Many hyperlinks are disabled.
Use anonymous login
to enable hyperlinks.
Overview
Comment: | continued work on btree (CVS 222) |
---|---|
Downloads: | Tarball | ZIP archive |
Timelines: | family | ancestors | descendants | both | trunk |
Files: | files | file ages | folders |
SHA1: |
d07e0e80a0b33081adda8651e9a6750b |
User & Date: | drh 2001-06-02 02:40:57.000 |
Context
2001-06-08
| ||
00:21 | incremental update (CVS 223) (check-in: 7108b699cc user: drh tags: trunk) | |
2001-06-02
| ||
02:40 | continued work on btree (CVS 222) (check-in: d07e0e80a0 user: drh tags: trunk) | |
2001-05-28
| ||
00:41 | :-) (CVS 1720) (check-in: d78febd197 user: drh tags: trunk) | |
Changes
Changes to src/btree.c.
︙ | ︙ | |||
17 18 19 20 21 22 23 | ** Boston, MA 02111-1307, USA. ** ** Author contact information: ** drh@hwaci.com ** http://www.hwaci.com/drh/ ** ************************************************************************* | | | > > > > > > > > | < > > > > > | < > | < < > | < < | | < < < < < > > > | > > | > > | | 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 | ** Boston, MA 02111-1307, USA. ** ** Author contact information: ** drh@hwaci.com ** http://www.hwaci.com/drh/ ** ************************************************************************* ** $Id: btree.c,v 1.10 2001/06/02 02:40:57 drh Exp $ */ #include "sqliteInt.h" #include "pager.h" #include "btree.h" #include <assert.h> /* ** Primitive data types. u32 must be 4 bytes and u16 must be 2 bytes. ** Change these typedefs when porting to new architectures. */ typedef unsigned int u32; typedef unsigned short int u16; typedef unsigned char u8; /* ** Forward declarations of structures used only in this file. */ typedef struct PageOne PageOne; typedef struct MemPage MemPage; typedef struct PageHdr PageHdr; typedef struct Cell Cell; typedef struct CellHdr CellHdr; typedef struct FreeBlk FreeBlk; typedef struct OverflowPage OverflowPage; /* ** All structures on a database page are aligned to 4-byte boundries. ** This routine rounds up a number of bytes to the next multiple of 4. ** ** This might need to change for computer architectures that require ** and 8-byte alignment boundry for structures. */ #define ROUNDUP(X) ((X+3) & ~3) /* ** This is a magic string that appears at the beginning of every ** SQLite database in order to identify the fail as a real database. */ static const char zMagicHeader[] = "** This file contains an SQLite 2.0 database **" #define MAGIC_SIZE (sizeof(zMagicHeader)) /* ** The first page of the database file contains a magic header string ** to identify the file as an SQLite database file. It also contains ** a pointer to the first free page of the file. Page 2 contains the ** root of the BTree. ** ** Remember that pages are numbered beginning with 1. (See pager.c ** for additional information.) Page 0 does not exist and a page ** number of 0 is used to mean "no such page". */ struct PageOne { char zMagic[MAGIC_SIZE]; /* String that identifies the file as a database */ Pgno firstList; /* First free page in a list of all free pages */ }; /* ** Each database page has a header that is an instance of this ** structure. ** ** MemPage.pHdr always points to the rightmost_pgno. First_free is ** 0 if there is no free space on this page. Otherwise, first_free is ** the index in MemPage.aDisk[] of a FreeBlk structure that describes ** the first block of free space. All free space is defined by a linked ** list of FreeBlk structures. ** ** Data is stored in a linked list of Cell structures. First_cell is ** the index into MemPage.aDisk[] of the first cell on the page. The ** Cells are kept in sorted order. */ struct PageHdr { Pgno rightChild; /* Child page that comes after all cells on this page */ u16 firstCell; /* Index in MemPage.aDisk[] of the first cell */ u16 firstFree; /* Index in MemPage.aDisk[] of the first free block */ }; /* ** Entries on a page of the database are called "Cells". Each Cell ** has a header and data. This structure defines the header. The ** key and data (collectively the "payload") follow this header on ** the database page. ** ** A definition of the complete Cell structure is given below. The ** header for the cell must be defined separately in order to do some ** of the sizing #defines that follow. */ struct CellHdr { Pgno leftChild; /* Child page that comes before this cell */ u16 nKey; /* Number of bytes in the key */ u16 iNext; /* Index in MemPage.aDisk[] of next cell in sorted order */ u32 nData; /* Number of bytes of data */ } /* ** The minimum size of a complete Cell. The Cell must contain a header ** and at least 4 bytes of payload. */ #define MIN_CELL_SIZE (sizeof(CellHdr)+4) /* ** The maximum number of database entries that can be held in a single ** page of the database. */ #define MX_CELL ((SQLITE_PAGE_SIZE-sizeof(PageHdr))/MIN_CELL_SIZE) /* ** The maximum amount of data (in bytes) that can be stored locally for a ** database entry. If the entry contains more data than this, the ** extra goes onto overflow pages. ** ** This number is chosen so that at least 4 cells will fit on every page. */ #define MX_LOCAL_PAYLOAD \ ((SQLITE_PAGE_SIZE-sizeof(PageHdr))/4-(sizeof(CellHdr)+sizeof(Pgno))) /* ** Data on a database page is stored as a linked list of Cell structures. ** Both the key and the data are stored in aPayload[]. The key always comes ** first. The aPayload[] field grows as necessary to hold the key and data, ** up to a maximum of MX_LOCAL_PAYLOAD bytes. If the size of the key and ** data combined exceeds MX_LOCAL_PAYLOAD bytes, then Cell.ovfl is the ** page number of the first overflow page. ** ** Though this structure is fixed in size, the Cell on the database ** page varies in size. Every cell has a CellHdr and at least 4 bytes ** of payload space. Additional payload bytes (up to the maximum of ** MX_LOCAL_PAYLOAD) and the Cell.ovfl value are allocated only as ** needed. */ struct Cell { CellHdr h; /* The cell header */ char aPayload[MX_LOCAL_PAYLOAD]; /* Key and data */ |
︙ | ︙ | |||
167 168 169 170 171 172 173 | /* ** When the key and data for a single entry in the BTree will not fit in ** the MX_LOACAL_PAYLOAD bytes of space available on the database page, ** then all extra data is written to a linked list of overflow pages. ** Each overflow page is an instance of the following structure. ** ** Unused pages in the database are also represented by instances of | | | | | | | | > | < < | | 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 | /* ** When the key and data for a single entry in the BTree will not fit in ** the MX_LOACAL_PAYLOAD bytes of space available on the database page, ** then all extra data is written to a linked list of overflow pages. ** Each overflow page is an instance of the following structure. ** ** Unused pages in the database are also represented by instances of ** the OverflowPage structure. The PageOne.freeList field is the ** page number of the first page in a linked list of unused database ** pages. */ struct OverflowPage { Pgno next; char aPayload[OVERFLOW_SIZE]; }; /* ** For every page in the database file, an instance of the following structure ** is stored in memory. The aDisk[] array contains the raw bits read from ** the disk. The rest is auxiliary information that held in memory only. The ** auxiliary info is only valid for regular database pages - it is not ** used for overflow pages and pages on the freelist. ** ** Of particular interest in the auxiliary info is the apCell[] entry. Each ** apCell[] entry is a pointer to a Cell structure in aDisk[]. The cells are ** put in this array so that they can be accessed in constant time, rather ** than in linear time which would be needed if we had to walk the linked ** list on every access. ** ** The pParent field points back to the parent page. This allows us to ** walk up the BTree from any leaf to the root. Care must be taken to ** unref() the parent page pointer when this page is no longer referenced. ** The pageDestructor() routine handles that chore. */ struct MemPage { char aDisk[SQLITE_PAGE_SIZE]; /* Page data stored on disk */ int isInit; /* True if auxiliary data is initialized */ MemPage *pParent; /* The parent of this page. NULL for root */ int nFree; /* Number of free bytes in aDisk[] */ int nCell; /* Number of entries on this page */ Cell *apCell[MX_CELL]; /* All data entires in sorted order */ } /* ** The in-memory image of a disk page has the auxiliary information appended ** to the end. EXTRA_SIZE is the number of bytes of space needed to hold ** that extra information. */ #define EXTRA_SIZE (sizeof(MemPage)-SQLITE_PAGE_SIZE) /* ** Everything we need to know about an open database */ struct Btree { Pager *pPager; /* The page cache */ BtCursor *pCursor; /* A list of all open cursors */ PageOne *page1; /* First page of the database */ int inTrans; /* True if a transaction is in progress */ }; typedef Btree Bt; /* ** A cursor is a pointer to a particular entry in the BTree. ** The entry is identified by its MemPage and the index in |
︙ | ︙ | |||
240 241 242 243 244 245 246 | u8 iMatch; /* compare result from last sqliteBtreeMoveto() */ }; /* ** Compute the total number of bytes that a Cell needs on the main ** database page. The number returned includes the Cell header, ** local payload storage, and the pointer to overflow pages (if | | | | 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 | u8 iMatch; /* compare result from last sqliteBtreeMoveto() */ }; /* ** Compute the total number of bytes that a Cell needs on the main ** database page. The number returned includes the Cell header, ** local payload storage, and the pointer to overflow pages (if ** applicable). Additional spaced allocated on overflow pages ** is NOT included in the value returned from this routine. */ static int cellSize(Cell *pCell){ int n = pCell->h.nKey + pCell->h.nData; if( n>MX_LOCAL_PAYLOAD ){ n = MX_LOCAL_PAYLOAD + sizeof(Pgno); }else{ n = ROUNDUP(n); |
︙ | ︙ | |||
265 266 267 268 269 270 271 | */ static void defragmentPage(MemPage *pPage){ int pc; int i, n; FreeBlk *pFBlk; char newPage[SQLITE_PAGE_SIZE]; | | | | | | | | | | | | | 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 | */ static void defragmentPage(MemPage *pPage){ int pc; int i, n; FreeBlk *pFBlk; char newPage[SQLITE_PAGE_SIZE]; pc = sizeof(PageHdr); ((PageHdr*)pPage)->firstCell = pc; memcpy(newPage, pPage->aDisk, pc); for(i=0; i<pPage->nCell; i++){ Cell *pCell = &pPage->apCell[i]; n = cellSize(pCell); pCell->h.iNext = i<pPage->nCell ? pc + n : 0; memcpy(&newPage[pc], pCell, n); pPage->apCell[i] = (Cell*)&pPage->aDisk[pc]; pc += n; } assert( pPage->nFree==SQLITE_PAGE_SIZE-pc ); memcpy(pPage->aDisk, newPage, pc); pFBlk = &pPage->aDisk[pc]; pFBlk->iSize = SQLITE_PAGE_SIZE - pc; pFBlk->iNext = 0; ((PageHdr*)pPage)->firstFree = pc; memset(&pFBlk[1], 0, SQLITE_PAGE_SIZE - pc - sizeof(FreeBlk)); } /* ** Allocate space on a page. The space needs to be at least ** nByte bytes in size. nByte must be a multiple of 4. ** ** Return the index into pPage->aDisk[] of the first byte of ** the new allocation. Or return 0 if there is not enough free ** space on the page to satisfy the allocation request. ** ** If the page contains nBytes of free space but does not contain ** nBytes of contiguous free space, then defragementPage() is ** called to consolidate all free space before allocating the ** new chunk. */ static int allocateSpace(MemPage *pPage, int nByte){ FreeBlk *p; u16 *pIdx; int start; assert( nByte==ROUNDUP(nByte) ); if( pPage->nFree<nByte ) return 0; pIdx = &((PageHdr*)pPage)->firstFree; p = (FreeBlk*)&pPage->aDisk[*pIdx]; while( p->iSize<nByte ){ if( p->iNext==0 ){ defragmentPage(pPage); pIdx = &((PageHdr*)pPage)->firstFree; }else{ pIdx = &p->iNext; } p = (FreeBlk*)&pPage->aDisk[*pIdx]; } if( p->iSize==nByte ){ start = *pIdx; |
︙ | ︙ | |||
347 348 349 350 351 352 353 | u16 *pIdx, idx; FreeBlk *pFBlk; FreeBlk *pNew; FreeBlk *pNext; assert( size == ROUNDUP(size) ); assert( start == ROUNDUP(start) ); | | | 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 | u16 *pIdx, idx; FreeBlk *pFBlk; FreeBlk *pNew; FreeBlk *pNext; assert( size == ROUNDUP(size) ); assert( start == ROUNDUP(start) ); pIdx = &((PageHdr*)pPage)->firstFree; idx = *pIdx; while( idx!=0 && idx<start ){ pFBlk = (FreeBlk*)&pPage->aDisk[idx]; if( idx + pFBlk->iSize == start ){ pFBlk->iSize += size; if( idx + pFBlk->iSize == pFBlk->iNext ){ pNext = (FreeBlk*)&pPage->aDisk[pFblk->iNext]; |
︙ | ︙ | |||
380 381 382 383 384 385 386 | *pIdx = start; pPage->nFree += size; } /* ** Initialize the auxiliary information for a disk block. ** | | > > | 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 | *pIdx = start; pPage->nFree += size; } /* ** Initialize the auxiliary information for a disk block. ** ** The pParent parameter must be a pointer to the MemPage which ** is the parent of the page being initialized. The root of the ** BTree (page 2) has no parent and so for that page, pParent==NULL. ** ** Return SQLITE_OK on success. If we see that the page does ** not contained a well-formed database page, then return ** SQLITE_CORRUPT. Note that a return of SQLITE_OK does not ** guarantee that the page is well-formed. It only shows that ** we failed to detect any corruption. */ |
︙ | ︙ | |||
404 405 406 407 408 409 410 | return SQLITE_OK; } if( pParent ){ pPage->pParent = pParent; sqlitepager_ref(pParent); } if( pPage->isInit ) return SQLITE_OK; | < < | | | | | > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > | 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 | return SQLITE_OK; } if( pParent ){ pPage->pParent = pParent; sqlitepager_ref(pParent); } if( pPage->isInit ) return SQLITE_OK; pPage->isInit = 1; pPage->nCell = 0; freeSpace = SQLITE_PAGE_SIZE - sizeof(PageHdr); idx = ((PageHdr*)pPage)->firstCell; while( idx!=0 ){ if( idx>SQLITE_PAGE_SIZE-MN_CELL_SIZE ) goto page_format_error; if( idx<sizeof(PageHdr) ) goto page_format_error; pCell = (Cell*)&pPage->aDisk[idx]; sz = cellSize(pCell); if( idx+sz > SQLITE_PAGE_SIZE ) goto page_format_error; freeSpace -= sz; pPage->apCell[pPage->nCell++] = pCell; idx = pCell->h.iNext; } pPage->nFree = 0; idx = ((PageHdr*)pPage)->firstFree; while( idx!=0 ){ if( idx>SQLITE_PAGE_SIZE-sizeof(FreeBlk) ) goto page_format_error; if( idx<sizeof(PageHdr) ) goto page_format_error; pFBlk = (FreeBlk*)&pPage->aDisk[idx]; pPage->nFree += pFBlk->iSize; if( pFBlk->iNext <= idx ) goto page_format_error; idx = pFBlk->iNext; } if( pPage->nFree!=freeSpace ) goto page_format_error; return SQLITE_OK; page_format_error: return SQLITE_CORRUPT; } /* ** Recompute the MemPage.apCell[], MemPage.nCell, and MemPage.nFree parameters ** for a cell after the content has be changed significantly. ** ** The computation here is similar to initPage() except that in this case ** the MemPage.aDisk[] field has been set up internally (instead of ** having been read from disk) so we do not need to do as much error ** checking. */ static void reinitPage(MemPage *pPage){ Cell *pCell; pPage->nCell = 0; idx = ((PageHdr*)pPage)->firstCell; while( idx!=0 ){ pCell = (Cell*)&pPage->aDisk[idx]; sz = cellSize(pCell); pPage->apCell[pPage->nCell++] = pCell; idx = pCell->h.iNext; } pPage->nFree = 0; idx = ((PageHdr*)pPage)->firstFree; while( idx!=0 ){ pFBlk = (FreeBlk*)&pPage->aDisk[idx]; pPage->nFree += pFBlk->iSize; idx = pFBlk->iNext; } return SQLITE_OK; } /* ** Initialize a database page so that it holds no entries at all. */ static void zeroPage(MemPage *pPage){ PageHdr *pHdr; FreeBlk *pFBlk; memset(pPage, 0, SQLITE_PAGE_SIZE); pHdr = (PageHdr*)pPage; pHdr->firstCell = 0; pHdr->firstFree = sizeof(*pHdr); pFBlk = (FreeBlk*)&pHdr[1]; pFBlk->iNext = 0; pFBlk->iSize = SQLITE_PAGE_SIZE - sizeof(*pHdr); } /* ** This routine is called when the reference count for a page ** reaches zero. We need to unref the pParent pointer when that ** happens. */ static void pageDestructor(void *pData){ |
︙ | ︙ | |||
507 508 509 510 511 512 513 | ** if there is a locking protocol violation. */ static int lockBtree(Btree *pBt){ int rc; if( pBt->page1 ) return SQLITE_OK; rc = sqlitepager_get(pBt->pPager, 1, &pBt->page1); if( rc!=SQLITE_OK ) return rc; | < < | | | 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 | ** if there is a locking protocol violation. */ static int lockBtree(Btree *pBt){ int rc; if( pBt->page1 ) return SQLITE_OK; rc = sqlitepager_get(pBt->pPager, 1, &pBt->page1); if( rc!=SQLITE_OK ) return rc; /* Do some checking to help insure the file we opened really is ** a valid database file. */ if( sqlitepager_pagecount(pBt->pPager)>0 ){ PageOne *pP1 = pBt->page1; if( strcmp(pP1->zMagic1,zMagicHeader)!=0 ){ rc = SQLITE_CORRUPT; goto page1_init_failed; } } return rc; page1_init_failed: |
︙ | ︙ | |||
603 604 605 606 607 608 609 | if( rc!=SQLITE_OK ){ *ppCur = 0; return rc; } } pCur = sqliteMalloc( sizeof(*pCur) ); if( pCur==0 ){ | < < | > > > > > > > > > < < < < < < < > > > > > > > > > | > < | < | | | | | | > > | 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688 689 690 691 692 693 694 695 696 697 698 699 700 701 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716 717 718 719 720 721 722 723 724 725 726 727 728 729 730 731 732 733 734 735 736 737 738 739 740 741 742 743 744 745 746 | if( rc!=SQLITE_OK ){ *ppCur = 0; return rc; } } pCur = sqliteMalloc( sizeof(*pCur) ); if( pCur==0 ){ rc = SQLITE_NOMEM; goto create_cursor_exception; } rc = sqlitepager_get(pBt->pPager, 2, &pCur->pPage); if( rc!=SQLITE_OK ){ goto create_cursor_exception; } rc = initPage(pCur->pPage, 2, 0); if( rc!=SQLITE_OK ){ goto create_cursor_exception; } pCur->pPrev = 0; pCur->pNext = pBt->pCursor; if( pCur->pNext ){ pCur->pNext->pPrev = pCur; } pBt->pCursor = pCur; pCur->pBt = pBt; pCur->idx = 0; *ppCur = pCur; return SQLITE_OK; create_cursor_exception: *ppCur = 0; if( pCur ){ if( pCur->pPage ) sqlitepager_unref(pCur->pPage); sqliteFree(pCur); } unlinkBtree(pBt); return rc; } /* ** Close a cursor. The lock on the database file is released ** when the last cursor is closed. */ int sqliteBtreeCloseCursor(BtCursor *pCur){ Btree *pBt = pCur->pBt; int i; if( pCur->pPrev ){ pCur->pPrev->pNext = pCur->pNext; }else{ pBt->pCursor = pCur->pNext; } if( pCur->pNext ){ pCur->pNext->pPrev = pCur->pPrev; } sqlitepager_unref(pCur->pPage); unlockBtree(pBt); sqliteFree(pCur); } /* ** Make a temporary cursor by filling in the fields of pTempCur. ** The temporary cursor is not on the cursor list for the Btree. */ static void CreateTemporaryCursor(BtCursor *pCur, BtCursor *pTempCur){ memcpy(pTempCur, pCur, sizeof(*pCur)); pTempCur->pNext = 0; pTempCur->pPrev = 0; sqlitepager_ref(pTempCur->pPage); } /* ** Delete a temporary cursor such as was made by the CreateTemporaryCursor() ** function above. */ static void DestroyTemporaryCursor(BeCursor *pCur){ sqlitepager_unref(pCur->pPage); } /* ** Set *pSize to the number of bytes of key in the entry the ** cursor currently points to. Always return SQLITE_OK. ** Failure is not possible. If the cursor is not currently ** pointing to an entry (which can happen, for example, if ** the database is empty) then *pSize is set to 0. */ int sqliteBtreeKeySize(BtCursor *pCur, int *pSize){ Cell *pCell; MemPage *pPage; pPage = pCur->pPage; assert( pPage!=0 ); |
︙ | ︙ | |||
712 713 714 715 716 717 718 | memcpy(zBuf, &aPayload[offset], a); if( a==amt ){ return SQLITE_OK; } offset += a; zBuf += a; amt -= a; | > | < | < < > | 775 776 777 778 779 780 781 782 783 784 785 786 787 788 789 790 791 792 793 794 795 796 797 798 799 800 801 802 803 804 805 806 807 808 809 | memcpy(zBuf, &aPayload[offset], a); if( a==amt ){ return SQLITE_OK; } offset += a; zBuf += a; amt -= a; } if( amt>0 ){ nextPage = pCur->pPage->apCell[pCur->idx].ovfl; } while( amt>0 && nextPage ){ OverflowPage *pOvfl; rc = sqlitepager_get(pCur->pBt->pPager, nextPage, &pOvfl); if( rc!=0 ){ return rc; } nextPage = pOvfl->next; if( offset<OVERFLOW_SIZE ){ int a = amt; if( a + offset > OVERFLOW_SIZE ){ a = OVERFLOW_SIZE - offset; } memcpy(zBuf, &pOvfl->aPayload[offset], a); amt -= a; zBuf += a; } offset -= OVERFLOW_SIZE; sqlitepager_unref(pOvfl); } return amt==0 ? SQLITE_OK : SQLITE_CORRUPT; } /* ** Read part of the key associated with cursor pCur. A total |
︙ | ︙ | |||
760 761 762 763 764 765 766 767 768 769 770 | pPage = pCur->pPage; assert( pPage!=0 ); if( pCur->idx >= pPage->nCell ){ return SQLITE_ERROR; } pCell = pPage->apCell[pCur->idx]; if( amt+offset > pCell->h.nKey ){ return getPayload(pCur, offset, amt, zBuf); } /* | > > | | | > > | 822 823 824 825 826 827 828 829 830 831 832 833 834 835 836 837 838 839 840 841 842 843 844 845 846 | pPage = pCur->pPage; assert( pPage!=0 ); if( pCur->idx >= pPage->nCell ){ return SQLITE_ERROR; } pCell = pPage->apCell[pCur->idx]; if( amt+offset > pCell->h.nKey ){ return SQLITE_ERROR; } return getPayload(pCur, offset, amt, zBuf); } /* ** Set *pSize to the number of bytes of data in the entry the ** cursor currently points to. Always return SQLITE_OK. ** Failure is not possible. If the cursor is not currently ** pointing to an entry (which can happen, for example, if ** the database is empty) then *pSize is set to 0. */ int sqliteBtreeDataSize(BtCursor *pCur, int *pSize){ Cell *pCell; MemPage *pPage; pPage = pCur->pPage; assert( pPage!=0 ); |
︙ | ︙ | |||
803 804 805 806 807 808 809 | if( amt==0 ) return SQLITE_OK; pPage = pCur->pPage; assert( pPage!=0 ); if( pCur->idx >= pPage->nCell ){ return SQLITE_ERROR; } pCell = pPage->apCell[pCur->idx]; | | > > | 869 870 871 872 873 874 875 876 877 878 879 880 881 882 883 884 885 | if( amt==0 ) return SQLITE_OK; pPage = pCur->pPage; assert( pPage!=0 ); if( pCur->idx >= pPage->nCell ){ return SQLITE_ERROR; } pCell = pPage->apCell[pCur->idx]; if( amt+offset > pCell->h.nData ){ return SQLITE_ERROR; } return getPayload(pCur, offset + pCell->h.nKey, amt, zBuf); } /* ** Compare the key for the entry that pCur points to against the ** given key (pKey,nKeyOrig). Put the comparison result in *pResult. ** The result is negative if pCur<pKey, zero if they are equal and |
︙ | ︙ | |||
826 827 828 829 830 831 832 | Pgno nextPage; int nKey = nKeyOrig; int n; Cell *pCell; assert( pCur->pPage ); assert( pCur->idx>=0 && pCur->idx<pCur->pPage->nCell ); | | | 894 895 896 897 898 899 900 901 902 903 904 905 906 907 908 | Pgno nextPage; int nKey = nKeyOrig; int n; Cell *pCell; assert( pCur->pPage ); assert( pCur->idx>=0 && pCur->idx<pCur->pPage->nCell ); pCell = pCur->pPage->apCell[pCur->idx]; if( nKey > pCell->h.nKey ){ nKey = pCell->h.nKey; } n = nKey; if( n>MX_LOCAL_PAYLOAD ){ n = MX_LOCAL_PAYLOAD; } |
︙ | ︙ | |||
894 895 896 897 898 899 900 | /* ** Move the cursor up to the parent page. ** ** pCur->idx is set to the cell index that contains the pointer ** to the page we are coming from. If we are coming from the ** right-most child page then pCur->idx is set to one more than | | > < < < > | | > | 962 963 964 965 966 967 968 969 970 971 972 973 974 975 976 977 978 979 980 981 982 983 984 985 986 987 988 989 990 991 992 993 994 995 996 997 998 999 1000 1001 1002 1003 1004 1005 1006 | /* ** Move the cursor up to the parent page. ** ** pCur->idx is set to the cell index that contains the pointer ** to the page we are coming from. If we are coming from the ** right-most child page then pCur->idx is set to one more than ** the largest cell index. */ static int moveToParent(BtCursor *pCur){ Pgno oldPgno; MemPage *pParent; pParent = pCur->pPage->pParent; if( pParent==0 ) return SQLITE_INTERNAL; oldPgno = sqlitepager_pagenumber(pCur->pPage); sqlitepager_ref(pParent); sqlitepager_unref(pCur->pPage); pCur->pPage = pParent; pCur->idx = pPage->nCell; for(i=0; i<pPage->nCell; i++){ if( pPage->apCell[i].h.leftChild==oldPgno ){ pCur->idx = i; break; } } return SQLITE_OK; } /* ** Move the cursor to the root page */ static int moveToRoot(BtCursor *pCur){ MemPage *pNew; int rc; rc = sqlitepager_get(pCur->pBt->pPager, 2, &pNew); if( rc ) return rc; sqlitepager_unref(pCur->pPage); pCur->pPage = pNew; pCur->idx = 0; return SQLITE_OK; } /* |
︙ | ︙ | |||
951 952 953 954 955 956 957 | } /* Move the cursor so that it points to an entry near pKey. ** Return a success code. ** ** If an exact match is not found, then the cursor is always | | | | > > > | | > | | > > | | 1019 1020 1021 1022 1023 1024 1025 1026 1027 1028 1029 1030 1031 1032 1033 1034 1035 1036 1037 1038 1039 1040 1041 1042 1043 1044 1045 1046 1047 1048 1049 | } /* Move the cursor so that it points to an entry near pKey. ** Return a success code. ** ** If an exact match is not found, then the cursor is always ** left pointing at a leaf page which would hold the entry if it ** were present. The cursor might point to an entry that comes ** before or after the key. ** ** The result of comparing the key with the entry to which the ** cursor is left pointing is stored in pCur->iMatch. The same ** value is also written to *pRes if pRes!=NULL. The meaning of ** this value is as follows: ** ** *pRes<0 The cursor is left pointing at an entry that ** is larger than pKey. ** ** *pRes==0 The cursor is left pointing at an entry that ** exactly matches pKey. ** ** *pRes>0 The cursor is left pointing at an entry that ** is smaller than pKey. */ int sqliteBtreeMoveto(BtCursor *pCur, void *pKey, int nKey, int *pRes){ int rc; rc = moveToRoot(pCur); if( rc ) return rc; for(;;){ int lwr, upr; |
︙ | ︙ | |||
991 992 993 994 995 996 997 | lwr = pCur->idx+1; }else{ upr = pCur->idx-1; } } assert( lwr==upr+1 ); if( lwr>=pPage->nCell ){ | | > | | | > | | | 1065 1066 1067 1068 1069 1070 1071 1072 1073 1074 1075 1076 1077 1078 1079 1080 1081 1082 1083 1084 1085 1086 1087 1088 1089 1090 1091 1092 1093 1094 1095 1096 1097 1098 1099 1100 1101 1102 1103 1104 1105 1106 1107 1108 1109 1110 | lwr = pCur->idx+1; }else{ upr = pCur->idx-1; } } assert( lwr==upr+1 ); if( lwr>=pPage->nCell ){ chldPg = ((PageHdr*)pPage)->rightChild; }else{ chldPg = pPage->apCell[lwr]->h.leftChild; } if( chldPg==0 ){ pCur->iMatch = c; if( pRes ) *pRes = c; return SQLITE_OK; } rc = moveToChild(pCur, chldPg); if( rc ) return rc; } /* NOT REACHED */ } /* ** Advance the cursor to the next entry in the database. If ** successful and pRes!=NULL then set *pRes=0. If the cursor ** was already pointing to the last entry in the database before ** this routine was called, then set *pRes=1 if pRes!=NULL. */ int sqliteBtreeNext(BtCursor *pCur, int *pRes){ int rc; if( pCur->bSkipNext ){ pCur->bSkipNext = 0; if( pRes ) *pRes = 0; return SQLITE_OK; } pCur->idx++; if( pCur->idx>=pCur->pPage->nCell ){ if( ((PageHdr*)pPage)->rightChild ){ rc = moveToChild(pCur, ((PageHdr*)pPage)->rightChild); if( rc ) return rc; rc = moveToLeftmost(pCur); if( rc ) return rc; if( pRes ) *pRes = 0; return SQLITE_OK; } do{ |
︙ | ︙ | |||
1057 1058 1059 1060 1061 1062 1063 | ** sqlitepager_unref() on the new page when it is done. ** ** SQLITE_OK is returned on success. Any other return value indicates ** an error. *ppPage and *pPgno are undefined in the event of an error. ** Do not invoke sqlitepager_unref() on *ppPage if an error is returned. */ static int allocatePage(Btree *pBt, MemPage **ppPage, Pgno *pPgno){ | | | 1133 1134 1135 1136 1137 1138 1139 1140 1141 1142 1143 1144 1145 1146 1147 | ** sqlitepager_unref() on the new page when it is done. ** ** SQLITE_OK is returned on success. Any other return value indicates ** an error. *ppPage and *pPgno are undefined in the event of an error. ** Do not invoke sqlitepager_unref() on *ppPage if an error is returned. */ static int allocatePage(Btree *pBt, MemPage **ppPage, Pgno *pPgno){ PageOne *pPage1 = pBt->page1; if( pPage1->freeList ){ OverflowPage *pOvfl; rc = sqlitepager_write(pPage1); if( rc ) return rc; *pPgno = pPage1->freeList; rc = sqlitepager_get(pBt->pPager, pPage1->freeList, &pOvfl); if( rc ) return rc; |
︙ | ︙ | |||
1089 1090 1091 1092 1093 1094 1095 | ** Add a page of the database file to the freelist. Either pgno or ** pPage but not both may be 0. ** ** sqlitepager_unref() is NOT called for pPage. The calling routine ** needs to do that. */ static int freePage(Btree *pBt, void *pPage, Pgno pgno){ | | | 1165 1166 1167 1168 1169 1170 1171 1172 1173 1174 1175 1176 1177 1178 1179 | ** Add a page of the database file to the freelist. Either pgno or ** pPage but not both may be 0. ** ** sqlitepager_unref() is NOT called for pPage. The calling routine ** needs to do that. */ static int freePage(Btree *pBt, void *pPage, Pgno pgno){ PageOne *pPage1 = pBt->page1; OverflowPage *pOvfl = (OverflowPage*)pPage; int rc; int needOvflUnref = 0; if( pgno==0 ){ assert( pOvfl!=0 ); pgno = sqlitepager_pagenumber(pOvfl); } |
︙ | ︙ | |||
1115 1116 1117 1118 1119 1120 1121 1122 1123 1124 1125 1126 1127 1128 1129 1130 1131 1132 | if( rc ){ if( needOvflUnref ) sqlitepager_unref(pOvfl); return rc; } pOvfl->next = pPage1->freeList; pPage1->freeList = pgno; memset(pOvfl->aPayload, 0, OVERFLOW_SIZE); rc = sqlitepager_unref(pOvfl); return rc; } /* ** Erase all the data out of a cell. This involves returning overflow ** pages back the freelist. */ static int clearCell(Btree *pBt, Cell *pCell){ Pager *pPager = pBt->pPager; OverflowPage *pOvfl; | > > < | > | 1191 1192 1193 1194 1195 1196 1197 1198 1199 1200 1201 1202 1203 1204 1205 1206 1207 1208 1209 1210 1211 1212 1213 1214 1215 1216 1217 1218 1219 1220 1221 1222 1223 1224 1225 1226 1227 1228 1229 1230 1231 | if( rc ){ if( needOvflUnref ) sqlitepager_unref(pOvfl); return rc; } pOvfl->next = pPage1->freeList; pPage1->freeList = pgno; memset(pOvfl->aPayload, 0, OVERFLOW_SIZE); pPage->isInit = 0; assert( pPage->pParent==0 ); rc = sqlitepager_unref(pOvfl); return rc; } /* ** Erase all the data out of a cell. This involves returning overflow ** pages back the freelist. */ static int clearCell(Btree *pBt, Cell *pCell){ Pager *pPager = pBt->pPager; OverflowPage *pOvfl; Pgno ovfl, nextOvfl; int rc; if( pCell->h.nKey + pCell->h.nData <= MX_LOCAL_PAYLOAD ){ return SQLITE_OK; } ovfl = pCell->ovfl; pCell->ovfl = 0; while( ovfl ){ rc = sqlitepager_get(pPager, ovfl, &pOvfl); if( rc ) return rc; nextOvfl = pOvfl->next; rc = freePage(pBt, pOvfl, ovfl); if( rc ) return rc; ovfl = nextOvfl; sqlitepager_unref(pOvfl); } return SQLITE_OK; } /* |
︙ | ︙ | |||
1203 1204 1205 1206 1207 1208 1209 1210 1211 1212 1213 1214 1215 1216 1217 1218 1219 1220 1221 1222 1223 1224 | pPayload += n; } spaceLeft -= n; pSpace += n; } return SQLITE_OK; } /* ** Attempt to move N or more bytes out of the page that the cursor ** points to into the left sibling page. (The left sibling page ** contains cells that are less than the cells on this page.) The ** entry that the cursor is pointing to cannot be moved. Return ** TRUE if successful and FALSE if not. ** ** Reasons for not being successful include: ** ** (1) there is no left sibling, ** (2) we could only move N-1 bytes or less, ** (3) some kind of file I/O error occurred ** ** Note that a partial rotation may have occurred even if this routine | > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > | | 1281 1282 1283 1284 1285 1286 1287 1288 1289 1290 1291 1292 1293 1294 1295 1296 1297 1298 1299 1300 1301 1302 1303 1304 1305 1306 1307 1308 1309 1310 1311 1312 1313 1314 1315 1316 1317 1318 1319 1320 1321 1322 1323 1324 1325 1326 1327 1328 1329 1330 1331 1332 1333 1334 1335 1336 1337 1338 1339 1340 1341 1342 1343 | pPayload += n; } spaceLeft -= n; pSpace += n; } return SQLITE_OK; } /* ** Change the MemPage.pParent pointer on the page whose number is ** given in the second argument sot that MemPage.pParent holds the ** pointer in the third argument. */ static void reparentPage(Pager *pPager, Pgno pgno, MemPage *pNewParent){ MemPage *pThis; assert( pPager!=0 && pgno!=0 ); pThis = sqlitepager_lookup(pPager, pgno); if( pThis && pThis->pParent!=pNewParent ){ if( pThis->pParent ) sqlitepager_unref(pThis->pParent); pThis->pParent = pNewParent; if( pNewParent ) sqlitepager_ref(pNewParent); } } /* ** Reparent all children of the given page to be the given page. ** In other words, for every child of pPage, invoke reparentPage() ** to make sure that child knows that pPage is its parent. ** ** This routine gets called after you memcpy() one page into ** another. */ static void reparentChildPages(Pager *pPager, Page *pPage){ int i; for(i=0; i<pPage->nCell; i++){ reparentPage(pPager, pPage->apCell[i]->leftChild, pPage); } reparentPage(pPager, ((PageHdr*)pPage)->rightChild, pPage); } /* ** Attempt to move N or more bytes out of the page that the cursor ** points to into the left sibling page. (The left sibling page ** contains cells that are less than the cells on this page.) The ** entry that the cursor is pointing to cannot be moved. Return ** TRUE if successful and FALSE if not. ** ** Reasons for not being successful include: ** ** (1) there is no left sibling, ** (2) we could only move N-1 bytes or less, ** (3) some kind of file I/O error occurred ** ** Note that a partial rotation may have occurred even if this routine ** returns FALSE. Failure means we could not rotation a full N bytes. ** If it is possible to rotation some smaller number M, then the ** rotation occurs but we still return false. ** ** Example: Consider a segment of the Btree that looks like the ** figure below prior to rotation. The cursor is pointing to the ** entry *. The sort order of the entries is A B C D E * F Y. ** |
︙ | ︙ | |||
1262 1263 1264 1265 1266 1267 1268 1269 1270 1271 1272 1273 1274 1275 | ** This routine is the same as rotateLeft() except that it move data ** to the right instead of to the left. See comments on the rotateLeft() ** routine for additional information. */ static int rotateRight(BtCursor *pCur, int N){ return 0; } /* ** Split a single database page into two roughly equal-sized pages. ** ** The input is an existing page and a new Cell. The Cell might contain ** a valid Cell.h.leftChild field pointing to a child page. ** | > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > | 1373 1374 1375 1376 1377 1378 1379 1380 1381 1382 1383 1384 1385 1386 1387 1388 1389 1390 1391 1392 1393 1394 1395 1396 1397 1398 1399 1400 1401 1402 1403 1404 1405 1406 1407 1408 1409 1410 1411 1412 1413 1414 1415 1416 | ** This routine is the same as rotateLeft() except that it move data ** to the right instead of to the left. See comments on the rotateLeft() ** routine for additional information. */ static int rotateRight(BtCursor *pCur, int N){ return 0; } /* ** Append a cell onto the end of a page. ** ** The child page of the cell is reparented if pPager!=NULL. */ static void appendCell( Pager *pPager, /* The page cache. Needed for reparenting */ Cell *pSrc, /* The Cell to be copied onto a new page */ MemPage *pPage /* The page into which the cell is copied */ ){ int pc; int sz; Cell *pDest; sz = cellSize(pSrc); pc = allocateSpace(pPage, sz); assert( pc>0 ){ pDest = pPage->apCell[pPage->nCell] = &pPage->aDisk[pc]; memcpy(pDest, pSrc, sz); pDest->h.iNext = 0; if( pPage->nCell>0 ){ pPage->apCell[pPage->nCell-1]->h.iNext = pc; }else{ ((PageHdr*)pPage)->firstCell = pc; } if( pPager && pDest->h.leftChild ){ reparentPage(pPager, pDest->h.leftChild, pPage); } } /* ** Split a single database page into two roughly equal-sized pages. ** ** The input is an existing page and a new Cell. The Cell might contain ** a valid Cell.h.leftChild field pointing to a child page. ** |
︙ | ︙ | |||
1287 1288 1289 1290 1291 1292 1293 | */ static int split( BtCursor *pCur, /* A cursor pointing at a cell on the page to be split */ Cell *pNewCell, /* A new cell to add to pIn before dividing it up */ Cell *pCenter, /* Write the cell that divides the two pages here */ MemPage **ppOut /* If not NULL, put larger cells in new page at *ppOut */ ){ | > > > > > > | > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > | 1428 1429 1430 1431 1432 1433 1434 1435 1436 1437 1438 1439 1440 1441 1442 1443 1444 1445 1446 1447 1448 1449 1450 1451 1452 1453 1454 1455 1456 1457 1458 1459 1460 1461 1462 1463 1464 1465 1466 1467 1468 1469 1470 1471 1472 1473 1474 1475 1476 1477 1478 1479 1480 1481 1482 1483 1484 1485 1486 1487 1488 1489 1490 1491 1492 1493 1494 1495 1496 1497 1498 1499 1500 1501 1502 1503 1504 1505 1506 1507 1508 1509 1510 1511 1512 1513 1514 1515 1516 1517 1518 1519 1520 1521 1522 1523 1524 1525 1526 1527 | */ static int split( BtCursor *pCur, /* A cursor pointing at a cell on the page to be split */ Cell *pNewCell, /* A new cell to add to pIn before dividing it up */ Cell *pCenter, /* Write the cell that divides the two pages here */ MemPage **ppOut /* If not NULL, put larger cells in new page at *ppOut */ ){ MemPage *pLeft, *pRight; Pgno pgnoLeft, pgnoRight; PageHdr *pHdr; int rc; Pager *pPager = pCur->pBt->pPager; MemPage tempPage; /* Allocate pages to hold cells after the split and make pRight and ** pLeft point to the newly allocated pages. */ rc = allocatePage(pCur->pBt, &pLeft, &pgnoLeft); if( rc ) return rc; if( ppOut ){ rc = allocatePage(pCur->pBt, &pRight, &pgnoRight); if( rc ){ freePage(pCur->pBt, pLeft, pgnoLeft); return rc; } *ppOut = pRight; }else{ *ppOut = tempPage; } /* Copy the smaller cells from the original page into the left page ** of the split. */ zeroPage(pLeft); if( pCur->idx==0 && pCur->match>0 ){ appendCell(pPager, pNewCell, pLeft); } do{ assert( i<pPage->nCell ); appendCell(pPager, pPage->apCell[i++], pLeft); if( pCur->idx==i && pCur->iMatch>0 ){ appendCell(pPager, pNewCell, Left); } }while( pc < SQLITE_PAGE_SIZE/2 ); /* Copy the middle entry into *pCenter */ assert( i<pPage->nCell ); memcpy(pCenter, pPage->aCell[i], cellSize(pPage->aCell[i])); i++; pHdr = (PageHdr*)pLeft; pHdr->rightChild = pCenter->h.leftChild; if( pHdr->rightChild ){ reparentPage(pPager, pHdr->rightChild, pLeft); } pCenter->h.leftChild = pgnoLeft; /* Copy the larger cells from the original page into the right ** page of the split */ zeroPage(pRight); while( i<pPage->nCell ){ appendCell(0, pPage->apCell[i++], pRight); } /* If ppOut==NULL then copy the temporary right page over top of ** the original input page. */ if( ppOut==0 ){ pRight->pParent = pPage->pParent; pRight->isInit = 1; memcpy(pPage, pRight, sizeof(*pPage)); } reparentChildPages(pPager, pPage); } /* ** Unlink a cell from a database page. Add the space used by the cell ** back to the freelist for the database page on which the cell used to ** reside. ** ** This operation overwrites the cell header and content. */ static void unlinkCell(BtCursor *pCur){ MemPage *pPage; /* Page containing cell to be unlinked */ int idx; /* The index of the cell to be unlinked */ Cell *pCell; /* Pointer to the cell to be unlinked */ u16 *piCell; /* iNext pointer from prior cell */ int iCell; /* Index in pPage->aDisk[] of cell to be unlinked */ int i; /* Loop counter */ pPage = pCur->pPage; sqlitepager_write(pPage); idx = pCur->idx; pCell = pPage->apCell[idx]; if( idx==0 ){ piCell = &pPage->pHdr->firstCell; }else{ piCell = &pPage->apCell[idx-1]->h.iNext; } |
︙ | ︙ | |||
1329 1330 1331 1332 1333 1334 1335 | /* ** Add a Cell to a database page at the spot indicated by the cursor. ** ** With this routine, we know that the Cell pNewCell will fit into the ** database page that pCur points to. The calling routine has made ** sure it will fit. All this routine needs to do is add the Cell ** to the page. The addToPage() routine should be used for cases | | | 1537 1538 1539 1540 1541 1542 1543 1544 1545 1546 1547 1548 1549 1550 1551 | /* ** Add a Cell to a database page at the spot indicated by the cursor. ** ** With this routine, we know that the Cell pNewCell will fit into the ** database page that pCur points to. The calling routine has made ** sure it will fit. All this routine needs to do is add the Cell ** to the page. The addToPage() routine should be used for cases ** were it is not known if the new cell will fit. ** ** The new cell is added to the page either before or after the cell ** to which the cursor is pointing. The new cell is added before ** the cursor cell if pCur->iMatch>0 and the new cell is added after ** the cursor cell if pCur->iMatch<0. pCur->iMatch should have been set ** by a prior call to sqliteBtreeMoveto() where the key was the key ** of the cell being inserted. If sqliteBtreeMoveto() ended up on a |
︙ | ︙ | |||
1403 1404 1405 1406 1407 1408 1409 1410 1411 1412 1413 1414 1415 1416 1417 1418 1419 1420 1421 1422 1423 1424 | */ static int addToPage(BtCursor *pCur, Cell *pNewCell){ Cell tempCell; Cell centerCell; for(;;){ MemPage *pPage = pCur->pPage; int sz = cellSize(pNewCell); if( sz<=pPage->nFree ){ insertCell(pCur, pNewCell); return SQLITE_OK; } if( pPage->pParent==0 ){ MemPage *pRight; PageHdr *pHdr; FreeBlk *pFBlk; int pc; rc = split(pCur, pNewCell, ¢erCell, &pRight); if( rc ) return rc; pHdr = pPage->pHdr; pHdr->right = sqlitepager_pagenumber(pRight); sqlitepager_unref(pRight); | > > | | 1611 1612 1613 1614 1615 1616 1617 1618 1619 1620 1621 1622 1623 1624 1625 1626 1627 1628 1629 1630 1631 1632 1633 1634 1635 1636 1637 1638 1639 1640 1641 1642 | */ static int addToPage(BtCursor *pCur, Cell *pNewCell){ Cell tempCell; Cell centerCell; for(;;){ MemPage *pPage = pCur->pPage; rc = sqlitepager_write(pPage); if( rc ) return rc; int sz = cellSize(pNewCell); if( sz<=pPage->nFree ){ insertCell(pCur, pNewCell); return SQLITE_OK; } if( pPage->pParent==0 ){ MemPage *pRight; PageHdr *pHdr; FreeBlk *pFBlk; int pc; rc = split(pCur, pNewCell, ¢erCell, &pRight); if( rc ) return rc; pHdr = pPage->pHdr; pHdr->right = sqlitepager_pagenumber(pRight); sqlitepager_unref(pRight); pHdr->firstCell = pc = sizeof(*pHdr); sz = cellSize(¢erCell); memcpy(&pPage->aDisk[pc], ¢erCell, sz); pc += sz; pHdr->firstFree = pc; pFBlk = (FreeBlk*)&pPage->aDisk[pc]; pFBlk->iSize = SQLITE_PAGE_SIZE - pc; pFBlk->iNext = 0; |
︙ | ︙ | |||
1461 1462 1463 1464 1465 1466 1467 1468 1469 1470 1471 1472 1473 1474 1475 1476 1477 1478 1479 1480 | Cell newCell; int rc; int loc; MemPage *pPage; Btree *pBt = pCur->pBt; rc = sqliteBtreeMoveTo(pCur, pKey, nKey, &loc); if( rc ) return rc; rc = fillInCell(pBt, &newCell, pKey, nKey, pData, nData); if( rc ) return rc; if( loc==0 ){ newCell.h.leftChild = pCur->pPage->apCell[pCur->idx]->h.leftChild; rc = clearCell(pBt, pCur->pPage->apCell[pCur->idx]); if( rc ) return rc; unlinkCell(pCur); } return addToPage(pCur, &newCell); } /* | > > | | > > > > > > > > > > > > > > > > > | | > > > > > > > > > > > | | | | | | | | > > > > > | | | | | 1671 1672 1673 1674 1675 1676 1677 1678 1679 1680 1681 1682 1683 1684 1685 1686 1687 1688 1689 1690 1691 1692 1693 1694 1695 1696 1697 1698 1699 1700 1701 1702 1703 1704 1705 1706 1707 1708 1709 1710 1711 1712 1713 1714 1715 1716 1717 1718 1719 1720 1721 1722 1723 1724 1725 1726 1727 1728 1729 1730 1731 1732 1733 1734 1735 1736 1737 1738 1739 1740 1741 1742 1743 1744 1745 1746 1747 1748 1749 1750 1751 1752 1753 1754 1755 1756 1757 1758 1759 1760 1761 1762 1763 1764 1765 1766 1767 1768 1769 1770 1771 1772 1773 1774 1775 1776 1777 1778 1779 1780 1781 1782 1783 1784 1785 1786 1787 1788 1789 1790 1791 1792 1793 1794 1795 1796 1797 1798 1799 1800 1801 1802 1803 1804 1805 1806 1807 1808 1809 1810 1811 1812 1813 | Cell newCell; int rc; int loc; MemPage *pPage; Btree *pBt = pCur->pBt; rc = sqliteBtreeMoveTo(pCur, pKey, nKey, &loc); if( rc ) return rc; rc = sqlitepager_write(pCur->pPage); if( rc ) return rc; rc = fillInCell(pBt, &newCell, pKey, nKey, pData, nData); if( rc ) return rc; if( loc==0 ){ newCell.h.leftChild = pCur->pPage->apCell[pCur->idx]->h.leftChild; rc = clearCell(pBt, pCur->pPage->apCell[pCur->idx]); if( rc ) return rc; unlinkCell(pCur); } return addToPage(pCur, &newCell); } /* ** Check the page at which the cursor points to see if it is less than ** half full. If it is less than half full, then try to increase ** its fill factor by grabbing cells from siblings or by merging ** the page with siblings. */ static int refillPage(BtCursor *pCur){ MemPage *pPage; BtCursor tempCur; int rc; Pager *pPager; pPage = pCur->pPage; if( pPage->nFree < SQLITE_PAGE_SIZE/2 ){ return SQLITE_OK; } rc = sqlitepager_write(pPage); if( rc ) return rc; pPager = pCur->pBt->pPager; if( pPage->nCell==0 ){ /* The page being refilled is the root of the BTree and it has ** no entries of its own. If there is a child page, then make the ** child become the new root. */ MemPage *pChild; Pgno pgnoChild; assert( pPage->pParent==0 ); assert( sqlitepager_pagenumber(pPage)==2 ); pgnoChild = ((PageHdr*)pPage)->rightChild; if( pgnoChild==0 ){ return SQLITE_OK; } rc = sqlitepager_get(pPager, pgno, &pChild); if( rc ) return rc; memcpy(pPage, pChild, SQLITE_PAGE_SIZE); memset(&pPage->aDisk[SQLITE_PAGE_SIZE], 0, EXTRA_SIZE); freePage(pCur->pBt, pChild, pgnoChild); sqlitepager_unref(pChild); rc = initPage(pPage, 2, 0); reparentChildPages(pPager, pPage); return SQLITE_OK; } /** merge with siblings **/ /** borrow from siblings **/ } /* ** Replace the content of the cell that pCur is pointing to with the content ** in pNewContent. The pCur cell is not unlinked or moved in the Btree, ** its content is just replaced. ** ** If the size of pNewContent is greater than the current size of the ** cursor cell then the page that cursor points to might have to split. */ static int ReplaceContentOfCell(BtCursor *pCur, Cell *pNewContent){ Cell *pCell; /* The cell whose content will be changed */ Pgno pgno; /* Temporary storage for a page number */ pCell = pCur->pPage->apCell[pCur->idx]; rc = clearCell(pCur->pBt, pCell); if( rc ) return rc; pgno = pNewCell->h.leftChild; pNewCell->h.leftChild = pCell->h.leftChild; unlinkCell(pCur); rc = addToPage(pCur, pNewCell); pNewCell->h.leftChild = pgno; return rc; } /* ** Delete the entry that the cursor is pointing to. ** ** The cursor is left pointing at either the next or the previous ** entry. If the cursor is left pointing to the next entry, then ** the pCur->bSkipNext flag is set which forces the next call to ** sqliteBtreeNext() to be a no-op. That way, you can always call ** sqliteBtreeNext() after a delete and the cursor will be left ** pointing to the first entry after the deleted entry. */ int sqliteBtreeDelete(BtCursor *pCur){ MemPage *pPage = pCur->pPage; Cell *pCell; int rc; if( pCur->idx >= pPage->nCell ){ return SQLITE_ERROR; /* The cursor is not pointing to anything */ } rc = sqlitepager_write(pPage); if( rc ) return rc; pCell = pPage->apCell[pCur->idx]; if( pPage->pHdr->rightChild ){ /* The entry to be deleted is not on a leaf page. Non-leaf entries ** cannot be deleted directly because they have to be present to ** hold pointers to subpages. So what we do is look at the next ** entry in sequence. The next entry is guaranteed to exist and ** be a leaf. We copy the payload from the next entry into this ** entry, then delete the next entry. */ BtCursor origCur; CreateTemporaryCursor(pCur, &origCur); rc = sqliteBtreeNext(pCur, 0); if( rc==SQLITE_OK ){ pPage = pCur->pPage; pCell = pPage->apCell[pCur->idx]; rc = ReplaceContentOfCell(&origCur, pCell); } DestroyTemporaryCursor(&origCur); if( rc ) return rc; } rc = clearCell(pCell); if( rc ) return rc; unlinkCell(pCur->pBt, pCell); if( pCur->idx == 0 ){ pCur->bSkipNext = 1; }else{ pCur->idx--; } rc = refillPage(pCur); return rc; } |
Changes to src/btree.h.
︙ | ︙ | |||
20 21 22 23 24 25 26 | ** drh@hwaci.com ** http://www.hwaci.com/drh/ ** ************************************************************************* ** This header file defines the interface that the sqlite B-Tree file ** subsystem. ** | | > > | | 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 | ** drh@hwaci.com ** http://www.hwaci.com/drh/ ** ************************************************************************* ** This header file defines the interface that the sqlite B-Tree file ** subsystem. ** ** @(#) $Id: btree.h,v 1.3 2001/06/02 02:40:57 drh Exp $ */ typedef struct Btree Btree; typedef struct BtCursor BtCursor; int sqliteBtreeOpen(const char *zFilename, int mode, Btree **ppBtree); int sqliteBtreeClose(Btree*); int sqliteBtreeBeginTrans(Btree*); int sqliteBtreeCommit(Btree*); int sqliteBtreeRollback(Btree*); int sqliteBtreeCreateTable(Btree*, int*); int sqliteBtreeDropTable(Btree*, int); int sqliteBtreeCursor(Btree*, int iTable, BtCursor **ppCur); int sqliteBtreeMoveto(BtCursor*, void *pKey, int nKey, *pRes); int sqliteBtreeDelete(BtCursor*); int sqliteBtreeInsert(BtCursor*, void *pKey, int nKey, void *pData, int nData); int sqliteBtreeNext(BtCursor*, int *pRes); int sqliteBtreeKeySize(BtCursor*, int *pSize); int sqliteBtreeKey(BtCursor*, int offset, int amt, char *zBuf); int sqliteBtreeDataSize(BtCursor*, int *pSize); |
︙ | ︙ |
Changes to src/pager.c.
︙ | ︙ | |||
23 24 25 26 27 28 29 | ************************************************************************* ** This is the implementation of the page cache subsystem. ** ** The page cache is used to access a database file. The pager journals ** all writes in order to support rollback. Locking is used to limit ** access to one or more reader or one writer. ** | | | 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 | ************************************************************************* ** This is the implementation of the page cache subsystem. ** ** The page cache is used to access a database file. The pager journals ** all writes in order to support rollback. Locking is used to limit ** access to one or more reader or one writer. ** ** @(#) $Id: pager.c,v 1.8 2001/06/02 02:40:57 drh Exp $ */ #include "sqliteInt.h" #include "pager.h" #include <fcntl.h> #include <sys/stat.h> #include <unistd.h> #include <assert.h> |
︙ | ︙ | |||
70 71 72 73 74 75 76 77 78 79 80 81 82 83 | #define SQLITE_UNLOCK 0 #define SQLITE_READLOCK 1 #define SQLITE_WRITELOCK 2 /* ** Each in-memory image of a page begins with the following header. */ typedef struct PgHdr PgHdr; struct PgHdr { Pager *pPager; /* The pager to which this page belongs */ Pgno pgno; /* The page number for this page */ PgHdr *pNextHash, *pPrevHash; /* Hash collision chain for PgHdr.pgno */ int nRef; /* Number of users of this page */ | > > | 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 | #define SQLITE_UNLOCK 0 #define SQLITE_READLOCK 1 #define SQLITE_WRITELOCK 2 /* ** Each in-memory image of a page begins with the following header. ** This header is only visible to this pager module. The client ** code that calls pager sees only the data that follows the header. */ typedef struct PgHdr PgHdr; struct PgHdr { Pager *pPager; /* The pager to which this page belongs */ Pgno pgno; /* The page number for this page */ PgHdr *pNextHash, *pPrevHash; /* Hash collision chain for PgHdr.pgno */ int nRef; /* Number of users of this page */ |
︙ | ︙ |
Added src/test3.c.
> > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 | /* ** Copyright (c) 2001 D. Richard Hipp ** ** This program is free software; you can redistribute it and/or ** modify it under the terms of the GNU General Public ** License as published by the Free Software Foundation; either ** version 2 of the License, or (at your option) any later version. ** ** This program is distributed in the hope that it will be useful, ** but WITHOUT ANY WARRANTY; without even the implied warranty of ** MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU ** General Public License for more details. ** ** You should have received a copy of the GNU General Public ** License along with this library; if not, write to the ** Free Software Foundation, Inc., 59 Temple Place - Suite 330, ** Boston, MA 02111-1307, USA. ** ** Author contact information: ** drh@hwaci.com ** http://www.hwaci.com/drh/ ** ************************************************************************* ** Code for testing the btree.c module in SQLite. This code ** is not included in the SQLite library. It is used for automated ** testing of the SQLite library. ** ** $Id: test3.c,v 1.1 2001/06/02 02:40:57 drh Exp $ */ #include "sqliteInt.h" #include "pager.h" #include "btree.h" #include "tcl.h" #include <stdlib.h> #include <string.h> /* ** Interpret an SQLite error number */ static char *errorName(int rc){ char *zName; switch( rc ){ case SQLITE_OK: zName = "SQLITE_OK"; break; case SQLITE_ERROR: zName = "SQLITE_ERROR"; break; case SQLITE_INTERNAL: zName = "SQLITE_INTERNAL"; break; case SQLITE_PERM: zName = "SQLITE_PERM"; break; case SQLITE_ABORT: zName = "SQLITE_ABORT"; break; case SQLITE_BUSY: zName = "SQLITE_BUSY"; break; case SQLITE_NOMEM: zName = "SQLITE_NOMEM"; break; case SQLITE_READONLY: zName = "SQLITE_READONLY"; break; case SQLITE_INTERRUPT: zName = "SQLITE_INTERRUPT"; break; case SQLITE_IOERR: zName = "SQLITE_IOERR"; break; case SQLITE_CORRUPT: zName = "SQLITE_CORRUPT"; break; case SQLITE_NOTFOUND: zName = "SQLITE_NOTFOUND"; break; case SQLITE_FULL: zName = "SQLITE_FULL"; break; case SQLITE_CANTOPEN: zName = "SQLITE_CANTOPEN"; break; case SQLITE_PROTOCOL: zName = "SQLITE_PROTOCOL"; break; default: zName = "SQLITE_Unknown"; break; } return zName; } /* ** Usage: btree_open FILENAME ** ** Open a new database */ static int btree_open( void *NotUsed, Tcl_Interp *interp, /* The TCL interpreter that invoked this command */ int argc, /* Number of arguments */ char **argv /* Text of each argument */ ){ BTree *pBt; int nPage; int rc; char zBuf[100]; if( argc!=2 ){ Tcl_AppendResult(interp, "wrong # args: should be \"", argv[0], " FILENAME\"", 0); return TCL_ERROR; } rc = sqliteBtreeOpen(argv[1], 0666, &pBt); if( rc!=SQLITE_OK ){ Tcl_AppendResult(interp, errorName(rc), 0); return TCL_ERROR; } sprintf(zBuf,"0x%x",(int)pBt); Tcl_AppendResult(interp, zBuf, 0); return TCL_OK; } /* ** Usage: btree_close ID ** ** Close the given database. */ static int btree_close( void *NotUsed, Tcl_Interp *interp, /* The TCL interpreter that invoked this command */ int argc, /* Number of arguments */ char **argv /* Text of each argument */ ){ Btree *pBt; int rc; if( argc!=2 ){ Tcl_AppendResult(interp, "wrong # args: should be \"", argv[0], " ID\"", 0); return TCL_ERROR; } if( Tcl_GetInt(interp, argv[1], (int*)&pBt) ) return TCL_ERROR; rc = sqliteBtreeClose(pBt); if( rc!=SQLITE_OK ){ Tcl_AppendResult(interp, errorName(rc), 0); return TCL_ERROR; } return TCL_OK; } /* ** Usage: btree_begin_transaction ID ** ** Start a new transaction */ static int btree_begin_transaction( void *NotUsed, Tcl_Interp *interp, /* The TCL interpreter that invoked this command */ int argc, /* Number of arguments */ char **argv /* Text of each argument */ ){ Btree *pBt; int rc; if( argc!=2 ){ Tcl_AppendResult(interp, "wrong # args: should be \"", argv[0], " ID\"", 0); return TCL_ERROR; } if( Tcl_GetInt(interp, argv[1], (int*)&pBt) ) return TCL_ERROR; rc = sqliteBtreeBeginTrans(pBt); if( rc!=SQLITE_OK ){ Tcl_AppendResult(interp, errorName(rc), 0); return TCL_ERROR; } return TCL_OK; } /* ** Usage: btree_rollback ID ** ** Rollback changes */ static int btree_rollback( void *NotUsed, Tcl_Interp *interp, /* The TCL interpreter that invoked this command */ int argc, /* Number of arguments */ char **argv /* Text of each argument */ ){ Btree *pBt int rc; if( argc!=2 ){ Tcl_AppendResult(interp, "wrong # args: should be \"", argv[0], " ID\"", 0); return TCL_ERROR; } if( Tcl_GetInt(interp, argv[1], (int*)&pBt) ) return TCL_ERROR; rc = sqliteBtreeRollback(pBt); if( rc!=SQLITE_OK ){ Tcl_AppendResult(interp, errorName(rc), 0); return TCL_ERROR; } return TCL_OK; } /* ** Usage: btree_commit ID ** ** Commit all changes */ static int btree_commit( void *NotUsed, Tcl_Interp *interp, /* The TCL interpreter that invoked this command */ int argc, /* Number of arguments */ char **argv /* Text of each argument */ ){ Btree *pBt; int rc; if( argc!=2 ){ Tcl_AppendResult(interp, "wrong # args: should be \"", argv[0], " ID\"", 0); return TCL_ERROR; } if( Tcl_GetInt(interp, argv[1], (int*)&pBt) ) return TCL_ERROR; rc = sqliteBtreeCommit(pBt); if( rc!=SQLITE_OK ){ Tcl_AppendResult(interp, errorName(rc), 0); return TCL_ERROR; } return TCL_OK; } /* ** Usage: btree_create_table ID ** ** Create a new table in the database */ static int btree_create_table( void *NotUsed, Tcl_Interp *interp, /* The TCL interpreter that invoked this command */ int argc, /* Number of arguments */ char **argv /* Text of each argument */ ){ Btree *pBt; int rc, iTable; char zBuf[30]; if( argc!=2 ){ Tcl_AppendResult(interp, "wrong # args: should be \"", argv[0], " ID\"", 0); return TCL_ERROR; } if( Tcl_GetInt(interp, argv[1], (int*)&pBt) ) return TCL_ERROR; rc = sqliteBtreeCreateTable(pBt, &iTable); if( rc!=SQLITE_OK ){ Tcl_AppendResult(interp, errorName(rc), 0); return TCL_ERROR; } sprintf(zBuf, "%d", iTable); Tcl_AppendResult(interp, zBuf, 0); return TCL_OK; } /* ** Usage: btree_drop_table ID TABLENUM ** ** Delete an entire table from the database */ static int btree_drop_table( void *NotUsed, Tcl_Interp *interp, /* The TCL interpreter that invoked this command */ int argc, /* Number of arguments */ char **argv /* Text of each argument */ ){ Pager *pPager; int iTable; char zBuf[100]; if( argc!=3 ){ Tcl_AppendResult(interp, "wrong # args: should be \"", argv[0], " ID TABLENUM\"", 0); return TCL_ERROR; } if( Tcl_GetInt(interp, argv[1], (int*)&pBt) ) return TCL_ERROR; if( Tcl_GetInt(interp, argv[2], &iTable ) return TCL_ERROR; rc = sqliteBtreeDropTable(pBt, iTable); if( rc!=SQLITE_OK ){ Tcl_AppendResult(interp, errorName(rc), 0); return TCL_ERROR; } return TCL_OK; } /* ** Register commands with the TCL interpreter. */ int Sqlitetest3_Init(Tcl_Interp *interp){ Tcl_CreateCommand(interp, "btree_open", btree_open, 0, 0); Tcl_CreateCommand(interp, "btree_close", btree_close, 0, 0); Tcl_CreateCommand(interp, "btree_begin_transaction", btree_begin_transaction, 0, 0); Tcl_CreateCommand(interp, "btree_commit", btree_commit, 0, 0); Tcl_CreateCommand(interp, "btree_rollback", btree_rollback, 0, 0); Tcl_CreateCommand(interp, "btree_create_table", btree_create_table, 0, 0); Tcl_CreateCommand(interp, "btree_drop_table", btree_drop_table, 0, 0); return TCL_OK; } |