SQLite

/*
** 2004 April 6
**
** The author disclaims copyright to this source code.  In place of
** a legal notice, here is a blessing:
**
**    May you do good and not evil.
**    May you find forgiveness for yourself and forgive others.
**    May you share freely, never taking more than you give.
**
*************************************************************************
** $Id: btree.c,v 1.694 2009/07/21 11:52:35 drh Exp $
**
** This file implements a external (disk-based) database using BTrees.
** See the header comment on "btreeInt.h" for additional information.
** Including a description of file format and an overview of operation.
*/
#include "btreeInt.h"

  
  assert( sqlite3PagerIswriteable(pPage->pDbPage) );
  assert( pPage->pBt );
  assert( sqlite3_mutex_held(pPage->pBt->mutex) );
  assert( nByte>=0 );  /* Minimum cell size is 4 */
  assert( pPage->nFree>=nByte );
  assert( pPage->nOverflow==0 );
  assert( nByte<pPage->pBt->usableSize-8 );

  nFrag = data[hdr+7];
  assert( pPage->cellOffset == hdr + 12 - 4*pPage->leaf );
  gap = pPage->cellOffset + 2*pPage->nCell;
  top = get2byte(&data[hdr+5]);
  if( gap>top ) return SQLITE_CORRUPT_BKPT;
  testcase( gap+2==top );

    if( rc ) return rc;
    top = get2byte(&data[hdr+5]);
    assert( gap+nByte<=top );
  }


  /* Allocate memory from the gap in between the cell pointer array
  ** and the cell content area.  The btreeInitPage() call has already
  ** validated the freelist.  Given that the freelist is valid, there
  ** is no way that the allocation can extend off the end of the page.
  ** The assert() below verifies the previous sentence.
  */
  top -= nByte;
  put2byte(&data[hdr+5], top);
  assert( top+nByte <= pPage->pBt->usableSize );
  *pIdx = top;
  return SQLITE_OK;
}

/*
** Return a section of the pPage->aData to the freelist.
** The first byte of the new free block is pPage->aDisk[start]

  if( pPage ){
    pPage->isInit = 0;
  }
  releasePage(pPage);
  releasePage(pTrunk);
  return rc;
}
static void freePage(MemPage *pPage, int *pRC){
  if( (*pRC)==SQLITE_OK ){
    *pRC = freePage2(pPage->pBt, pPage, pPage->pgno);
  }
}

/*
** Free any overflow pages associated with the given Cell.
*/
static int clearCell(MemPage *pPage, unsigned char *pCell){
  BtShared *pBt = pPage->pBt;

*/
static void dropCell(MemPage *pPage, int idx, int sz, int *pRC){
  int i;          /* Loop counter */
  int pc;         /* Offset to cell content of cell being deleted */
  u8 *data;       /* pPage->aData */
  u8 *ptr;        /* Used to move bytes around within data[] */
  int rc;         /* The return code */
  int hdr;        /* Beginning of the header.  0 most pages.  100 page 1 */

  if( *pRC ) return;

  assert( idx>=0 && idx<pPage->nCell );
  assert( sz==cellSize(pPage, idx) );
  assert( sqlite3PagerIswriteable(pPage->pDbPage) );
  assert( sqlite3_mutex_held(pPage->pBt->mutex) );
  data = pPage->aData;
  ptr = &data[pPage->cellOffset + 2*idx];
  pc = get2byte(ptr);
  hdr = pPage->hdrOffset;
  testcase( pc==get2byte(&data[hdr+5]) );
  testcase( pc+sz==pPage->pBt->usableSize );
  if( pc < get2byte(&data[hdr+5]) || pc+sz > pPage->pBt->usableSize ){
    *pRC = SQLITE_CORRUPT_BKPT;
    return;
  }
  rc = freeSpace(pPage, pc, sz);
  if( rc ){
    *pRC = rc;
    return;
  }
  for(i=idx+1; i<pPage->nCell; i++, ptr+=2){
    ptr[0] = ptr[2];
    ptr[1] = ptr[3];
  }
  pPage->nCell--;
  put2byte(&data[hdr+3], pPage->nCell);
  pPage->nFree += 2;
}

/*
** Insert a new cell on pPage at cell index "i".  pCell points to the
** content of the cell.
**

    assert( sqlite3PagerIswriteable(pPage->pDbPage) );
    data = pPage->aData;
    cellOffset = pPage->cellOffset;
    end = cellOffset + 2*pPage->nCell;
    ins = cellOffset + 2*i;
    rc = allocateSpace(pPage, sz, &idx);
    if( rc ){ *pRC = rc; return; }
    /* The allocateSpace() routine guarantees the following two properties
    ** if it returns success */
    assert( idx >= end+2 );
    assert( idx+sz <= pPage->pBt->usableSize );



    pPage->nCell++;
    pPage->nFree -= (u16)(2 + sz);
    memcpy(&data[idx+nSkip], pCell+nSkip, sz-nSkip);
    if( iChild ){
      put4byte(&data[idx], iChild);
    }
    for(j=end, ptr=&data[j]; j>ins; j-=2, ptr-=2){

#ifndef SQLITE_OMIT_QUICKBALANCE
/*
** This version of balance() handles the common special case where
** a new entry is being inserted on the extreme right-end of the
** tree, in other words, when the new entry will become the largest
** entry in the tree.
**
** Instead of trying to balance the 3 right-most leaf pages, just add
** a new page to the right-hand side and put the one new entry in
** that page.  This leaves the right side of the tree somewhat
** unbalanced.  But odds are that we will be inserting new entries
** at the end soon afterwards so the nearly empty page will quickly
** fill up.  On average.
**
** pPage is the leaf page which is the right-most page in the tree.

  int rc;                              /* Return Code */
  Pgno pgnoNew;                        /* Page number of pNew */

  assert( sqlite3_mutex_held(pPage->pBt->mutex) );
  assert( sqlite3PagerIswriteable(pParent->pDbPage) );
  assert( pPage->nOverflow==1 );

  if( NEVER(pPage->nCell<=0) ) return SQLITE_CORRUPT_BKPT;

  /* Allocate a new page. This page will become the right-sibling of 
  ** pPage. Make the parent page writable, so that the new divider cell
  ** may be inserted. If both these operations are successful, proceed.
  */
  rc = allocateBtreePage(pBt, &pNew, &pgnoNew, 0, 0);

**
** Before returning, page pTo is reinitialized using btreeInitPage().
**
** The performance of this function is not critical. It is only used by 
** the balance_shallower() and balance_deeper() procedures, neither of
** which are called often under normal circumstances.
*/
static void copyNodeContent(MemPage *pFrom, MemPage *pTo, int *pRC){
  if( (*pRC)==SQLITE_OK ){
    BtShared * const pBt = pFrom->pBt;
    u8 * const aFrom = pFrom->aData;
    u8 * const aTo = pTo->aData;
    int const iFromHdr = pFrom->hdrOffset;
    int const iToHdr = ((pTo->pgno==1) ? 100 : 0);
    TESTONLY(int rc;)
    int iData;
  
  
    assert( pFrom->isInit );
    assert( pFrom->nFree>=iToHdr );
    assert( get2byte(&aFrom[iFromHdr+5])<=pBt->usableSize );
  
    /* Copy the b-tree node content from page pFrom to page pTo. */
    iData = get2byte(&aFrom[iFromHdr+5]);
    memcpy(&aTo[iData], &aFrom[iData], pBt->usableSize-iData);
    memcpy(&aTo[iToHdr], &aFrom[iFromHdr], pFrom->cellOffset + 2*pFrom->nCell);
  
    /* Reinitialize page pTo so that the contents of the MemPage structure
    ** match the new data. The initialization of pTo "cannot" fail, as the
    ** data copied from pFrom is known to be valid.  */
    pTo->isInit = 0;
    TESTONLY(rc = ) btreeInitPage(pTo);
    assert( rc==SQLITE_OK );
  
    /* If this is an auto-vacuum database, update the pointer-map entries
    ** for any b-tree or overflow pages that pTo now contains the pointers to.
    */
    if( ISAUTOVACUUM ){
      *pRC = setChildPtrmaps(pTo);
    }

  }
}

/*
** This routine redistributes cells on the iParentIdx'th child of pParent
** (hereafter "the page") and up to 2 siblings so that all pages have about the
** same amount of free space. Usually a single sibling on either side of the
** page are used in the balancing, though both siblings might come from one

      }
    }
  }

  /* Free any old pages that were not reused as new pages.
  */
  while( i<nOld ){
    freePage(apOld[i], &rc);
    if( rc ) goto balance_cleanup;
    releasePage(apOld[i]);
    apOld[i] = 0;
    i++;
  }

  /*

    ** (it must be, as it was just reconstructed using assemblePage()). This
    ** is important if the parent page happens to be page 1 of the database
    ** image.  */
    assert( nNew==1 );
    assert( apNew[0]->nFree == 
        (get2byte(&apNew[0]->aData[5])-apNew[0]->cellOffset-apNew[0]->nCell*2) 
    );
    copyNodeContent(apNew[0], pParent, &rc);
    freePage(apNew[0], &rc);

  }else if( ISAUTOVACUUM ){
    /* Fix the pointer-map entries for all the cells that were shifted around. 
    ** There are several different types of pointer-map entries that need to
    ** be dealt with by this routine. Some of these have been set already, but
    ** many have not. The following is a summary:
    **
    **   1) The entries associated with new sibling pages that were not

    MemPage *pNew = apNew[0];
    MemPage *pOld = apCopy[0];
    int nOverflow = pOld->nOverflow;
    int iNextOld = pOld->nCell + nOverflow;
    int iOverflow = (nOverflow ? pOld->aOvfl[0].idx : -1);
    j = 0;                             /* Current 'old' sibling page */
    k = 0;                             /* Current 'new' sibling page */
    for(i=0; i<nCell; i++){
      int isDivider = 0;
      while( i==iNextOld ){
        /* Cell i is the cell immediately following the last cell on old
        ** sibling page j. If the siblings are not leaf pages of an
        ** intkey b-tree, then cell i was a divider cell. */
        pOld = apCopy[++j];
        iNextOld = i + !leafData + pOld->nCell + pOld->nOverflow;

  /* Make pRoot, the root page of the b-tree, writable. Allocate a new 
  ** page that will become the new right-child of pPage. Copy the contents
  ** of the node stored on pRoot into the new child page.
  */
  rc = sqlite3PagerWrite(pRoot->pDbPage);
  if( rc==SQLITE_OK ){
    rc = allocateBtreePage(pBt,&pChild,&pgnoChild,pRoot->pgno,0);

    copyNodeContent(pRoot, pChild, &rc);
    if( ISAUTOVACUUM ){
      ptrmapPut(pBt, pgnoChild, PTRMAP_BTREE, pRoot->pgno, &rc);

    }
  }
  if( rc ){
    *ppChild = 0;
    releasePage(pChild);
    return rc;
  }

    rc = clearDatabasePage(pBt, get4byte(&pPage->aData[8]), 1, pnChange);
    if( rc ) goto cleardatabasepage_out;
  }else if( pnChange ){
    assert( pPage->intKey );
    *pnChange += pPage->nCell;
  }
  if( freePageFlag ){
    freePage(pPage, &rc);
  }else if( (rc = sqlite3PagerWrite(pPage->pDbPage))==0 ){
    zeroPage(pPage, pPage->aData[0] | PTF_LEAF);
  }

cleardatabasepage_out:
  releasePage(pPage);
  return rc;

    return rc;
  }

  *piMoved = 0;

  if( iTable>1 ){
#ifdef SQLITE_OMIT_AUTOVACUUM
    freePage(pPage, &rc);
    releasePage(pPage);
#else
    if( pBt->autoVacuum ){
      Pgno maxRootPgno;
      sqlite3BtreeGetMeta(p, BTREE_LARGEST_ROOT_PAGE, &maxRootPgno);

      if( iTable==maxRootPgno ){
        /* If the table being dropped is the table with the largest root-page
        ** number in the database, put the root page on the free list. 
        */
        freePage(pPage, &rc);
        releasePage(pPage);
        if( rc!=SQLITE_OK ){
          return rc;
        }
      }else{
        /* The table being dropped does not have the largest root-page
        ** number in the database. So move the page that does into the 

        }
        rc = relocatePage(pBt, pMove, PTRMAP_ROOTPAGE, 0, iTable, 0);
        releasePage(pMove);
        if( rc!=SQLITE_OK ){
          return rc;
        }
        rc = btreeGetPage(pBt, maxRootPgno, &pMove, 0);



        freePage(pMove, &rc);
        releasePage(pMove);
        if( rc!=SQLITE_OK ){
          return rc;
        }
        *piMoved = maxRootPgno;
      }

             || PTRMAP_ISPAGE(pBt, maxRootPgno) ){
        maxRootPgno--;
      }
      assert( maxRootPgno!=PENDING_BYTE_PAGE(pBt) );

      rc = sqlite3BtreeUpdateMeta(p, 4, maxRootPgno);
    }else{
      freePage(pPage, &rc);
      releasePage(pPage);
    }
#endif
  }else{
    /* If sqlite3BtreeDropTable was called on page 1.
    ** This really never should happen except in a corrupt
    ** database.