0x01000000 /* Output will appear in correct order */ 261 #define WHERE_REVERSE 0x02000000 /* Scan in reverse order */ 262 #define WHERE_UNIQUE 0x04000000 /* Selects no more than one row */ 263 #define WHERE_VIRTUALTABLE 0x08000000 /* Use virtual-table processing */ 264 #define WHERE_MULTI_OR 0x10000000 /* OR using multiple indices */ 265 #define WHERE_TEMP_INDEX 0x20000000 /* Uses an ephemeral index */ 266 #define WHERE_DISTINCT 0x40000000 /* Correct order for DISTINCT */ 267 268 /* 269 ** Initialize a preallocated WhereClause structure. 270 */ 271 static void whereClauseInit( 272 WhereClause *pWC, /* The WhereClause to be initialized */ 273 Parse *pParse, /* The parsing context */ 274 WhereMaskSet *pMaskSet, /* Mapping from table cursor numbers to bitmasks */ 275 u16 wctrlFlags /* Might include WHERE_AND_ONLY */ 276 ){ 277 pWC->pParse = pParse; 278 pWC->pMaskSet = pMaskSet; 279 pWC->pOuter = 0; 280 pWC->nTerm = 0; 281 pWC->nSlot = ArraySize(pWC->aStatic); 282 pWC->a = pWC->aStatic; 283 pWC->vmask = 0; 284 pWC->wctrlFlags = wctrlFlags; 285 } 286 287 /* Forward reference */ 288 static void whereClauseClear(WhereClause*); 289 290 /* 291 ** Deallocate all memory associated with a WhereOrInfo object. 292 */ 293 static void whereOrInfoDelete(sqlite3 *db, WhereOrInfo *p){ 294 whereClauseClear(&p->wc); 295 sqlite3DbFree(db, p); 296 } 297 298 /* 299 ** Deallocate all memory associated with a WhereAndInfo object. 300 */ 301 static void whereAndInfoDelete(sqlite3 *db, WhereAndInfo *p){ 302 whereClauseClear(&p->wc); 303 sqlite3DbFree(db, p); 304 } 305 306 /* 307 ** Deallocate a WhereClause structure. The WhereClause structure 308 ** itself is not freed. This routine is the inverse of whereClauseInit(). 309 */ 310 static void whereClauseClear(WhereClause *pWC){ 311 int i; 312 WhereTerm *a; 313 sqlite3 *db = pWC->pParse->db; 314 for(i=pWC->nTerm-1, a=pWC->a; i>=0; i--, a++){ 315 if( a->wtFlags & TERM_DYNAMIC ){ 316 sqlite3ExprDelete(db, a->pExpr); 317 } 318 if( a->wtFlags & TERM_ORINFO ){ 319 whereOrInfoDelete(db, a->u.pOrInfo); 320 }else if( a->wtFlags & TERM_ANDINFO ){ 321 whereAndInfoDelete(db, a->u.pAndInfo); 322 } 323 } 324 if( pWC->a!=pWC->aStatic ){ 325 sqlite3DbFree(db, pWC->a); 326 } 327 } 328 329 /* 330 ** Add a single new WhereTerm entry to the WhereClause object pWC. 331 ** The new WhereTerm object is constructed from Expr p and with wtFlags. 332 ** The index in pWC->a[] of the new WhereTerm is returned on success. 333 ** 0 is returned if the new WhereTerm could not be added due to a memory 334 ** allocation error. The memory allocation failure will be recorded in 335 ** the db->mallocFailed flag so that higher-level functions can detect it. 336 ** 337 ** This routine will increase the size of the pWC->a[] array as necessary. 338 ** 339 ** If the wtFlags argument includes TERM_DYNAMIC, then responsibility 340 ** for freeing the expression p is assumed by the WhereClause object pWC. 341 ** This is true even if this routine fails to allocate a new WhereTerm. 342 ** 343 ** WARNING: This routine might reallocate the space used to store 344 ** WhereTerms. All pointers to WhereTerms should be invalidated after 345 ** calling this routine. Such pointers may be reinitialized by referencing 346 ** the pWC->a[] array. 347 */ 348 static int whereClauseInsert(WhereClause *pWC, Expr *p, u8 wtFlags){ 349 WhereTerm *pTerm; 350 int idx; 351 testcase( wtFlags & TERM_VIRTUAL ); /* EV: R-00211-15100 */ 352 if( pWC->nTerm>=pWC->nSlot ){ 353 WhereTerm *pOld = pWC->a; 354 sqlite3 *db = pWC->pParse->db; 355 pWC->a = sqlite3DbMallocRaw(db, sizeof(pWC->a[0])*pWC->nSlot*2 ); 356 if( pWC->a==0 ){ 357 if( wtFlags & TERM_DYNAMIC ){ 358 sqlite3ExprDelete(db, p); 359 } 360 pWC->a = pOld; 361 return 0; 362 } 363 memcpy(pWC->a, pOld, sizeof(pWC->a[0])*pWC->nTerm); 364 if( pOld!=pWC->aStatic ){ 365 sqlite3DbFree(db, pOld); 366 } 367 pWC->nSlot = sqlite3DbMallocSize(db, pWC->a)/sizeof(pWC->a[0]); 368 } 369 pTerm = &pWC->a[idx = pWC->nTerm++]; 370 pTerm->pExpr = p; 371 pTerm->wtFlags = wtFlags; 372 pTerm->pWC = pWC; 373 pTerm->iParent = -1; 374 return idx; 375 } 376 377 /* 378 ** This routine identifies subexpressions in the WHERE clause where 379 ** each subexpression is separated by the AND operator or some other 380 ** operator specified in the op parameter. The WhereClause structure 381 ** is filled with pointers to subexpressions. For example: 382 ** 383 ** WHERE a=='hello' AND coalesce(b,11)<10 AND (c+12!=d OR c==22) 384 ** \________/ \_______________/ \________________/ 385 ** slot[0] slot[1] slot[2] 386 ** 387 ** The original WHERE clause in pExpr is unaltered. All this routine 388 ** does is make slot[] entries point to substructure within pExpr. 389 ** 390 ** In the previous sentence and in the diagram, "slot[]" refers to 391 ** the WhereClause.a[] array. The slot[] array grows as needed to contain 392 ** all terms of the WHERE clause. 393 */ 394 static void whereSplit(WhereClause *pWC, Expr *pExpr, int op){ 395 pWC->op = (u8)op; 396 if( pExpr==0 ) return; 397 if( pExpr->op!=op ){ 398 whereClauseInsert(pWC, pExpr, 0); 399 }else{ 400 whereSplit(pWC, pExpr->pLeft, op); 401 whereSplit(pWC, pExpr->pRight, op); 402 } 403 } 404 405 /* 406 ** Initialize an expression mask set (a WhereMaskSet object) 407 */ 408 #define initMaskSet(P) memset(P, 0, sizeof(*P)) 409 410 /* 411 ** Return the bitmask for the given cursor number. Return 0 if 412 ** iCursor is not in the set. 413 */ 414 static Bitmask getMask(WhereMaskSet *pMaskSet, int iCursor){ 415 int i; 416 assert( pMaskSet->n<=(int)sizeof(Bitmask)*8 ); 417 for(i=0; i<pMaskSet->n; i++){ 418 if( pMaskSet->ix[i]==iCursor ){ 419 return ((Bitmask)1)<<i; 420 } 421 } 422 return 0; 423 } 424 425 /* 426 ** Create a new mask for cursor iCursor. 427 ** 428 ** There is one cursor per table in the FROM clause. The number of 429 ** tables in the FROM clause is limited by a test early in the 430 ** sqlite3WhereBegin() routine. So we know that the pMaskSet->ix[] 431 ** array will never overflow. 432 */ 433 static void createMask(WhereMaskSet *pMaskSet, int iCursor){ 434 assert( pMaskSet->n < ArraySize(pMaskSet->ix) ); 435 pMaskSet->ix[pMaskSet->n++] = iCursor; 436 } 437 438 /* 439 ** This routine walks (recursively) an expression tree and generates 440 ** a bitmask indicating which tables are used in that expression 441 ** tree. 442 ** 443 ** In order for this routine to work, the calling function must have 444 ** previously invoked sqlite3ResolveExprNames() on the expression. See 445 ** the header comment on that routine for additional information. 446 ** The sqlite3ResolveExprNames() routines looks for column names and 447 ** sets their opcodes to TK_COLUMN and their Expr.iTable fields to 448 ** the VDBE cursor number of the table. This routine just has to 449 ** translate the cursor numbers into bitmask values and OR all 450 ** the bitmasks together. 451 */ 452 static Bitmask exprListTableUsage(WhereMaskSet*, ExprList*); 453 static Bitmask exprSelectTableUsage(WhereMaskSet*, Select*); 454 static Bitmask exprTableUsage(WhereMaskSet *pMaskSet, Expr *p){ 455 Bitmask mask = 0; 456 if( p==0 ) return 0; 457 if( p->op==TK_COLUMN ){ 458 mask = getMask(pMaskSet, p->iTable); 459 return mask; 460 } 461 mask = exprTableUsage(pMaskSet, p->pRight); 462 mask |= exprTableUsage(pMaskSet, p->pLeft); 463 if( ExprHasProperty(p, EP_xIsSelect) ){ 464 mask |= exprSelectTableUsage(pMaskSet, p->x.pSelect); 465 }else{ 466 mask |= exprListTableUsage(pMaskSet, p->x.pList); 467 } 468 return mask; 469 } 470 static Bitmask exprListTableUsage(WhereMaskSet *pMaskSet, ExprList *pList){ 471 int i; 472 Bitmask mask = 0; 473 if( pList ){ 474 for(i=0; i<pList->nExpr; i++){ 475 mask |= exprTableUsage(pMaskSet, pList->a[i].pExpr); 476 } 477 } 478 return mask; 479 } 480 static Bitmask exprSelectTableUsage(WhereMaskSet *pMaskSet, Select *pS){ 481 Bitmask mask = 0; 482 while( pS ){ 483 SrcList *pSrc = pS->pSrc; 484 mask |= exprListTableUsage(pMaskSet, pS->pEList); 485 mask |= exprListTableUsage(pMaskSet, pS->pGroupBy); 486 mask |= exprListTableUsage(pMaskSet, pS->pOrderBy); 487 mask |= exprTableUsage(pMaskSet, pS->pWhere); 488 mask |= exprTableUsage(pMaskSet, pS->pHaving); 489 if( ALWAYS(pSrc!=0) ){ 490 int i; 491 for(i=0; i<pSrc->nSrc; i++){ 492 mask |= exprSelectTableUsage(pMaskSet, pSrc->a[i].pSelect); 493 mask |= exprTableUsage(pMaskSet, pSrc->a[i].pOn); 494 } 495 } 496 pS = pS->pPrior; 497 } 498 return mask; 499 } 500 501 /* 502 ** Return TRUE if the given operator is one of the operators that is 503 ** allowed for an indexable WHERE clause term. The allowed operators are 504 ** "=", "<", ">", "<=", ">=", and "IN". 505 ** 506 ** IMPLEMENTATION-OF: R-59926-26393 To be usable by an index a term must be 507 ** of one of the following forms: column = expression column > expression 508 ** column >= expression column < expression column <= expression 509 ** expression = column expression > column expression >= column 510 ** expression < column expression <= column column IN 511 ** (expression-list) column IN (subquery) column IS NULL 512 */ 513 static int allowedOp(int op){ 514 assert( TK_GT>TK_EQ && TK_GT<TK_GE ); 515 assert( TK_LT>TK_EQ && TK_LT<TK_GE ); 516 assert( TK_LE>TK_EQ && TK_LE<TK_GE ); 517 assert( TK_GE==TK_EQ+4 ); 518 return op==TK_IN || (op>=TK_EQ && op<=TK_GE) || op==TK_ISNULL; 519 } 520 521 /* 522 ** Swap two objects of type TYPE. 523 */ 524 #define SWAP(TYPE,A,B) {TYPE t=A; A=B; B=t;} 525 526 /* 527 ** Commute a comparison operator. Expressions of the form "X op Y" 528 ** are converted into "Y op X". 529 ** 530 ** If a collation sequence is associated with either the left or right 531 ** side of the comparison, it remains associated with the same side after 532 ** the commutation. So "Y collate NOCASE op X" becomes 533 ** "X collate NOCASE op Y". This is because any collation sequence on 534 ** the left hand side of a comparison overrides any collation sequence 535 ** attached to the right. For the same reason the EP_ExpCollate flag 536 ** is not commuted. 537 */ 538 static void exprCommute(Parse *pParse, Expr *pExpr){ 539 u16 expRight = (pExpr->pRight->flags & EP_ExpCollate); 540 u16 expLeft = (pExpr->pLeft->flags & EP_ExpCollate); 541 assert( allowedOp(pExpr->op) && pExpr->op!=TK_IN ); 542 pExpr->pRight->pColl = sqlite3ExprCollSeq(pParse, pExpr->pRight); 543 pExpr->pLeft->pColl = sqlite3ExprCollSeq(pParse, pExpr->pLeft); 544 SWAP(CollSeq*,pExpr->pRight->pColl,pExpr->pLeft->pColl); 545 pExpr->pRight->flags = (pExpr->pRight->flags & ~EP_ExpCollate) | expLeft; 546 pExpr->pLeft->flags = (pExpr->pLeft->flags & ~EP_ExpCollate) | expRight; 547 SWAP(Expr*,pExpr->pRight,pExpr->pLeft); 548 if( pExpr->op>=TK_GT ){ 549 assert( TK_LT==TK_GT+2 ); 550 assert( TK_GE==TK_LE+2 ); 551 assert( TK_GT>TK_EQ ); 552 assert( TK_GT<TK_LE ); 553 assert( pExpr->op>=TK_GT && pExpr->op<=TK_GE ); 554 pExpr->op = ((pExpr->op-TK_GT)^2)+TK_GT; 555 } 556 } 557 558 /* 559 ** Translate from TK_xx operator to WO_xx bitmask. 560 */ 561 static u16 operatorMask(int op){ 562 u16 c; 563 assert( allowedOp(op) ); 564 if( op==TK_IN ){ 565 c = WO_IN; 566 }else if( op==TK_ISNULL ){ 567 c = WO_ISNULL; 568 }else{ 569 assert( (WO_EQ<<(op-TK_EQ)) < 0x7fff ); 570 c = (u16)(WO_EQ<<(op-TK_EQ)); 571 } 572 assert( op!=TK_ISNULL || c==WO_ISNULL ); 573 assert( op!=TK_IN || c==WO_IN ); 574 assert( op!=TK_EQ || c==WO_EQ ); 575 assert( op!=TK_LT || c==WO_LT ); 576 assert( op!=TK_LE || c==WO_LE ); 577 assert( op!=TK_GT || c==WO_GT ); 578 assert( op!=TK_GE || c==WO_GE ); 579 return c; 580 } 581 582 /* 583 ** Search for a term in the WHERE clause that is of the form "X <op> <expr>" 584 ** where X is a reference to the iColumn of table iCur and <op> is one of 585 ** the WO_xx operator codes specified by the op parameter. 586 ** Return a pointer to the term. Return 0 if not found. 587 */ 588 static WhereTerm *findTerm( 589 WhereClause *pWC, /* The WHERE clause to be searched */ 590 int iCur, /* Cursor number of LHS */ 591 int iColumn, /* Column number of LHS */ 592 Bitmask notReady, /* RHS must not overlap with this mask */ 593 u32 op, /* Mask of WO_xx values describing operator */ 594 Index *pIdx /* Must be compatible with this index, if not NULL */ 595 ){ 596 WhereTerm *pTerm; 597 int k; 598 assert( iCur>=0 ); 599 op &= WO_ALL; 600 for(; pWC; pWC=pWC->pOuter){ 601 for(pTerm=pWC->a, k=pWC->nTerm; k; k--, pTerm++){ 602 if( pTerm->leftCursor==iCur 603 && (pTerm->prereqRight & notReady)==0 604 && pTerm->u.leftColumn==iColumn 605 && (pTerm->eOperator & op)!=0 606 ){ 607 if( iColumn>=0 && pIdx && pTerm->eOperator!=WO_ISNULL ){ 608 Expr *pX = pTerm->pExpr; 609 CollSeq *pColl; 610 char idxaff; 611 int j; 612 Parse *pParse = pWC->pParse; 613 614 idxaff = pIdx->pTable->aCol[iColumn].affinity; 615 if( !sqlite3IndexAffinityOk(pX, idxaff) ) continue; 616 617 /* Figure out the collation sequence required from an index for 618 ** it to be useful for optimising expression pX. Store this 619 ** value in variable pColl. 620 */ 621 assert(pX->pLeft); 622 pColl = sqlite3BinaryCompareCollSeq(pParse, pX->pLeft, pX->pRight); 623 assert(pColl || pParse->nErr); 624 625 for(j=0; pIdx->aiColumn[j]!=iColumn; j++){ 626 if( NEVER(j>=pIdx->nColumn) ) return 0; 627 } 628 if( pColl && sqlite3StrICmp(pColl->zName, pIdx->azColl[j]) ) continue; 629 } 630 return pTerm; 631 } 632 } 633 } 634 return 0; 635 } 636 637 /* Forward reference */ 638 static void exprAnalyze(SrcList*, WhereClause*, int); 639 640 /* 641 ** Call exprAnalyze on all terms in a WHERE clause. 642 ** 643 ** 644 */ 645 static void exprAnalyzeAll( 646 SrcList *pTabList, /* the FROM clause */ 647 WhereClause *pWC /* the WHERE clause to be analyzed */ 648 ){ 649 int i; 650 for(i=pWC->nTerm-1; i>=0; i--){ 651 exprAnalyze(pTabList, pWC, i); 652 } 653 } 654 655 #ifndef SQLITE_OMIT_LIKE_OPTIMIZATION 656 /* 657 ** Check to see if the given expression is a LIKE or GLOB operator that 658 ** can be optimized using inequality constraints. Return TRUE if it is 659 ** so and false if not. 660 ** 661 ** In order for the operator to be optimizible, the RHS must be a string 662 ** literal that does not begin with a wildcard. 663 */ 664 static int isLikeOrGlob( 665 Parse *pParse, /* Parsing and code generating context */ 666 Expr *pExpr, /* Test this expression */ 667 Expr **ppPrefix, /* Pointer to TK_STRING expression with pattern prefix */ 668 int *pisComplete, /* True if the only wildcard is % in the last character */ 669 int *pnoCase /* True if uppercase is equivalent to lowercase */ 670 ){ 671 const char *z = 0; /* String on RHS of LIKE operator */ 672 Expr *pRight, *pLeft; /* Right and left size of LIKE operator */ 673 ExprList *pList; /* List of operands to the LIKE operator */ 674 int c; /* One character in z[] */ 675 int cnt; /* Number of non-wildcard prefix characters */ 676 char wc[3]; /* Wildcard characters */ 677 sqlite3 *db = pParse->db; /* Database connection */ 678 sqlite3_value *pVal = 0; 679 int op; /* Opcode of pRight */ 680 681 if( !sqlite3IsLikeFunction(db, pExpr, pnoCase, wc) ){ 682 return 0; 683 } 684 #ifdef SQLITE_EBCDIC 685 if( *pnoCase ) return 0; 686 #endif 687 pList = pExpr->x.pList; 688 pLeft = pList->a[1].pExpr; 689 if( pLeft->op!=TK_COLUMN || sqlite3ExprAffinity(pLeft)!=SQLITE_AFF_TEXT ){ 690 /* IMP: R-02065-49465 The left-hand side of the LIKE or GLOB operator must 691 ** be the name of an indexed column with TEXT affinity. */ 692 return 0; 693 } 694 assert( pLeft->iColumn!=(-1) ); /* Because IPK never has AFF_TEXT */ 695 696 pRight = pList->a[0].pExpr; 697 op = pRight->op; 698 if( op==TK_REGISTER ){ 699 op = pRight->op2; 700 } 701 if( op==TK_VARIABLE ){ 702 Vdbe *pReprepare = pParse->pReprepare; 703 int iCol = pRight->iColumn; 704 pVal = sqlite3VdbeGetValue(pReprepare, iCol, SQLITE_AFF_NONE); 705 if( pVal && sqlite3_value_type(pVal)==SQLITE_TEXT ){ 706 z = (char *)sqlite3_value_text(pVal); 707 } 708 sqlite3VdbeSetVarmask(pParse->pVdbe, iCol); 709 assert( pRight->op==TK_VARIABLE || pRight->op==TK_REGISTER ); 710 }else if( op==TK_STRING ){ 711 z = pRight->u.zToken; 712 } 713 if( z ){ 714 cnt = 0; 715 while( (c=z[cnt])!=0 && c!=wc[0] && c!=wc[1] && c!=wc[2] ){ 716 cnt++; 717 } 718 if( cnt!=0 && 255!=(u8)z[cnt-1] ){ 719 Expr *pPrefix; 720 *pisComplete = c==wc[0] && z[cnt+1]==0; 721 pPrefix = sqlite3Expr(db, TK_STRING, z); 722 if( pPrefix ) pPrefix->u.zToken[cnt] = 0; 723 *ppPrefix = pPrefix; 724 if( op==TK_VARIABLE ){ 725 Vdbe *v = pParse->pVdbe; 726 sqlite3VdbeSetVarmask(v, pRight->iColumn); 727 if( *pisComplete && pRight->u.zToken[1] ){ 728 /* If the rhs of the LIKE expression is a variable, and the current 729 ** value of the variable means there is no need to invoke the LIKE 730 ** function, then no OP_Variable will be added to the program. 731 ** This causes problems for the sqlite3_bind_parameter_name() 732 ** API. To workaround them, add a dummy OP_Variable here. 733 */ 734 int r1 = sqlite3GetTempReg(pParse); 735 sqlite3ExprCodeTarget(pParse, pRight, r1); 736 sqlite3VdbeChangeP3(v, sqlite3VdbeCurrentAddr(v)-1, 0); 737 sqlite3ReleaseTempReg(pParse, r1); 738 } 739 } 740 }else{ 741 z = 0; 742 } 743 } 744 745 sqlite3ValueFree(pVal); 746 return (z!=0); 747 } 748 #endif /* SQLITE_OMIT_LIKE_OPTIMIZATION */ 749 750 751 #ifndef SQLITE_OMIT_VIRTUALTABLE 752 /* 753 ** Check to see if the given expression is of the form 754 ** 755 ** column MATCH expr 756 ** 757 ** If it is then return TRUE. If not, return FALSE. 758 */ 759 static int isMatchOfColumn( 760 Expr *pExpr /* Test this expression */ 761 ){ 762 ExprList *pList; 763 764 if( pExpr->op!=TK_FUNCTION ){ 765 return 0; 766 } 767 if( sqlite3StrICmp(pExpr->u.zToken,"match")!=0 ){ 768 return 0; 769 } 770 pList = pExpr->x.pList; 771 if( pList->nExpr!=2 ){ 772 return 0; 773 } 774 if( pList->a[1].pExpr->op != TK_COLUMN ){ 775 return 0; 776 } 777 return 1; 778 } 779 #endif /* SQLITE_OMIT_VIRTUALTABLE */ 780 781 /* 782 ** If the pBase expression originated in the ON or USING clause of 783 ** a join, then transfer the appropriate markings over to derived. 784 */ 785 static void transferJoinMarkings(Expr *pDerived, Expr *pBase){ 786 pDerived->flags |= pBase->flags & EP_FromJoin; 787 pDerived->iRightJoinTable = pBase->iRightJoinTable; 788 } 789 790 #if !defined(SQLITE_OMIT_OR_OPTIMIZATION) && !defined(SQLITE_OMIT_SUBQUERY) 791 /* 792 ** Analyze a term that consists of two or more OR-connected 793 ** subterms. So in: 794 ** 795 ** ... WHERE (a=5) AND (b=7 OR c=9 OR d=13) AND (d=13) 796 ** ^^^^^^^^^^^^^^^^^^^^ 797 ** 798 ** This routine analyzes terms such as the middle term in the above example. 799 ** A WhereOrTerm object is computed and attached to the term under 800 ** analysis, regardless of the outcome of the analysis. Hence: 801 ** 802 ** WhereTerm.wtFlags |= TERM_ORINFO 803 ** WhereTerm.u.pOrInfo = a dynamically allocated WhereOrTerm object 804 ** 805 ** The term being analyzed must have two or more of OR-connected subterms. 806 ** A single subterm might be a set of AND-connected sub-subterms. 807 ** Examples of terms under analysis: 808 ** 809 ** (A) t1.x=t2.y OR t1.x=t2.z OR t1.y=15 OR t1.z=t3.a+5 810 ** (B) x=expr1 OR expr2=x OR x=expr3 811 ** (C) t1.x=t2.y OR (t1.x=t2.z AND t1.y=15) 812 ** (D) x=expr1 OR (y>11 AND y<22 AND z LIKE '*hello*') 813 ** (E) (p.a=1 AND q.b=2 AND r.c=3) OR (p.x=4 AND q.y=5 AND r.z=6) 814 ** 815 ** CASE 1: 816 ** 817 ** If all subterms are of the form T.C=expr for some single column of C 818 ** a single table T (as shown in example B above) then create a new virtual 819 ** term that is an equivalent IN expression. In other words, if the term 820 ** being analyzed is: 821 ** 822 ** x = expr1 OR expr2 = x OR x = expr3 823 ** 824 ** then create a new virtual term like this: 825 ** 826 ** x IN (expr1,expr2,expr3) 827 ** 828 ** CASE 2: 829 ** 830 ** If all subterms are indexable by a single table T, then set 831 ** 832 ** WhereTerm.eOperator = WO_OR 833 ** WhereTerm.u.pOrInfo->indexable |= the cursor number for table T 834 ** 835 ** A subterm is "indexable" if it is of the form 836 ** "T.C <op> <expr>" where C is any column of table T and 837 ** <op> is one of "=", "<", "<=", ">", ">=", "IS NULL", or "IN". 838 ** A subterm is also indexable if it is an AND of two or more 839 ** subsubterms at least one of which is indexable. Indexable AND 840 ** subterms have their eOperator set to WO_AND and they have 841 ** u.pAndInfo set to a dynamically allocated WhereAndTerm object. 842 ** 843 ** From another point of view, "indexable" means that the subterm could 844 ** potentially be used with an index if an appropriate index exists. 845 ** This analysis does not consider whether or not the index exists; that 846 ** is something the bestIndex() routine will determine. This analysis 847 ** only looks at whether subterms appropriate for indexing exist. 848 ** 849 ** All examples A through E above all satisfy case 2. But if a term 850 ** also statisfies case 1 (such as B) we know that the optimizer will 851 ** always prefer case 1, so in that case we pretend that case 2 is not 852 ** satisfied. 853 ** 854 ** It might be the case that multiple tables are indexable. For example, 855 ** (E) above is indexable on tables P, Q, and R. 856 ** 857 ** Terms that satisfy case 2 are candidates for lookup by using 858 ** separate indices to find rowids for each subterm and composing 859 ** the union of all rowids using a RowSet object. This is similar 860 ** to "bitmap indices" in other database engines. 861 ** 862 ** OTHERWISE: 863 ** 864 ** If neither case 1 nor case 2 apply, then leave the eOperator set to 865 ** zero. This term is not useful for search. 866 */ 867 static void exprAnalyzeOrTerm( 868 SrcList *pSrc, /* the FROM clause */ 869 WhereClause *pWC, /* the complete WHERE clause */ 870 int idxTerm /* Index of the OR-term to be analyzed */ 871 ){ 872 Parse *pParse = pWC->pParse; /* Parser context */ 873 sqlite3 *db = pParse->db; /* Database connection */ 874 WhereTerm *pTerm = &pWC->a[idxTerm]; /* The term to be analyzed */ 875 Expr *pExpr = pTerm->pExpr; /* The expression of the term */ 876 WhereMaskSet *pMaskSet = pWC->pMaskSet; /* Table use masks */ 877 int i; /* Loop counters */ 878 WhereClause *pOrWc; /* Breakup of pTerm into subterms */ 879 WhereTerm *pOrTerm; /* A Sub-term within the pOrWc */ 880 WhereOrInfo *pOrInfo; /* Additional information associated with pTerm */ 881 Bitmask chngToIN; /* Tables that might satisfy case 1 */ 882 Bitmask indexable; /* Tables that are indexable, satisfying case 2 */ 883 884 /* 885 ** Break the OR clause into its separate subterms. The subterms are 886 ** stored in a WhereClause structure containing within the WhereOrInfo 887 ** object that is attached to the original OR clause term. 888 */ 889 assert( (pTerm->wtFlags & (TERM_DYNAMIC|TERM_ORINFO|TERM_ANDINFO))==0 ); 890 assert( pExpr->op==TK_OR ); 891 pTerm->u.pOrInfo = pOrInfo = sqlite3DbMallocZero(db, sizeof(*pOrInfo)); 892 if( pOrInfo==0 ) return; 893 pTerm->wtFlags |= TERM_ORINFO; 894 pOrWc = &pOrInfo->wc; 895 whereClauseInit(pOrWc, pWC->pParse, pMaskSet, pWC->wctrlFlags); 896 whereSplit(pOrWc, pExpr, TK_OR); 897 exprAnalyzeAll(pSrc, pOrWc); 898 if( db->mallocFailed ) return; 899 assert( pOrWc->nTerm>=2 ); 900 901 /* 902 ** Compute the set of tables that might satisfy cases 1 or 2. 903 */ 904 indexable = ~(Bitmask)0; 905 chngToIN = ~(pWC->vmask); 906 for(i=pOrWc->nTerm-1, pOrTerm=pOrWc->a; i>=0 && indexable; i--, pOrTerm++){ 907 if( (pOrTerm->eOperator & WO_SINGLE)==0 ){ 908 WhereAndInfo *pAndInfo; 909 assert( pOrTerm->eOperator==0 ); 910 assert( (pOrTerm->wtFlags & (TERM_ANDINFO|TERM_ORINFO))==0 ); 911 chngToIN = 0; 912 pAndInfo = sqlite3DbMallocRaw(db, sizeof(*pAndInfo)); 913 if( pAndInfo ){ 914 WhereClause *pAndWC; 915 WhereTerm *pAndTerm; 916 int j; 917 Bitmask b = 0; 918 pOrTerm->u.pAndInfo = pAndInfo; 919 pOrTerm->wtFlags |= TERM_ANDINFO; 920 pOrTerm->eOperator = WO_AND; 921 pAndWC = &pAndInfo->wc; 922 whereClauseInit(pAndWC, pWC->pParse, pMaskSet, pWC->wctrlFlags); 923 whereSplit(pAndWC, pOrTerm->pExpr, TK_AND); 924 exprAnalyzeAll(pSrc, pAndWC); 925 pAndWC->pOuter = pWC; 926 testcase( db->mallocFailed ); 927 if( !db->mallocFailed ){ 928 for(j=0, pAndTerm=pAndWC->a; j<pAndWC->nTerm; j++, pAndTerm++){ 929 assert( pAndTerm->pExpr ); 930 if( allowedOp(pAndTerm->pExpr->op) ){ 931 b |= getMask(pMaskSet, pAndTerm->leftCursor); 932 } 933 } 934 } 935 indexable &= b; 936 } 937 }else if( pOrTerm->wtFlags & TERM_COPIED ){ 938 /* Skip this term for now. We revisit it when we process the 939 ** corresponding TERM_VIRTUAL term */ 940 }else{ 941 Bitmask b; 942 b = getMask(pMaskSet, pOrTerm->leftCursor); 943 if( pOrTerm->wtFlags & TERM_VIRTUAL ){ 944 WhereTerm *pOther = &pOrWc->a[pOrTerm->iParent]; 945 b |= getMask(pMaskSet, pOther->leftCursor); 946 } 947 indexable &= b; 948 if( pOrTerm->eOperator!=WO_EQ ){ 949 chngToIN = 0; 950 }else{ 951 chngToIN &= b; 952 } 953 } 954 } 955 956 /* 957 ** Record the set of tables that satisfy case 2. The set might be 958 ** empty. 959 */ 960 pOrInfo->indexable = indexable; 961 pTerm->eOperator = indexable==0 ? 0 : WO_OR; 962 963 /* 964 ** chngToIN holds a set of tables that *might* satisfy case 1. But 965 ** we have to do some additional checking to see if case 1 really 966 ** is satisfied. 967 ** 968 ** chngToIN will hold either 0, 1, or 2 bits. The 0-bit case means 969 ** that there is no possibility of transforming the OR clause into an 970 ** IN operator because one or more terms in the OR clause contain 971 ** something other than == on a column in the single table. The 1-bit 972 ** case means that every term of the OR clause is of the form 973 ** "table.column=expr" for some single table. The one bit that is set 974 ** will correspond to the common table. We still need to check to make 975 ** sure the same column is used on all terms. The 2-bit case is when 976 ** the all terms are of the form "table1.column=table2.column". It 977 ** might be possible to form an IN operator with either table1.column 978 ** or table2.column as the LHS if either is common to every term of 979 ** the OR clause. 980 ** 981 ** Note that terms of the form "table.column1=table.column2" (the 982 ** same table on both sizes of the ==) cannot be optimized. 983 */ 984 if( chngToIN ){ 985 int okToChngToIN = 0; /* True if the conversion to IN is valid */ 986 int iColumn = -1; /* Column index on lhs of IN operator */ 987 int iCursor = -1; /* Table cursor common to all terms */ 988 int j = 0; /* Loop counter */ 989 990 /* Search for a table and column that appears on one side or the 991 ** other of the == operator in every subterm. That table and column 992 ** will be recorded in iCursor and iColumn. There might not be any 993 ** such table and column. Set okToChngToIN if an appropriate table 994 ** and column is found but leave okToChngToIN false if not found. 995 */ 996 for(j=0; j<2 && !okToChngToIN; j++){ 997 pOrTerm = pOrWc->a; 998 for(i=pOrWc->nTerm-1; i>=0; i--, pOrTerm++){ 999 assert( pOrTerm->eOperator==WO_EQ ); 1000 pOrTerm->wtFlags &= ~TERM_OR_OK; 1001 if( pOrTerm->leftCursor==iCursor ){ 1002 /* This is the 2-bit case and we are on the second iteration and 1003 ** current term is from the first iteration. So skip this term. */ 1004 assert( j==1 ); 1005 continue; 1006 } 1007 if( (chngToIN & getMask(pMaskSet, pOrTerm->leftCursor))==0 ){ 1008 /* This term must be of the form t1.a==t2.b where t2 is in the 1009 ** chngToIN set but t1 is not. This term will be either preceeded 1010 ** or follwed by an inverted copy (t2.b==t1.a). Skip this term 1011 ** and use its inversion. */ 1012 testcase( pOrTerm->wtFlags & TERM_COPIED ); 1013 testcase( pOrTerm->wtFlags & TERM_VIRTUAL ); 1014 assert( pOrTerm->wtFlags & (TERM_COPIED|TERM_VIRTUAL) ); 1015 continue; 1016 } 1017 iColumn = pOrTerm->u.leftColumn; 1018 iCursor = pOrTerm->leftCursor; 1019 break; 1020 } 1021 if( i<0 ){ 1022 /* No candidate table+column was found. This can only occur 1023 ** on the second iteration */ 1024 assert( j==1 ); 1025 assert( (chngToIN&(chngToIN-1))==0 ); 1026 assert( chngToIN==getMask(pMaskSet, iCursor) ); 1027 break; 1028 } 1029 testcase( j==1 ); 1030 1031 /* We have found a candidate table and column. Check to see if that 1032 ** table and column is common to every term in the OR clause */ 1033 okToChngToIN = 1; 1034 for(; i>=0 && okToChngToIN; i--, pOrTerm++){ 1035 assert( pOrTerm->eOperator==WO_EQ ); 1036 if( pOrTerm->leftCursor!=iCursor ){ 1037 pOrTerm->wtFlags &= ~TERM_OR_OK; 1038 }else if( pOrTerm->u.leftColumn!=iColumn ){ 1039 okToChngToIN = 0; 1040 }else{ 1041 int affLeft, affRight; 1042 /* If the right-hand side is also a column, then the affinities 1043 ** of both right and left sides must be such that no type 1044 ** conversions are required on the right. (Ticket #2249) 1045 */ 1046 affRight = sqlite3ExprAffinity(pOrTerm->pExpr->pRight); 1047 affLeft = sqlite3ExprAffinity(pOrTerm->pExpr->pLeft); 1048 if( affRight!=0 && affRight!=affLeft ){ 1049 okToChngToIN = 0; 1050 }else{ 1051 pOrTerm->wtFlags |= TERM_OR_OK; 1052 } 1053 } 1054 } 1055 } 1056 1057 /* At this point, okToChngToIN is true if original pTerm satisfies 1058 ** case 1. In that case, construct a new virtual term that is 1059 ** pTerm converted into an IN operator. 1060 ** 1061 ** EV: R-00211-15100 1062 */ 1063 if( okToChngToIN ){ 1064 Expr *pDup; /* A transient duplicate expression */ 1065 ExprList *pList = 0; /* The RHS of the IN operator */ 1066 Expr *pLeft = 0; /* The LHS of the IN operator */ 1067 Expr *pNew; /* The complete IN operator */ 1068 1069 for(i=pOrWc->nTerm-1, pOrTerm=pOrWc->a; i>=0; i--, pOrTerm++){ 1070 if( (pOrTerm->wtFlags & TERM_OR_OK)==0 ) continue; 1071 assert( pOrTerm->eOperator==WO_EQ ); 1072 assert( pOrTerm->leftCursor==iCursor ); 1073 assert( pOrTerm->u.leftColumn==iColumn ); 1074 pDup = sqlite3ExprDup(db, pOrTerm->pExpr->pRight, 0); 1075 pList = sqlite3ExprListAppend(pWC->pParse, pList, pDup); 1076 pLeft = pOrTerm->pExpr->pLeft; 1077 } 1078 assert( pLeft!=0 ); 1079 pDup = sqlite3ExprDup(db, pLeft, 0); 1080 pNew = sqlite3PExpr(pParse, TK_IN, pDup, 0, 0); 1081 if( pNew ){ 1082 int idxNew; 1083 transferJoinMarkings(pNew, pExpr); 1084 assert( !ExprHasProperty(pNew, EP_xIsSelect) ); 1085 pNew->x.pList = pList; 1086 idxNew = whereClauseInsert(pWC, pNew, TERM_VIRTUAL|TERM_DYNAMIC); 1087 testcase( idxNew==0 ); 1088 exprAnalyze(pSrc, pWC, idxNew); 1089 pTerm = &pWC->a[idxTerm]; 1090 pWC->a[idxNew].iParent = idxTerm; 1091 pTerm->nChild = 1; 1092 }else{ 1093 sqlite3ExprListDelete(db, pList); 1094 } 1095 pTerm->eOperator = WO_NOOP; /* case 1 trumps case 2 */ 1096 } 1097 } 1098 } 1099 #endif /* !SQLITE_OMIT_OR_OPTIMIZATION && !SQLITE_OMIT_SUBQUERY */ 1100 1101 1102 /* 1103 ** The input to this routine is an WhereTerm structure with only the 1104 ** "pExpr" field filled in. The job of this routine is to analyze the 1105 ** subexpression and populate all the other fields of the WhereTerm 1106 ** structure. 1107 ** 1108 ** If the expression is of the form "<expr> <op> X" it gets commuted 1109 ** to the standard form of "X <op> <expr>". 1110 ** 1111 ** If the expression is of the form "X <op> Y" where both X and Y are 1112 ** columns, then the original expression is unchanged and a new virtual 1113 ** term of the form "Y <op> X" is added to the WHERE clause and 1114 ** analyzed separately. The original term is marked with TERM_COPIED 1115 ** and the new term is marked with TERM_DYNAMIC (because it's pExpr 1116 ** needs to be freed with the WhereClause) and TERM_VIRTUAL (because it 1117 ** is a commuted copy of a prior term.) The original term has nChild=1 1118 ** and the copy has idxParent set to the index of the original term. 1119 */ 1120 static void exprAnalyze( 1121 SrcList *pSrc, /* the FROM clause */ 1122 WhereClause *pWC, /* the WHERE clause */ 1123 int idxTerm /* Index of the term to be analyzed */ 1124 ){ 1125 WhereTerm *pTerm; /* The term to be analyzed */ 1126 WhereMaskSet *pMaskSet; /* Set of table index masks */ 1127 Expr *pExpr; /* The expression to be analyzed */ 1128 Bitmask prereqLeft; /* Prerequesites of the pExpr->pLeft */ 1129 Bitmask prereqAll; /* Prerequesites of pExpr */ 1130 Bitmask extraRight = 0; /* Extra dependencies on LEFT JOIN */ 1131 Expr *pStr1 = 0; /* RHS of LIKE/GLOB operator */ 1132 int isComplete = 0; /* RHS of LIKE/GLOB ends with wildcard */ 1133 int noCase = 0; /* LIKE/GLOB distinguishes case */ 1134 int op; /* Top-level operator. pExpr->op */ 1135 Parse *pParse = pWC->pParse; /* Parsing context */ 1136 sqlite3 *db = pParse->db; /* Database connection */ 1137 1138 if( db->mallocFailed ){ 1139 return; 1140 } 1141 pTerm = &pWC->a[idxTerm]; 1142 pMaskSet = pWC->pMaskSet; 1143 pExpr = pTerm->pExpr; 1144 prereqLeft = exprTableUsage(pMaskSet, pExpr->pLeft); 1145 op = pExpr->op; 1146 if( op==TK_IN ){ 1147 assert( pExpr->pRight==0 ); 1148 if( ExprHasProperty(pExpr, EP_xIsSelect) ){ 1149 pTerm->prereqRight = exprSelectTableUsage(pMaskSet, pExpr->x.pSelect); 1150 }else{ 1151 pTerm->prereqRight = exprListTableUsage(pMaskSet, pExpr->x.pList); 1152 } 1153 }else if( op==TK_ISNULL ){ 1154 pTerm->prereqRight = 0; 1155 }else{ 1156 pTerm->prereqRight = exprTableUsage(pMaskSet, pExpr->pRight); 1157 } 1158 prereqAll = exprTableUsage(pMaskSet, pExpr); 1159 if( ExprHasProperty(pExpr, EP_FromJoin) ){ 1160 Bitmask x = getMask(pMaskSet, pExpr->iRightJoinTable); 1161 prereqAll |= x; 1162 extraRight = x-1; /* ON clause terms may not be used with an index 1163 ** on left table of a LEFT JOIN. Ticket #3015 */ 1164 } 1165 pTerm->prereqAll = prereqAll; 1166 pTerm->leftCursor = -1; 1167 pTerm->iParent = -1; 1168 pTerm->eOperator = 0; 1169 if( allowedOp(op) && (pTerm->prereqRight & prereqLeft)==0 ){ 1170 Expr *pLeft = pExpr->pLeft; 1171 Expr *pRight = pExpr->pRight; 1172 if( pLeft->op==TK_COLUMN ){ 1173 pTerm->leftCursor = pLeft->iTable; 1174 pTerm->u.leftColumn = pLeft->iColumn; 1175 pTerm->eOperator = operatorMask(op); 1176 } 1177 if( pRight && pRight->op==TK_COLUMN ){ 1178 WhereTerm *pNew; 1179 Expr *pDup; 1180 if( pTerm->leftCursor>=0 ){ 1181 int idxNew; 1182 pDup = sqlite3ExprDup(db, pExpr, 0); 1183 if( db->mallocFailed ){ 1184 sqlite3ExprDelete(db, pDup); 1185 return; 1186 } 1187 idxNew = whereClauseInsert(pWC, pDup, TERM_VIRTUAL|TERM_DYNAMIC); 1188 if( idxNew==0 ) return; 1189 pNew = &pWC->a[idxNew]; 1190 pNew->iParent = idxTerm; 1191 pTerm = &pWC->a[idxTerm]; 1192 pTerm->nChild = 1; 1193 pTerm->wtFlags |= TERM_COPIED; 1194 }else{ 1195 pDup = pExpr; 1196 pNew = pTerm; 1197 } 1198 exprCommute(pParse, pDup); 1199 pLeft = pDup->pLeft; 1200 pNew->leftCursor = pLeft->iTable; 1201 pNew->u.leftColumn = pLeft->iColumn; 1202 testcase( (prereqLeft | extraRight) != prereqLeft ); 1203 pNew->prereqRight = prereqLeft | extraRight; 1204 pNew->prereqAll = prereqAll; 1205 pNew->eOperator = operatorMask(pDup->op); 1206 } 1207 } 1208 1209 #ifndef SQLITE_OMIT_BETWEEN_OPTIMIZATION 1210 /* If a term is the BETWEEN operator, create two new virtual terms 1211 ** that define the range that the BETWEEN implements. For example: 1212 ** 1213 ** a BETWEEN b AND c 1214 ** 1215 ** is converted into: 1216 ** 1217 ** (a BETWEEN b AND c) AND (a>=b) AND (a<=c) 1218 ** 1219 ** The two new terms are added onto the end of the WhereClause object. 1220 ** The new terms are "dynamic" and are children of the original BETWEEN 1221 ** term. That means that if the BETWEEN term is coded, the children are 1222 ** skipped. Or, if the children are satisfied by an index, the original 1223 ** BETWEEN term is skipped. 1224 */ 1225 else if( pExpr->op==TK_BETWEEN && pWC->op==TK_AND ){ 1226 ExprList *pList = pExpr->x.pList; 1227 int i; 1228 static const u8 ops[] = {TK_GE, TK_LE}; 1229 assert( pList!=0 ); 1230 assert( pList->nExpr==2 ); 1231 for(i=0; i<2; i++){ 1232 Expr *pNewExpr; 1233 int idxNew; 1234 pNewExpr = sqlite3PExpr(pParse, ops[i], 1235 sqlite3ExprDup(db, pExpr->pLeft, 0), 1236 sqlite3ExprDup(db, pList->a[i].pExpr, 0), 0); 1237 idxNew = whereClauseInsert(pWC, pNewExpr, TERM_VIRTUAL|TERM_DYNAMIC); 1238 testcase( idxNew==0 ); 1239 exprAnalyze(pSrc, pWC, idxNew); 1240 pTerm = &pWC->a[idxTerm]; 1241 pWC->a[idxNew].iParent = idxTerm; 1242 } 1243 pTerm->nChild = 2; 1244 } 1245 #endif /* SQLITE_OMIT_BETWEEN_OPTIMIZATION */ 1246 1247 #if !defined(SQLITE_OMIT_OR_OPTIMIZATION) && !defined(SQLITE_OMIT_SUBQUERY) 1248 /* Analyze a term that is composed of two or more subterms connected by 1249 ** an OR operator. 1250 */ 1251 else if( pExpr->op==TK_OR ){ 1252 assert( pWC->op==TK_AND ); 1253 exprAnalyzeOrTerm(pSrc, pWC, idxTerm); 1254 pTerm = &pWC->a[idxTerm]; 1255 } 1256 #endif /* SQLITE_OMIT_OR_OPTIMIZATION */ 1257 1258 #ifndef SQLITE_OMIT_LIKE_OPTIMIZATION 1259 /* Add constraints to reduce the search space on a LIKE or GLOB 1260 ** operator. 1261 ** 1262 ** A like pattern of the form "x LIKE 'abc%'" is changed into constraints 1263 ** 1264 ** x>='abc' AND x<'abd' AND x LIKE 'abc%' 1265 ** 1266 ** The last character of the prefix "abc" is incremented to form the 1267 ** termination condition "abd". 1268 */ 1269 if( pWC->op==TK_AND 1270 && isLikeOrGlob(pParse, pExpr, &pStr1, &isComplete, &noCase) 1271 ){ 1272 Expr *pLeft; /* LHS of LIKE/GLOB operator */ 1273 Expr *pStr2; /* Copy of pStr1 - RHS of LIKE/GLOB operator */ 1274 Expr *pNewExpr1; 1275 Expr *pNewExpr2; 1276 int idxNew1; 1277 int idxNew2; 1278 CollSeq *pColl; /* Collating sequence to use */ 1279 1280 pLeft = pExpr->x.pList->a[1].pExpr; 1281 pStr2 = sqlite3ExprDup(db, pStr1, 0); 1282 if( !db->mallocFailed ){ 1283 u8 c, *pC; /* Last character before the first wildcard */ 1284 pC = (u8*)&pStr2->u.zToken[sqlite3Strlen30(pStr2->u.zToken)-1]; 1285 c = *pC; 1286 if( noCase ){ 1287 /* The point is to increment the last character before the first 1288 ** wildcard. But if we increment '@', that will push it into the 1289 ** alphabetic range where case conversions will mess up the 1290 ** inequality. To avoid this, make sure to also run the full 1291 ** LIKE on all candidate expressions by clearing the isComplete flag 1292 */ 1293 if( c=='A'-1 ) isComplete = 0; /* EV: R-64339-08207 */ 1294 1295 1296 c = sqlite3UpperToLower[c]; 1297 } 1298 *pC = c + 1; 1299 } 1300 pColl = sqlite3FindCollSeq(db, SQLITE_UTF8, noCase ? "NOCASE" : "BINARY",0); 1301 pNewExpr1 = sqlite3PExpr(pParse, TK_GE, 1302 sqlite3ExprSetColl(sqlite3ExprDup(db,pLeft,0), pColl), 1303 pStr1, 0); 1304 idxNew1 = whereClauseInsert(pWC, pNewExpr1, TERM_VIRTUAL|TERM_DYNAMIC); 1305 testcase( idxNew1==0 ); 1306 exprAnalyze(pSrc, pWC, idxNew1); 1307 pNewExpr2 = sqlite3PExpr(pParse, TK_LT, 1308 sqlite3ExprSetColl(sqlite3ExprDup(db,pLeft,0), pColl), 1309 pStr2, 0); 1310 idxNew2 = whereClauseInsert(pWC, pNewExpr2, TERM_VIRTUAL|TERM_DYNAMIC); 1311 testcase( idxNew2==0 ); 1312 exprAnalyze(pSrc, pWC, idxNew2); 1313 pTerm = &pWC->a[idxTerm]; 1314 if( isComplete ){ 1315 pWC->a[idxNew1].iParent = idxTerm; 1316 pWC->a[idxNew2].iParent = idxTerm; 1317 pTerm->nChild = 2; 1318 } 1319 } 1320 #endif /* SQLITE_OMIT_LIKE_OPTIMIZATION */ 1321 1322 #ifndef SQLITE_OMIT_VIRTUALTABLE 1323 /* Add a WO_MATCH auxiliary term to the constraint set if the 1324 ** current expression is of the form: column MATCH expr. 1325 ** This information is used by the xBestIndex methods of 1326 ** virtual tables. The native query optimizer does not attempt 1327 ** to do anything with MATCH functions. 1328 */ 1329 if( isMatchOfColumn(pExpr) ){ 1330 int idxNew; 1331 Expr *pRight, *pLeft; 1332 WhereTerm *pNewTerm; 1333 Bitmask prereqColumn, prereqExpr; 1334 1335 pRight = pExpr->x.pList->a[0].pExpr; 1336 pLeft = pExpr->x.pList->a[1].pExpr; 1337 prereqExpr = exprTableUsage(pMaskSet, pRight); 1338 prereqColumn = exprTableUsage(pMaskSet, pLeft); 1339 if( (prereqExpr & prereqColumn)==0 ){ 1340 Expr *pNewExpr; 1341 pNewExpr = sqlite3PExpr(pParse, TK_MATCH, 1342 0, sqlite3ExprDup(db, pRight, 0), 0); 1343 idxNew = whereClauseInsert(pWC, pNewExpr, TERM_VIRTUAL|TERM_DYNAMIC); 1344 testcase( idxNew==0 ); 1345 pNewTerm = &pWC->a[idxNew]; 1346 pNewTerm->prereqRight = prereqExpr; 1347 pNewTerm->leftCursor = pLeft->iTable; 1348 pNewTerm->u.leftColumn = pLeft->iColumn; 1349 pNewTerm->eOperator = WO_MATCH; 1350 pNewTerm->iParent = idxTerm; 1351 pTerm = &pWC->a[idxTerm]; 1352 pTerm->nChild = 1; 1353 pTerm->wtFlags |= TERM_COPIED; 1354 pNewTerm->prereqAll = pTerm->prereqAll; 1355 } 1356 } 1357 #endif /* SQLITE_OMIT_VIRTUALTABLE */ 1358 1359 #ifdef SQLITE_ENABLE_STAT3 1360 /* When sqlite_stat3 histogram data is available an operator of the 1361 ** form "x IS NOT NULL" can sometimes be evaluated more efficiently 1362 ** as "x>NULL" if x is not an INTEGER PRIMARY KEY. So construct a 1363 ** virtual term of that form. 1364 ** 1365 ** Note that the virtual term must be tagged with TERM_VNULL. This 1366 ** TERM_VNULL tag will suppress the not-null check at the beginning 1367 ** of the loop. Without the TERM_VNULL flag, the not-null check at 1368 ** the start of the loop will prevent any results from being returned. 1369 */ 1370 if( pExpr->op==TK_NOTNULL 1371 && pExpr->pLeft->op==TK_COLUMN 1372 && pExpr->pLeft->iColumn>=0 1373 ){ 1374 Expr *pNewExpr; 1375 Expr *pLeft = pExpr->pLeft; 1376 int idxNew; 1377 WhereTerm *pNewTerm; 1378 1379 pNewExpr = sqlite3PExpr(pParse, TK_GT, 1380 sqlite3ExprDup(db, pLeft, 0), 1381 sqlite3PExpr(pParse, TK_NULL, 0, 0, 0), 0); 1382 1383 idxNew = whereClauseInsert(pWC, pNewExpr, 1384 TERM_VIRTUAL|TERM_DYNAMIC|TERM_VNULL); 1385 if( idxNew ){ 1386 pNewTerm = &pWC->a[idxNew]; 1387 pNewTerm->prereqRight = 0; 1388 pNewTerm->leftCursor = pLeft->iTable; 1389 pNewTerm->u.leftColumn = pLeft->iColumn; 1390 pNewTerm->eOperator = WO_GT; 1391 pNewTerm->iParent = idxTerm; 1392 pTerm = &pWC->a[idxTerm]; 1393 pTerm->nChild = 1; 1394 pTerm->wtFlags |= TERM_COPIED; 1395 pNewTerm->prereqAll = pTerm->prereqAll; 1396 } 1397 } 1398 #endif /* SQLITE_ENABLE_STAT */ 1399 1400 /* Prevent ON clause terms of a LEFT JOIN from being used to drive 1401 ** an index for tables to the left of the join. 1402 */ 1403 pTerm->prereqRight |= extraRight; 1404 } 1405 1406 /* 1407 ** Return TRUE if any of the expressions in pList->a[iFirst...] contain 1408 ** a reference to any table other than the iBase table. 1409 */ 1410 static int referencesOtherTables( 1411 ExprList *pList, /* Search expressions in ths list */ 1412 WhereMaskSet *pMaskSet, /* Mapping from tables to bitmaps */ 1413 int iFirst, /* Be searching with the iFirst-th expression */ 1414 int iBase /* Ignore references to this table */ 1415 ){ 1416 Bitmask allowed = ~getMask(pMaskSet, iBase); 1417 while( iFirst<pList->nExpr ){ 1418 if( (exprTableUsage(pMaskSet, pList->a[iFirst++].pExpr)&allowed)!=0 ){ 1419 return 1; 1420 } 1421 } 1422 return 0; 1423 } 1424 1425 /* 1426 ** This function searches the expression list passed as the second argument 1427 ** for an expression of type TK_COLUMN that refers to the same column and 1428 ** uses the same collation sequence as the iCol'th column of index pIdx. 1429 ** Argument iBase is the cursor number used for the table that pIdx refers 1430 ** to. 1431 ** 1432 ** If such an expression is found, its index in pList->a[] is returned. If 1433 ** no expression is found, -1 is returned. 1434 */ 1435 static int findIndexCol( 1436 Parse *pParse, /* Parse context */ 1437 ExprList *pList, /* Expression list to search */ 1438 int iBase, /* Cursor for table associated with pIdx */ 1439 Index *pIdx, /* Index to match column of */ 1440 int iCol /* Column of index to match */ 1441 ){ 1442 int i; 1443 const char *zColl = pIdx->azColl[iCol]; 1444 1445 for(i=0; i<pList->nExpr; i++){ 1446 Expr *p = pList->a[i].pExpr; 1447 if( p->op==TK_COLUMN 1448 && p->iColumn==pIdx->aiColumn[iCol] 1449 && p->iTable==iBase 1450 ){ 1451 CollSeq *pColl = sqlite3ExprCollSeq(pParse, p); 1452 if( ALWAYS(pColl) && 0==sqlite3StrICmp(pColl->zName, zColl) ){ 1453 return i; 1454 } 1455 } 1456 } 1457 1458 return -1; 1459 } 1460 1461 /* 1462 ** This routine determines if pIdx can be used to assist in processing a 1463 ** DISTINCT qualifier. In other words, it tests whether or not using this 1464 ** index for the outer loop guarantees that rows with equal values for 1465 ** all expressions in the pDistinct list are delivered grouped together. 1466 ** 1467 ** For example, the query 1468 ** 1469 ** SELECT DISTINCT a, b, c FROM tbl WHERE a = ? 1470 ** 1471 ** can benefit from any index on columns "b" and "c". 1472 */ 1473 static int isDistinctIndex( 1474 Parse *pParse, /* Parsing context */ 1475 WhereClause *pWC, /* The WHERE clause */ 1476 Index *pIdx, /* The index being considered */ 1477 int base, /* Cursor number for the table pIdx is on */ 1478 ExprList *pDistinct, /* The DISTINCT expressions */ 1479 int nEqCol /* Number of index columns with == */ 1480 ){ 1481 Bitmask mask = 0; /* Mask of unaccounted for pDistinct exprs */ 1482 int i; /* Iterator variable */ 1483 1484 if( pIdx->zName==0 || pDistinct==0 || pDistinct->nExpr>=BMS ) return 0; 1485 testcase( pDistinct->nExpr==BMS-1 ); 1486 1487 /* Loop through all the expressions in the distinct list. If any of them 1488 ** are not simple column references, return early. Otherwise, test if the 1489 ** WHERE clause contains a "col=X" clause. If it does, the expression 1490 ** can be ignored. If it does not, and the column does not belong to the 1491 ** same table as index pIdx, return early. Finally, if there is no 1492 ** matching "col=X" expression and the column is on the same table as pIdx, 1493 ** set the corresponding bit in variable mask. 1494 */ 1495 for(i=0; i<pDistinct->nExpr; i++){ 1496 WhereTerm *pTerm; 1497 Expr *p = pDistinct->a[i].pExpr; 1498 if( p->op!=TK_COLUMN ) return 0; 1499 pTerm = findTerm(pWC, p->iTable, p->iColumn, ~(Bitmask)0, WO_EQ, 0); 1500 if( pTerm ){ 1501 Expr *pX = pTerm->pExpr; 1502 CollSeq *p1 = sqlite3BinaryCompareCollSeq(pParse, pX->pLeft, pX->pRight); 1503 CollSeq *p2 = sqlite3ExprCollSeq(pParse, p); 1504 if( p1==p2 ) continue; 1505 } 1506 if( p->iTable!=base ) return 0; 1507 mask |= (((Bitmask)1) << i); 1508 } 1509 1510 for(i=nEqCol; mask && i<pIdx->nColumn; i++){ 1511 int iExpr = findIndexCol(pParse, pDistinct, base, pIdx, i); 1512 if( iExpr<0 ) break; 1513 mask &= ~(((Bitmask)1) << iExpr); 1514 } 1515 1516 return (mask==0); 1517 } 1518 1519 1520 /* 1521 ** Return true if the DISTINCT expression-list passed as the third argument 1522 ** is redundant. A DISTINCT list is redundant if the database contains a 1523 ** UNIQUE index that guarantees that the result of the query will be distinct 1524 ** anyway. 1525 */ 1526 static int isDistinctRedundant( 1527 Parse *pParse, 1528 SrcList *pTabList, 1529 WhereClause *pWC, 1530 ExprList *pDistinct 1531 ){ 1532 Table *pTab; 1533 Index *pIdx; 1534 int i; 1535 int iBase; 1536 1537 /* If there is more than one table or sub-select in the FROM clause of 1538 ** this query, then it will not be possible to show that the DISTINCT 1539 ** clause is redundant. */ 1540 if( pTabList->nSrc!=1 ) return 0; 1541 iBase = pTabList->a[0].iCursor; 1542 pTab = pTabList->a[0].pTab; 1543 1544 /* If any of the expressions is an IPK column on table iBase, then return 1545 ** true. Note: The (p->iTable==iBase) part of this test may be false if the 1546 ** current SELECT is a correlated sub-query. 1547 */ 1548 for(i=0; i<pDistinct->nExpr; i++){ 1549 Expr *p = pDistinct->a[i].pExpr; 1550 if( p->op==TK_COLUMN && p->iTable==iBase && p->iColumn<0 ) return 1; 1551 } 1552 1553 /* Loop through all indices on the table, checking each to see if it makes 1554 ** the DISTINCT qualifier redundant. It does so if: 1555 ** 1556 ** 1. The index is itself UNIQUE, and 1557 ** 1558 ** 2. All of the columns in the index are either part of the pDistinct 1559 ** list, or else the WHERE clause contains a term of the form "col=X", 1560 ** where X is a constant value. The collation sequences of the 1561 ** comparison and select-list expressions must match those of the index. 1562 */ 1563 for(pIdx=pTab->pIndex; pIdx; pIdx=pIdx->pNext){ 1564 if( pIdx->onError==OE_None ) continue; 1565 for(i=0; i<pIdx->nColumn; i++){ 1566 int iCol = pIdx->aiColumn[i]; 1567 if( 0==findTerm(pWC, iBase, iCol, ~(Bitmask)0, WO_EQ, pIdx) 1568 && 0>findIndexCol(pParse, pDistinct, iBase, pIdx, i) 1569 ){ 1570 break; 1571 } 1572 } 1573 if( i==pIdx->nColumn ){ 1574 /* This index implies that the DISTINCT qualifier is redundant. */ 1575 return 1; 1576 } 1577 } 1578 1579 return 0; 1580 } 1581 1582 /* 1583 ** This routine decides if pIdx can be used to satisfy the ORDER BY 1584 ** clause. If it can, it returns 1. If pIdx cannot satisfy the 1585 ** ORDER BY clause, this routine returns 0. 1586 ** 1587 ** pOrderBy is an ORDER BY clause from a SELECT statement. pTab is the 1588 ** left-most table in the FROM clause of that same SELECT statement and 1589 ** the table has a cursor number of "base". pIdx is an index on pTab. 1590 ** 1591 ** nEqCol is the number of columns of pIdx that are used as equality 1592 ** constraints. Any of these columns may be missing from the ORDER BY 1593 ** clause and the match can still be a success. 1594 ** 1595 ** All terms of the ORDER BY that match against the index must be either 1596 ** ASC or DESC. (Terms of the ORDER BY clause past the end of a UNIQUE 1597 ** index do not need to satisfy this constraint.) The *pbRev value is 1598 ** set to 1 if the ORDER BY clause is all DESC and it is set to 0 if 1599 ** the ORDER BY clause is all ASC. 1600 */ 1601 static int isSortingIndex( 1602 Parse *pParse, /* Parsing context */ 1603 WhereMaskSet *pMaskSet, /* Mapping from table cursor numbers to bitmaps */ 1604 Index *pIdx, /* The index we are testing */ 1605 int base, /* Cursor number for the table to be sorted */ 1606 ExprList *pOrderBy, /* The ORDER BY clause */ 1607 int nEqCol, /* Number of index columns with == constraints */ 1608 int wsFlags, /* Index usages flags */ 1609 int *pbRev /* Set to 1 if ORDER BY is DESC */ 1610 ){ 1611 int i, j; /* Loop counters */ 1612 int sortOrder = 0; /* XOR of index and ORDER BY sort direction */ 1613 int nTerm; /* Number of ORDER BY terms */ 1614 struct ExprList_item *pTerm; /* A term of the ORDER BY clause */ 1615 sqlite3 *db = pParse->db; 1616 1617 if( !pOrderBy ) return 0; 1618 if( wsFlags & WHERE_COLUMN_IN ) return 0; 1619 if( pIdx->bUnordered ) return 0; 1620 1621 nTerm = pOrderBy->nExpr; 1622 assert( nTerm>0 ); 1623 1624 /* Argument pIdx must either point to a 'real' named index structure, 1625 ** or an index structure allocated on the stack by bestBtreeIndex() to 1626 ** represent the rowid index that is part of every table. */ 1627 assert( pIdx->zName || (pIdx->nColumn==1 && pIdx->aiColumn[0]==-1) ); 1628 1629 /* Match terms of the ORDER BY clause against columns of 1630 ** the index. 1631 ** 1632 ** Note that indices have pIdx->nColumn regular columns plus 1633 ** one additional column containing the rowid. The rowid column 1634 ** of the index is also allowed to match against the ORDER BY 1635 ** clause. 1636 */ 1637 for(i=j=0, pTerm=pOrderBy->a; j<nTerm && i<=pIdx->nColumn; i++){ 1638 Expr *pExpr; /* The expression of the ORDER BY pTerm */ 1639 CollSeq *pColl; /* The collating sequence of pExpr */ 1640 int termSortOrder; /* Sort order for this term */ 1641 int iColumn; /* The i-th column of the index. -1 for rowid */ 1642 int iSortOrder; /* 1 for DESC, 0 for ASC on the i-th index term */ 1643 const char *zColl; /* Name of the collating sequence for i-th index term */ 1644 1645 pExpr = pTerm->pExpr; 1646 if( pExpr->op!=TK_COLUMN || pExpr->iTable!=base ){ 1647 /* Can not use an index sort on anything that is not a column in the 1648 ** left-most table of the FROM clause */ 1649 break; 1650 } 1651 pColl = sqlite3ExprCollSeq(pParse, pExpr); 1652 if( !pColl ){ 1653 pColl = db->pDfltColl; 1654 } 1655 if( pIdx->zName && i<pIdx->nColumn ){ 1656 iColumn = pIdx->aiColumn[i]; 1657 if( iColumn==pIdx->pTable->iPKey ){ 1658 iColumn = -1; 1659 } 1660 iSortOrder = pIdx->aSortOrder[i]; 1661 zColl = pIdx->azColl[i]; 1662 }else{ 1663 iColumn = -1; 1664 iSortOrder = 0; 1665 zColl = pColl->zName; 1666 } 1667 if( pExpr->iColumn!=iColumn || sqlite3StrICmp(pColl->zName, zColl) ){ 1668 /* Term j of the ORDER BY clause does not match column i of the index */ 1669 if( i<nEqCol ){ 1670 /* If an index column that is constrained by == fails to match an 1671 ** ORDER BY term, that is OK. Just ignore that column of the index 1672 */ 1673 continue; 1674 }else if( i==pIdx->nColumn ){ 1675 /* Index column i is the rowid. All other terms match. */ 1676 break; 1677 }else{ 1678 /* If an index column fails to match and is not constrained by == 1679 ** then the index cannot satisfy the ORDER BY constraint. 1680 */ 1681 return 0; 1682 } 1683 } 1684 assert( pIdx->aSortOrder!=0 || iColumn==-1 ); 1685 assert( pTerm->sortOrder==0 || pTerm->sortOrder==1 ); 1686 assert( iSortOrder==0 || iSortOrder==1 ); 1687 termSortOrder = iSortOrder ^ pTerm->sortOrder; 1688 if( i>nEqCol ){ 1689 if( termSortOrder!=sortOrder ){ 1690 /* Indices can only be used if all ORDER BY terms past the 1691 ** equality constraints are all either DESC or ASC. */ 1692 return 0; 1693 } 1694 }else{ 1695 sortOrder = termSortOrder; 1696 } 1697 j++; 1698 pTerm++; 1699 if( iColumn<0 && !referencesOtherTables(pOrderBy, pMaskSet, j, base) ){ 1700 /* If the indexed column is the primary key and everything matches 1701 ** so far and none of the ORDER BY terms to the right reference other 1702 ** tables in the join, then we are assured that the index can be used 1703 ** to sort because the primary key is unique and so none of the other 1704 ** columns will make any difference 1705 */ 1706 j = nTerm; 1707 } 1708 } 1709 1710 *pbRev = sortOrder!=0; 1711 if( j>=nTerm ){ 1712 /* All terms of the ORDER BY clause are covered by this index so 1713 ** this index can be used for sorting. */ 1714 return 1; 1715 } 1716 if( pIdx->onError!=OE_None && i==pIdx->nColumn 1717 && (wsFlags & WHERE_COLUMN_NULL)==0 1718 && !referencesOtherTables(pOrderBy, pMaskSet, j, base) ){ 1719 /* All terms of this index match some prefix of the ORDER BY clause 1720 ** and the index is UNIQUE and no terms on the tail of the ORDER BY 1721 ** clause reference other tables in a join. If this is all true then 1722 ** the order by clause is superfluous. Not that if the matching 1723 ** condition is IS NULL then the result is not necessarily unique 1724 ** even on a UNIQUE index, so disallow those cases. */ 1725 return 1; 1726 } 1727 return 0; 1728 } 1729 1730 /* 1731 ** Prepare a crude estimate of the logarithm of the input value. 1732 ** The results need not be exact. This is only used for estimating 1733 ** the total cost of performing operations with O(logN) or O(NlogN) 1734 ** complexity. Because N is just a guess, it is no great tragedy if 1735 ** logN is a little off. 1736 */ 1737 static double estLog(double N){ 1738 double logN = 1; 1739 double x = 10; 1740 while( N>x ){ 1741 logN += 1; 1742 x *= 10; 1743 } 1744 return logN; 1745 } 1746 1747 /* 1748 ** Two routines for printing the content of an sqlite3_index_info 1749 ** structure. Used for testing and debugging only. If neither 1750 ** SQLITE_TEST or SQLITE_DEBUG are defined, then these routines 1751 ** are no-ops. 1752 */ 1753 #if !defined(SQLITE_OMIT_VIRTUALTABLE) && defined(SQLITE_DEBUG) 1754 static void TRACE_IDX_INPUTS(sqlite3_index_info *p){ 1755 int i; 1756 if( !sqlite3WhereTrace ) return; 1757 for(i=0; i<p->nConstraint; i++){ 1758 sqlite3DebugPrintf(" constraint[%d]: col=%d termid=%d op=%d usabled=%d\n", 1759 i, 1760 p->aConstraint[i].iColumn, 1761 p->aConstraint[i].iTermOffset, 1762 p->aConstraint[i].op, 1763 p->aConstraint[i].usable); 1764 } 1765 for(i=0; i<p->nOrderBy; i++){ 1766 sqlite3DebugPrintf(" orderby[%d]: col=%d desc=%d\n", 1767 i, 1768 p->aOrderBy[i].iColumn, 1769 p->aOrderBy[i].desc); 1770 } 1771 } 1772 static void TRACE_IDX_OUTPUTS(sqlite3_index_info *p){ 1773 int i; 1774 if( !sqlite3WhereTrace ) return; 1775 for(i=0; i<p->nConstraint; i++){ 1776 sqlite3DebugPrintf(" usage[%d]: argvIdx=%d omit=%d\n", 1777 i, 1778 p->aConstraintUsage[i].argvIndex, 1779 p->aConstraintUsage[i].omit); 1780 } 1781 sqlite3DebugPrintf(" idxNum=%d\n", p->idxNum); 1782 sqlite3DebugPrintf(" idxStr=%s\n", p->idxStr); 1783 sqlite3DebugPrintf(" orderByConsumed=%d\n", p->orderByConsumed); 1784 sqlite3DebugPrintf(" estimatedCost=%g\n", p->estimatedCost); 1785 } 1786 #else 1787 #define TRACE_IDX_INPUTS(A) 1788 #define TRACE_IDX_OUTPUTS(A) 1789 #endif 1790 1791 /* 1792 ** Required because bestIndex() is called by bestOrClauseIndex() 1793 */ 1794 static void bestIndex( 1795 Parse*, WhereClause*, struct SrcList_item*, 1796 Bitmask, Bitmask, ExprList*, WhereCost*); 1797 1798 /* 1799 ** This routine attempts to find an scanning strategy that can be used 1800 ** to optimize an 'OR' expression that is part of a WHERE clause. 1801 ** 1802 ** The table associated with FROM clause term pSrc may be either a 1803 ** regular B-Tree table or a virtual table. 1804 */ 1805 static void bestOrClauseIndex( 1806 Parse *pParse, /* The parsing context */ 1807 WhereClause *pWC, /* The WHERE clause */ 1808 struct SrcList_item *pSrc, /* The FROM clause term to search */ 1809 Bitmask notReady, /* Mask of cursors not available for indexing */ 1810 Bitmask notValid, /* Cursors not available for any purpose */ 1811 ExprList *pOrderBy, /* The ORDER BY clause */ 1812 WhereCost *pCost /* Lowest cost query plan */ 1813 ){ 1814 #ifndef SQLITE_OMIT_OR_OPTIMIZATION 1815 const int iCur = pSrc->iCursor; /* The cursor of the table to be accessed */ 1816 const Bitmask maskSrc = getMask(pWC->pMaskSet, iCur); /* Bitmask for pSrc */ 1817 WhereTerm * const pWCEnd = &pWC->a[pWC->nTerm]; /* End of pWC->a[] */ 1818 WhereTerm *pTerm; /* A single term of the WHERE clause */ 1819 1820 /* The OR-clause optimization is disallowed if the INDEXED BY or 1821 ** NOT INDEXED clauses are used or if the WHERE_AND_ONLY bit is set. */ 1822 if( pSrc->notIndexed || pSrc->pIndex!=0 ){ 1823 return; 1824 } 1825 if( pWC->wctrlFlags & WHERE_AND_ONLY ){ 1826 return; 1827 } 1828 1829 /* Search the WHERE clause terms for a usable WO_OR term. */ 1830 for(pTerm=pWC->a; pTerm<pWCEnd; pTerm++){ 1831 if( pTerm->eOperator==WO_OR 1832 && ((pTerm->prereqAll & ~maskSrc) & notReady)==0 1833 && (pTerm->u.pOrInfo->indexable & maskSrc)!=0 1834 ){ 1835 WhereClause * const pOrWC = &pTerm->u.pOrInfo->wc; 1836 WhereTerm * const pOrWCEnd = &pOrWC->a[pOrWC->nTerm]; 1837 WhereTerm *pOrTerm; 1838 int flags = WHERE_MULTI_OR; 1839 double rTotal = 0; 1840 double nRow = 0; 1841 Bitmask used = 0; 1842 1843 for(pOrTerm=pOrWC->a; pOrTerm<pOrWCEnd; pOrTerm++){ 1844 WhereCost sTermCost; 1845 WHERETRACE(("... Multi-index OR testing for term %d of %d....\n", 1846 (pOrTerm - pOrWC->a), (pTerm - pWC->a) 1847 )); 1848 if( pOrTerm->eOperator==WO_AND ){ 1849 WhereClause *pAndWC = &pOrTerm->u.pAndInfo->wc; 1850 bestIndex(pParse, pAndWC, pSrc, notReady, notValid, 0, &sTermCost); 1851 }else if( pOrTerm->leftCursor==iCur ){ 1852 WhereClause tempWC; 1853 tempWC.pParse = pWC->pParse; 1854 tempWC.pMaskSet = pWC->pMaskSet; 1855 tempWC.pOuter = pWC; 1856 tempWC.op = TK_AND; 1857 tempWC.a = pOrTerm; 1858 tempWC.wctrlFlags = 0; 1859 tempWC.nTerm = 1; 1860 bestIndex(pParse, &tempWC, pSrc, notReady, notValid, 0, &sTermCost); 1861 }else{ 1862 continue; 1863 } 1864 rTotal += sTermCost.rCost; 1865 nRow += sTermCost.plan.nRow; 1866 used |= sTermCost.used; 1867 if( rTotal>=pCost->rCost ) break; 1868 } 1869 1870 /* If there is an ORDER BY clause, increase the scan cost to account 1871 ** for the cost of the sort. */ 1872 if( pOrderBy!=0 ){ 1873 WHERETRACE(("... sorting increases OR cost %.9g to %.9g\n", 1874 rTotal, rTotal+nRow*estLog(nRow))); 1875 rTotal += nRow*estLog(nRow); 1876 } 1877 1878 /* If the cost of scanning using this OR term for optimization is 1879 ** less than the current cost stored in pCost, replace the contents 1880 ** of pCost. */ 1881 WHERETRACE(("... multi-index OR cost=%.9g nrow=%.9g\n", rTotal, nRow)); 1882 if( rTotal<pCost->rCost ){ 1883 pCost->rCost = rTotal; 1884 pCost->used = used; 1885 pCost->plan.nRow = nRow; 1886 pCost->plan.wsFlags = flags; 1887 pCost->plan.u.pTerm = pTerm; 1888 } 1889 } 1890 } 1891 #endif /* SQLITE_OMIT_OR_OPTIMIZATION */ 1892 } 1893 1894 #ifndef SQLITE_OMIT_AUTOMATIC_INDEX 1895 /* 1896 ** Return TRUE if the WHERE clause term pTerm is of a form where it 1897 ** could be used with an index to access pSrc, assuming an appropriate 1898 ** index existed. 1899 */ 1900 static int termCanDriveIndex( 1901 WhereTerm *pTerm, /* WHERE clause term to check */ 1902 struct SrcList_item *pSrc, /* Table we are trying to access */ 1903 Bitmask notReady /* Tables in outer loops of the join */ 1904 ){ 1905 char aff; 1906 if( pTerm->leftCursor!=pSrc->iCursor ) return 0; 1907 if( pTerm->eOperator!=WO_EQ ) return 0; 1908 if( (pTerm->prereqRight & notReady)!=0 ) return 0; 1909 aff = pSrc->pTab->aCol[pTerm->u.leftColumn].affinity; 1910 if( !sqlite3IndexAffinityOk(pTerm->pExpr, aff) ) return 0; 1911 return 1; 1912 } 1913 #endif 1914 1915 #ifndef SQLITE_OMIT_AUTOMATIC_INDEX 1916 /* 1917 ** If the query plan for pSrc specified in pCost is a full table scan 1918 ** and indexing is allows (if there is no NOT INDEXED clause) and it 1919 ** possible to construct a transient index that would perform better 1920 ** than a full table scan even when the cost of constructing the index 1921 ** is taken into account, then alter the query plan to use the 1922 ** transient index. 1923 */ 1924 static void bestAutomaticIndex( 1925 Parse *pParse, /* The parsing context */ 1926 WhereClause *pWC, /* The WHERE clause */ 1927 struct SrcList_item *pSrc, /* The FROM clause term to search */ 1928 Bitmask notReady, /* Mask of cursors that are not available */ 1929 WhereCost *pCost /* Lowest cost query plan */ 1930 ){ 1931 double nTableRow; /* Rows in the input table */ 1932 double logN; /* log(nTableRow) */ 1933 double costTempIdx; /* per-query cost of the transient index */ 1934 WhereTerm *pTerm; /* A single term of the WHERE clause */ 1935 WhereTerm *pWCEnd; /* End of pWC->a[] */ 1936 Table *pTable; /* Table tht might be indexed */ 1937 1938 if( pParse->nQueryLoop<=(double)1 ){ 1939 /* There is no point in building an automatic index for a single scan */ 1940 return; 1941 } 1942 if( (pParse->db->flags & SQLITE_AutoIndex)==0 ){ 1943 /* Automatic indices are disabled at run-time */ 1944 return; 1945 } 1946 if( (pCost->plan.wsFlags & WHERE_NOT_FULLSCAN)!=0 ){ 1947 /* We already have some kind of index in use for this query. */ 1948 return; 1949 } 1950 if( pSrc->notIndexed ){ 1951 /* The NOT INDEXED clause appears in the SQL. */ 1952 return; 1953 } 1954 if( pSrc->isCorrelated ){ 1955 /* The source is a correlated sub-query. No point in indexing it. */ 1956 return; 1957 } 1958 1959 assert( pParse->nQueryLoop >= (double)1 ); 1960 pTable = pSrc->pTab; 1961 nTableRow = pTable->nRowEst; 1962 logN = estLog(nTableRow); 1963 costTempIdx = 2*logN*(nTableRow/pParse->nQueryLoop + 1); 1964 if( costTempIdx>=pCost->rCost ){ 1965 /* The cost of creating the transient table would be greater than 1966 ** doing the full table scan */ 1967 return; 1968 } 1969 1970 /* Search for any equality comparison term */ 1971 pWCEnd = &pWC->a[pWC->nTerm]; 1972 for(pTerm=pWC->a; pTerm<pWCEnd; pTerm++){ 1973 if( termCanDriveIndex(pTerm, pSrc, notReady) ){ 1974 WHERETRACE(("auto-index reduces cost from %.1f to %.1f\n", 1975 pCost->rCost, costTempIdx)); 1976 pCost->rCost = costTempIdx; 1977 pCost->plan.nRow = logN + 1; 1978 pCost->plan.wsFlags = WHERE_TEMP_INDEX; 1979 pCost->used = pTerm->prereqRight; 1980 break; 1981 } 1982 } 1983 } 1984 #else 1985 # define bestAutomaticIndex(A,B,C,D,E) /* no-op */ 1986 #endif /* SQLITE_OMIT_AUTOMATIC_INDEX */ 1987 1988 1989 #ifndef SQLITE_OMIT_AUTOMATIC_INDEX 1990 /* 1991 ** Generate code to construct the Index object for an automatic index 1992 ** and to set up the WhereLevel object pLevel so that the code generator 1993 ** makes use of the automatic index. 1994 */ 1995 static void constructAutomaticIndex( 1996 Parse *pParse, /* The parsing context */ 1997 WhereClause *pWC, /* The WHERE clause */ 1998 struct SrcList_item *pSrc, /* The FROM clause term to get the next index */ 1999 Bitmask notReady, /* Mask of cursors that are not available */ 2000 WhereLevel *pLevel /* Write new index here */ 2001 ){ 2002 int nColumn; /* Number of columns in the constructed index */ 2003 WhereTerm *pTerm; /* A single term of the WHERE clause */ 2004 WhereTerm *pWCEnd; /* End of pWC->a[] */ 2005 int nByte; /* Byte of memory needed for pIdx */ 2006 Index *pIdx; /* Object describing the transient index */ 2007 Vdbe *v; /* Prepared statement under construction */ 2008 int addrInit; /* Address of the initialization bypass jump */ 2009 Table *pTable; /* The table being indexed */ 2010 KeyInfo *pKeyinfo; /* Key information for the index */ 2011 int addrTop; /* Top of the index fill loop */ 2012 int regRecord; /* Register holding an index record */ 2013 int n; /* Column counter */ 2014 int i; /* Loop counter */ 2015 int mxBitCol; /* Maximum column in pSrc->colUsed */ 2016 CollSeq *pColl; /* Collating sequence to on a column */ 2017 Bitmask idxCols; /* Bitmap of columns used for indexing */ 2018 Bitmask extraCols; /* Bitmap of additional columns */ 2019 2020 /* Generate code to skip over the creation and initialization of the 2021 ** transient index on 2nd and subsequent iterations of the loop. */ 2022 v = pParse->pVdbe; 2023 assert( v!=0 ); 2024 addrInit = sqlite3CodeOnce(pParse); 2025 2026 /* Count the number of columns that will be added to the index 2027 ** and used to match WHERE clause constraints */ 2028 nColumn = 0; 2029 pTable = pSrc->pTab; 2030 pWCEnd = &pWC->a[pWC->nTerm]; 2031 idxCols = 0; 2032 for(pTerm=pWC->a; pTerm<pWCEnd; pTerm++){ 2033 if( termCanDriveIndex(pTerm, pSrc, notReady) ){ 2034 int iCol = pTerm->u.leftColumn; 2035 Bitmask cMask = iCol>=BMS ? ((Bitmask)1)<<(BMS-1) : ((Bitmask)1)<<iCol; 2036 testcase( iCol==BMS ); 2037 testcase( iCol==BMS-1 ); 2038 if( (idxCols & cMask)==0 ){ 2039 nColumn++; 2040 idxCols |= cMask; 2041 } 2042 } 2043 } 2044 assert( nColumn>0 ); 2045 pLevel->plan.nEq = nColumn; 2046 2047 /* Count the number of additional columns needed to create a 2048 ** covering index. A "covering index" is an index that contains all 2049 ** columns that are needed by the query. With a covering index, the 2050 ** original table never needs to be accessed. Automatic indices must 2051 ** be a covering index because the index will not be updated if the 2052 ** original table changes and the index and table cannot both be used 2053 ** if they go out of sync. 2054 */ 2055 extraCols = pSrc->colUsed & (~idxCols | (((Bitmask)1)<<(BMS-1))); 2056 mxBitCol = (pTable->nCol >= BMS-1) ? BMS-1 : pTable->nCol; 2057 testcase( pTable->nCol==BMS-1 ); 2058 testcase( pTable->nCol==BMS-2 ); 2059 for(i=0; i<mxBitCol; i++){ 2060 if( extraCols & (((Bitmask)1)<<i) ) nColumn++; 2061 } 2062 if( pSrc->colUsed & (((Bitmask)1)<<(BMS-1)) ){ 2063 nColumn += pTable->nCol - BMS + 1; 2064 } 2065 pLevel->plan.wsFlags |= WHERE_COLUMN_EQ | WHERE_IDX_ONLY | WO_EQ; 2066 2067 /* Construct the Index object to describe this index */ 2068 nByte = sizeof(Index); 2069 nByte += nColumn*sizeof(int); /* Index.aiColumn */ 2070 nByte += nColumn*sizeof(char*); /* Index.azColl */ 2071 nByte += nColumn; /* Index.aSortOrder */ 2072 pIdx = sqlite3DbMallocZero(pParse->db, nByte); 2073 if( pIdx==0 ) return; 2074 pLevel->plan.u.pIdx = pIdx; 2075 pIdx->azColl = (char**)&pIdx[1]; 2076 pIdx->aiColumn = (int*)&pIdx->azColl[nColumn]; 2077 pIdx->aSortOrder = (u8*)&pIdx->aiColumn[nColumn]; 2078 pIdx->zName = "auto-index"; 2079 pIdx->nColumn = nColumn; 2080 pIdx->pTable = pTable; 2081 n = 0; 2082 idxCols = 0; 2083 for(pTerm=pWC->a; pTerm<pWCEnd; pTerm++){ 2084 if( termCanDriveIndex(pTerm, pSrc, notReady) ){ 2085 int iCol = pTerm->u.leftColumn; 2086 Bitmask cMask = iCol>=BMS ? ((Bitmask)1)<<(BMS-1) : ((Bitmask)1)<<iCol; 2087 if( (idxCols & cMask)==0 ){ 2088 Expr *pX = pTerm->pExpr; 2089 idxCols |= cMask; 2090 pIdx->aiColumn[n] = pTerm->u.leftColumn; 2091 pColl = sqlite3BinaryCompareCollSeq(pParse, pX->pLeft, pX->pRight); 2092 pIdx->azColl[n] = ALWAYS(pColl) ? pColl->zName : "BINARY"; 2093 n++; 2094 } 2095 } 2096 } 2097 assert( (u32)n==pLevel->plan.nEq ); 2098 2099 /* Add additional columns needed to make the automatic index into 2100 ** a covering index */ 2101 for(i=0; i<mxBitCol; i++){ 2102 if( extraCols & (((Bitmask)1)<<i) ){ 2103 pIdx->aiColumn[n] = i; 2104 pIdx->azColl[n] = "BINARY"; 2105 n++; 2106 } 2107 } 2108 if( pSrc->colUsed & (((Bitmask)1)<<(BMS-1)) ){ 2109 for(i=BMS-1; i<pTable->nCol; i++){ 2110 pIdx->aiColumn[n] = i; 2111 pIdx->azColl[n] = "BINARY"; 2112 n++; 2113 } 2114 } 2115 assert( n==nColumn ); 2116 2117 /* Create the automatic index */ 2118 pKeyinfo = sqlite3IndexKeyinfo(pParse, pIdx); 2119 assert( pLevel->iIdxCur>=0 ); 2120 sqlite3VdbeAddOp4(v, OP_OpenAutoindex, pLevel->iIdxCur, nColumn+1, 0, 2121 (char*)pKeyinfo, P4_KEYINFO_HANDOFF); 2122 VdbeComment((v, "for %s", pTable->zName)); 2123 2124 /* Fill the automatic index with content */ 2125 addrTop = sqlite3VdbeAddOp1(v, OP_Rewind, pLevel->iTabCur); 2126 regRecord = sqlite3GetTempReg(pParse); 2127 sqlite3GenerateIndexKey(pParse, pIdx, pLevel->iTabCur, regRecord, 1); 2128 sqlite3VdbeAddOp2(v, OP_IdxInsert, pLevel->iIdxCur, regRecord); 2129 sqlite3VdbeChangeP5(v, OPFLAG_USESEEKRESULT); 2130 sqlite3VdbeAddOp2(v, OP_Next, pLevel->iTabCur, addrTop+1); 2131 sqlite3VdbeChangeP5(v, SQLITE_STMTSTATUS_AUTOINDEX); 2132 sqlite3VdbeJumpHere(v, addrTop); 2133 sqlite3ReleaseTempReg(pParse, regRecord); 2134 2135 /* Jump here when skipping the initialization */ 2136 sqlite3VdbeJumpHere(v, addrInit); 2137 } 2138 #endif /* SQLITE_OMIT_AUTOMATIC_INDEX */ 2139 2140 #ifndef SQLITE_OMIT_VIRTUALTABLE 2141 /* 2142 ** Allocate and populate an sqlite3_index_info structure. It is the 2143 ** responsibility of the caller to eventually release the structure 2144 ** by passing the pointer returned by this function to sqlite3_free(). 2145 */ 2146 static sqlite3_index_info *allocateIndexInfo( 2147 Parse *pParse, 2148 WhereClause *pWC, 2149 struct SrcList_item *pSrc, 2150 ExprList *pOrderBy 2151 ){ 2152 int i, j; 2153 int nTerm; 2154 struct sqlite3_index_constraint *pIdxCons; 2155 struct sqlite3_index_orderby *pIdxOrderBy; 2156 struct sqlite3_index_constraint_usage *pUsage; 2157 WhereTerm *pTerm; 2158 int nOrderBy; 2159 sqlite3_index_info *pIdxInfo; 2160 2161 WHERETRACE(("Recomputing index info for %s...\n", pSrc->pTab->zName)); 2162 2163 /* Count the number of possible WHERE clause constraints referring 2164 ** to this virtual table */ 2165 for(i=nTerm=0, pTerm=pWC->a; i<pWC->nTerm; i++, pTerm++){ 2166 if( pTerm->leftCursor != pSrc->iCursor ) continue; 2167 assert( (pTerm->eOperator&(pTerm->eOperator-1))==0 ); 2168 testcase( pTerm->eOperator==WO_IN ); 2169 testcase( pTerm->eOperator==WO_ISNULL ); 2170 if( pTerm->eOperator & (WO_IN|WO_ISNULL) ) continue; 2171 if( pTerm->wtFlags & TERM_VNULL ) continue; 2172 nTerm++; 2173 } 2174 2175 /* If the ORDER BY clause contains only columns in the current 2176 ** virtual table then allocate space for the aOrderBy part of 2177 ** the sqlite3_index_info structure. 2178 */ 2179 nOrderBy = 0; 2180 if( pOrderBy ){ 2181 for(i=0; i<pOrderBy->nExpr; i++){ 2182 Expr *pExpr = pOrderBy->a[i].pExpr; 2183 if( pExpr->op!=TK_COLUMN || pExpr->iTable!=pSrc->iCursor ) break; 2184 } 2185 if( i==pOrderBy->nExpr ){ 2186 nOrderBy = pOrderBy->nExpr; 2187 } 2188 } 2189 2190 /* Allocate the sqlite3_index_info structure 2191 */ 2192 pIdxInfo = sqlite3DbMallocZero(pParse->db, sizeof(*pIdxInfo) 2193 + (sizeof(*pIdxCons) + sizeof(*pUsage))*nTerm 2194 + sizeof(*pIdxOrderBy)*nOrderBy ); 2195 if( pIdxInfo==0 ){ 2196 sqlite3ErrorMsg(pParse, "out of memory"); 2197 /* (double)0 In case of SQLITE_OMIT_FLOATING_POINT... */ 2198 return 0; 2199 } 2200 2201 /* Initialize the structure. The sqlite3_index_info structure contains 2202 ** many fields that are declared "const" to prevent xBestIndex from 2203 ** changing them. We have to do some funky casting in order to 2204 ** initialize those fields. 2205 */ 2206 pIdxCons = (struct sqlite3_index_constraint*)&pIdxInfo[1]; 2207 pIdxOrderBy = (struct sqlite3_index_orderby*)&pIdxCons[nTerm]; 2208 pUsage = (struct sqlite3_index_constraint_usage*)&pIdxOrderBy[nOrderBy]; 2209 *(int*)&pIdxInfo->nConstraint = nTerm; 2210 *(int*)&pIdxInfo->nOrderBy = nOrderBy; 2211 *(struct sqlite3_index_constraint**)&pIdxInfo->aConstraint = pIdxCons; 2212 *(struct sqlite3_index_orderby**)&pIdxInfo->aOrderBy = pIdxOrderBy; 2213 *(struct sqlite3_index_constraint_usage**)&pIdxInfo->aConstraintUsage = 2214 pUsage; 2215 2216 for(i=j=0, pTerm=pWC->a; i<pWC->nTerm; i++, pTerm++){ 2217 if( pTerm->leftCursor != pSrc->iCursor ) continue; 2218 assert( (pTerm->eOperator&(pTerm->eOperator-1))==0 ); 2219 testcase( pTerm->eOperator==WO_IN ); 2220 testcase( pTerm->eOperator==WO_ISNULL ); 2221 if( pTerm->eOperator & (WO_IN|WO_ISNULL) ) continue; 2222 if( pTerm->wtFlags & TERM_VNULL ) continue; 2223 pIdxCons[j].iColumn = pTerm->u.leftColumn; 2224 pIdxCons[j].iTermOffset = i; 2225 pIdxCons[j].op = (u8)pTerm->eOperator; 2226 /* The direct assignment in the previous line is possible only because 2227 ** the WO_ and SQLITE_INDEX_CONSTRAINT_ codes are identical. The 2228 ** following asserts verify this fact. */ 2229 assert( WO_EQ==SQLITE_INDEX_CONSTRAINT_EQ ); 2230 assert( WO_LT==SQLITE_INDEX_CONSTRAINT_LT ); 2231 assert( WO_LE==SQLITE_INDEX_CONSTRAINT_LE ); 2232 assert( WO_GT==SQLITE_INDEX_CONSTRAINT_GT ); 2233 assert( WO_GE==SQLITE_INDEX_CONSTRAINT_GE ); 2234 assert( WO_MATCH==SQLITE_INDEX_CONSTRAINT_MATCH ); 2235 assert( pTerm->eOperator & (WO_EQ|WO_LT|WO_LE|WO_GT|WO_GE|WO_MATCH) ); 2236 j++; 2237 } 2238 for(i=0; i<nOrderBy; i++){ 2239 Expr *pExpr = pOrderBy->a[i].pExpr; 2240 pIdxOrderBy[i].iColumn = pExpr->iColumn; 2241 pIdxOrderBy[i].desc = pOrderBy->a[i].sortOrder; 2242 } 2243 2244 return pIdxInfo; 2245 } 2246 2247 /* 2248 ** The table object reference passed as the second argument to this function 2249 ** must represent a virtual table. This function invokes the xBestIndex() 2250 ** method of the virtual table with the sqlite3_index_info pointer passed 2251 ** as the argument. 2252 ** 2253 ** If an error occurs, pParse is populated with an error message and a 2254 ** non-zero value is returned. Otherwise, 0 is returned and the output 2255 ** part of the sqlite3_index_info structure is left populated. 2256 ** 2257 ** Whether or not an error is returned, it is the responsibility of the 2258 ** caller to eventually free p->idxStr if p->needToFreeIdxStr indicates 2259 ** that this is required. 2260 */ 2261 static int vtabBestIndex(Parse *pParse, Table *pTab, sqlite3_index_info *p){ 2262 sqlite3_vtab *pVtab = sqlite3GetVTable(pParse->db, pTab)->pVtab; 2263 int i; 2264 int rc; 2265 2266 WHERETRACE(("xBestIndex for %s\n", pTab->zName)); 2267 TRACE_IDX_INPUTS(p); 2268 rc = pVtab->pModule->xBestIndex(pVtab, p); 2269 TRACE_IDX_OUTPUTS(p); 2270 2271 if( rc!=SQLITE_OK ){ 2272 if( rc==SQLITE_NOMEM ){ 2273 pParse->db->mallocFailed = 1; 2274 }else if( !pVtab->zErrMsg ){ 2275 sqlite3ErrorMsg(pParse, "%s", sqlite3ErrStr(rc)); 2276 }else{ 2277 sqlite3ErrorMsg(pParse, "%s", pVtab->zErrMsg); 2278 } 2279 } 2280 sqlite3_free(pVtab->zErrMsg); 2281 pVtab->zErrMsg = 0; 2282 2283 for(i=0; i<p->nConstraint; i++){ 2284 if( !p->aConstraint[i].usable && p->aConstraintUsage[i].argvIndex>0 ){ 2285 sqlite3ErrorMsg(pParse, 2286 "table %s: xBestIndex returned an invalid plan", pTab->zName); 2287 } 2288 } 2289 2290 return pParse->nErr; 2291 } 2292 2293 2294 /* 2295 ** Compute the best index for a virtual table. 2296 ** 2297 ** The best index is computed by the xBestIndex method of the virtual 2298 ** table module. This routine is really just a wrapper that sets up 2299 ** the sqlite3_index_info structure that is used to communicate with 2300 ** xBestIndex. 2301 ** 2302 ** In a join, this routine might be called multiple times for the 2303 ** same virtual table. The sqlite3_index_info structure is created 2304 ** and initialized on the first invocation and reused on all subsequent 2305 ** invocations. The sqlite3_index_info structure is also used when 2306 ** code is generated to access the virtual table. The whereInfoDelete() 2307 ** routine takes care of freeing the sqlite3_index_info structure after 2308 ** everybody has finished with it. 2309 */ 2310 static void bestVirtualIndex( 2311 Parse *pParse, /* The parsing context */ 2312 WhereClause *pWC, /* The WHERE clause */ 2313 struct SrcList_item *pSrc, /* The FROM clause term to search */ 2314 Bitmask notReady, /* Mask of cursors not available for index */ 2315 Bitmask notValid, /* Cursors not valid for any purpose */ 2316 ExprList *pOrderBy, /* The order by clause */ 2317 WhereCost *pCost, /* Lowest cost query plan */ 2318 sqlite3_index_info **ppIdxInfo /* Index information passed to xBestIndex */ 2319 ){ 2320 Table *pTab = pSrc->pTab; 2321 sqlite3_index_info *pIdxInfo; 2322 struct sqlite3_index_constraint *pIdxCons; 2323 struct sqlite3_index_constraint_usage *pUsage; 2324 WhereTerm *pTerm; 2325 int i, j; 2326 int nOrderBy; 2327 double rCost; 2328 2329 /* Make sure wsFlags is initialized to some sane value. Otherwise, if the 2330 ** malloc in allocateIndexInfo() fails and this function returns leaving 2331 ** wsFlags in an uninitialized state, the caller may behave unpredictably. 2332 */ 2333 memset(pCost, 0, sizeof(*pCost)); 2334 pCost->plan.wsFlags = WHERE_VIRTUALTABLE; 2335 2336 /* If the sqlite3_index_info structure has not been previously 2337 ** allocated and initialized, then allocate and initialize it now. 2338 */ 2339 pIdxInfo = *ppIdxInfo; 2340 if( pIdxInfo==0 ){ 2341 *ppIdxInfo = pIdxInfo = allocateIndexInfo(pParse, pWC, pSrc, pOrderBy); 2342 } 2343 if( pIdxInfo==0 ){ 2344 return; 2345 } 2346 2347 /* At this point, the sqlite3_index_info structure that pIdxInfo points 2348 ** to will have been initialized, either during the current invocation or 2349 ** during some prior invocation. Now we just have to customize the 2350 ** details of pIdxInfo for the current invocation and pass it to 2351 ** xBestIndex. 2352 */ 2353 2354 /* The module name must be defined. Also, by this point there must 2355 ** be a pointer to an sqlite3_vtab structure. Otherwise 2356 ** sqlite3ViewGetColumnNames() would have picked up the error. 2357 */ 2358 assert( pTab->azModuleArg && pTab->azModuleArg[0] ); 2359 assert( sqlite3GetVTable(pParse->db, pTab) ); 2360 2361 /* Set the aConstraint[].usable fields and initialize all 2362 ** output variables to zero. 2363 ** 2364 ** aConstraint[].usable is true for constraints where the right-hand 2365 ** side contains only references to tables to the left of the current 2366 ** table. In other words, if the constraint is of the form: 2367 ** 2368 ** column = expr 2369 ** 2370 ** and we are evaluating a join, then the constraint on column is 2371 ** only valid if all tables referenced in expr occur to the left 2372 ** of the table containing column. 2373 ** 2374 ** The aConstraints[] array contains entries for all constraints 2375 ** on the current table. That way we only have to compute it once 2376 ** even though we might try to pick the best index multiple times. 2377 ** For each attempt at picking an index, the order of tables in the 2378 ** join might be different so we have to recompute the usable flag 2379 ** each time. 2380 */ 2381 pIdxCons = *(struct sqlite3_index_constraint**)&pIdxInfo->aConstraint; 2382 pUsage = pIdxInfo->aConstraintUsage; 2383 for(i=0; i<pIdxInfo->nConstraint; i++, pIdxCons++){ 2384 j = pIdxCons->iTermOffset; 2385 pTerm = &pWC->a[j]; 2386 pIdxCons->usable = (pTerm->prereqRight¬Ready) ? 0 : 1; 2387 } 2388 memset(pUsage, 0, sizeof(pUsage[0])*pIdxInfo->nConstraint); 2389 if( pIdxInfo->needToFreeIdxStr ){ 2390 sqlite3_free(pIdxInfo->idxStr); 2391 } 2392 pIdxInfo->idxStr = 0; 2393 pIdxInfo->idxNum = 0; 2394 pIdxInfo->needToFreeIdxStr = 0; 2395 pIdxInfo->orderByConsumed = 0; 2396 /* ((double)2) In case of SQLITE_OMIT_FLOATING_POINT... */ 2397 pIdxInfo->estimatedCost = SQLITE_BIG_DBL / ((double)2); 2398 nOrderBy = pIdxInfo->nOrderBy; 2399 if( !pOrderBy ){ 2400 pIdxInfo->nOrderBy = 0; 2401 } 2402 2403 if( vtabBestIndex(pParse, pTab, pIdxInfo) ){ 2404 return; 2405 } 2406 2407 pIdxCons = *(struct sqlite3_index_constraint**)&pIdxInfo->aConstraint; 2408 for(i=0; i<pIdxInfo->nConstraint; i++){ 2409 if( pUsage[i].argvIndex>0 ){ 2410 pCost->used |= pWC->a[pIdxCons[i].iTermOffset].prereqRight; 2411 } 2412 } 2413 2414 /* If there is an ORDER BY clause, and the selected virtual table index 2415 ** does not satisfy it, increase the cost of the scan accordingly. This 2416 ** matches the processing for non-virtual tables in bestBtreeIndex(). 2417 */ 2418 rCost = pIdxInfo->estimatedCost; 2419 if( pOrderBy && pIdxInfo->orderByConsumed==0 ){ 2420 rCost += estLog(rCost)*rCost; 2421 } 2422 2423 /* The cost is not allowed to be larger than SQLITE_BIG_DBL (the 2424 ** inital value of lowestCost in this loop. If it is, then the 2425 ** (cost<lowestCost) test below will never be true. 2426 ** 2427 ** Use "(double)2" instead of "2.0" in case OMIT_FLOATING_POINT 2428 ** is defined. 2429 */ 2430 if( (SQLITE_BIG_DBL/((double)2))<rCost ){ 2431 pCost->rCost = (SQLITE_BIG_DBL/((double)2)); 2432 }else{ 2433 pCost->rCost = rCost; 2434 } 2435 pCost->plan.u.pVtabIdx = pIdxInfo; 2436 if( pIdxInfo->orderByConsumed ){ 2437 pCost->plan.wsFlags |= WHERE_ORDERBY; 2438 } 2439 pCost->plan.nEq = 0; 2440 pIdxInfo->nOrderBy = nOrderBy; 2441 2442 /* Try to find a more efficient access pattern by using multiple indexes 2443 ** to optimize an OR expression within the WHERE clause. 2444 */ 2445 bestOrClauseIndex(pParse, pWC, pSrc, notReady, notValid, pOrderBy, pCost); 2446 } 2447 #endif /* SQLITE_OMIT_VIRTUALTABLE */ 2448 2449 #ifdef SQLITE_ENABLE_STAT3 2450 /* 2451 ** Estimate the location of a particular key among all keys in an 2452 ** index. Store the results in aStat as follows: 2453 ** 2454 ** aStat[0] Est. number of rows less than pVal 2455 ** aStat[1] Est. number of rows equal to pVal 2456 ** 2457 ** Return SQLITE_OK on success. 2458 */ 2459 static int whereKeyStats( 2460 Parse *pParse, /* Database connection */ 2461 Index *pIdx, /* Index to consider domain of */ 2462 sqlite3_value *pVal, /* Value to consider */ 2463 int roundUp, /* Round up if true. Round down if false */ 2464 tRowcnt *aStat /* OUT: stats written here */ 2465 ){ 2466 tRowcnt n; 2467 IndexSample *aSample; 2468 int i, eType; 2469 int isEq = 0; 2470 i64 v; 2471 double r, rS; 2472 2473 assert( roundUp==0 || roundUp==1 ); 2474 assert( pIdx->nSample>0 ); 2475 if( pVal==0 ) return SQLITE_ERROR; 2476 n = pIdx->aiRowEst[0]; 2477 aSample = pIdx->aSample; 2478 eType = sqlite3_value_type(pVal); 2479 2480 if( eType==SQLITE_INTEGER ){ 2481 v = sqlite3_value_int64(pVal); 2482 r = (i64)v; 2483 for(i=0; i<pIdx->nSample; i++){ 2484 if( aSample[i].eType==SQLITE_NULL ) continue; 2485 if( aSample[i].eType>=SQLITE_TEXT ) break; 2486 if( aSample[i].eType==SQLITE_INTEGER ){ 2487 if( aSample[i].u.i>=v ){ 2488 isEq = aSample[i].u.i==v; 2489 break; 2490 } 2491 }else{ 2492 assert( aSample[i].eType==SQLITE_FLOAT ); 2493 if( aSample[i].u.r>=r ){ 2494 isEq = aSample[i].u.r==r; 2495 break; 2496 } 2497 } 2498 } 2499 }else if( eType==SQLITE_FLOAT ){ 2500 r = sqlite3_value_double(pVal); 2501 for(i=0; i<pIdx->nSample; i++){ 2502 if( aSample[i].eType==SQLITE_NULL ) continue; 2503 if( aSample[i].eType>=SQLITE_TEXT ) break; 2504 if( aSample[i].eType==SQLITE_FLOAT ){ 2505 rS = aSample[i].u.r; 2506 }else{ 2507 rS = aSample[i].u.i; 2508 } 2509 if( rS>=r ){ 2510 isEq = rS==r; 2511 break; 2512 } 2513 } 2514 }else if( eType==SQLITE_NULL ){ 2515 i = 0; 2516 if( aSample[0].eType==SQLITE_NULL ) isEq = 1; 2517 }else{ 2518 assert( eType==SQLITE_TEXT || eType==SQLITE_BLOB ); 2519 for(i=0; i<pIdx->nSample; i++){ 2520 if( aSample[i].eType==SQLITE_TEXT || aSample[i].eType==SQLITE_BLOB ){ 2521 break; 2522 } 2523 } 2524 if( i<pIdx->nSample ){ 2525 sqlite3 *db = pParse->db; 2526 CollSeq *pColl; 2527 const u8 *z; 2528 if( eType==SQLITE_BLOB ){ 2529 z = (const u8 *)sqlite3_value_blob(pVal); 2530 pColl = db->pDfltColl; 2531 assert( pColl->enc==SQLITE_UTF8 ); 2532 }else{ 2533 pColl = sqlite3GetCollSeq(db, SQLITE_UTF8, 0, *pIdx->azColl); 2534 if( pColl==0 ){ 2535 sqlite3ErrorMsg(pParse, "no such collation sequence: %s", 2536 *pIdx->azColl); 2537 return SQLITE_ERROR; 2538 } 2539 z = (const u8 *)sqlite3ValueText(pVal, pColl->enc); 2540 if( !z ){ 2541 return SQLITE_NOMEM; 2542 } 2543 assert( z && pColl && pColl->xCmp ); 2544 } 2545 n = sqlite3ValueBytes(pVal, pColl->enc); 2546 2547 for(; i<pIdx->nSample; i++){ 2548 int c; 2549 int eSampletype = aSample[i].eType; 2550 if( eSampletype<eType ) continue; 2551 if( eSampletype!=eType ) break; 2552 #ifndef SQLITE_OMIT_UTF16 2553 if( pColl->enc!=SQLITE_UTF8 ){ 2554 int nSample; 2555 char *zSample = sqlite3Utf8to16( 2556 db, pColl->enc, aSample[i].u.z, aSample[i].nByte, &nSample 2557 ); 2558 if( !zSample ){ 2559 assert( db->mallocFailed ); 2560 return SQLITE_NOMEM; 2561 } 2562 c = pColl->xCmp(pColl->pUser, nSample, zSample, n, z); 2563 sqlite3DbFree(db, zSample); 2564 }else 2565 #endif 2566 { 2567 c = pColl->xCmp(pColl->pUser, aSample[i].nByte, aSample[i].u.z, n, z); 2568 } 2569 if( c>=0 ){ 2570 if( c==0 ) isEq = 1; 2571 break; 2572 } 2573 } 2574 } 2575 } 2576 2577 /* At this point, aSample[i] is the first sample that is greater than 2578 ** or equal to pVal. Or if i==pIdx->nSample, then all samples are less 2579 ** than pVal. If aSample[i]==pVal, then isEq==1. 2580 */ 2581 if( isEq ){ 2582 assert( i<pIdx->nSample ); 2583 aStat[0] = aSample[i].nLt; 2584 aStat[1] = aSample[i].nEq; 2585 }else{ 2586 tRowcnt iLower, iUpper, iGap; 2587 if( i==0 ){ 2588 iLower = 0; 2589 iUpper = aSample[0].nLt; 2590 }else{ 2591 iUpper = i>=pIdx->nSample ? n : aSample[i].nLt; 2592 iLower = aSample[i-1].nEq + aSample[i-1].nLt; 2593 } 2594 aStat[1] = pIdx->avgEq; 2595 if( iLower>=iUpper ){ 2596 iGap = 0; 2597 }else{ 2598 iGap = iUpper - iLower; 2599 } 2600 if( roundUp ){ 2601 iGap = (iGap*2)/3; 2602 }else{ 2603 iGap = iGap/3; 2604 } 2605 aStat[0] = iLower + iGap; 2606 } 2607 return SQLITE_OK; 2608 } 2609 #endif /* SQLITE_ENABLE_STAT3 */ 2610 2611 /* 2612 ** If expression pExpr represents a literal value, set *pp to point to 2613 ** an sqlite3_value structure containing the same value, with affinity 2614 ** aff applied to it, before returning. It is the responsibility of the 2615 ** caller to eventually release this structure by passing it to 2616 ** sqlite3ValueFree(). 2617 ** 2618 ** If the current parse is a recompile (sqlite3Reprepare()) and pExpr 2619 ** is an SQL variable that currently has a non-NULL value bound to it, 2620 ** create an sqlite3_value structure containing this value, again with 2621 ** affinity aff applied to it, instead. 2622 ** 2623 ** If neither of the above apply, set *pp to NULL. 2624 ** 2625 ** If an error occurs, return an error code. Otherwise, SQLITE_OK. 2626 */ 2627 #ifdef SQLITE_ENABLE_STAT3 2628 static int valueFromExpr( 2629 Parse *pParse, 2630 Expr *pExpr, 2631 u8 aff, 2632 sqlite3_value **pp 2633 ){ 2634 if( pExpr->op==TK_VARIABLE 2635 || (pExpr->op==TK_REGISTER && pExpr->op2==TK_VARIABLE) 2636 ){ 2637 int iVar = pExpr->iColumn; 2638 sqlite3VdbeSetVarmask(pParse->pVdbe, iVar); 2639 *pp = sqlite3VdbeGetValue(pParse->pReprepare, iVar, aff); 2640 return SQLITE_OK; 2641 } 2642 return sqlite3ValueFromExpr(pParse->db, pExpr, SQLITE_UTF8, aff, pp); 2643 } 2644 #endif 2645 2646 /* 2647 ** This function is used to estimate the number of rows that will be visited 2648 ** by scanning an index for a range of values. The range may have an upper 2649 ** bound, a lower bound, or both. The WHERE clause terms that set the upper 2650 ** and lower bounds are represented by pLower and pUpper respectively. For 2651 ** example, assuming that index p is on t1(a): 2652 ** 2653 ** ... FROM t1 WHERE a > ? AND a < ? ... 2654 ** |_____| |_____| 2655 ** | | 2656 ** pLower pUpper 2657 ** 2658 ** If either of the upper or lower bound is not present, then NULL is passed in 2659 ** place of the corresponding WhereTerm. 2660 ** 2661 ** The nEq parameter is passed the index of the index column subject to the 2662 ** range constraint. Or, equivalently, the number of equality constraints 2663 ** optimized by the proposed index scan. For example, assuming index p is 2664 ** on t1(a, b), and the SQL query is: 2665 ** 2666 ** ... FROM t1 WHERE a = ? AND b > ? AND b < ? ... 2667 ** 2668 ** then nEq should be passed the value 1 (as the range restricted column, 2669 ** b, is the second left-most column of the index). Or, if the query is: 2670 ** 2671 ** ... FROM t1 WHERE a > ? AND a < ? ... 2672 ** 2673 ** then nEq should be passed 0. 2674 ** 2675 ** The returned value is an integer divisor to reduce the estimated 2676 ** search space. A return value of 1 means that range constraints are 2677 ** no help at all. A return value of 2 means range constraints are 2678 ** expected to reduce the search space by half. And so forth... 2679 ** 2680 ** In the absence of sqlite_stat3 ANALYZE data, each range inequality 2681 ** reduces the search space by a factor of 4. Hence a single constraint (x>?) 2682 ** results in a return of 4 and a range constraint (x>? AND x<?) results 2683 ** in a return of 16. 2684 */ 2685 static int whereRangeScanEst( 2686 Parse *pParse, /* Parsing & code generating context */ 2687 Index *p, /* The index containing the range-compared column; "x" */ 2688 int nEq, /* index into p->aCol[] of the range-compared column */ 2689 WhereTerm *pLower, /* Lower bound on the range. ex: "x>123" Might be NULL */ 2690 WhereTerm *pUpper, /* Upper bound on the range. ex: "x<455" Might be NULL */ 2691 double *pRangeDiv /* OUT: Reduce search space by this divisor */ 2692 ){ 2693 int rc = SQLITE_OK; 2694 2695 #ifdef SQLITE_ENABLE_STAT3 2696 2697 if( nEq==0 && p->nSample ){ 2698 sqlite3_value *pRangeVal; 2699 tRowcnt iLower = 0; 2700 tRowcnt iUpper = p->aiRowEst[0]; 2701 tRowcnt a[2]; 2702 u8 aff = p->pTable->aCol[p->aiColumn[0]].affinity; 2703 2704 if( pLower ){ 2705 Expr *pExpr = pLower->pExpr->pRight; 2706 rc = valueFromExpr(pParse, pExpr, aff, &pRangeVal); 2707 assert( pLower->eOperator==WO_GT || pLower->eOperator==WO_GE ); 2708 if( rc==SQLITE_OK 2709 && whereKeyStats(pParse, p, pRangeVal, 0, a)==SQLITE_OK 2710 ){ 2711 iLower = a[0]; 2712 if( pLower->eOperator==WO_GT ) iLower += a[1]; 2713 } 2714 sqlite3ValueFree(pRangeVal); 2715 } 2716 if( rc==SQLITE_OK && pUpper ){ 2717 Expr *pExpr = pUpper->pExpr->pRight; 2718 rc = valueFromExpr(pParse, pExpr, aff, &pRangeVal); 2719 assert( pUpper->eOperator==WO_LT || pUpper->eOperator==WO_LE ); 2720 if( rc==SQLITE_OK 2721 && whereKeyStats(pParse, p, pRangeVal, 1, a)==SQLITE_OK 2722 ){ 2723 iUpper = a[0]; 2724 if( pUpper->eOperator==WO_LE ) iUpper += a[1]; 2725 } 2726 sqlite3ValueFree(pRangeVal); 2727 } 2728 if( rc==SQLITE_OK ){ 2729 if( iUpper<=iLower ){ 2730 *pRangeDiv = (double)p->aiRowEst[0]; 2731 }else{ 2732 *pRangeDiv = (double)p->aiRowEst[0]/(double)(iUpper - iLower); 2733 } 2734 WHERETRACE(("range scan regions: %u..%u div=%g\n", 2735 (u32)iLower, (u32)iUpper, *pRangeDiv)); 2736 return SQLITE_OK; 2737 } 2738 } 2739 #else 2740 UNUSED_PARAMETER(pParse); 2741 UNUSED_PARAMETER(p); 2742 UNUSED_PARAMETER(nEq); 2743 #endif 2744 assert( pLower || pUpper ); 2745 *pRangeDiv = (double)1; 2746 if( pLower && (pLower->wtFlags & TERM_VNULL)==0 ) *pRangeDiv *= (double)4; 2747 if( pUpper ) *pRangeDiv *= (double)4; 2748 return rc; 2749 } 2750 2751 #ifdef SQLITE_ENABLE_STAT3 2752 /* 2753 ** Estimate the number of rows that will be returned based on 2754 ** an equality constraint x=VALUE and where that VALUE occurs in 2755 ** the histogram data. This only works when x is the left-most 2756 ** column of an index and sqlite_stat3 histogram data is available 2757 ** for that index. When pExpr==NULL that means the constraint is 2758 ** "x IS NULL" instead of "x=VALUE". 2759 ** 2760 ** Write the estimated row count into *pnRow and return SQLITE_OK. 2761 ** If unable to make an estimate, leave *pnRow unchanged and return 2762 ** non-zero. 2763 ** 2764 ** This routine can fail if it is unable to load a collating sequence 2765 ** required for string comparison, or if unable to allocate memory 2766 ** for a UTF conversion required for comparison. The error is stored 2767 ** in the pParse structure. 2768 */ 2769 static int whereEqualScanEst( 2770 Parse *pParse, /* Parsing & code generating context */ 2771 Index *p, /* The index whose left-most column is pTerm */ 2772 Expr *pExpr, /* Expression for VALUE in the x=VALUE constraint */ 2773 double *pnRow /* Write the revised row estimate here */ 2774 ){ 2775 sqlite3_value *pRhs = 0; /* VALUE on right-hand side of pTerm */ 2776 u8 aff; /* Column affinity */ 2777 int rc; /* Subfunction return code */ 2778 tRowcnt a[2]; /* Statistics */ 2779 2780 assert( p->aSample!=0 ); 2781 assert( p->nSample>0 ); 2782 aff = p->pTable->aCol[p->aiColumn[0]].affinity; 2783 if( pExpr ){ 2784 rc = valueFromExpr(pParse, pExpr, aff, &pRhs); 2785 if( rc ) goto whereEqualScanEst_cancel; 2786 }else{ 2787 pRhs = sqlite3ValueNew(pParse->db); 2788 } 2789 if( pRhs==0 ) return SQLITE_NOTFOUND; 2790 rc = whereKeyStats(pParse, p, pRhs, 0, a); 2791 if( rc==SQLITE_OK ){ 2792 WHERETRACE(("equality scan regions: %d\n", (int)a[1])); 2793 *pnRow = a[1]; 2794 } 2795 whereEqualScanEst_cancel: 2796 sqlite3ValueFree(pRhs); 2797 return rc; 2798 } 2799 #endif /* defined(SQLITE_ENABLE_STAT3) */ 2800 2801 #ifdef SQLITE_ENABLE_STAT3 2802 /* 2803 ** Estimate the number of rows that will be returned based on 2804 ** an IN constraint where the right-hand side of the IN operator 2805 ** is a list of values. Example: 2806 ** 2807 ** WHERE x IN (1,2,3,4) 2808 ** 2809 ** Write the estimated row count into *pnRow and return SQLITE_OK. 2810 ** If unable to make an estimate, leave *pnRow unchanged and return 2811 ** non-zero. 2812 ** 2813 ** This routine can fail if it is unable to load a collating sequence 2814 ** required for string comparison, or if unable to allocate memory 2815 ** for a UTF conversion required for comparison. The error is stored 2816 ** in the pParse structure. 2817 */ 2818 static int whereInScanEst( 2819 Parse *pParse, /* Parsing & code generating context */ 2820 Index *p, /* The index whose left-most column is pTerm */ 2821 ExprList *pList, /* The value list on the RHS of "x IN (v1,v2,v3,...)" */ 2822 double *pnRow /* Write the revised row estimate here */ 2823 ){ 2824 int rc = SQLITE_OK; /* Subfunction return code */ 2825 double nEst; /* Number of rows for a single term */ 2826 double nRowEst = (double)0; /* New estimate of the number of rows */ 2827 int i; /* Loop counter */ 2828 2829 assert( p->aSample!=0 ); 2830 for(i=0; rc==SQLITE_OK && i<pList->nExpr; i++){ 2831 nEst = p->aiRowEst[0]; 2832 rc = whereEqualScanEst(pParse, p, pList->a[i].pExpr, &nEst); 2833 nRowEst += nEst; 2834 } 2835 if( rc==SQLITE_OK ){ 2836 if( nRowEst > p->aiRowEst[0] ) nRowEst = p->aiRowEst[0]; 2837 *pnRow = nRowEst; 2838 WHERETRACE(("IN row estimate: est=%g\n", nRowEst)); 2839 } 2840 return rc; 2841 } 2842 #endif /* defined(SQLITE_ENABLE_STAT3) */ 2843 2844 2845 /* 2846 ** Find the best query plan for accessing a particular table. Write the 2847 ** best query plan and its cost into the WhereCost object supplied as the 2848 ** last parameter. 2849 ** 2850 ** The lowest cost plan wins. The cost is an estimate of the amount of 2851 ** CPU and disk I/O needed to process the requested result. 2852 ** Factors that influence cost include: 2853 ** 2854 ** * The estimated number of rows that will be retrieved. (The 2855 ** fewer the better.) 2856 ** 2857 ** * Whether or not sorting must occur. 2858 ** 2859 ** * Whether or not there must be separate lookups in the 2860 ** index and in the main table. 2861 ** 2862 ** If there was an INDEXED BY clause (pSrc->pIndex) attached to the table in 2863 ** the SQL statement, then this function only considers plans using the 2864 ** named index. If no such plan is found, then the returned cost is 2865 ** SQLITE_BIG_DBL. If a plan is found that uses the named index, 2866 ** then the cost is calculated in the usual way. 2867 ** 2868 ** If a NOT INDEXED clause (pSrc->notIndexed!=0) was attached to the table 2869 ** in the SELECT statement, then no indexes are considered. However, the 2870 ** selected plan may still take advantage of the built-in rowid primary key 2871 ** index. 2872 */ 2873 static void bestBtreeIndex( 2874 Parse *pParse, /* The parsing context */ 2875 WhereClause *pWC, /* The WHERE clause */ 2876 struct SrcList_item *pSrc, /* The FROM clause term to search */ 2877 Bitmask notReady, /* Mask of cursors not available for indexing */ 2878 Bitmask notValid, /* Cursors not available for any purpose */ 2879 ExprList *pOrderBy, /* The ORDER BY clause */ 2880 ExprList *pDistinct, /* The select-list if query is DISTINCT */ 2881 WhereCost *pCost /* Lowest cost query plan */ 2882 ){ 2883 int iCur = pSrc->iCursor; /* The cursor of the table to be accessed */ 2884 Index *pProbe; /* An index we are evaluating */ 2885 Index *pIdx; /* Copy of pProbe, or zero for IPK index */ 2886 int eqTermMask; /* Current mask of valid equality operators */ 2887 int idxEqTermMask; /* Index mask of valid equality operators */ 2888 Index sPk; /* A fake index object for the primary key */ 2889 tRowcnt aiRowEstPk[2]; /* The aiRowEst[] value for the sPk index */ 2890 int aiColumnPk = -1; /* The aColumn[] value for the sPk index */ 2891 int wsFlagMask; /* Allowed flags in pCost->plan.wsFlag */ 2892 2893 /* Initialize the cost to a worst-case value */ 2894 memset(pCost, 0, sizeof(*pCost)); 2895 pCost->rCost = SQLITE_BIG_DBL; 2896 2897 /* If the pSrc table is the right table of a LEFT JOIN then we may not 2898 ** use an index to satisfy IS NULL constraints on that table. This is 2899 ** because columns might end up being NULL if the table does not match - 2900 ** a circumstance which the index cannot help us discover. Ticket #2177. 2901 */ 2902 if( pSrc->jointype & JT_LEFT ){ 2903 idxEqTermMask = WO_EQ|WO_IN; 2904 }else{ 2905 idxEqTermMask = WO_EQ|WO_IN|WO_ISNULL; 2906 } 2907 2908 if( pSrc->pIndex ){ 2909 /* An INDEXED BY clause specifies a particular index to use */ 2910 pIdx = pProbe = pSrc->pIndex; 2911 wsFlagMask = ~(WHERE_ROWID_EQ|WHERE_ROWID_RANGE); 2912 eqTermMask = idxEqTermMask; 2913 }else{ 2914 /* There is no INDEXED BY clause. Create a fake Index object in local 2915 ** variable sPk to represent the rowid primary key index. Make this 2916 ** fake index the first in a chain of Index objects with all of the real 2917 ** indices to follow */ 2918 Index *pFirst; /* First of real indices on the table */ 2919 memset(&sPk, 0, sizeof(Index)); 2920 sPk.nColumn = 1; 2921 sPk.aiColumn = &aiColumnPk; 2922 sPk.aiRowEst = aiRowEstPk; 2923 sPk.onError = OE_Replace; 2924 sPk.pTable = pSrc->pTab; 2925 aiRowEstPk[0] = pSrc->pTab->nRowEst; 2926 aiRowEstPk[1] = 1; 2927 pFirst = pSrc->pTab->pIndex; 2928 if( pSrc->notIndexed==0 ){ 2929 /* The real indices of the table are only considered if the 2930 ** NOT INDEXED qualifier is omitted from the FROM clause */ 2931 sPk.pNext = pFirst; 2932 } 2933 pProbe = &sPk; 2934 wsFlagMask = ~( 2935 WHERE_COLUMN_IN|WHERE_COLUMN_EQ|WHERE_COLUMN_NULL|WHERE_COLUMN_RANGE 2936 ); 2937 eqTermMask = WO_EQ|WO_IN; 2938 pIdx = 0; 2939 } 2940 2941 /* Loop over all indices looking for the best one to use 2942 */ 2943 for(; pProbe; pIdx=pProbe=pProbe->pNext){ 2944 const tRowcnt * const aiRowEst = pProbe->aiRowEst; 2945 double cost; /* Cost of using pProbe */ 2946 double nRow; /* Estimated number of rows in result set */ 2947 double log10N = (double)1; /* base-10 logarithm of nRow (inexact) */ 2948 int rev; /* True to scan in reverse order */ 2949 int wsFlags = 0; 2950 Bitmask used = 0; 2951 2952 /* The following variables are populated based on the properties of 2953 ** index being evaluated. They are then used to determine the expected 2954 ** cost and number of rows returned. 2955 ** 2956 ** nEq: 2957 ** Number of equality terms that can be implemented using the index. 2958 ** In other words, the number of initial fields in the index that 2959 ** are used in == or IN or NOT NULL constraints of the WHERE clause. 2960 ** 2961 ** nInMul: 2962 ** The "in-multiplier". This is an estimate of how many seek operations 2963 ** SQLite must perform on the index in question. For example, if the 2964 ** WHERE clause is: 2965 ** 2966 ** WHERE a IN (1, 2, 3) AND b IN (4, 5, 6) 2967 ** 2968 ** SQLite must perform 9 lookups on an index on (a, b), so nInMul is 2969 ** set to 9. Given the same schema and either of the following WHERE 2970 ** clauses: 2971 ** 2972 ** WHERE a = 1 2973 ** WHERE a >= 2 2974 ** 2975 ** nInMul is set to 1. 2976 ** 2977 ** If there exists a WHERE term of the form "x IN (SELECT ...)", then 2978 ** the sub-select is assumed to return 25 rows for the purposes of 2979 ** determining nInMul. 2980 ** 2981 ** bInEst: 2982 ** Set to true if there was at least one "x IN (SELECT ...)" term used 2983 ** in determining the value of nInMul. Note that the RHS of the 2984 ** IN operator must be a SELECT, not a value list, for this variable 2985 ** to be true. 2986 ** 2987 ** rangeDiv: 2988 ** An estimate of a divisor by which to reduce the search space due 2989 ** to inequality constraints. In the absence of sqlite_stat3 ANALYZE 2990 ** data, a single inequality reduces the search space to 1/4rd its 2991 ** original size (rangeDiv==4). Two inequalities reduce the search 2992 ** space to 1/16th of its original size (rangeDiv==16). 2993 ** 2994 ** bSort: 2995 ** Boolean. True if there is an ORDER BY clause that will require an 2996 ** external sort (i.e. scanning the index being evaluated will not 2997 ** correctly order records). 2998 ** 2999 ** bLookup: 3000 ** Boolean. True if a table lookup is required for each index entry 3001 ** visited. In other words, true if this is not a covering index. 3002 ** This is always false for the rowid primary key index of a table. 3003 ** For other indexes, it is true unless all the columns of the table 3004 ** used by the SELECT statement are present in the index (such an 3005 ** index is sometimes described as a covering index). 3006 ** For example, given the index on (a, b), the second of the following 3007 ** two queries requires table b-tree lookups in order to find the value 3008 ** of column c, but the first does not because columns a and b are 3009 ** both available in the index. 3010 ** 3011 ** SELECT a, b FROM tbl WHERE a = 1; 3012 ** SELECT a, b, c FROM tbl WHERE a = 1; 3013 */ 3014 int nEq; /* Number of == or IN terms matching index */ 3015 int bInEst = 0; /* True if "x IN (SELECT...)" seen */ 3016 int nInMul = 1; /* Number of distinct equalities to lookup */ 3017 double rangeDiv = (double)1; /* Estimated reduction in search space */ 3018 int nBound = 0; /* Number of range constraints seen */ 3019 int bSort = !!pOrderBy; /* True if external sort required */ 3020 int bDist = !!pDistinct; /* True if index cannot help with DISTINCT */ 3021 int bLookup = 0; /* True if not a covering index */ 3022 WhereTerm *pTerm; /* A single term of the WHERE clause */ 3023 #ifdef SQLITE_ENABLE_STAT3 3024 WhereTerm *pFirstTerm = 0; /* First term matching the index */ 3025 #endif 3026 3027 /* Determine the values of nEq and nInMul */ 3028 for(nEq=0; nEq<pProbe->nColumn; nEq++){ 3029 int j = pProbe->aiColumn[nEq]; 3030 pTerm = findTerm(pWC, iCur, j, notReady, eqTermMask, pIdx); 3031 if( pTerm==0 ) break; 3032 wsFlags |= (WHERE_COLUMN_EQ|WHERE_ROWID_EQ); 3033 testcase( pTerm->pWC!=pWC ); 3034 if( pTerm->eOperator & WO_IN ){ 3035 Expr *pExpr = pTerm->pExpr; 3036 wsFlags |= WHERE_COLUMN_IN; 3037 if( ExprHasProperty(pExpr, EP_xIsSelect) ){ 3038 /* "x IN (SELECT ...)": Assume the SELECT returns 25 rows */ 3039 nInMul *= 25; 3040 bInEst = 1; 3041 }else if( ALWAYS(pExpr->x.pList && pExpr->x.pList->nExpr) ){ 3042 /* "x IN (value, value, ...)" */ 3043 nInMul *= pExpr->x.pList->nExpr; 3044 } 3045 }else if( pTerm->eOperator & WO_ISNULL ){ 3046 wsFlags |= WHERE_COLUMN_NULL; 3047 } 3048 #ifdef SQLITE_ENABLE_STAT3 3049 if( nEq==0 && pProbe->aSample ) pFirstTerm = pTerm; 3050 #endif 3051 used |= pTerm->prereqRight; 3052 } 3053 3054 /* If the index being considered is UNIQUE, and there is an equality 3055 ** constraint for all columns in the index, then this search will find 3056 ** at most a single row. In this case set the WHERE_UNIQUE flag to 3057 ** indicate this to the caller. 3058 ** 3059 ** Otherwise, if the search may find more than one row, test to see if 3060 ** there is a range constraint on indexed column (nEq+1) that can be 3061 ** optimized using the index. 3062 */ 3063 if( nEq==pProbe->nColumn && pProbe->onError!=OE_None ){ 3064 testcase( wsFlags & WHERE_COLUMN_IN ); 3065 testcase( wsFlags & WHERE_COLUMN_NULL ); 3066 if( (wsFlags & (WHERE_COLUMN_IN|WHERE_COLUMN_NULL))==0 ){ 3067 wsFlags |= WHERE_UNIQUE; 3068 } 3069 }else if( pProbe->bUnordered==0 ){ 3070 int j = (nEq==pProbe->nColumn ? -1 : pProbe->aiColumn[nEq]); 3071 if( findTerm(pWC, iCur, j, notReady, WO_LT|WO_LE|WO_GT|WO_GE, pIdx) ){ 3072 WhereTerm *pTop = findTerm(pWC, iCur, j, notReady, WO_LT|WO_LE, pIdx); 3073 WhereTerm *pBtm = findTerm(pWC, iCur, j, notReady, WO_GT|WO_GE, pIdx); 3074 whereRangeScanEst(pParse, pProbe, nEq, pBtm, pTop, &rangeDiv); 3075 if( pTop ){ 3076 nBound = 1; 3077 wsFlags |= WHERE_TOP_LIMIT; 3078 used |= pTop->prereqRight; 3079 testcase( pTop->pWC!=pWC ); 3080 } 3081 if( pBtm ){ 3082 nBound++; 3083 wsFlags |= WHERE_BTM_LIMIT; 3084 used |= pBtm->prereqRight; 3085 testcase( pBtm->pWC!=pWC ); 3086 } 3087 wsFlags |= (WHERE_COLUMN_RANGE|WHERE_ROWID_RANGE); 3088 } 3089 } 3090 3091 /* If there is an ORDER BY clause and the index being considered will 3092 ** naturally scan rows in the required order, set the appropriate flags 3093 ** in wsFlags. Otherwise, if there is an ORDER BY clause but the index 3094 ** will scan rows in a different order, set the bSort variable. */ 3095 if( isSortingIndex( 3096 pParse, pWC->pMaskSet, pProbe, iCur, pOrderBy, nEq, wsFlags, &rev) 3097 ){ 3098 bSort = 0; 3099 wsFlags |= WHERE_ROWID_RANGE|WHERE_COLUMN_RANGE|WHERE_ORDERBY; 3100 wsFlags |= (rev ? WHERE_REVERSE : 0); 3101 } 3102 3103 /* If there is a DISTINCT qualifier and this index will scan rows in 3104 ** order of the DISTINCT expressions, clear bDist and set the appropriate 3105 ** flags in wsFlags. */ 3106 if( isDistinctIndex(pParse, pWC, pProbe, iCur, pDistinct, nEq) ){ 3107 bDist = 0; 3108 wsFlags |= WHERE_ROWID_RANGE|WHERE_COLUMN_RANGE|WHERE_DISTINCT; 3109 } 3110 3111 /* If currently calculating the cost of using an index (not the IPK 3112 ** index), determine if all required column data may be obtained without 3113 ** using the main table (i.e. if the index is a covering 3114 ** index for this query). If it is, set the WHERE_IDX_ONLY flag in 3115 ** wsFlags. Otherwise, set the bLookup variable to true. */ 3116 if( pIdx && wsFlags ){ 3117 Bitmask m = pSrc->colUsed; 3118 int j; 3119 for(j=0; j<pIdx->nColumn; j++){ 3120 int x = pIdx->aiColumn[j]; 3121 if( x<BMS-1 ){ 3122 m &= ~(((Bitmask)1)<<x); 3123 } 3124 } 3125 if( m==0 ){ 3126 wsFlags |= WHERE_IDX_ONLY; 3127 }else{ 3128 bLookup = 1; 3129 } 3130 } 3131 3132 /* 3133 ** Estimate the number of rows of output. For an "x IN (SELECT...)" 3134 ** constraint, do not let the estimate exceed half the rows in the table. 3135 */ 3136 nRow = (double)(aiRowEst[nEq] * nInMul); 3137 if( bInEst && nRow*2>aiRowEst[0] ){ 3138 nRow = aiRowEst[0]/2; 3139 nInMul = (int)(nRow / aiRowEst[nEq]); 3140 } 3141 3142 #ifdef SQLITE_ENABLE_STAT3 3143 /* If the constraint is of the form x=VALUE or x IN (E1,E2,...) 3144 ** and we do not think that values of x are unique and if histogram 3145 ** data is available for column x, then it might be possible 3146 ** to get a better estimate on the number of rows based on 3147 ** VALUE and how common that value is according to the histogram. 3148 */ 3149 if( nRow>(double)1 && nEq==1 && pFirstTerm!=0 && aiRowEst[1]>1 ){ 3150 assert( (pFirstTerm->eOperator & (WO_EQ|WO_ISNULL|WO_IN))!=0 ); 3151 if( pFirstTerm->eOperator & (WO_EQ|WO_ISNULL) ){ 3152 testcase( pFirstTerm->eOperator==WO_EQ ); 3153 testcase( pFirstTerm->eOperator==WO_ISNULL ); 3154 whereEqualScanEst(pParse, pProbe, pFirstTerm->pExpr->pRight, &nRow); 3155 }else if( bInEst==0 ){ 3156 assert( pFirstTerm->eOperator==WO_IN ); 3157 whereInScanEst(pParse, pProbe, pFirstTerm->pExpr->x.pList, &nRow); 3158 } 3159 } 3160 #endif /* SQLITE_ENABLE_STAT3 */ 3161 3162 /* Adjust the number of output rows and downward to reflect rows 3163 ** that are excluded by range constraints. 3164 */ 3165 nRow = nRow/rangeDiv; 3166 if( nRow<1 ) nRow = 1; 3167 3168 /* Experiments run on real SQLite databases show that the time needed 3169 ** to do a binary search to locate a row in a table or index is roughly 3170 ** log10(N) times the time to move from one row to the next row within 3171 ** a table or index. The actual times can vary, with the size of 3172 ** records being an important factor. Both moves and searches are 3173 ** slower with larger records, presumably because fewer records fit 3174 ** on one page and hence more pages have to be fetched. 3175 ** 3176 ** The ANALYZE command and the sqlite_stat1 and sqlite_stat3 tables do 3177 ** not give us data on the relative sizes of table and index records. 3178 ** So this computation assumes table records are about twice as big 3179 ** as index records 3180 */ 3181 if( (wsFlags & WHERE_NOT_FULLSCAN)==0 ){ 3182 /* The cost of a full table scan is a number of move operations equal 3183 ** to the number of rows in the table. 3184 ** 3185 ** We add an additional 4x penalty to full table scans. This causes 3186 ** the cost function to err on the side of choosing an index over 3187 ** choosing a full scan. This 4x full-scan penalty is an arguable 3188 ** decision and one which we expect to revisit in the future. But 3189 ** it seems to be working well enough at the moment. 3190 */ 3191 cost = aiRowEst[0]*4; 3192 }else{ 3193 log10N = estLog(aiRowEst[0]); 3194 cost = nRow; 3195 if( pIdx ){ 3196 if( bLookup ){ 3197 /* For an index lookup followed by a table lookup: 3198 ** nInMul index searches to find the start of each index range 3199 ** + nRow steps through the index 3200 ** + nRow table searches to lookup the table entry using the rowid 3201 */ 3202 cost += (nInMul + nRow)*log10N; 3203 }else{ 3204 /* For a covering index: 3205 ** nInMul index searches to find the initial entry 3206 ** + nRow steps through the index 3207 */ 3208 cost += nInMul*log10N; 3209 } 3210 }else{ 3211 /* For a rowid primary key lookup: 3212 ** nInMult table searches to find the initial entry for each range 3213 ** + nRow steps through the table 3214 */ 3215 cost += nInMul*log10N; 3216 } 3217 } 3218 3219 /* Add in the estimated cost of sorting the result. Actual experimental 3220 ** measurements of sorting performance in SQLite show that sorting time 3221 ** adds C*N*log10(N) to the cost, where N is the number of rows to be 3222 ** sorted and C is a factor between 1.95 and 4.3. We will split the 3223 ** difference and select C of 3.0. 3224 */ 3225 if( bSort ){ 3226 cost += nRow*estLog(nRow)*3; 3227 } 3228 if( bDist ){ 3229 cost += nRow*estLog(nRow)*3; 3230 } 3231 3232 /**** Cost of using this index has now been computed ****/ 3233 3234 /* If there are additional constraints on this table that cannot 3235 ** be used with the current index, but which might lower the number 3236 ** of output rows, adjust the nRow value accordingly. This only 3237 ** matters if the current index is the least costly, so do not bother 3238 ** with this step if we already know this index will not be chosen. 3239 ** Also, never reduce the output row count below 2 using this step. 3240 ** 3241 ** It is critical that the notValid mask be used here instead of 3242 ** the notReady mask. When computing an "optimal" index, the notReady 3243 ** mask will only have one bit set - the bit for the current table. 3244 ** The notValid mask, on the other hand, always has all bits set for 3245 ** tables that are not in outer loops. If notReady is used here instead 3246 ** of notValid, then a optimal index that depends on inner joins loops 3247 ** might be selected even when there exists an optimal index that has 3248 ** no such dependency. 3249 */ 3250 if( nRow>2 && cost<=pCost->rCost ){ 3251 int k; /* Loop counter */ 3252 int nSkipEq = nEq; /* Number of == constraints to skip */ 3253 int nSkipRange = nBound; /* Number of < constraints to skip */ 3254 Bitmask thisTab; /* Bitmap for pSrc */ 3255 3256 thisTab = getMask(pWC->pMaskSet, iCur); 3257 for(pTerm=pWC->a, k=pWC->nTerm; nRow>2 && k; k--, pTerm++){ 3258 if( pTerm->wtFlags & TERM_VIRTUAL ) continue; 3259 if( (pTerm->prereqAll & notValid)!=thisTab ) continue; 3260 if( pTerm->eOperator & (WO_EQ|WO_IN|WO_ISNULL) ){ 3261 if( nSkipEq ){ 3262 /* Ignore the first nEq equality matches since the index 3263 ** has already accounted for these */ 3264 nSkipEq--; 3265 }else{ 3266 /* Assume each additional equality match reduces the result 3267 ** set size by a factor of 10 */ 3268 nRow /= 10; 3269 } 3270 }else if( pTerm->eOperator & (WO_LT|WO_LE|WO_GT|WO_GE) ){ 3271 if( nSkipRange ){ 3272 /* Ignore the first nSkipRange range constraints since the index 3273 ** has already accounted for these */ 3274 nSkipRange--; 3275 }else{ 3276 /* Assume each additional range constraint reduces the result 3277 ** set size by a factor of 3. Indexed range constraints reduce 3278 ** the search space by a larger factor: 4. We make indexed range 3279 ** more selective intentionally because of the subjective 3280 ** observation that indexed range constraints really are more 3281 ** selective in practice, on average. */ 3282 nRow /= 3; 3283 } 3284 }else if( pTerm->eOperator!=WO_NOOP ){ 3285 /* Any other expression lowers the output row count by half */ 3286 nRow /= 2; 3287 } 3288 } 3289 if( nRow<2 ) nRow = 2; 3290 } 3291 3292 3293 WHERETRACE(( 3294 "%s(%s): nEq=%d nInMul=%d rangeDiv=%d bSort=%d bLookup=%d wsFlags=0x%x\n" 3295 " notReady=0x%llx log10N=%.1f nRow=%.1f cost=%.1f used=0x%llx\n", 3296 pSrc->pTab->zName, (pIdx ? pIdx->zName : "ipk"), 3297 nEq, nInMul, (int)rangeDiv, bSort, bLookup, wsFlags, 3298 notReady, log10N, nRow, cost, used 3299 )); 3300 3301 /* If this index is the best we have seen so far, then record this 3302 ** index and its cost in the pCost structure. 3303 */ 3304 if( (!pIdx || wsFlags) 3305 && (cost<pCost->rCost || (cost<=pCost->rCost && nRow<pCost->plan.nRow)) 3306 ){ 3307 pCost->rCost = cost; 3308 pCost->used = used; 3309 pCost->plan.nRow = nRow; 3310 pCost->plan.wsFlags = (wsFlags&wsFlagMask); 3311 pCost->plan.nEq = nEq; 3312 pCost->plan.u.pIdx = pIdx; 3313 } 3314 3315 /* If there was an INDEXED BY clause, then only that one index is 3316 ** considered. */ 3317 if( pSrc->pIndex ) break; 3318 3319 /* Reset masks for the next index in the loop */ 3320 wsFlagMask = ~(WHERE_ROWID_EQ|WHERE_ROWID_RANGE); 3321 eqTermMask = idxEqTermMask; 3322 } 3323 3324 /* If there is no ORDER BY clause and the SQLITE_ReverseOrder flag 3325 ** is set, then reverse the order that the index will be scanned 3326 ** in. This is used for application testing, to help find cases 3327 ** where application behaviour depends on the (undefined) order that 3328 ** SQLite outputs rows in in the absence of an ORDER BY clause. */ 3329 if( !pOrderBy && pParse->db->flags & SQLITE_ReverseOrder ){ 3330 pCost->plan.wsFlags |= WHERE_REVERSE; 3331 } 3332 3333 assert( pOrderBy || (pCost->plan.wsFlags&WHERE_ORDERBY)==0 ); 3334 assert( pCost->plan.u.pIdx==0 || (pCost->plan.wsFlags&WHERE_ROWID_EQ)==0 ); 3335 assert( pSrc->pIndex==0 3336 || pCost->plan.u.pIdx==0 3337 || pCost->plan.u.pIdx==pSrc->pIndex 3338 ); 3339 3340 WHERETRACE(("best index is: %s\n", 3341 ((pCost->plan.wsFlags & WHERE_NOT_FULLSCAN)==0 ? "none" : 3342 pCost->plan.u.pIdx ? pCost->plan.u.pIdx->zName : "ipk") 3343 )); 3344 3345 bestOrClauseIndex(pParse, pWC, pSrc, notReady, notValid, pOrderBy, pCost); 3346 bestAutomaticIndex(pParse, pWC, pSrc, notReady, pCost); 3347 pCost->plan.wsFlags |= eqTermMask; 3348 } 3349 3350 /* 3351 ** Find the query plan for accessing table pSrc->pTab. Write the 3352 ** best query plan and its cost into the WhereCost object supplied 3353 ** as the last parameter. This function may calculate the cost of 3354 ** both real and virtual table scans. 3355 */ 3356 static void bestIndex( 3357 Parse *pParse, /* The parsing context */ 3358 WhereClause *pWC, /* The WHERE clause */ 3359 struct SrcList_item *pSrc, /* The FROM clause term to search */ 3360 Bitmask notReady, /* Mask of cursors not available for indexing */ 3361 Bitmask notValid, /* Cursors not available for any purpose */ 3362 ExprList *pOrderBy, /* The ORDER BY clause */ 3363 WhereCost *pCost /* Lowest cost query plan */ 3364 ){ 3365 #ifndef SQLITE_OMIT_VIRTUALTABLE 3366 if( IsVirtual(pSrc->pTab) ){ 3367 sqlite3_index_info *p = 0; 3368 bestVirtualIndex(pParse, pWC, pSrc, notReady, notValid, pOrderBy, pCost,&p); 3369 if( p->needToFreeIdxStr ){ 3370 sqlite3_free(p->idxStr); 3371 } 3372 sqlite3DbFree(pParse->db, p); 3373 }else 3374 #endif 3375 { 3376 bestBtreeIndex(pParse, pWC, pSrc, notReady, notValid, pOrderBy, 0, pCost); 3377 } 3378 } 3379 3380 /* 3381 ** Disable a term in the WHERE clause. Except, do not disable the term 3382 ** if it controls a LEFT OUTER JOIN and it did not originate in the ON 3383 ** or USING clause of that join. 3384 ** 3385 ** Consider the term t2.z='ok' in the following queries: 3386 ** 3387 ** (1) SELECT * FROM t1 LEFT JOIN t2 ON t1.a=t2.x WHERE t2.z='ok' 3388 ** (2) SELECT * FROM t1 LEFT JOIN t2 ON t1.a=t2.x AND t2.z='ok' 3389 ** (3) SELECT * FROM t1, t2 WHERE t1.a=t2.x AND t2.z='ok' 3390 ** 3391 ** The t2.z='ok' is disabled in the in (2) because it originates 3392 ** in the ON clause. The term is disabled in (3) because it is not part 3393 ** of a LEFT OUTER JOIN. In (1), the term is not disabled. 3394 ** 3395 ** IMPLEMENTATION-OF: R-24597-58655 No tests are done for terms that are 3396 ** completely satisfied by indices. 3397 ** 3398 ** Disabling a term causes that term to not be tested in the inner loop 3399 ** of the join. Disabling is an optimization. When terms are satisfied 3400 ** by indices, we disable them to prevent redundant tests in the inner 3401 ** loop. We would get the correct results if nothing were ever disabled, 3402 ** but joins might run a little slower. The trick is to disable as much 3403 ** as we can without disabling too much. If we disabled in (1), we'd get 3404 ** the wrong answer. See ticket #813. 3405 */ 3406 static void disableTerm(WhereLevel *pLevel, WhereTerm *pTerm){ 3407 if( pTerm 3408 && (pTerm->wtFlags & TERM_CODED)==0 3409 && (pLevel->iLeftJoin==0 || ExprHasProperty(pTerm->pExpr, EP_FromJoin)) 3410 ){ 3411 pTerm->wtFlags |= TERM_CODED; 3412 if( pTerm->iParent>=0 ){ 3413 WhereTerm *pOther = &pTerm->pWC->a[pTerm->iParent]; 3414 if( (--pOther->nChild)==0 ){ 3415 disableTerm(pLevel, pOther); 3416 } 3417 } 3418 } 3419 } 3420 3421 /* 3422 ** Code an OP_Affinity opcode to apply the column affinity string zAff 3423 ** to the n registers starting at base. 3424 ** 3425 ** As an optimization, SQLITE_AFF_NONE entries (which are no-ops) at the 3426 ** beginning and end of zAff are ignored. If all entries in zAff are 3427 ** SQLITE_AFF_NONE, then no code gets generated. 3428 ** 3429 ** This routine makes its own copy of zAff so that the caller is free 3430 ** to modify zAff after this routine returns. 3431 */ 3432 static void codeApplyAffinity(Parse *pParse, int base, int n, char *zAff){ 3433 Vdbe *v = pParse->pVdbe; 3434 if( zAff==0 ){ 3435 assert( pParse->db->mallocFailed ); 3436 return; 3437 } 3438 assert( v!=0 ); 3439 3440 /* Adjust base and n to skip over SQLITE_AFF_NONE entries at the beginning 3441 ** and end of the affinity string. 3442 */ 3443 while( n>0 && zAff[0]==SQLITE_AFF_NONE ){ 3444 n--; 3445 base++; 3446 zAff++; 3447 } 3448 while( n>1 && zAff[n-1]==SQLITE_AFF_NONE ){ 3449 n--; 3450 } 3451 3452 /* Code the OP_Affinity opcode if there is anything left to do. */ 3453 if( n>0 ){ 3454 sqlite3VdbeAddOp2(v, OP_Affinity, base, n); 3455 sqlite3VdbeChangeP4(v, -1, zAff, n); 3456 sqlite3ExprCacheAffinityChange(pParse, base, n); 3457 } 3458 } 3459 3460 3461 /* 3462 ** Generate code for a single equality term of the WHERE clause. An equality 3463 ** term can be either X=expr or X IN (...). pTerm is the term to be 3464 ** coded. 3465 ** 3466 ** The current value for the constraint is left in register iReg. 3467 ** 3468 ** For a constraint of the form X=expr, the expression is evaluated and its 3469 ** result is left on the stack. For constraints of the form X IN (...) 3470 ** this routine sets up a loop that will iterate over all values of X. 3471 */ 3472 static int codeEqualityTerm( 3473 Parse *pParse, /* The parsing context */ 3474 WhereTerm *pTerm, /* The term of the WHERE clause to be coded */ 3475 WhereLevel *pLevel, /* When level of the FROM clause we are working on */ 3476 int iTarget /* Attempt to leave results in this register */ 3477 ){ 3478 Expr *pX = pTerm->pExpr; 3479 Vdbe *v = pParse->pVdbe; 3480 int iReg; /* Register holding results */ 3481 3482 assert( iTarget>0 ); 3483 if( pX->op==TK_EQ ){ 3484 iReg = sqlite3ExprCodeTarget(pParse, pX->pRight, iTarget); 3485 }else if( pX->op==TK_ISNULL ){ 3486 iReg = iTarget; 3487 sqlite3VdbeAddOp2(v, OP_Null, 0, iReg); 3488 #ifndef SQLITE_OMIT_SUBQUERY 3489 }else{ 3490 int eType; 3491 int iTab; 3492 struct InLoop *pIn; 3493 3494 assert( pX->op==TK_IN ); 3495 iReg = iTarget; 3496 eType = sqlite3FindInIndex(pParse, pX, 0); 3497 iTab = pX->iTable; 3498 sqlite3VdbeAddOp2(v, OP_Rewind, iTab, 0); 3499 assert( pLevel->plan.wsFlags & WHERE_IN_ABLE ); 3500 if( pLevel->u.in.nIn==0 ){ 3501 pLevel->addrNxt = sqlite3VdbeMakeLabel(v); 3502 } 3503 pLevel->u.in.nIn++; 3504 pLevel->u.in.aInLoop = 3505 sqlite3DbReallocOrFree(pParse->db, pLevel->u.in.aInLoop, 3506 sizeof(pLevel->u.in.aInLoop[0])*pLevel->u.in.nIn); 3507 pIn = pLevel->u.in.aInLoop; 3508 if( pIn ){ 3509 pIn += pLevel->u.in.nIn - 1; 3510 pIn->iCur = iTab; 3511 if( eType==IN_INDEX_ROWID ){ 3512 pIn->addrInTop = sqlite3VdbeAddOp2(v, OP_Rowid, iTab, iReg); 3513 }else{ 3514 pIn->addrInTop = sqlite3VdbeAddOp3(v, OP_Column, iTab, 0, iReg); 3515 } 3516 sqlite3VdbeAddOp1(v, OP_IsNull, iReg); 3517 }else{ 3518 pLevel->u.in.nIn = 0; 3519 } 3520 #endif 3521 } 3522 disableTerm(pLevel, pTerm); 3523 return iReg; 3524 } 3525 3526 /* 3527 ** Generate code that will evaluate all == and IN constraints for an 3528 ** index. 3529 ** 3530 ** For example, consider table t1(a,b,c,d,e,f) with index i1(a,b,c). 3531 ** Suppose the WHERE clause is this: a==5 AND b IN (1,2,3) AND c>5 AND c<10 3532 ** The index has as many as three equality constraints, but in this 3533 ** example, the third "c" value is an inequality. So only two 3534 ** constraints are coded. This routine will generate code to evaluate 3535 ** a==5 and b IN (1,2,3). The current values for a and b will be stored 3536 ** in consecutive registers and the index of the first register is returned. 3537 ** 3538 ** In the example above nEq==2. But this subroutine works for any value 3539 ** of nEq including 0. If nEq==0, this routine is nearly a no-op. 3540 ** The only thing it does is allocate the pLevel->iMem memory cell and 3541 ** compute the affinity string. 3542 ** 3543 ** This routine always allocates at least one memory cell and returns 3544 ** the index of that memory cell. The code that 3545 ** calls this routine will use that memory cell to store the termination 3546 ** key value of the loop. If one or more IN operators appear, then 3547 ** this routine allocates an additional nEq memory cells for internal 3548 ** use. 3549 ** 3550 ** Before returning, *pzAff is set to point to a buffer containing a 3551 ** copy of the column affinity string of the index allocated using 3552 ** sqlite3DbMalloc(). Except, entries in the copy of the string associated 3553 ** with equality constraints that use NONE affinity are set to 3554 ** SQLITE_AFF_NONE. This is to deal with SQL such as the following: 3555 ** 3556 ** CREATE TABLE t1(a TEXT PRIMARY KEY, b); 3557 ** SELECT ... FROM t1 AS t2, t1 WHERE t1.a = t2.b; 3558 ** 3559 ** In the example above, the index on t1(a) has TEXT affinity. But since 3560 ** the right hand side of the equality constraint (t2.b) has NONE affinity, 3561 ** no conversion should be attempted before using a t2.b value as part of 3562 ** a key to search the index. Hence the first byte in the returned affinity 3563 ** string in this example would be set to SQLITE_AFF_NONE. 3564 */ 3565 static int codeAllEqualityTerms( 3566 Parse *pParse, /* Parsing context */ 3567 WhereLevel *pLevel, /* Which nested loop of the FROM we are coding */ 3568 WhereClause *pWC, /* The WHERE clause */ 3569 Bitmask notReady, /* Which parts of FROM have not yet been coded */ 3570 int nExtraReg, /* Number of extra registers to allocate */ 3571 char **pzAff /* OUT: Set to point to affinity string */ 3572 ){ 3573 int nEq = pLevel->plan.nEq; /* The number of == or IN constraints to code */ 3574 Vdbe *v = pParse->pVdbe; /* The vm under construction */ 3575 Index *pIdx; /* The index being used for this loop */ 3576 int iCur = pLevel->iTabCur; /* The cursor of the table */ 3577 WhereTerm *pTerm; /* A single constraint term */ 3578 int j; /* Loop counter */ 3579 int regBase; /* Base register */ 3580 int nReg; /* Number of registers to allocate */ 3581 char *zAff; /* Affinity string to return */ 3582 3583 /* This module is only called on query plans that use an index. */ 3584 assert( pLevel->plan.wsFlags & WHERE_INDEXED ); 3585 pIdx = pLevel->plan.u.pIdx; 3586 3587 /* Figure out how many memory cells we will need then allocate them. 3588 */ 3589 regBase = pParse->nMem + 1; 3590 nReg = pLevel->plan.nEq + nExtraReg; 3591 pParse->nMem += nReg; 3592 3593 zAff = sqlite3DbStrDup(pParse->db, sqlite3IndexAffinityStr(v, pIdx)); 3594 if( !zAff ){ 3595 pParse->db->mallocFailed = 1; 3596 } 3597 3598 /* Evaluate the equality constraints 3599 */ 3600 assert( pIdx->nColumn>=nEq ); 3601 for(j=0; j<nEq; j++){ 3602 int r1; 3603 int k = pIdx->aiColumn[j]; 3604 pTerm = findTerm(pWC, iCur, k, notReady, pLevel->plan.wsFlags, pIdx); 3605 if( NEVER(pTerm==0) ) break; 3606 /* The following true for indices with redundant columns. 3607 ** Ex: CREATE INDEX i1 ON t1(a,b,a); SELECT * FROM t1 WHERE a=0 AND b=0; */ 3608 testcase( (pTerm->wtFlags & TERM_CODED)!=0 ); 3609 testcase( pTerm->wtFlags & TERM_VIRTUAL ); /* EV: R-30575-11662 */ 3610 r1 = codeEqualityTerm(pParse, pTerm, pLevel, regBase+j); 3611 if( r1!=regBase+j ){ 3612 if( nReg==1 ){ 3613 sqlite3ReleaseTempReg(pParse, regBase); 3614 regBase = r1; 3615 }else{ 3616 sqlite3VdbeAddOp2(v, OP_SCopy, r1, regBase+j); 3617 } 3618 } 3619 testcase( pTerm->eOperator & WO_ISNULL ); 3620 testcase( pTerm->eOperator & WO_IN ); 3621 if( (pTerm->eOperator & (WO_ISNULL|WO_IN))==0 ){ 3622 Expr *pRight = pTerm->pExpr->pRight; 3623 sqlite3ExprCodeIsNullJump(v, pRight, regBase+j, pLevel->addrBrk); 3624 if( zAff ){ 3625 if( sqlite3CompareAffinity(pRight, zAff[j])==SQLITE_AFF_NONE ){ 3626 zAff[j] = SQLITE_AFF_NONE; 3627 } 3628 if( sqlite3ExprNeedsNoAffinityChange(pRight, zAff[j]) ){ 3629 zAff[j] = SQLITE_AFF_NONE; 3630 } 3631 } 3632 } 3633 } 3634 *pzAff = zAff; 3635 return regBase; 3636 } 3637 3638 #ifndef SQLITE_OMIT_EXPLAIN 3639 /* 3640 ** This routine is a helper for explainIndexRange() below 3641 ** 3642 ** pStr holds the text of an expression that we are building up one term 3643 ** at a time. This routine adds a new term to the end of the expression. 3644 ** Terms are separated by AND so add the "AND" text for second and subsequent 3645 ** terms only. 3646 */ 3647 static void explainAppendTerm( 3648 StrAccum *pStr, /* The text expression being built */ 3649 int iTerm, /* Index of this term. First is zero */ 3650 const char *zColumn, /* Name of the column */ 3651 const char *zOp /* Name of the operator */ 3652 ){ 3653 if( iTerm ) sqlite3StrAccumAppend(pStr, " AND ", 5); 3654 sqlite3StrAccumAppend(pStr, zColumn, -1); 3655 sqlite3StrAccumAppend(pStr, zOp, 1); 3656 sqlite3StrAccumAppend(pStr, "?", 1); 3657 } 3658 3659 /* 3660 ** Argument pLevel describes a strategy for scanning table pTab. This 3661 ** function returns a pointer to a string buffer containing a description 3662 ** of the subset of table rows scanned by the strategy in the form of an 3663 ** SQL expression. Or, if all rows are scanned, NULL is returned. 3664 ** 3665 ** For example, if the query: 3666 ** 3667 ** SELECT * FROM t1 WHERE a=1 AND b>2; 3668 ** 3669 ** is run and there is an index on (a, b), then this function returns a 3670 ** string similar to: 3671 ** 3672 ** "a=? AND b>?" 3673 ** 3674 ** The returned pointer points to memory obtained from sqlite3DbMalloc(). 3675 ** It is the responsibility of the caller to free the buffer when it is 3676 ** no longer required. 3677 */ 3678 static char *explainIndexRange(sqlite3 *db, WhereLevel *pLevel, Table *pTab){ 3679 WherePlan *pPlan = &pLevel->plan; 3680 Index *pIndex = pPlan->u.pIdx; 3681 int nEq = pPlan->nEq; 3682 int i, j; 3683 Column *aCol = pTab->aCol; 3684 int *aiColumn = pIndex->aiColumn; 3685 StrAccum txt; 3686 3687 if( nEq==0 && (pPlan->wsFlags & (WHERE_BTM_LIMIT|WHERE_TOP_LIMIT))==0 ){ 3688 return 0; 3689 } 3690 sqlite3StrAccumInit(&txt, 0, 0, SQLITE_MAX_LENGTH); 3691 txt.db = db; 3692 sqlite3StrAccumAppend(&txt, " (", 2); 3693 for(i=0; i<nEq; i++){ 3694 explainAppendTerm(&txt, i, aCol[aiColumn[i]].zName, "="); 3695 } 3696 3697 j = i; 3698 if( pPlan->wsFlags&WHERE_BTM_LIMIT ){ 3699 char *z = (j==pIndex->nColumn ) ? "rowid" : aCol[aiColumn[j]].zName; 3700 explainAppendTerm(&txt, i++, z, ">"); 3701 } 3702 if( pPlan->wsFlags&WHERE_TOP_LIMIT ){ 3703 char *z = (j==pIndex->nColumn ) ? "rowid" : aCol[aiColumn[j]].zName; 3704 explainAppendTerm(&txt, i, z, "<"); 3705 } 3706 sqlite3StrAccumAppend(&txt, ")", 1); 3707 return sqlite3StrAccumFinish(&txt); 3708 } 3709 3710 /* 3711 ** This function is a no-op unless currently processing an EXPLAIN QUERY PLAN 3712 ** command. If the query being compiled is an EXPLAIN QUERY PLAN, a single 3713 ** record is added to the output to describe the table scan strategy in 3714 ** pLevel. 3715 */ 3716 static void explainOneScan( 3717 Parse *pParse, /* Parse context */ 3718 SrcList *pTabList, /* Table list this loop refers to */ 3719 WhereLevel *pLevel, /* Scan to write OP_Explain opcode for */ 3720 int iLevel, /* Value for "level" column of output */ 3721 int iFrom, /* Value for "from" column of output */ 3722 u16 wctrlFlags /* Flags passed to sqlite3WhereBegin() */ 3723 ){ 3724 if( pParse->explain==2 ){ 3725 u32 flags = pLevel->plan.wsFlags; 3726 struct SrcList_item *pItem = &pTabList->a[pLevel->iFrom]; 3727 Vdbe *v = pParse->pVdbe; /* VM being constructed */ 3728 sqlite3 *db = pParse->db; /* Database handle */ 3729 char *zMsg; /* Text to add to EQP output */ 3730 sqlite3_int64 nRow; /* Expected number of rows visited by scan */ 3731 int iId = pParse->iSelectId; /* Select id (left-most output column) */ 3732 int isSearch; /* True for a SEARCH. False for SCAN. */ 3733 3734 if( (flags&WHERE_MULTI_OR) || (wctrlFlags&WHERE_ONETABLE_ONLY) ) return; 3735 3736 isSearch = (pLevel->plan.nEq>0) 3737 || (flags&(WHERE_BTM_LIMIT|WHERE_TOP_LIMIT))!=0 3738 || (wctrlFlags&(WHERE_ORDERBY_MIN|WHERE_ORDERBY_MAX)); 3739 3740 zMsg = sqlite3MPrintf(db, "%s", isSearch?"SEARCH":"SCAN"); 3741 if( pItem->pSelect ){ 3742 zMsg = sqlite3MAppendf(db, zMsg, "%s SUBQUERY %d", zMsg,pItem->iSelectId); 3743 }else{ 3744 zMsg = sqlite3MAppendf(db, zMsg, "%s TABLE %s", zMsg, pItem->zName); 3745 } 3746 3747 if( pItem->zAlias ){ 3748 zMsg = sqlite3MAppendf(db, zMsg, "%s AS %s", zMsg, pItem->zAlias); 3749 } 3750 if( (flags & WHERE_INDEXED)!=0 ){ 3751 char *zWhere = explainIndexRange(db, pLevel, pItem->pTab); 3752 zMsg = sqlite3MAppendf(db, zMsg, "%s USING %s%sINDEX%s%s%s", zMsg, 3753 ((flags & WHERE_TEMP_INDEX)?"AUTOMATIC ":""), 3754 ((flags & WHERE_IDX_ONLY)?"COVERING ":""), 3755 ((flags & WHERE_TEMP_INDEX)?"":" "), 3756 ((flags & WHERE_TEMP_INDEX)?"": pLevel->plan.u.pIdx->zName), 3757 zWhere 3758 ); 3759 sqlite3DbFree(db, zWhere); 3760 }else if( flags & (WHERE_ROWID_EQ|WHERE_ROWID_RANGE) ){ 3761 zMsg = sqlite3MAppendf(db, zMsg, "%s USING INTEGER PRIMARY KEY", zMsg); 3762 3763 if( flags&WHERE_ROWID_EQ ){ 3764 zMsg = sqlite3MAppendf(db, zMsg, "%s (rowid=?)", zMsg); 3765 }else if( (flags&WHERE_BOTH_LIMIT)==WHERE_BOTH_LIMIT ){ 3766 zMsg = sqlite3MAppendf(db, zMsg, "%s (rowid>? AND rowid<?)", zMsg); 3767 }else if( flags&WHERE_BTM_LIMIT ){ 3768 zMsg = sqlite3MAppendf(db, zMsg, "%s (rowid>?)", zMsg); 3769 }else if( flags&WHERE_TOP_LIMIT ){ 3770 zMsg = sqlite3MAppendf(db, zMsg, "%s (rowid<?)", zMsg); 3771 } 3772 } 3773 #ifndef SQLITE_OMIT_VIRTUALTABLE 3774 else if( (flags & WHERE_VIRTUALTABLE)!=0 ){ 3775 sqlite3_index_info *pVtabIdx = pLevel->plan.u.pVtabIdx; 3776 zMsg = sqlite3MAppendf(db, zMsg, "%s VIRTUAL TABLE INDEX %d:%s", zMsg, 3777 pVtabIdx->idxNum, pVtabIdx->idxStr); 3778 } 3779 #endif 3780 if( wctrlFlags&(WHERE_ORDERBY_MIN|WHERE_ORDERBY_MAX) ){ 3781 testcase( wctrlFlags & WHERE_ORDERBY_MIN ); 3782 nRow = 1; 3783 }else{ 3784 nRow = (sqlite3_int64)pLevel->plan.nRow; 3785 } 3786 zMsg = sqlite3MAppendf(db, zMsg, "%s (~%lld rows)", zMsg, nRow); 3787 sqlite3VdbeAddOp4(v, OP_Explain, iId, iLevel, iFrom, zMsg, P4_DYNAMIC); 3788 } 3789 } 3790 #else 3791 # define explainOneScan(u,v,w,x,y,z) 3792 #endif /* SQLITE_OMIT_EXPLAIN */ 3793 3794 3795 /* 3796 ** Generate code for the start of the iLevel-th loop in the WHERE clause 3797 ** implementation described by pWInfo. 3798 */ 3799 static Bitmask codeOneLoopStart( 3800 WhereInfo *pWInfo, /* Complete information about the WHERE clause */ 3801 int iLevel, /* Which level of pWInfo->a[] should be coded */ 3802 u16 wctrlFlags, /* One of the WHERE_* flags defined in sqliteInt.h */ 3803 Bitmask notReady, /* Which tables are currently available */ 3804 Expr *pWhere /* Complete WHERE clause */ 3805 ){ 3806 int j, k; /* Loop counters */ 3807 int iCur; /* The VDBE cursor for the table */ 3808 int addrNxt; /* Where to jump to continue with the next IN case */ 3809 int omitTable; /* True if we use the index only */ 3810 int bRev; /* True if we need to scan in reverse order */ 3811 WhereLevel *pLevel; /* The where level to be coded */ 3812 WhereClause *pWC; /* Decomposition of the entire WHERE clause */ 3813 WhereTerm *pTerm; /* A WHERE clause term */ 3814 Parse *pParse; /* Parsing context */ 3815 Vdbe *v; /* The prepared stmt under constructions */ 3816 struct SrcList_item *pTabItem; /* FROM clause term being coded */ 3817 int addrBrk; /* Jump here to break out of the loop */ 3818 int addrCont; /* Jump here to continue with next cycle */ 3819 int iRowidReg = 0; /* Rowid is stored in this register, if not zero */ 3820 int iReleaseReg = 0; /* Temp register to free before returning */ 3821 3822 pParse = pWInfo->pParse; 3823 v = pParse->pVdbe; 3824 pWC = pWInfo->pWC; 3825 pLevel = &pWInfo->a[iLevel]; 3826 pTabItem = &pWInfo->pTabList->a[pLevel->iFrom]; 3827 iCur = pTabItem->iCursor; 3828 bRev = (pLevel->plan.wsFlags & WHERE_REVERSE)!=0; 3829 omitTable = (pLevel->plan.wsFlags & WHERE_IDX_ONLY)!=0 3830 && (wctrlFlags & WHERE_FORCE_TABLE)==0; 3831 3832 /* Create labels for the "break" and "continue" instructions 3833 ** for the current loop. Jump to addrBrk to break out of a loop. 3834 ** Jump to cont to go immediately to the next iteration of the 3835 ** loop. 3836 ** 3837 ** When there is an IN operator, we also have a "addrNxt" label that 3838 ** means to continue with the next IN value combination. When 3839 ** there are no IN operators in the constraints, the "addrNxt" label 3840 ** is the same as "addrBrk". 3841 */ 3842 addrBrk = pLevel->addrBrk = pLevel->addrNxt = sqlite3VdbeMakeLabel(v); 3843 addrCont = pLevel->addrCont = sqlite3VdbeMakeLabel(v); 3844 3845 /* If this is the right table of a LEFT OUTER JOIN, allocate and 3846 ** initialize a memory cell that records if this table matches any 3847 ** row of the left table of the join. 3848 */ 3849 if( pLevel->iFrom>0 && (pTabItem[0].jointype & JT_LEFT)!=0 ){ 3850 pLevel->iLeftJoin = ++pParse->nMem; 3851 sqlite3VdbeAddOp2(v, OP_Integer, 0, pLevel->iLeftJoin); 3852 VdbeComment((v, "init LEFT JOIN no-match flag")); 3853 } 3854 3855 #ifndef SQLITE_OMIT_VIRTUALTABLE 3856 if( (pLevel->plan.wsFlags & WHERE_VIRTUALTABLE)!=0 ){ 3857 /* Case 0: The table is a virtual-table. Use the VFilter and VNext 3858 ** to access the data. 3859 */ 3860 int iReg; /* P3 Value for OP_VFilter */ 3861 sqlite3_index_info *pVtabIdx = pLevel->plan.u.pVtabIdx; 3862 int nConstraint = pVtabIdx->nConstraint; 3863 struct sqlite3_index_constraint_usage *aUsage = 3864 pVtabIdx->aConstraintUsage; 3865 const struct sqlite3_index_constraint *aConstraint = 3866 pVtabIdx->aConstraint; 3867 3868 sqlite3ExprCachePush(pParse); 3869 iReg = sqlite3GetTempRange(pParse, nConstraint+2); 3870 for(j=1; j<=nConstraint; j++){ 3871 for(k=0; k<nConstraint; k++){ 3872 if( aUsage[k].argvIndex==j ){ 3873 int iTerm = aConstraint[k].iTermOffset; 3874 sqlite3ExprCode(pParse, pWC->a[iTerm].pExpr->pRight, iReg+j+1); 3875 break; 3876 } 3877 } 3878 if( k==nConstraint ) break; 3879 } 3880 sqlite3VdbeAddOp2(v, OP_Integer, pVtabIdx->idxNum, iReg); 3881 sqlite3VdbeAddOp2(v, OP_Integer, j-1, iReg+1); 3882 sqlite3VdbeAddOp4(v, OP_VFilter, iCur, addrBrk, iReg, pVtabIdx->idxStr, 3883 pVtabIdx->needToFreeIdxStr ? P4_MPRINTF : P4_STATIC); 3884 pVtabIdx->needToFreeIdxStr = 0; 3885 for(j=0; j<nConstraint; j++){ 3886 if( aUsage[j].omit ){ 3887 int iTerm = aConstraint[j].iTermOffset; 3888 disableTerm(pLevel, &pWC->a[iTerm]); 3889 } 3890 } 3891 pLevel->op = OP_VNext; 3892 pLevel->p1 = iCur; 3893 pLevel->p2 = sqlite3VdbeCurrentAddr(v); 3894 sqlite3ReleaseTempRange(pParse, iReg, nConstraint+2); 3895 sqlite3ExprCachePop(pParse, 1); 3896 }else 3897 #endif /* SQLITE_OMIT_VIRTUALTABLE */ 3898 3899 if( pLevel->plan.wsFlags & WHERE_ROWID_EQ ){ 3900 /* Case 1: We can directly reference a single row using an 3901 ** equality comparison against the ROWID field. Or 3902 ** we reference multiple rows using a "rowid IN (...)" 3903 ** construct. 3904 */ 3905 iReleaseReg = sqlite3GetTempReg(pParse); 3906 pTerm = findTerm(pWC, iCur, -1, notReady, WO_EQ|WO_IN, 0); 3907 assert( pTerm!=0 ); 3908 assert( pTerm->pExpr!=0 ); 3909 assert( pTerm->leftCursor==iCur ); 3910 assert( omitTable==0 ); 3911 testcase( pTerm->wtFlags & TERM_VIRTUAL ); /* EV: R-30575-11662 */ 3912 iRowidReg = codeEqualityTerm(pParse, pTerm, pLevel, iReleaseReg); 3913 addrNxt = pLevel->addrNxt; 3914 sqlite3VdbeAddOp2(v, OP_MustBeInt, iRowidReg, addrNxt); 3915 sqlite3VdbeAddOp3(v, OP_NotExists, iCur, addrNxt, iRowidReg); 3916 sqlite3ExprCacheStore(pParse, iCur, -1, iRowidReg); 3917 VdbeComment((v, "pk")); 3918 pLevel->op = OP_Noop; 3919 }else if( pLevel->plan.wsFlags & WHERE_ROWID_RANGE ){ 3920 /* Case 2: We have an inequality comparison against the ROWID field. 3921 */ 3922 int testOp = OP_Noop; 3923 int start; 3924 int memEndValue = 0; 3925 WhereTerm *pStart, *pEnd; 3926 3927 assert( omitTable==0 ); 3928 pStart = findTerm(pWC, iCur, -1, notReady, WO_GT|WO_GE, 0); 3929 pEnd = findTerm(pWC, iCur, -1, notReady, WO_LT|WO_LE, 0); 3930 if( bRev ){ 3931 pTerm = pStart; 3932 pStart = pEnd; 3933 pEnd = pTerm; 3934 } 3935 if( pStart ){ 3936 Expr *pX; /* The expression that defines the start bound */ 3937 int r1, rTemp; /* Registers for holding the start boundary */ 3938 3939 /* The following constant maps TK_xx codes into corresponding 3940 ** seek opcodes. It depends on a particular ordering of TK_xx 3941 */ 3942 const u8 aMoveOp[] = { 3943 /* TK_GT */ OP_SeekGt, 3944 /* TK_LE */ OP_SeekLe, 3945 /* TK_LT */ OP_SeekLt, 3946 /* TK_GE */ OP_SeekGe 3947 }; 3948 assert( TK_LE==TK_GT+1 ); /* Make sure the ordering.. */ 3949 assert( TK_LT==TK_GT+2 ); /* ... of the TK_xx values... */ 3950 assert( TK_GE==TK_GT+3 ); /* ... is correcct. */ 3951 3952 testcase( pStart->wtFlags & TERM_VIRTUAL ); /* EV: R-30575-11662 */ 3953 pX = pStart->pExpr; 3954 assert( pX!=0 ); 3955 assert( pStart->leftCursor==iCur ); 3956 r1 = sqlite3ExprCodeTemp(pParse, pX->pRight, &rTemp); 3957 sqlite3VdbeAddOp3(v, aMoveOp[pX->op-TK_GT], iCur, addrBrk, r1); 3958 VdbeComment((v, "pk")); 3959 sqlite3ExprCacheAffinityChange(pParse, r1, 1); 3960 sqlite3ReleaseTempReg(pParse, rTemp); 3961 disableTerm(pLevel, pStart); 3962 }else{ 3963 sqlite3VdbeAddOp2(v, bRev ? OP_Last : OP_Rewind, iCur, addrBrk); 3964 } 3965 if( pEnd ){ 3966 Expr *pX; 3967 pX = pEnd->pExpr; 3968 assert( pX!=0 ); 3969 assert( pEnd->leftCursor==iCur ); 3970 testcase( pEnd->wtFlags & TERM_VIRTUAL ); /* EV: R-30575-11662 */ 3971 memEndValue = ++pParse->nMem; 3972 sqlite3ExprCode(pParse, pX->pRight, memEndValue); 3973 if( pX->op==TK_LT || pX->op==TK_GT ){ 3974 testOp = bRev ? OP_Le : OP_Ge; 3975 }else{ 3976 testOp = bRev ? OP_Lt : OP_Gt; 3977 } 3978 disableTerm(pLevel, pEnd); 3979 } 3980 start = sqlite3VdbeCurrentAddr(v); 3981 pLevel->op = bRev ? OP_Prev : OP_Next; 3982 pLevel->p1 = iCur; 3983 pLevel->p2 = start; 3984 if( pStart==0 && pEnd==0 ){ 3985 pLevel->p5 = SQLITE_STMTSTATUS_FULLSCAN_STEP; 3986 }else{ 3987 assert( pLevel->p5==0 ); 3988 } 3989 if( testOp!=OP_Noop ){ 3990 iRowidReg = iReleaseReg = sqlite3GetTempReg(pParse); 3991 sqlite3VdbeAddOp2(v, OP_Rowid, iCur, iRowidReg); 3992 sqlite3ExprCacheStore(pParse, iCur, -1, iRowidReg); 3993 sqlite3VdbeAddOp3(v, testOp, memEndValue, addrBrk, iRowidReg); 3994 sqlite3VdbeChangeP5(v, SQLITE_AFF_NUMERIC | SQLITE_JUMPIFNULL); 3995 } 3996 }else if( pLevel->plan.wsFlags & (WHERE_COLUMN_RANGE|WHERE_COLUMN_EQ) ){ 3997 /* Case 3: A scan using an index. 3998 ** 3999 ** The WHERE clause may contain zero or more equality 4000 ** terms ("==" or "IN" operators) that refer to the N 4001 ** left-most columns of the index. It may also contain 4002 ** inequality constraints (>, <, >= or <=) on the indexed 4003 ** column that immediately follows the N equalities. Only 4004 ** the right-most column can be an inequality - the rest must 4005 ** use the "==" and "IN" operators. For example, if the 4006 ** index is on (x,y,z), then the following clauses are all 4007 ** optimized: 4008 ** 4009 ** x=5 4010 ** x=5 AND y=10 4011 ** x=5 AND y<10 4012 ** x=5 AND y>5 AND y<10 4013 ** x=5 AND y=5 AND z<=10 4014 ** 4015 ** The z<10 term of the following cannot be used, only 4016 ** the x=5 term: 4017 ** 4018 ** x=5 AND z<10 4019 ** 4020 ** N may be zero if there are inequality constraints. 4021 ** If there are no inequality constraints, then N is at 4022 ** least one. 4023 ** 4024 ** This case is also used when there are no WHERE clause 4025 ** constraints but an index is selected anyway, in order 4026 ** to force the output order to conform to an ORDER BY. 4027 */ 4028 static const u8 aStartOp[] = { 4029 0, 4030 0, 4031 OP_Rewind, /* 2: (!start_constraints && startEq && !bRev) */ 4032 OP_Last, /* 3: (!start_constraints && startEq && bRev) */ 4033 OP_SeekGt, /* 4: (start_constraints && !startEq && !bRev) */ 4034 OP_SeekLt, /* 5: (start_constraints && !startEq && bRev) */ 4035 OP_SeekGe, /* 6: (start_constraints && startEq && !bRev) */ 4036 OP_SeekLe /* 7: (start_constraints && startEq && bRev) */ 4037 }; 4038 static const u8 aEndOp[] = { 4039 OP_Noop, /* 0: (!end_constraints) */ 4040 OP_IdxGE, /* 1: (end_constraints && !bRev) */ 4041 OP_IdxLT /* 2: (end_constraints && bRev) */ 4042 }; 4043 int nEq = pLevel->plan.nEq; /* Number of == or IN terms */ 4044 int isMinQuery = 0; /* If this is an optimized SELECT min(x).. */ 4045 int regBase; /* Base register holding constraint values */ 4046 int r1; /* Temp register */ 4047 WhereTerm *pRangeStart = 0; /* Inequality constraint at range start */ 4048 WhereTerm *pRangeEnd = 0; /* Inequality constraint at range end */ 4049 int startEq; /* True if range start uses ==, >= or <= */ 4050 int endEq; /* True if range end uses ==, >= or <= */ 4051 int start_constraints; /* Start of range is constrained */ 4052 int nConstraint; /* Number of constraint terms */ 4053 Index *pIdx; /* The index we will be using */ 4054 int iIdxCur; /* The VDBE cursor for the index */ 4055 int nExtraReg = 0; /* Number of extra registers needed */ 4056 int op; /* Instruction opcode */ 4057 char *zStartAff; /* Affinity for start of range constraint */ 4058 char *zEndAff; /* Affinity for end of range constraint */ 4059 4060 pIdx = pLevel->plan.u.pIdx; 4061 iIdxCur = pLevel->iIdxCur; 4062 k = (nEq==pIdx->nColumn ? -1 : pIdx->aiColumn[nEq]); 4063 4064 /* If this loop satisfies a sort order (pOrderBy) request that 4065 ** was passed to this function to implement a "SELECT min(x) ..." 4066 ** query, then the caller will only allow the loop to run for 4067 ** a single iteration. This means that the first row returned 4068 ** should not have a NULL value stored in 'x'. If column 'x' is 4069 ** the first one after the nEq equality constraints in the index, 4070 ** this requires some special handling. 4071 */ 4072 if( (wctrlFlags&WHERE_ORDERBY_MIN)!=0 4073 && (pLevel->plan.wsFlags&WHERE_ORDERBY) 4074 && (pIdx->nColumn>nEq) 4075 ){ 4076 /* assert( pOrderBy->nExpr==1 ); */ 4077 /* assert( pOrderBy->a[0].pExpr->iColumn==pIdx->aiColumn[nEq] ); */ 4078 isMinQuery = 1; 4079 nExtraReg = 1; 4080 } 4081 4082 /* Find any inequality constraint terms for the start and end 4083 ** of the range. 4084 */ 4085 if( pLevel->plan.wsFlags & WHERE_TOP_LIMIT ){ 4086 pRangeEnd = findTerm(pWC, iCur, k, notReady, (WO_LT|WO_LE), pIdx); 4087 nExtraReg = 1; 4088 } 4089 if( pLevel->plan.wsFlags & WHERE_BTM_LIMIT ){ 4090 pRangeStart = findTerm(pWC, iCur, k, notReady, (WO_GT|WO_GE), pIdx); 4091 nExtraReg = 1; 4092 } 4093 4094 /* Generate code to evaluate all constraint terms using == or IN 4095 ** and store the values of those terms in an array of registers 4096 ** starting at regBase. 4097 */ 4098 regBase = codeAllEqualityTerms( 4099 pParse, pLevel, pWC, notReady, nExtraReg, &zStartAff 4100 ); 4101 zEndAff = sqlite3DbStrDup(pParse->db, zStartAff); 4102 addrNxt = pLevel->addrNxt; 4103 4104 /* If we are doing a reverse order scan on an ascending index, or 4105 ** a forward order scan on a descending index, interchange the 4106 ** start and end terms (pRangeStart and pRangeEnd). 4107 */ 4108 if( (nEq<pIdx->nColumn && bRev==(pIdx->aSortOrder[nEq]==SQLITE_SO_ASC)) 4109 || (bRev && pIdx->nColumn==nEq) 4110 ){ 4111 SWAP(WhereTerm *, pRangeEnd, pRangeStart); 4112 } 4113 4114 testcase( pRangeStart && pRangeStart->eOperator & WO_LE ); 4115 testcase( pRangeStart && pRangeStart->eOperator & WO_GE ); 4116 testcase( pRangeEnd && pRangeEnd->eOperator & WO_LE ); 4117 testcase( pRangeEnd && pRangeEnd->eOperator & WO_GE ); 4118 startEq = !pRangeStart || pRangeStart->eOperator & (WO_LE|WO_GE); 4119 endEq = !pRangeEnd || pRangeEnd->eOperator & (WO_LE|WO_GE); 4120 start_constraints = pRangeStart || nEq>0; 4121 4122 /* Seek the index cursor to the start of the range. */ 4123 nConstraint = nEq; 4124 if( pRangeStart ){ 4125 Expr *pRight = pRangeStart->pExpr->pRight; 4126 sqlite3ExprCode(pParse, pRight, regBase+nEq); 4127 if( (pRangeStart->wtFlags & TERM_VNULL)==0 ){ 4128 sqlite3ExprCodeIsNullJump(v, pRight, regBase+nEq, addrNxt); 4129 } 4130 if( zStartAff ){ 4131 if( sqlite3CompareAffinity(pRight, zStartAff[nEq])==SQLITE_AFF_NONE){ 4132 /* Since the comparison is to be performed with no conversions 4133 ** applied to the operands, set the affinity to apply to pRight to 4134 ** SQLITE_AFF_NONE. */ 4135 zStartAff[nEq] = SQLITE_AFF_NONE; 4136 } 4137 if( sqlite3ExprNeedsNoAffinityChange(pRight, zStartAff[nEq]) ){ 4138 zStartAff[nEq] = SQLITE_AFF_NONE; 4139 } 4140 } 4141 nConstraint++; 4142 testcase( pRangeStart->wtFlags & TERM_VIRTUAL ); /* EV: R-30575-11662 */ 4143 }else if( isMinQuery ){ 4144 sqlite3VdbeAddOp2(v, OP_Null, 0, regBase+nEq); 4145 nConstraint++; 4146 startEq = 0; 4147 start_constraints = 1; 4148 } 4149 codeApplyAffinity(pParse, regBase, nConstraint, zStartAff); 4150 op = aStartOp[(start_constraints<<2) + (startEq<<1) + bRev]; 4151 assert( op!=0 ); 4152 testcase( op==OP_Rewind ); 4153 testcase( op==OP_Last ); 4154 testcase( op==OP_SeekGt ); 4155 testcase( op==OP_SeekGe ); 4156 testcase( op==OP_SeekLe ); 4157 testcase( op==OP_SeekLt ); 4158 sqlite3VdbeAddOp4Int(v, op, iIdxCur, addrNxt, regBase, nConstraint); 4159 4160 /* Load the value for the inequality constraint at the end of the 4161 ** range (if any). 4162 */ 4163 nConstraint = nEq; 4164 if( pRangeEnd ){ 4165 Expr *pRight = pRangeEnd->pExpr->pRight; 4166 sqlite3ExprCacheRemove(pParse, regBase+nEq, 1); 4167 sqlite3ExprCode(pParse, pRight, regBase+nEq); 4168 if( (pRangeEnd->wtFlags & TERM_VNULL)==0 ){ 4169 sqlite3ExprCodeIsNullJump(v, pRight, regBase+nEq, addrNxt); 4170 } 4171 if( zEndAff ){ 4172 if( sqlite3CompareAffinity(pRight, zEndAff[nEq])==SQLITE_AFF_NONE){ 4173 /* Since the comparison is to be performed with no conversions 4174 ** applied to the operands, set the affinity to apply to pRight to 4175 ** SQLITE_AFF_NONE. */ 4176 zEndAff[nEq] = SQLITE_AFF_NONE; 4177 } 4178 if( sqlite3ExprNeedsNoAffinityChange(pRight, zEndAff[nEq]) ){ 4179 zEndAff[nEq] = SQLITE_AFF_NONE; 4180 } 4181 } 4182 codeApplyAffinity(pParse, regBase, nEq+1, zEndAff); 4183 nConstraint++; 4184 testcase( pRangeEnd->wtFlags & TERM_VIRTUAL ); /* EV: R-30575-11662 */ 4185 } 4186 sqlite3DbFree(pParse->db, zStartAff); 4187 sqlite3DbFree(pParse->db, zEndAff); 4188 4189 /* Top of the loop body */ 4190 pLevel->p2 = sqlite3VdbeCurrentAddr(v); 4191 4192 /* Check if the index cursor is past the end of the range. */ 4193 op = aEndOp[(pRangeEnd || nEq) * (1 + bRev)]; 4194 testcase( op==OP_Noop ); 4195 testcase( op==OP_IdxGE ); 4196 testcase( op==OP_IdxLT ); 4197 if( op!=OP_Noop ){ 4198 sqlite3VdbeAddOp4Int(v, op, iIdxCur, addrNxt, regBase, nConstraint); 4199 sqlite3VdbeChangeP5(v, endEq!=bRev ?1:0); 4200 } 4201 4202 /* If there are inequality constraints, check that the value 4203 ** of the table column that the inequality contrains is not NULL. 4204 ** If it is, jump to the next iteration of the loop. 4205 */ 4206 r1 = sqlite3GetTempReg(pParse); 4207 testcase( pLevel->plan.wsFlags & WHERE_BTM_LIMIT ); 4208 testcase( pLevel->plan.wsFlags & WHERE_TOP_LIMIT ); 4209 if( (pLevel->plan.wsFlags & (WHERE_BTM_LIMIT|WHERE_TOP_LIMIT))!=0 ){ 4210 sqlite3VdbeAddOp3(v, OP_Column, iIdxCur, nEq, r1); 4211 sqlite3VdbeAddOp2(v, OP_IsNull, r1, addrCont); 4212 } 4213 sqlite3ReleaseTempReg(pParse, r1); 4214 4215 /* Seek the table cursor, if required */ 4216 disableTerm(pLevel, pRangeStart); 4217 disableTerm(pLevel, pRangeEnd); 4218 if( !omitTable ){ 4219 iRowidReg = iReleaseReg = sqlite3GetTempReg(pParse); 4220 sqlite3VdbeAddOp2(v, OP_IdxRowid, iIdxCur, iRowidReg); 4221 sqlite3ExprCacheStore(pParse, iCur, -1, iRowidReg); 4222 sqlite3VdbeAddOp2(v, OP_Seek, iCur, iRowidReg); /* Deferred seek */ 4223 } 4224 4225 /* Record the instruction used to terminate the loop. Disable 4226 ** WHERE clause terms made redundant by the index range scan. 4227 */ 4228 if( pLevel->plan.wsFlags & WHERE_UNIQUE ){ 4229 pLevel->op = OP_Noop; 4230 }else if( bRev ){ 4231 pLevel->op = OP_Prev; 4232 }else{ 4233 pLevel->op = OP_Next; 4234 } 4235 pLevel->p1 = iIdxCur; 4236 }else 4237 4238 #ifndef SQLITE_OMIT_OR_OPTIMIZATION 4239 if( pLevel->plan.wsFlags & WHERE_MULTI_OR ){ 4240 /* Case 4: Two or more separately indexed terms connected by OR 4241 ** 4242 ** Example: 4243 ** 4244 ** CREATE TABLE t1(a,b,c,d); 4245 ** CREATE INDEX i1 ON t1(a); 4246 ** CREATE INDEX i2 ON t1(b); 4247 ** CREATE INDEX i3 ON t1(c); 4248 ** 4249 ** SELECT * FROM t1 WHERE a=5 OR b=7 OR (c=11 AND d=13) 4250 ** 4251 ** In the example, there are three indexed terms connected by OR. 4252 ** The top of the loop looks like this: 4253 ** 4254 ** Null 1 # Zero the rowset in reg 1 4255 ** 4256 ** Then, for each indexed term, the following. The arguments to 4257 ** RowSetTest are such that the rowid of the current row is inserted 4258 ** into the RowSet. If it is already present, control skips the 4259 ** Gosub opcode and jumps straight to the code generated by WhereEnd(). 4260 ** 4261 ** sqlite3WhereBegin(<term>) 4262 ** RowSetTest # Insert rowid into rowset 4263 ** Gosub 2 A 4264 ** sqlite3WhereEnd() 4265 ** 4266 ** Following the above, code to terminate the loop. Label A, the target 4267 ** of the Gosub above, jumps to the instruction right after the Goto. 4268 ** 4269 ** Null 1 # Zero the rowset in reg 1 4270 ** Goto B # The loop is finished. 4271 ** 4272 ** A: <loop body> # Return data, whatever. 4273 ** 4274 ** Return 2 # Jump back to the Gosub 4275 ** 4276 ** B: <after the loop> 4277 ** 4278 */ 4279 WhereClause *pOrWc; /* The OR-clause broken out into subterms */ 4280 SrcList *pOrTab; /* Shortened table list or OR-clause generation */ 4281 4282 int regReturn = ++pParse->nMem; /* Register used with OP_Gosub */ 4283 int regRowset = 0; /* Register for RowSet object */ 4284 int regRowid = 0; /* Register holding rowid */ 4285 int iLoopBody = sqlite3VdbeMakeLabel(v); /* Start of loop body */ 4286 int iRetInit; /* Address of regReturn init */ 4287 int untestedTerms = 0; /* Some terms not completely tested */ 4288 int ii; /* Loop counter */ 4289 Expr *pAndExpr = 0; /* An ".. AND (...)" expression */ 4290 4291 pTerm = pLevel->plan.u.pTerm; 4292 assert( pTerm!=0 ); 4293 assert( pTerm->eOperator==WO_OR ); 4294 assert( (pTerm->wtFlags & TERM_ORINFO)!=0 ); 4295 pOrWc = &pTerm->u.pOrInfo->wc; 4296 pLevel->op = OP_Return; 4297 pLevel->p1 = regReturn; 4298 4299 /* Set up a new SrcList ni pOrTab containing the table being scanned 4300 ** by this loop in the a[0] slot and all notReady tables in a[1..] slots. 4301 ** This becomes the SrcList in the recursive call to sqlite3WhereBegin(). 4302 */ 4303 if( pWInfo->nLevel>1 ){ 4304 int nNotReady; /* The number of notReady tables */ 4305 struct SrcList_item *origSrc; /* Original list of tables */ 4306 nNotReady = pWInfo->nLevel - iLevel - 1; 4307 pOrTab = sqlite3StackAllocRaw(pParse->db, 4308 sizeof(*pOrTab)+ nNotReady*sizeof(pOrTab->a[0])); 4309 if( pOrTab==0 ) return notReady; 4310 pOrTab->nAlloc = (i16)(nNotReady + 1); 4311 pOrTab->nSrc = pOrTab->nAlloc; 4312 memcpy(pOrTab->a, pTabItem, sizeof(*pTabItem)); 4313 origSrc = pWInfo->pTabList->a; 4314 for(k=1; k<=nNotReady; k++){ 4315 memcpy(&pOrTab->a[k], &origSrc[pLevel[k].iFrom], sizeof(pOrTab->a[k])); 4316 } 4317 }else{ 4318 pOrTab = pWInfo->pTabList; 4319 } 4320 4321 /* Initialize the rowset register to contain NULL. An SQL NULL is 4322 ** equivalent to an empty rowset. 4323 ** 4324 ** Also initialize regReturn to contain the address of the instruction 4325 ** immediately following the OP_Return at the bottom of the loop. This 4326 ** is required in a few obscure LEFT JOIN cases where control jumps 4327 ** over the top of the loop into the body of it. In this case the 4328 ** correct response for the end-of-loop code (the OP_Return) is to 4329 ** fall through to the next instruction, just as an OP_Next does if 4330 ** called on an uninitialized cursor. 4331 */ 4332 if( (wctrlFlags & WHERE_DUPLICATES_OK)==0 ){ 4333 regRowset = ++pParse->nMem; 4334 regRowid = ++pParse->nMem; 4335 sqlite3VdbeAddOp2(v, OP_Null, 0, regRowset); 4336 } 4337 iRetInit = sqlite3VdbeAddOp2(v, OP_Integer, 0, regReturn); 4338 4339 /* If the original WHERE clause is z of the form: (x1 OR x2 OR ...) AND y 4340 ** Then for every term xN, evaluate as the subexpression: xN AND z 4341 ** That way, terms in y that are factored into the disjunction will 4342 ** be picked up by the recursive calls to sqlite3WhereBegin() below. 4343 */ 4344 if( pWC->nTerm>1 ){ 4345 pAndExpr = sqlite3ExprAlloc(pParse->db, TK_AND, 0, 0); 4346 pAndExpr->pRight = pWhere; 4347 } 4348 4349 for(ii=0; ii<pOrWc->nTerm; ii++){ 4350 WhereTerm *pOrTerm = &pOrWc->a[ii]; 4351 if( pOrTerm->leftCursor==iCur || pOrTerm->eOperator==WO_AND ){ 4352 WhereInfo *pSubWInfo; /* Info for single OR-term scan */ 4353 Expr *pOrExpr = pOrTerm->pExpr; 4354 if( pAndExpr ){ 4355 pAndExpr->pLeft = pOrExpr; 4356 pOrExpr = pAndExpr; 4357 } 4358 /* Loop through table entries that match term pOrTerm. */ 4359 pSubWInfo = sqlite3WhereBegin(pParse, pOrTab, pOrExpr, 0, 0, 4360 WHERE_OMIT_OPEN_CLOSE | WHERE_AND_ONLY | 4361 WHERE_FORCE_TABLE | WHERE_ONETABLE_ONLY); 4362 if( pSubWInfo ){ 4363 explainOneScan( 4364 pParse, pOrTab, &pSubWInfo->a[0], iLevel, pLevel->iFrom, 0 4365 ); 4366 if( (wctrlFlags & WHERE_DUPLICATES_OK)==0 ){ 4367 int iSet = ((ii==pOrWc->nTerm-1)?-1:ii); 4368 int r; 4369 r = sqlite3ExprCodeGetColumn(pParse, pTabItem->pTab, -1, iCur, 4370 regRowid); 4371 sqlite3VdbeAddOp4Int(v, OP_RowSetTest, regRowset, 4372 sqlite3VdbeCurrentAddr(v)+2, r, iSet); 4373 } 4374 sqlite3VdbeAddOp2(v, OP_Gosub, regReturn, iLoopBody); 4375 4376 /* The pSubWInfo->untestedTerms flag means that this OR term 4377 ** contained one or more AND term from a notReady table. The 4378 ** terms from the notReady table could not be tested and will 4379 ** need to be tested later. 4380 */ 4381 if( pSubWInfo->untestedTerms ) untestedTerms = 1; 4382 4383 /* Finish the loop through table entries that match term pOrTerm. */ 4384 sqlite3WhereEnd(pSubWInfo); 4385 } 4386 } 4387 } 4388 sqlite3DbFree(pParse->db, pAndExpr); 4389 sqlite3VdbeChangeP1(v, iRetInit, sqlite3VdbeCurrentAddr(v)); 4390 sqlite3VdbeAddOp2(v, OP_Goto, 0, pLevel->addrBrk); 4391 sqlite3VdbeResolveLabel(v, iLoopBody); 4392 4393 if( pWInfo->nLevel>1 ) sqlite3StackFree(pParse->db, pOrTab); 4394 if( !untestedTerms ) disableTerm(pLevel, pTerm); 4395 }else 4396 #endif /* SQLITE_OMIT_OR_OPTIMIZATION */ 4397 4398 { 4399 /* Case 5: There is no usable index. We must do a complete 4400 ** scan of the entire table. 4401 */ 4402 static const u8 aStep[] = { OP_Next, OP_Prev }; 4403 static const u8 aStart[] = { OP_Rewind, OP_Last }; 4404 assert( bRev==0 || bRev==1 ); 4405 assert( omitTable==0 ); 4406 pLevel->op = aStep[bRev]; 4407 pLevel->p1 = iCur; 4408 pLevel->p2 = 1 + sqlite3VdbeAddOp2(v, aStart[bRev], iCur, addrBrk); 4409 pLevel->p5 = SQLITE_STMTSTATUS_FULLSCAN_STEP; 4410 } 4411 notReady &= ~getMask(pWC->pMaskSet, iCur); 4412 4413 /* Insert code to test every subexpression that can be completely 4414 ** computed using the current set of tables. 4415 ** 4416 ** IMPLEMENTATION-OF: R-49525-50935 Terms that cannot be satisfied through 4417 ** the use of indices become tests that are evaluated against each row of 4418 ** the relevant input tables. 4419 */ 4420 for(pTerm=pWC->a, j=pWC->nTerm; j>0; j--, pTerm++){ 4421 Expr *pE; 4422 testcase( pTerm->wtFlags & TERM_VIRTUAL ); /* IMP: R-30575-11662 */ 4423 testcase( pTerm->wtFlags & TERM_CODED ); 4424 if( pTerm->wtFlags & (TERM_VIRTUAL|TERM_CODED) ) continue; 4425 if( (pTerm->prereqAll & notReady)!=0 ){ 4426 testcase( pWInfo->untestedTerms==0 4427 && (pWInfo->wctrlFlags & WHERE_ONETABLE_ONLY)!=0 ); 4428 pWInfo->untestedTerms = 1; 4429 continue; 4430 } 4431 pE = pTerm->pExpr; 4432 assert( pE!=0 ); 4433 if( pLevel->iLeftJoin && !ExprHasProperty(pE, EP_FromJoin) ){ 4434 continue; 4435 } 4436 sqlite3ExprIfFalse(pParse, pE, addrCont, SQLITE_JUMPIFNULL); 4437 pTerm->wtFlags |= TERM_CODED; 4438 } 4439 4440 /* For a LEFT OUTER JOIN, generate code that will record the fact that 4441 ** at least one row of the right table has matched the left table. 4442 */ 4443 if( pLevel->iLeftJoin ){ 4444 pLevel->addrFirst = sqlite3VdbeCurrentAddr(v); 4445 sqlite3VdbeAddOp2(v, OP_Integer, 1, pLevel->iLeftJoin); 4446 VdbeComment((v, "record LEFT JOIN hit")); 4447 sqlite3ExprCacheClear(pParse); 4448 for(pTerm=pWC->a, j=0; j<pWC->nTerm; j++, pTerm++){ 4449 testcase( pTerm->wtFlags & TERM_VIRTUAL ); /* IMP: R-30575-11662 */ 4450 testcase( pTerm->wtFlags & TERM_CODED ); 4451 if( pTerm->wtFlags & (TERM_VIRTUAL|TERM_CODED) ) continue; 4452 if( (pTerm->prereqAll & notReady)!=0 ){ 4453 assert( pWInfo->untestedTerms ); 4454 continue; 4455 } 4456 assert( pTerm->pExpr ); 4457 sqlite3ExprIfFalse(pParse, pTerm->pExpr, addrCont, SQLITE_JUMPIFNULL); 4458 pTerm->wtFlags |= TERM_CODED; 4459 } 4460 } 4461 sqlite3ReleaseTempReg(pParse, iReleaseReg); 4462 4463 return notReady; 4464 } 4465 4466 #if defined(SQLITE_TEST) 4467 /* 4468 ** The following variable holds a text description of query plan generated 4469 ** by the most recent call to sqlite3WhereBegin(). Each call to WhereBegin 4470 ** overwrites the previous. This information is used for testing and 4471 ** analysis only. 4472 */ 4473 char sqlite3_query_plan[BMS*2*40]; /* Text of the join */ 4474 static int nQPlan = 0; /* Next free slow in _query_plan[] */ 4475 4476 #endif /* SQLITE_TEST */ 4477 4478 4479 /* 4480 ** Free a WhereInfo structure 4481 */ 4482 static void whereInfoFree(sqlite3 *db, WhereInfo *pWInfo){ 4483 if( ALWAYS(pWInfo) ){ 4484 int i; 4485 for(i=0; i<pWInfo->nLevel; i++){ 4486 sqlite3_index_info *pInfo = pWInfo->a[i].pIdxInfo; 4487 if( pInfo ){ 4488 /* assert( pInfo->needToFreeIdxStr==0 || db->mallocFailed ); */ 4489 if( pInfo->needToFreeIdxStr ){ 4490 sqlite3_free(pInfo->idxStr); 4491 } 4492 sqlite3DbFree(db, pInfo); 4493 } 4494 if( pWInfo->a[i].plan.wsFlags & WHERE_TEMP_INDEX ){ 4495 Index *pIdx = pWInfo->a[i].plan.u.pIdx; 4496 if( pIdx ){ 4497 sqlite3DbFree(db, pIdx->zColAff); 4498 sqlite3DbFree(db, pIdx); 4499 } 4500 } 4501 } 4502 whereClauseClear(pWInfo->pWC); 4503 sqlite3DbFree(db, pWInfo); 4504 } 4505 } 4506 4507 4508 /* 4509 ** Generate the beginning of the loop used for WHERE clause processing. 4510 ** The return value is a pointer to an opaque structure that contains 4511 ** information needed to terminate the loop. Later, the calling routine 4512 ** should invoke sqlite3WhereEnd() with the return value of this function 4513 ** in order to complete the WHERE clause processing. 4514 ** 4515 ** If an error occurs, this routine returns NULL. 4516 ** 4517 ** The basic idea is to do a nested loop, one loop for each table in 4518 ** the FROM clause of a select. (INSERT and UPDATE statements are the 4519 ** same as a SELECT with only a single table in the FROM clause.) For 4520 ** example, if the SQL is this: 4521 ** 4522 ** SELECT * FROM t1, t2, t3 WHERE ...; 4523 ** 4524 ** Then the code generated is conceptually like the following: 4525 ** 4526 ** foreach row1 in t1 do \ Code generated 4527 ** foreach row2 in t2 do |-- by sqlite3WhereBegin() 4528 ** foreach row3 in t3 do / 4529 ** ... 4530 ** end \ Code generated 4531 ** end |-- by sqlite3WhereEnd() 4532 ** end / 4533 ** 4534 ** Note that the loops might not be nested in the order in which they 4535 ** appear in the FROM clause if a different order is better able to make 4536 ** use of indices. Note also that when the IN operator appears in 4537 ** the WHERE clause, it might result in additional nested loops for 4538 ** scanning through all values on the right-hand side of the IN. 4539 ** 4540 ** There are Btree cursors associated with each table. t1 uses cursor 4541 ** number pTabList->a[0].iCursor. t2 uses the cursor pTabList->a[1].iCursor. 4542 ** And so forth. This routine generates code to open those VDBE cursors 4543 ** and sqlite3WhereEnd() generates the code to close them. 4544 ** 4545 ** The code that sqlite3WhereBegin() generates leaves the cursors named 4546 ** in pTabList pointing at their appropriate entries. The [...] code 4547 ** can use OP_Column and OP_Rowid opcodes on these cursors to extract 4548 ** data from the various tables of the loop. 4549 ** 4550 ** If the WHERE clause is empty, the foreach loops must each scan their 4551 ** entire tables. Thus a three-way join is an O(N^3) operation. But if 4552 ** the tables have indices and there are terms in the WHERE clause that 4553 ** refer to those indices, a complete table scan can be avoided and the 4554 ** code will run much faster. Most of the work of this routine is checking 4555 ** to see if there are indices that can be used to speed up the loop. 4556 ** 4557 ** Terms of the WHERE clause are also used to limit which rows actually 4558 ** make it to the "..." in the middle of the loop. After each "foreach", 4559 ** terms of the WHERE clause that use only terms in that loop and outer 4560 ** loops are evaluated and if false a jump is made around all subsequent 4561 ** inner loops (or around the "..." if the test occurs within the inner- 4562 ** most loop) 4563 ** 4564 ** OUTER JOINS 4565 ** 4566 ** An outer join of tables t1 and t2 is conceptally coded as follows: 4567 ** 4568 ** foreach row1 in t1 do 4569 ** flag = 0 4570 ** foreach row2 in t2 do 4571 ** start: 4572 ** ... 4573 ** flag = 1 4574 ** end 4575 ** if flag==0 then 4576 ** move the row2 cursor to a null row 4577 ** goto start 4578 ** fi 4579 ** end 4580 ** 4581 ** ORDER BY CLAUSE PROCESSING 4582 ** 4583 ** *ppOrderBy is a pointer to the ORDER BY clause of a SELECT statement, 4584 ** if there is one. If there is no ORDER BY clause or if this routine 4585 ** is called from an UPDATE or DELETE statement, then ppOrderBy is NULL. 4586 ** 4587 ** If an index can be used so that the natural output order of the table 4588 ** scan is correct for the ORDER BY clause, then that index is used and 4589 ** *ppOrderBy is set to NULL. This is an optimization that prevents an 4590 ** unnecessary sort of the result set if an index appropriate for the 4591 ** ORDER BY clause already exists. 4592 ** 4593 ** If the where clause loops cannot be arranged to provide the correct 4594 ** output order, then the *ppOrderBy is unchanged. 4595 */ 4596 WhereInfo *sqlite3WhereBegin( 4597 Parse *pParse, /* The parser context */ 4598 SrcList *pTabList, /* A list of all tables to be scanned */ 4599 Expr *pWhere, /* The WHERE clause */ 4600 ExprList **ppOrderBy, /* An ORDER BY clause, or NULL */ 4601 ExprList *pDistinct, /* The select-list for DISTINCT queries - or NULL */ 4602 u16 wctrlFlags /* One of the WHERE_* flags defined in sqliteInt.h */ 4603 ){ 4604 int i; /* Loop counter */ 4605 int nByteWInfo; /* Num. bytes allocated for WhereInfo struct */ 4606 int nTabList; /* Number of elements in pTabList */ 4607 WhereInfo *pWInfo; /* Will become the return value of this function */ 4608 Vdbe *v = pParse->pVdbe; /* The virtual database engine */ 4609 Bitmask notReady; /* Cursors that are not yet positioned */ 4610 WhereMaskSet *pMaskSet; /* The expression mask set */ 4611 WhereClause *pWC; /* Decomposition of the WHERE clause */ 4612 struct SrcList_item *pTabItem; /* A single entry from pTabList */ 4613 WhereLevel *pLevel; /* A single level in the pWInfo list */ 4614 int iFrom; /* First unused FROM clause element */ 4615 int andFlags; /* AND-ed combination of all pWC->a[].wtFlags */ 4616 sqlite3 *db; /* Database connection */ 4617 4618 /* The number of tables in the FROM clause is limited by the number of 4619 ** bits in a Bitmask 4620 */ 4621 testcase( pTabList->nSrc==BMS ); 4622 if( pTabList->nSrc>BMS ){ 4623 sqlite3ErrorMsg(pParse, "at most %d tables in a join", BMS); 4624 return 0; 4625 } 4626 4627 /* This function normally generates a nested loop for all tables in 4628 ** pTabList. But if the WHERE_ONETABLE_ONLY flag is set, then we should 4629 ** only generate code for the first table in pTabList and assume that 4630 ** any cursors associated with subsequent tables are uninitialized. 4631 */ 4632 nTabList = (wctrlFlags & WHERE_ONETABLE_ONLY) ? 1 : pTabList->nSrc; 4633 4634 /* Allocate and initialize the WhereInfo structure that will become the 4635 ** return value. A single allocation is used to store the WhereInfo 4636 ** struct, the contents of WhereInfo.a[], the WhereClause structure 4637 ** and the WhereMaskSet structure. Since WhereClause contains an 8-byte 4638 ** field (type Bitmask) it must be aligned on an 8-byte boundary on 4639 ** some architectures. Hence the ROUND8() below. 4640 */ 4641 db = pParse->db; 4642 nByteWInfo = ROUND8(sizeof(WhereInfo)+(nTabList-1)*sizeof(WhereLevel)); 4643 pWInfo = sqlite3DbMallocZero(db, 4644 nByteWInfo + 4645 sizeof(WhereClause) + 4646 sizeof(WhereMaskSet) 4647 ); 4648 if( db->mallocFailed ){ 4649 sqlite3DbFree(db, pWInfo); 4650 pWInfo = 0; 4651 goto whereBeginError; 4652 } 4653 pWInfo->nLevel = nTabList; 4654 pWInfo->pParse = pParse; 4655 pWInfo->pTabList = pTabList; 4656 pWInfo->iBreak = sqlite3VdbeMakeLabel(v); 4657 pWInfo->pWC = pWC = (WhereClause *)&((u8 *)pWInfo)[nByteWInfo]; 4658 pWInfo->wctrlFlags = wctrlFlags; 4659 pWInfo->savedNQueryLoop = pParse->nQueryLoop; 4660 pMaskSet = (WhereMaskSet*)&pWC[1]; 4661 4662 /* Disable the DISTINCT optimization if SQLITE_DistinctOpt is set via 4663 ** sqlite3_test_ctrl(SQLITE_TESTCTRL_OPTIMIZATIONS,...) */ 4664 if( db->flags & SQLITE_DistinctOpt ) pDistinct = 0; 4665 4666 /* Split the WHERE clause into separate subexpressions where each 4667 ** subexpression is separated by an AND operator. 4668 */ 4669 initMaskSet(pMaskSet); 4670 whereClauseInit(pWC, pParse, pMaskSet, wctrlFlags); 4671 sqlite3ExprCodeConstants(pParse, pWhere); 4672 whereSplit(pWC, pWhere, TK_AND); /* IMP: R-15842-53296 */ 4673 4674 /* Special case: a WHERE clause that is constant. Evaluate the 4675 ** expression and either jump over all of the code or fall thru. 4676 */ 4677 if( pWhere && (nTabList==0 || sqlite3ExprIsConstantNotJoin(pWhere)) ){ 4678 sqlite3ExprIfFalse(pParse, pWhere, pWInfo->iBreak, SQLITE_JUMPIFNULL); 4679 pWhere = 0; 4680 } 4681 4682 /* Assign a bit from the bitmask to every term in the FROM clause. 4683 ** 4684 ** When assigning bitmask values to FROM clause cursors, it must be 4685 ** the case that if X is the bitmask for the N-th FROM clause term then 4686 ** the bitmask for all FROM clause terms to the left of the N-th term 4687 ** is (X-1). An expression from the ON clause of a LEFT JOIN can use 4688 ** its Expr.iRightJoinTable value to find the bitmask of the right table 4689 ** of the join. Subtracting one from the right table bitmask gives a 4690 ** bitmask for all tables to the left of the join. Knowing the bitmask 4691 ** for all tables to the left of a left join is important. Ticket #3015. 4692 ** 4693 ** Configure the WhereClause.vmask variable so that bits that correspond 4694 ** to virtual table cursors are set. This is used to selectively disable 4695 ** the OR-to-IN transformation in exprAnalyzeOrTerm(). It is not helpful 4696 ** with virtual tables. 4697 ** 4698 ** Note that bitmasks are created for all pTabList->nSrc tables in 4699 ** pTabList, not just the first nTabList tables. nTabList is normally 4700 ** equal to pTabList->nSrc but might be shortened to 1 if the 4701 ** WHERE_ONETABLE_ONLY flag is set. 4702 */ 4703 assert( pWC->vmask==0 && pMaskSet->n==0 ); 4704 for(i=0; i<pTabList->nSrc; i++){ 4705 createMask(pMaskSet, pTabList->a[i].iCursor); 4706 #ifndef SQLITE_OMIT_VIRTUALTABLE 4707 if( ALWAYS(pTabList->a[i].pTab) && IsVirtual(pTabList->a[i].pTab) ){ 4708 pWC->vmask |= ((Bitmask)1 << i); 4709 } 4710 #endif 4711 } 4712 #ifndef NDEBUG 4713 { 4714 Bitmask toTheLeft = 0; 4715 for(i=0; i<pTabList->nSrc; i++){ 4716 Bitmask m = getMask(pMaskSet, pTabList->a[i].iCursor); 4717 assert( (m-1)==toTheLeft ); 4718 toTheLeft |= m; 4719 } 4720 } 4721 #endif 4722 4723 /* Analyze all of the subexpressions. Note that exprAnalyze() might 4724 ** add new virtual terms onto the end of the WHERE clause. We do not 4725 ** want to analyze these virtual terms, so start analyzing at the end 4726 ** and work forward so that the added virtual terms are never processed. 4727 */ 4728 exprAnalyzeAll(pTabList, pWC); 4729 if( db->mallocFailed ){ 4730 goto whereBeginError; 4731 } 4732 4733 /* Check if the DISTINCT qualifier, if there is one, is redundant. 4734 ** If it is, then set pDistinct to NULL and WhereInfo.eDistinct to 4735 ** WHERE_DISTINCT_UNIQUE to tell the caller to ignore the DISTINCT. 4736 */ 4737 if( pDistinct && isDistinctRedundant(pParse, pTabList, pWC, pDistinct) ){ 4738 pDistinct = 0; 4739 pWInfo->eDistinct = WHERE_DISTINCT_UNIQUE; 4740 } 4741 4742 /* Chose the best index to use for each table in the FROM clause. 4743 ** 4744 ** This loop fills in the following fields: 4745 ** 4746 ** pWInfo->a[].pIdx The index to use for this level of the loop. 4747 ** pWInfo->a[].wsFlags WHERE_xxx flags associated with pIdx 4748 ** pWInfo->a[].nEq The number of == and IN constraints 4749 ** pWInfo->a[].iFrom Which term of the FROM clause is being coded 4750 ** pWInfo->a[].iTabCur The VDBE cursor for the database table 4751 ** pWInfo->a[].iIdxCur The VDBE cursor for the index 4752 ** pWInfo->a[].pTerm When wsFlags==WO_OR, the OR-clause term 4753 ** 4754 ** This loop also figures out the nesting order of tables in the FROM 4755 ** clause. 4756 */ 4757 notReady = ~(Bitmask)0; 4758 andFlags = ~0; 4759 WHERETRACE(("*** Optimizer Start ***\n")); 4760 for(i=iFrom=0, pLevel=pWInfo->a; i<nTabList; i++, pLevel++){ 4761 WhereCost bestPlan; /* Most efficient plan seen so far */ 4762 Index *pIdx; /* Index for FROM table at pTabItem */ 4763 int j; /* For looping over FROM tables */ 4764 int bestJ = -1; /* The value of j */ 4765 Bitmask m; /* Bitmask value for j or bestJ */ 4766 int isOptimal; /* Iterator for optimal/non-optimal search */ 4767 int nUnconstrained; /* Number tables without INDEXED BY */ 4768 Bitmask notIndexed; /* Mask of tables that cannot use an index */ 4769 4770 memset(&bestPlan, 0, sizeof(bestPlan)); 4771 bestPlan.rCost = SQLITE_BIG_DBL; 4772 WHERETRACE(("*** Begin search for loop %d ***\n", i)); 4773 4774 /* Loop through the remaining entries in the FROM clause to find the 4775 ** next nested loop. The loop tests all FROM clause entries 4776 ** either once or twice. 4777 ** 4778 ** The first test is always performed if there are two or more entries 4779 ** remaining and never performed if there is only one FROM clause entry 4780 ** to choose from. The first test looks for an "optimal" scan. In 4781 ** this context an optimal scan is one that uses the same strategy 4782 ** for the given FROM clause entry as would be selected if the entry 4783 ** were used as the innermost nested loop. In other words, a table 4784 ** is chosen such that the cost of running that table cannot be reduced 4785 ** by waiting for other tables to run first. This "optimal" test works 4786 ** by first assuming that the FROM clause is on the inner loop and finding 4787 ** its query plan, then checking to see if that query plan uses any 4788 ** other FROM clause terms that are notReady. If no notReady terms are 4789 ** used then the "optimal" query plan works. 4790 ** 4791 ** Note that the WhereCost.nRow parameter for an optimal scan might 4792 ** not be as small as it would be if the table really were the innermost 4793 ** join. The nRow value can be reduced by WHERE clause constraints 4794 ** that do not use indices. But this nRow reduction only happens if the 4795 ** table really is the innermost join. 4796 ** 4797 ** The second loop iteration is only performed if no optimal scan 4798 ** strategies were found by the first iteration. This second iteration 4799 ** is used to search for the lowest cost scan overall. 4800 ** 4801 ** Previous versions of SQLite performed only the second iteration - 4802 ** the next outermost loop was always that with the lowest overall 4803 ** cost. However, this meant that SQLite could select the wrong plan 4804 ** for scripts such as the following: 4805 ** 4806 ** CREATE TABLE t1(a, b); 4807 ** CREATE TABLE t2(c, d); 4808 ** SELECT * FROM t2, t1 WHERE t2.rowid = t1.a; 4809 ** 4810 ** The best strategy is to iterate through table t1 first. However it 4811 ** is not possible to determine this with a simple greedy algorithm. 4812 ** Since the cost of a linear scan through table t2 is the same 4813 ** as the cost of a linear scan through table t1, a simple greedy 4814 ** algorithm may choose to use t2 for the outer loop, which is a much 4815 ** costlier approach. 4816 */ 4817 nUnconstrained = 0; 4818 notIndexed = 0; 4819 for(isOptimal=(iFrom<nTabList-1); isOptimal>=0 && bestJ<0; isOptimal--){ 4820 Bitmask mask; /* Mask of tables not yet ready */ 4821 for(j=iFrom, pTabItem=&pTabList->a[j]; j<nTabList; j++, pTabItem++){ 4822 int doNotReorder; /* True if this table should not be reordered */ 4823 WhereCost sCost; /* Cost information from best[Virtual]Index() */ 4824 ExprList *pOrderBy; /* ORDER BY clause for index to optimize */ 4825 ExprList *pDist; /* DISTINCT clause for index to optimize */ 4826 4827 doNotReorder = (pTabItem->jointype & (JT_LEFT|JT_CROSS))!=0; 4828 if( j!=iFrom && doNotReorder ) break; 4829 m = getMask(pMaskSet, pTabItem->iCursor); 4830 if( (m & notReady)==0 ){ 4831 if( j==iFrom ) iFrom++; 4832 continue; 4833 } 4834 mask = (isOptimal ? m : notReady); 4835 pOrderBy = ((i==0 && ppOrderBy )?*ppOrderBy:0); 4836 pDist = (i==0 ? pDistinct : 0); 4837 if( pTabItem->pIndex==0 ) nUnconstrained++; 4838 4839 WHERETRACE(("=== trying table %d with isOptimal=%d ===\n", 4840 j, isOptimal)); 4841 assert( pTabItem->pTab ); 4842 #ifndef SQLITE_OMIT_VIRTUALTABLE 4843 if( IsVirtual(pTabItem->pTab) ){ 4844 sqlite3_index_info **pp = &pWInfo->a[j].pIdxInfo; 4845 bestVirtualIndex(pParse, pWC, pTabItem, mask, notReady, pOrderBy, 4846 &sCost, pp); 4847 }else 4848 #endif 4849 { 4850 bestBtreeIndex(pParse, pWC, pTabItem, mask, notReady, pOrderBy, 4851 pDist, &sCost); 4852 } 4853 assert( isOptimal || (sCost.used¬Ready)==0 ); 4854 4855 /* If an INDEXED BY clause is present, then the plan must use that 4856 ** index if it uses any index at all */ 4857 assert( pTabItem->pIndex==0 4858 || (sCost.plan.wsFlags & WHERE_NOT_FULLSCAN)==0 4859 || sCost.plan.u.pIdx==pTabItem->pIndex ); 4860 4861 if( isOptimal && (sCost.plan.wsFlags & WHERE_NOT_FULLSCAN)==0 ){ 4862 notIndexed |= m; 4863 } 4864 4865 /* Conditions under which this table becomes the best so far: 4866 ** 4867 ** (1) The table must not depend on other tables that have not 4868 ** yet run. 4869 ** 4870 ** (2) A full-table-scan plan cannot supercede indexed plan unless 4871 ** the full-table-scan is an "optimal" plan as defined above. 4872 ** 4873 ** (3) All tables have an INDEXED BY clause or this table lacks an 4874 ** INDEXED BY clause or this table uses the specific 4875 ** index specified by its INDEXED BY clause. This rule ensures 4876 ** that a best-so-far is always selected even if an impossible 4877 ** combination of INDEXED BY clauses are given. The error 4878 ** will be detected and relayed back to the application later. 4879 ** The NEVER() comes about because rule (2) above prevents 4880 ** An indexable full-table-scan from reaching rule (3). 4881 ** 4882 ** (4) The plan cost must be lower than prior plans or else the 4883 ** cost must be the same and the number of rows must be lower. 4884 */ 4885 if( (sCost.used¬Ready)==0 /* (1) */ 4886 && (bestJ<0 || (notIndexed&m)!=0 /* (2) */ 4887 || (bestPlan.plan.wsFlags & WHERE_NOT_FULLSCAN)==0 4888 || (sCost.plan.wsFlags & WHERE_NOT_FULLSCAN)!=0) 4889 && (nUnconstrained==0 || pTabItem->pIndex==0 /* (3) */ 4890 || NEVER((sCost.plan.wsFlags & WHERE_NOT_FULLSCAN)!=0)) 4891 && (bestJ<0 || sCost.rCost<bestPlan.rCost /* (4) */ 4892 || (sCost.rCost<=bestPlan.rCost 4893 && sCost.plan.nRow<bestPlan.plan.nRow)) 4894 ){ 4895 WHERETRACE(("=== table %d is best so far" 4896 " with cost=%g and nRow=%g\n", 4897 j, sCost.rCost, sCost.plan.nRow)); 4898 bestPlan = sCost; 4899 bestJ = j; 4900 } 4901 if( doNotReorder ) break; 4902 } 4903 } 4904 assert( bestJ>=0 ); 4905 assert( notReady & getMask(pMaskSet, pTabList->a[bestJ].iCursor) ); 4906 WHERETRACE(("*** Optimizer selects table %d for loop %d" 4907 " with cost=%g and nRow=%g\n", 4908 bestJ, pLevel-pWInfo->a, bestPlan.rCost, bestPlan.plan.nRow)); 4909 /* The ALWAYS() that follows was added to hush up clang scan-build */ 4910 if( (bestPlan.plan.wsFlags & WHERE_ORDERBY)!=0 && ALWAYS(ppOrderBy) ){ 4911 *ppOrderBy = 0; 4912 } 4913 if( (bestPlan.plan.wsFlags & WHERE_DISTINCT)!=0 ){ 4914 assert( pWInfo->eDistinct==0 ); 4915 pWInfo->eDistinct = WHERE_DISTINCT_ORDERED; 4916 } 4917 andFlags &= bestPlan.plan.wsFlags; 4918 pLevel->plan = bestPlan.plan; 4919 testcase( bestPlan.plan.wsFlags & WHERE_INDEXED ); 4920 testcase( bestPlan.plan.wsFlags & WHERE_TEMP_INDEX ); 4921 if( bestPlan.plan.wsFlags & (WHERE_INDEXED|WHERE_TEMP_INDEX) ){ 4922 pLevel->iIdxCur = pParse->nTab++; 4923 }else{ 4924 pLevel->iIdxCur = -1; 4925 } 4926 notReady &= ~getMask(pMaskSet, pTabList->a[bestJ].iCursor); 4927 pLevel->iFrom = (u8)bestJ; 4928 if( bestPlan.plan.nRow>=(double)1 ){ 4929 pParse->nQueryLoop *= bestPlan.plan.nRow; 4930 } 4931 4932 /* Check that if the table scanned by this loop iteration had an 4933 ** INDEXED BY clause attached to it, that the named index is being 4934 ** used for the scan. If not, then query compilation has failed. 4935 ** Return an error. 4936 */ 4937 pIdx = pTabList->a[bestJ].pIndex; 4938 if( pIdx ){ 4939 if( (bestPlan.plan.wsFlags & WHERE_INDEXED)==0 ){ 4940 sqlite3ErrorMsg(pParse, "cannot use index: %s", pIdx->zName); 4941 goto whereBeginError; 4942 }else{ 4943 /* If an INDEXED BY clause is used, the bestIndex() function is 4944 ** guaranteed to find the index specified in the INDEXED BY clause 4945 ** if it find an index at all. */ 4946 assert( bestPlan.plan.u.pIdx==pIdx ); 4947 } 4948 } 4949 } 4950 WHERETRACE(("*** Optimizer Finished ***\n")); 4951 if( pParse->nErr || db->mallocFailed ){ 4952 goto whereBeginError; 4953 } 4954 4955 /* If the total query only selects a single row, then the ORDER BY 4956 ** clause is irrelevant. 4957 */ 4958 if( (andFlags & WHERE_UNIQUE)!=0 && ppOrderBy ){ 4959 *ppOrderBy = 0; 4960 } 4961 4962 /* If the caller is an UPDATE or DELETE statement that is requesting 4963 ** to use a one-pass algorithm, determine if this is appropriate. 4964 ** The one-pass algorithm only works if the WHERE clause constraints 4965 ** the statement to update a single row. 4966 */ 4967 assert( (wctrlFlags & WHERE_ONEPASS_DESIRED)==0 || pWInfo->nLevel==1 ); 4968 if( (wctrlFlags & WHERE_ONEPASS_DESIRED)!=0 && (andFlags & WHERE_UNIQUE)!=0 ){ 4969 pWInfo->okOnePass = 1; 4970 pWInfo->a[0].plan.wsFlags &= ~WHERE_IDX_ONLY; 4971 } 4972 4973 /* Open all tables in the pTabList and any indices selected for 4974 ** searching those tables. 4975 */ 4976 sqlite3CodeVerifySchema(pParse, -1); /* Insert the cookie verifier Goto */ 4977 notReady = ~(Bitmask)0; 4978 pWInfo->nRowOut = (double)1; 4979 for(i=0, pLevel=pWInfo->a; i<nTabList; i++, pLevel++){ 4980 Table *pTab; /* Table to open */ 4981 int iDb; /* Index of database containing table/index */ 4982 4983 pTabItem = &pTabList->a[pLevel->iFrom]; 4984 pTab = pTabItem->pTab; 4985 pLevel->iTabCur = pTabItem->iCursor; 4986 pWInfo->nRowOut *= pLevel->plan.nRow; 4987 iDb = sqlite3SchemaToIndex(db, pTab->pSchema); 4988 if( (pTab->tabFlags & TF_Ephemeral)!=0 || pTab->pSelect ){ 4989 /* Do nothing */ 4990 }else 4991 #ifndef SQLITE_OMIT_VIRTUALTABLE 4992 if( (pLevel->plan.wsFlags & WHERE_VIRTUALTABLE)!=0 ){ 4993 const char *pVTab = (const char *)sqlite3GetVTable(db, pTab); 4994 int iCur = pTabItem->iCursor; 4995 sqlite3VdbeAddOp4(v, OP_VOpen, iCur, 0, 0, pVTab, P4_VTAB); 4996 }else 4997 #endif 4998 if( (pLevel->plan.wsFlags & WHERE_IDX_ONLY)==0 4999 && (wctrlFlags & WHERE_OMIT_OPEN_CLOSE)==0 ){ 5000 int op = pWInfo->okOnePass ? OP_OpenWrite : OP_OpenRead; 5001 sqlite3OpenTable(pParse, pTabItem->iCursor, iDb, pTab, op); 5002 testcase( pTab->nCol==BMS-1 ); 5003 testcase( pTab->nCol==BMS ); 5004 if( !pWInfo->okOnePass && pTab->nCol<BMS ){ 5005 Bitmask b = pTabItem->colUsed; 5006 int n = 0; 5007 for(; b; b=b>>1, n++){} 5008 sqlite3VdbeChangeP4(v, sqlite3VdbeCurrentAddr(v)-1, 5009 SQLITE_INT_TO_PTR(n), P4_INT32); 5010 assert( n<=pTab->nCol ); 5011 } 5012 }else{ 5013 sqlite3TableLock(pParse, iDb, pTab->tnum, 0, pTab->zName); 5014 } 5015 #ifndef SQLITE_OMIT_AUTOMATIC_INDEX 5016 if( (pLevel->plan.wsFlags & WHERE_TEMP_INDEX)!=0 ){ 5017 constructAutomaticIndex(pParse, pWC, pTabItem, notReady, pLevel); 5018 }else 5019 #endif 5020 if( (pLevel->plan.wsFlags & WHERE_INDEXED)!=0 ){ 5021 Index *pIx = pLevel->plan.u.pIdx; 5022 KeyInfo *pKey = sqlite3IndexKeyinfo(pParse, pIx); 5023 int iIdxCur = pLevel->iIdxCur; 5024 assert( pIx->pSchema==pTab->pSchema ); 5025 assert( iIdxCur>=0 ); 5026 sqlite3VdbeAddOp4(v, OP_OpenRead, iIdxCur, pIx->tnum, iDb, 5027 (char*)pKey, P4_KEYINFO_HANDOFF); 5028 VdbeComment((v, "%s", pIx->zName)); 5029 } 5030 sqlite3CodeVerifySchema(pParse, iDb); 5031 notReady &= ~getMask(pWC->pMaskSet, pTabItem->iCursor); 5032 } 5033 pWInfo->iTop = sqlite3VdbeCurrentAddr(v); 5034 if( db->mallocFailed ) goto whereBeginError; 5035 5036 /* Generate the code to do the search. Each iteration of the for 5037 ** loop below generates code for a single nested loop of the VM 5038 ** program. 5039 */ 5040 notReady = ~(Bitmask)0; 5041 for(i=0; i<nTabList; i++){ 5042 pLevel = &pWInfo->a[i]; 5043 explainOneScan(pParse, pTabList, pLevel, i, pLevel->iFrom, wctrlFlags); 5044 notReady = codeOneLoopStart(pWInfo, i, wctrlFlags, notReady, pWhere); 5045 pWInfo->iContinue = pLevel->addrCont; 5046 } 5047 5048 #ifdef SQLITE_TEST /* For testing and debugging use only */ 5049 /* Record in the query plan information about the current table 5050 ** and the index used to access it (if any). If the table itself 5051 ** is not used, its name is just '{}'. If no index is used 5052 ** the index is listed as "{}". If the primary key is used the 5053 ** index name is '*'. 5054 */ 5055 for(i=0; i<nTabList; i++){ 5056 char *z; 5057 int n; 5058 pLevel = &pWInfo->a[i]; 5059 pTabItem = &pTabList->a[pLevel->iFrom]; 5060 z = pTabItem->zAlias; 5061 if( z==0 ) z = pTabItem->pTab->zName; 5062 n = sqlite3Strlen30(z); 5063 if( n+nQPlan < sizeof(sqlite3_query_plan)-10 ){ 5064 if( pLevel->plan.wsFlags & WHERE_IDX_ONLY ){ 5065 memcpy(&sqlite3_query_plan[nQPlan], "{}", 2); 5066 nQPlan += 2; 5067 }else{ 5068 memcpy(&sqlite3_query_plan[nQPlan], z, n); 5069 nQPlan += n; 5070 } 5071 sqlite3_query_plan[nQPlan++] = ' '; 5072 } 5073 testcase( pLevel->plan.wsFlags & WHERE_ROWID_EQ ); 5074 testcase( pLevel->plan.wsFlags & WHERE_ROWID_RANGE ); 5075 if( pLevel->plan.wsFlags & (WHERE_ROWID_EQ|WHERE_ROWID_RANGE) ){ 5076 memcpy(&sqlite3_query_plan[nQPlan], "* ", 2); 5077 nQPlan += 2; 5078 }else if( (pLevel->plan.wsFlags & WHERE_INDEXED)!=0 ){ 5079 n = sqlite3Strlen30(pLevel->plan.u.pIdx->zName); 5080 if( n+nQPlan < sizeof(sqlite3_query_plan)-2 ){ 5081 memcpy(&sqlite3_query_plan[nQPlan], pLevel->plan.u.pIdx->zName, n); 5082 nQPlan += n; 5083 sqlite3_query_plan[nQPlan++] = ' '; 5084 } 5085 }else{ 5086 memcpy(&sqlite3_query_plan[nQPlan], "{} ", 3); 5087 nQPlan += 3; 5088 } 5089 } 5090 while( nQPlan>0 && sqlite3_query_plan[nQPlan-1]==' ' ){ 5091 sqlite3_query_plan[--nQPlan] = 0; 5092 } 5093 sqlite3_query_plan[nQPlan] = 0; 5094 nQPlan = 0; 5095 #endif /* SQLITE_TEST // Testing and debugging use only */ 5096 5097 /* Record the continuation address in the WhereInfo structure. Then 5098 ** clean up and return. 5099 */ 5100 return pWInfo; 5101 5102 /* Jump here if malloc fails */ 5103 whereBeginError: 5104 if( pWInfo ){ 5105 pParse->nQueryLoop = pWInfo->savedNQueryLoop; 5106 whereInfoFree(db, pWInfo); 5107 } 5108 return 0; 5109 } 5110 5111 /* 5112 ** Generate the end of the WHERE loop. See comments on 5113 ** sqlite3WhereBegin() for additional information. 5114 */ 5115 void sqlite3WhereEnd(WhereInfo *pWInfo){ 5116 Parse *pParse = pWInfo->pParse; 5117 Vdbe *v = pParse->pVdbe; 5118 int i; 5119 WhereLevel *pLevel; 5120 SrcList *pTabList = pWInfo->pTabList; 5121 sqlite3 *db = pParse->db; 5122 5123 /* Generate loop termination code. 5124 */ 5125 sqlite3ExprCacheClear(pParse); 5126 for(i=pWInfo->nLevel-1; i>=0; i--){ 5127 pLevel = &pWInfo->a[i]; 5128 sqlite3VdbeResolveLabel(v, pLevel->addrCont); 5129 if( pLevel->op!=OP_Noop ){ 5130 sqlite3VdbeAddOp2(v, pLevel->op, pLevel->p1, pLevel->p2); 5131 sqlite3VdbeChangeP5(v, pLevel->p5); 5132 } 5133 if( pLevel->plan.wsFlags & WHERE_IN_ABLE && pLevel->u.in.nIn>0 ){ 5134 struct InLoop *pIn; 5135 int j; 5136 sqlite3VdbeResolveLabel(v, pLevel->addrNxt); 5137 for(j=pLevel->u.in.nIn, pIn=&pLevel->u.in.aInLoop[j-1]; j>0; j--, pIn--){ 5138 sqlite3VdbeJumpHere(v, pIn->addrInTop+1); 5139 sqlite3VdbeAddOp2(v, OP_Next, pIn->iCur, pIn->addrInTop); 5140 sqlite3VdbeJumpHere(v, pIn->addrInTop-1); 5141 } 5142 sqlite3DbFree(db, pLevel->u.in.aInLoop); 5143 } 5144 sqlite3VdbeResolveLabel(v, pLevel->addrBrk); 5145 if( pLevel->iLeftJoin ){ 5146 int addr; 5147 addr = sqlite3VdbeAddOp1(v, OP_IfPos, pLevel->iLeftJoin); 5148 assert( (pLevel->plan.wsFlags & WHERE_IDX_ONLY)==0 5149 || (pLevel->plan.wsFlags & WHERE_INDEXED)!=0 ); 5150 if( (pLevel->plan.wsFlags & WHERE_IDX_ONLY)==0 ){ 5151 sqlite3VdbeAddOp1(v, OP_NullRow, pTabList->a[i].iCursor); 5152 } 5153 if( pLevel->iIdxCur>=0 ){ 5154 sqlite3VdbeAddOp1(v, OP_NullRow, pLevel->iIdxCur); 5155 } 5156 if( pLevel->op==OP_Return ){ 5157 sqlite3VdbeAddOp2(v, OP_Gosub, pLevel->p1, pLevel->addrFirst); 5158 }else{ 5159 sqlite3VdbeAddOp2(v, OP_Goto, 0, pLevel->addrFirst); 5160 } 5161 sqlite3VdbeJumpHere(v, addr); 5162 } 5163 } 5164 5165 /* The "break" point is here, just past the end of the outer loop. 5166 ** Set it. 5167 */ 5168 sqlite3VdbeResolveLabel(v, pWInfo->iBreak); 5169 5170 /* Close all of the cursors that were opened by sqlite3WhereBegin. 5171 */ 5172 assert( pWInfo->nLevel==1 || pWInfo->nLevel==pTabList->nSrc ); 5173 for(i=0, pLevel=pWInfo->a; i<pWInfo->nLevel; i++, pLevel++){ 5174 struct SrcList_item *pTabItem = &pTabList->a[pLevel->iFrom]; 5175 Table *pTab = pTabItem->pTab; 5176 assert( pTab!=0 ); 5177 if( (pTab->tabFlags & TF_Ephemeral)==0 5178 && pTab->pSelect==0 5179 && (pWInfo->wctrlFlags & WHERE_OMIT_OPEN_CLOSE)==0 5180 ){ 5181 int ws = pLevel->plan.wsFlags; 5182 if( !pWInfo->okOnePass && (ws & WHERE_IDX_ONLY)==0 ){ 5183 sqlite3VdbeAddOp1(v, OP_Close, pTabItem->iCursor); 5184 } 5185 if( (ws & WHERE_INDEXED)!=0 && (ws & WHERE_TEMP_INDEX)==0 ){ 5186 sqlite3VdbeAddOp1(v, OP_Close, pLevel->iIdxCur); 5187 } 5188 } 5189 5190 /* If this scan uses an index, make code substitutions to read data 5191 ** from the index in preference to the table. Sometimes, this means 5192 ** the table need never be read from. This is a performance boost, 5193 ** as the vdbe level waits until the table is read before actually 5194 ** seeking the table cursor to the record corresponding to the current 5195 ** position in the index. 5196 ** 5197 ** Calls to the code generator in between sqlite3WhereBegin and 5198 ** sqlite3WhereEnd will have created code that references the table 5199 ** directly. This loop scans all that code looking for opcodes 5200 ** that reference the table and converts them into opcodes that 5201 ** reference the index. 5202 */ 5203 if( (pLevel->plan.wsFlags & WHERE_INDEXED)!=0 && !db->mallocFailed){ 5204 int k, j, last; 5205 VdbeOp *pOp; 5206 Index *pIdx = pLevel->plan.u.pIdx; 5207 5208 assert( pIdx!=0 ); 5209 pOp = sqlite3VdbeGetOp(v, pWInfo->iTop); 5210 last = sqlite3VdbeCurrentAddr(v); 5211 for(k=pWInfo->iTop; k<last; k++, pOp++){ 5212 if( pOp->p1!=pLevel->iTabCur ) continue; 5213 if( pOp->opcode==OP_Column ){ 5214 for(j=0; j<pIdx->nColumn; j++){ 5215 if( pOp->p2==pIdx->aiColumn[j] ){ 5216 pOp->p2 = j; 5217 pOp->p1 = pLevel->iIdxCur; 5218 break; 5219 } 5220 } 5221 assert( (pLevel->plan.wsFlags & WHERE_IDX_ONLY)==0 5222 || j<pIdx->nColumn ); 5223 }else if( pOp->opcode==OP_Rowid ){ 5224 pOp->p1 = pLevel->iIdxCur; 5225 pOp->opcode = OP_IdxRowid; 5226 } 5227 } 5228 } 5229 } 5230 5231 /* Final cleanup 5232 */ 5233 pParse->nQueryLoop = pWInfo->savedNQueryLoop; 5234 whereInfoFree(db, pWInfo); 5235 return; 5236 } 5237