Add support for the WINDOW clause.
[sqlite.git] / src / rowset.c
blobaa81607b9fdc1864c5d542973fb2b632055995ab
1 /*
2 ** 2008 December 3
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.
11 *************************************************************************
13 ** This module implements an object we call a "RowSet".
15 ** The RowSet object is a collection of rowids. Rowids
16 ** are inserted into the RowSet in an arbitrary order. Inserts
17 ** can be intermixed with tests to see if a given rowid has been
18 ** previously inserted into the RowSet.
20 ** After all inserts are finished, it is possible to extract the
21 ** elements of the RowSet in sorted order. Once this extraction
22 ** process has started, no new elements may be inserted.
24 ** Hence, the primitive operations for a RowSet are:
26 ** CREATE
27 ** INSERT
28 ** TEST
29 ** SMALLEST
30 ** DESTROY
32 ** The CREATE and DESTROY primitives are the constructor and destructor,
33 ** obviously. The INSERT primitive adds a new element to the RowSet.
34 ** TEST checks to see if an element is already in the RowSet. SMALLEST
35 ** extracts the least value from the RowSet.
37 ** The INSERT primitive might allocate additional memory. Memory is
38 ** allocated in chunks so most INSERTs do no allocation. There is an
39 ** upper bound on the size of allocated memory. No memory is freed
40 ** until DESTROY.
42 ** The TEST primitive includes a "batch" number. The TEST primitive
43 ** will only see elements that were inserted before the last change
44 ** in the batch number. In other words, if an INSERT occurs between
45 ** two TESTs where the TESTs have the same batch nubmer, then the
46 ** value added by the INSERT will not be visible to the second TEST.
47 ** The initial batch number is zero, so if the very first TEST contains
48 ** a non-zero batch number, it will see all prior INSERTs.
50 ** No INSERTs may occurs after a SMALLEST. An assertion will fail if
51 ** that is attempted.
53 ** The cost of an INSERT is roughly constant. (Sometimes new memory
54 ** has to be allocated on an INSERT.) The cost of a TEST with a new
55 ** batch number is O(NlogN) where N is the number of elements in the RowSet.
56 ** The cost of a TEST using the same batch number is O(logN). The cost
57 ** of the first SMALLEST is O(NlogN). Second and subsequent SMALLEST
58 ** primitives are constant time. The cost of DESTROY is O(N).
60 ** TEST and SMALLEST may not be used by the same RowSet. This used to
61 ** be possible, but the feature was not used, so it was removed in order
62 ** to simplify the code.
64 #include "sqliteInt.h"
68 ** Target size for allocation chunks.
70 #define ROWSET_ALLOCATION_SIZE 1024
73 ** The number of rowset entries per allocation chunk.
75 #define ROWSET_ENTRY_PER_CHUNK \
76 ((ROWSET_ALLOCATION_SIZE-8)/sizeof(struct RowSetEntry))
79 ** Each entry in a RowSet is an instance of the following object.
81 ** This same object is reused to store a linked list of trees of RowSetEntry
82 ** objects. In that alternative use, pRight points to the next entry
83 ** in the list, pLeft points to the tree, and v is unused. The
84 ** RowSet.pForest value points to the head of this forest list.
86 struct RowSetEntry {
87 i64 v; /* ROWID value for this entry */
88 struct RowSetEntry *pRight; /* Right subtree (larger entries) or list */
89 struct RowSetEntry *pLeft; /* Left subtree (smaller entries) */
93 ** RowSetEntry objects are allocated in large chunks (instances of the
94 ** following structure) to reduce memory allocation overhead. The
95 ** chunks are kept on a linked list so that they can be deallocated
96 ** when the RowSet is destroyed.
98 struct RowSetChunk {
99 struct RowSetChunk *pNextChunk; /* Next chunk on list of them all */
100 struct RowSetEntry aEntry[ROWSET_ENTRY_PER_CHUNK]; /* Allocated entries */
104 ** A RowSet in an instance of the following structure.
106 ** A typedef of this structure if found in sqliteInt.h.
108 struct RowSet {
109 struct RowSetChunk *pChunk; /* List of all chunk allocations */
110 sqlite3 *db; /* The database connection */
111 struct RowSetEntry *pEntry; /* List of entries using pRight */
112 struct RowSetEntry *pLast; /* Last entry on the pEntry list */
113 struct RowSetEntry *pFresh; /* Source of new entry objects */
114 struct RowSetEntry *pForest; /* List of binary trees of entries */
115 u16 nFresh; /* Number of objects on pFresh */
116 u16 rsFlags; /* Various flags */
117 int iBatch; /* Current insert batch */
121 ** Allowed values for RowSet.rsFlags
123 #define ROWSET_SORTED 0x01 /* True if RowSet.pEntry is sorted */
124 #define ROWSET_NEXT 0x02 /* True if sqlite3RowSetNext() has been called */
127 ** Turn bulk memory into a RowSet object. N bytes of memory
128 ** are available at pSpace. The db pointer is used as a memory context
129 ** for any subsequent allocations that need to occur.
130 ** Return a pointer to the new RowSet object.
132 ** It must be the case that N is sufficient to make a Rowset. If not
133 ** an assertion fault occurs.
135 ** If N is larger than the minimum, use the surplus as an initial
136 ** allocation of entries available to be filled.
138 RowSet *sqlite3RowSetInit(sqlite3 *db, void *pSpace, unsigned int N){
139 RowSet *p;
140 assert( N >= ROUND8(sizeof(*p)) );
141 p = pSpace;
142 p->pChunk = 0;
143 p->db = db;
144 p->pEntry = 0;
145 p->pLast = 0;
146 p->pForest = 0;
147 p->pFresh = (struct RowSetEntry*)(ROUND8(sizeof(*p)) + (char*)p);
148 p->nFresh = (u16)((N - ROUND8(sizeof(*p)))/sizeof(struct RowSetEntry));
149 p->rsFlags = ROWSET_SORTED;
150 p->iBatch = 0;
151 return p;
155 ** Deallocate all chunks from a RowSet. This frees all memory that
156 ** the RowSet has allocated over its lifetime. This routine is
157 ** the destructor for the RowSet.
159 void sqlite3RowSetClear(RowSet *p){
160 struct RowSetChunk *pChunk, *pNextChunk;
161 for(pChunk=p->pChunk; pChunk; pChunk = pNextChunk){
162 pNextChunk = pChunk->pNextChunk;
163 sqlite3DbFree(p->db, pChunk);
165 p->pChunk = 0;
166 p->nFresh = 0;
167 p->pEntry = 0;
168 p->pLast = 0;
169 p->pForest = 0;
170 p->rsFlags = ROWSET_SORTED;
174 ** Allocate a new RowSetEntry object that is associated with the
175 ** given RowSet. Return a pointer to the new and completely uninitialized
176 ** objected.
178 ** In an OOM situation, the RowSet.db->mallocFailed flag is set and this
179 ** routine returns NULL.
181 static struct RowSetEntry *rowSetEntryAlloc(RowSet *p){
182 assert( p!=0 );
183 if( p->nFresh==0 ){ /*OPTIMIZATION-IF-FALSE*/
184 /* We could allocate a fresh RowSetEntry each time one is needed, but it
185 ** is more efficient to pull a preallocated entry from the pool */
186 struct RowSetChunk *pNew;
187 pNew = sqlite3DbMallocRawNN(p->db, sizeof(*pNew));
188 if( pNew==0 ){
189 return 0;
191 pNew->pNextChunk = p->pChunk;
192 p->pChunk = pNew;
193 p->pFresh = pNew->aEntry;
194 p->nFresh = ROWSET_ENTRY_PER_CHUNK;
196 p->nFresh--;
197 return p->pFresh++;
201 ** Insert a new value into a RowSet.
203 ** The mallocFailed flag of the database connection is set if a
204 ** memory allocation fails.
206 void sqlite3RowSetInsert(RowSet *p, i64 rowid){
207 struct RowSetEntry *pEntry; /* The new entry */
208 struct RowSetEntry *pLast; /* The last prior entry */
210 /* This routine is never called after sqlite3RowSetNext() */
211 assert( p!=0 && (p->rsFlags & ROWSET_NEXT)==0 );
213 pEntry = rowSetEntryAlloc(p);
214 if( pEntry==0 ) return;
215 pEntry->v = rowid;
216 pEntry->pRight = 0;
217 pLast = p->pLast;
218 if( pLast ){
219 if( rowid<=pLast->v ){ /*OPTIMIZATION-IF-FALSE*/
220 /* Avoid unnecessary sorts by preserving the ROWSET_SORTED flags
221 ** where possible */
222 p->rsFlags &= ~ROWSET_SORTED;
224 pLast->pRight = pEntry;
225 }else{
226 p->pEntry = pEntry;
228 p->pLast = pEntry;
232 ** Merge two lists of RowSetEntry objects. Remove duplicates.
234 ** The input lists are connected via pRight pointers and are
235 ** assumed to each already be in sorted order.
237 static struct RowSetEntry *rowSetEntryMerge(
238 struct RowSetEntry *pA, /* First sorted list to be merged */
239 struct RowSetEntry *pB /* Second sorted list to be merged */
241 struct RowSetEntry head;
242 struct RowSetEntry *pTail;
244 pTail = &head;
245 assert( pA!=0 && pB!=0 );
246 for(;;){
247 assert( pA->pRight==0 || pA->v<=pA->pRight->v );
248 assert( pB->pRight==0 || pB->v<=pB->pRight->v );
249 if( pA->v<=pB->v ){
250 if( pA->v<pB->v ) pTail = pTail->pRight = pA;
251 pA = pA->pRight;
252 if( pA==0 ){
253 pTail->pRight = pB;
254 break;
256 }else{
257 pTail = pTail->pRight = pB;
258 pB = pB->pRight;
259 if( pB==0 ){
260 pTail->pRight = pA;
261 break;
265 return head.pRight;
269 ** Sort all elements on the list of RowSetEntry objects into order of
270 ** increasing v.
272 static struct RowSetEntry *rowSetEntrySort(struct RowSetEntry *pIn){
273 unsigned int i;
274 struct RowSetEntry *pNext, *aBucket[40];
276 memset(aBucket, 0, sizeof(aBucket));
277 while( pIn ){
278 pNext = pIn->pRight;
279 pIn->pRight = 0;
280 for(i=0; aBucket[i]; i++){
281 pIn = rowSetEntryMerge(aBucket[i], pIn);
282 aBucket[i] = 0;
284 aBucket[i] = pIn;
285 pIn = pNext;
287 pIn = aBucket[0];
288 for(i=1; i<sizeof(aBucket)/sizeof(aBucket[0]); i++){
289 if( aBucket[i]==0 ) continue;
290 pIn = pIn ? rowSetEntryMerge(pIn, aBucket[i]) : aBucket[i];
292 return pIn;
297 ** The input, pIn, is a binary tree (or subtree) of RowSetEntry objects.
298 ** Convert this tree into a linked list connected by the pRight pointers
299 ** and return pointers to the first and last elements of the new list.
301 static void rowSetTreeToList(
302 struct RowSetEntry *pIn, /* Root of the input tree */
303 struct RowSetEntry **ppFirst, /* Write head of the output list here */
304 struct RowSetEntry **ppLast /* Write tail of the output list here */
306 assert( pIn!=0 );
307 if( pIn->pLeft ){
308 struct RowSetEntry *p;
309 rowSetTreeToList(pIn->pLeft, ppFirst, &p);
310 p->pRight = pIn;
311 }else{
312 *ppFirst = pIn;
314 if( pIn->pRight ){
315 rowSetTreeToList(pIn->pRight, &pIn->pRight, ppLast);
316 }else{
317 *ppLast = pIn;
319 assert( (*ppLast)->pRight==0 );
324 ** Convert a sorted list of elements (connected by pRight) into a binary
325 ** tree with depth of iDepth. A depth of 1 means the tree contains a single
326 ** node taken from the head of *ppList. A depth of 2 means a tree with
327 ** three nodes. And so forth.
329 ** Use as many entries from the input list as required and update the
330 ** *ppList to point to the unused elements of the list. If the input
331 ** list contains too few elements, then construct an incomplete tree
332 ** and leave *ppList set to NULL.
334 ** Return a pointer to the root of the constructed binary tree.
336 static struct RowSetEntry *rowSetNDeepTree(
337 struct RowSetEntry **ppList,
338 int iDepth
340 struct RowSetEntry *p; /* Root of the new tree */
341 struct RowSetEntry *pLeft; /* Left subtree */
342 if( *ppList==0 ){ /*OPTIMIZATION-IF-TRUE*/
343 /* Prevent unnecessary deep recursion when we run out of entries */
344 return 0;
346 if( iDepth>1 ){ /*OPTIMIZATION-IF-TRUE*/
347 /* This branch causes a *balanced* tree to be generated. A valid tree
348 ** is still generated without this branch, but the tree is wildly
349 ** unbalanced and inefficient. */
350 pLeft = rowSetNDeepTree(ppList, iDepth-1);
351 p = *ppList;
352 if( p==0 ){ /*OPTIMIZATION-IF-FALSE*/
353 /* It is safe to always return here, but the resulting tree
354 ** would be unbalanced */
355 return pLeft;
357 p->pLeft = pLeft;
358 *ppList = p->pRight;
359 p->pRight = rowSetNDeepTree(ppList, iDepth-1);
360 }else{
361 p = *ppList;
362 *ppList = p->pRight;
363 p->pLeft = p->pRight = 0;
365 return p;
369 ** Convert a sorted list of elements into a binary tree. Make the tree
370 ** as deep as it needs to be in order to contain the entire list.
372 static struct RowSetEntry *rowSetListToTree(struct RowSetEntry *pList){
373 int iDepth; /* Depth of the tree so far */
374 struct RowSetEntry *p; /* Current tree root */
375 struct RowSetEntry *pLeft; /* Left subtree */
377 assert( pList!=0 );
378 p = pList;
379 pList = p->pRight;
380 p->pLeft = p->pRight = 0;
381 for(iDepth=1; pList; iDepth++){
382 pLeft = p;
383 p = pList;
384 pList = p->pRight;
385 p->pLeft = pLeft;
386 p->pRight = rowSetNDeepTree(&pList, iDepth);
388 return p;
392 ** Extract the smallest element from the RowSet.
393 ** Write the element into *pRowid. Return 1 on success. Return
394 ** 0 if the RowSet is already empty.
396 ** After this routine has been called, the sqlite3RowSetInsert()
397 ** routine may not be called again.
399 ** This routine may not be called after sqlite3RowSetTest() has
400 ** been used. Older versions of RowSet allowed that, but as the
401 ** capability was not used by the code generator, it was removed
402 ** for code economy.
404 int sqlite3RowSetNext(RowSet *p, i64 *pRowid){
405 assert( p!=0 );
406 assert( p->pForest==0 ); /* Cannot be used with sqlite3RowSetText() */
408 /* Merge the forest into a single sorted list on first call */
409 if( (p->rsFlags & ROWSET_NEXT)==0 ){ /*OPTIMIZATION-IF-FALSE*/
410 if( (p->rsFlags & ROWSET_SORTED)==0 ){ /*OPTIMIZATION-IF-FALSE*/
411 p->pEntry = rowSetEntrySort(p->pEntry);
413 p->rsFlags |= ROWSET_SORTED|ROWSET_NEXT;
416 /* Return the next entry on the list */
417 if( p->pEntry ){
418 *pRowid = p->pEntry->v;
419 p->pEntry = p->pEntry->pRight;
420 if( p->pEntry==0 ){ /*OPTIMIZATION-IF-TRUE*/
421 /* Free memory immediately, rather than waiting on sqlite3_finalize() */
422 sqlite3RowSetClear(p);
424 return 1;
425 }else{
426 return 0;
431 ** Check to see if element iRowid was inserted into the rowset as
432 ** part of any insert batch prior to iBatch. Return 1 or 0.
434 ** If this is the first test of a new batch and if there exist entries
435 ** on pRowSet->pEntry, then sort those entries into the forest at
436 ** pRowSet->pForest so that they can be tested.
438 int sqlite3RowSetTest(RowSet *pRowSet, int iBatch, sqlite3_int64 iRowid){
439 struct RowSetEntry *p, *pTree;
441 /* This routine is never called after sqlite3RowSetNext() */
442 assert( pRowSet!=0 && (pRowSet->rsFlags & ROWSET_NEXT)==0 );
444 /* Sort entries into the forest on the first test of a new batch.
445 ** To save unnecessary work, only do this when the batch number changes.
447 if( iBatch!=pRowSet->iBatch ){ /*OPTIMIZATION-IF-FALSE*/
448 p = pRowSet->pEntry;
449 if( p ){
450 struct RowSetEntry **ppPrevTree = &pRowSet->pForest;
451 if( (pRowSet->rsFlags & ROWSET_SORTED)==0 ){ /*OPTIMIZATION-IF-FALSE*/
452 /* Only sort the current set of entiries if they need it */
453 p = rowSetEntrySort(p);
455 for(pTree = pRowSet->pForest; pTree; pTree=pTree->pRight){
456 ppPrevTree = &pTree->pRight;
457 if( pTree->pLeft==0 ){
458 pTree->pLeft = rowSetListToTree(p);
459 break;
460 }else{
461 struct RowSetEntry *pAux, *pTail;
462 rowSetTreeToList(pTree->pLeft, &pAux, &pTail);
463 pTree->pLeft = 0;
464 p = rowSetEntryMerge(pAux, p);
467 if( pTree==0 ){
468 *ppPrevTree = pTree = rowSetEntryAlloc(pRowSet);
469 if( pTree ){
470 pTree->v = 0;
471 pTree->pRight = 0;
472 pTree->pLeft = rowSetListToTree(p);
475 pRowSet->pEntry = 0;
476 pRowSet->pLast = 0;
477 pRowSet->rsFlags |= ROWSET_SORTED;
479 pRowSet->iBatch = iBatch;
482 /* Test to see if the iRowid value appears anywhere in the forest.
483 ** Return 1 if it does and 0 if not.
485 for(pTree = pRowSet->pForest; pTree; pTree=pTree->pRight){
486 p = pTree->pLeft;
487 while( p ){
488 if( p->v<iRowid ){
489 p = p->pRight;
490 }else if( p->v>iRowid ){
491 p = p->pLeft;
492 }else{
493 return 1;
497 return 0;