/ Check-in [e62aab5e]
Login

Many hyperlinks are disabled.
Use anonymous login to enable hyperlinks.

Overview
Comment:Fix typos in comments. No code changes.
Downloads: Tarball | ZIP archive | SQL archive
Timelines: family | ancestors | descendants | both | trunk
Files: files | file ages | folders
SHA1:e62aab5e9290503869e1f4d5e0fefd2b4dee0a69
User & Date: peter.d.reid 2014-09-06 16:39:46
Context
2014-09-06
17:06
Fixes to os_unix.c to support database (and other) files larger than 2GiB on Android. check-in: ad7063aa user: dan tags: trunk
16:52
Merge latest trunk changes with this branch. Closed-Leaf check-in: 9dca7ce5 user: dan tags: android-large-filles
16:39
Fix typos in comments. No code changes. check-in: e62aab5e user: peter.d.reid tags: trunk
03:38
Do not record a page-size change if the attempt to change the page size failed due to an OOM error. check-in: 4d4fb197 user: drh tags: trunk
Changes
Hide Diffs Unified Diffs Ignore Whitespace Patch

Changes to src/alter.c.

170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
  int len = 0;
  char *zRet;
  sqlite3 *db = sqlite3_context_db_handle(context);

  UNUSED_PARAMETER(NotUsed);

  /* The principle used to locate the table name in the CREATE TRIGGER 
  ** statement is that the table name is the first token that is immediatedly
  ** preceded by either TK_ON or TK_DOT and immediatedly followed by one
  ** of TK_WHEN, TK_BEGIN or TK_FOR.
  */
  if( zSql ){
    do {

      if( !*zCsr ){
        /* Ran out of input before finding the table name. Return NULL. */







|
|







170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
  int len = 0;
  char *zRet;
  sqlite3 *db = sqlite3_context_db_handle(context);

  UNUSED_PARAMETER(NotUsed);

  /* The principle used to locate the table name in the CREATE TRIGGER 
  ** statement is that the table name is the first token that is immediately
  ** preceded by either TK_ON or TK_DOT and immediately followed by one
  ** of TK_WHEN, TK_BEGIN or TK_FOR.
  */
  if( zSql ){
    do {

      if( !*zCsr ){
        /* Ran out of input before finding the table name. Return NULL. */

Changes to src/analyze.c.

31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
...
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
....
1579
1580
1581
1582
1583
1584
1585
1586
1587
1588
1589
1590
1591
1592
1593
** SQLITE_ENABLE_STAT3 defined.  The functionality of sqlite_stat3
** is a superset of sqlite_stat2.  The sqlite_stat4 is an enhanced
** version of sqlite_stat3 and is only available when compiled with
** SQLITE_ENABLE_STAT4 and in SQLite versions 3.8.1 and later.  It is
** not possible to enable both STAT3 and STAT4 at the same time.  If they
** are both enabled, then STAT4 takes precedence.
**
** For most applications, sqlite_stat1 provides all the statisics required
** for the query planner to make good choices.
**
** Format of sqlite_stat1:
**
** There is normally one row per index, with the index identified by the
** name in the idx column.  The tbl column is the name of the table to
** which the index belongs.  In each such row, the stat column will be
................................................................................
**
** For indexes on ordinary rowid tables, N==K+1.  But for indexes on
** WITHOUT ROWID tables, N=K+P where P is the number of columns in the
** PRIMARY KEY of the table.  The covering index that implements the
** original WITHOUT ROWID table as N==K as a special case.
**
** This routine allocates the Stat4Accum object in heap memory. The return 
** value is a pointer to the the Stat4Accum object.  The datatype of the
** return value is BLOB, but it is really just a pointer to the Stat4Accum
** object.
*/
static void statInit(
  sqlite3_context *context,
  int argc,
  sqlite3_value **argv
................................................................................
      tRowcnt nSum = 0;         /* Number of terms contributing to sumEq */
      tRowcnt avgEq = 0;
      tRowcnt nDLt = pFinal->anDLt[iCol];

      /* Set nSum to the number of distinct (iCol+1) field prefixes that
      ** occur in the stat4 table for this index before pFinal. Set
      ** sumEq to the sum of the nEq values for column iCol for the same
      ** set (adding the value only once where there exist dupicate 
      ** prefixes).  */
      for(i=0; i<(pIdx->nSample-1); i++){
        if( aSample[i].anDLt[iCol]!=aSample[i+1].anDLt[iCol] ){
          sumEq += aSample[i].anEq[iCol];
          nSum++;
        }
      }







|







 







|







 







|







31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
...
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
....
1579
1580
1581
1582
1583
1584
1585
1586
1587
1588
1589
1590
1591
1592
1593
** SQLITE_ENABLE_STAT3 defined.  The functionality of sqlite_stat3
** is a superset of sqlite_stat2.  The sqlite_stat4 is an enhanced
** version of sqlite_stat3 and is only available when compiled with
** SQLITE_ENABLE_STAT4 and in SQLite versions 3.8.1 and later.  It is
** not possible to enable both STAT3 and STAT4 at the same time.  If they
** are both enabled, then STAT4 takes precedence.
**
** For most applications, sqlite_stat1 provides all the statistics required
** for the query planner to make good choices.
**
** Format of sqlite_stat1:
**
** There is normally one row per index, with the index identified by the
** name in the idx column.  The tbl column is the name of the table to
** which the index belongs.  In each such row, the stat column will be
................................................................................
**
** For indexes on ordinary rowid tables, N==K+1.  But for indexes on
** WITHOUT ROWID tables, N=K+P where P is the number of columns in the
** PRIMARY KEY of the table.  The covering index that implements the
** original WITHOUT ROWID table as N==K as a special case.
**
** This routine allocates the Stat4Accum object in heap memory. The return 
** value is a pointer to the Stat4Accum object.  The datatype of the
** return value is BLOB, but it is really just a pointer to the Stat4Accum
** object.
*/
static void statInit(
  sqlite3_context *context,
  int argc,
  sqlite3_value **argv
................................................................................
      tRowcnt nSum = 0;         /* Number of terms contributing to sumEq */
      tRowcnt avgEq = 0;
      tRowcnt nDLt = pFinal->anDLt[iCol];

      /* Set nSum to the number of distinct (iCol+1) field prefixes that
      ** occur in the stat4 table for this index before pFinal. Set
      ** sumEq to the sum of the nEq values for column iCol for the same
      ** set (adding the value only once where there exist duplicate 
      ** prefixes).  */
      for(i=0; i<(pIdx->nSample-1); i++){
        if( aSample[i].anDLt[iCol]!=aSample[i+1].anDLt[iCol] ){
          sumEq += aSample[i].anEq[iCol];
          nSum++;
        }
      }

Changes to src/btmutex.c.

102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
** and thus help the sqlite3BtreeLock() routine to run much faster
** in the common case.
*/
static void SQLITE_NOINLINE btreeLockCarefully(Btree *p){
  Btree *pLater;

  /* In most cases, we should be able to acquire the lock we
  ** want without having to go throught the ascending lock
  ** procedure that follows.  Just be sure not to block.
  */
  if( sqlite3_mutex_try(p->pBt->mutex)==SQLITE_OK ){
    p->pBt->db = p->db;
    p->locked = 1;
    return;
  }







|







102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
** and thus help the sqlite3BtreeLock() routine to run much faster
** in the common case.
*/
static void SQLITE_NOINLINE btreeLockCarefully(Btree *p){
  Btree *pLater;

  /* In most cases, we should be able to acquire the lock we
  ** want without having to go through the ascending lock
  ** procedure that follows.  Just be sure not to block.
  */
  if( sqlite3_mutex_try(p->pBt->mutex)==SQLITE_OK ){
    p->pBt->db = p->db;
    p->locked = 1;
    return;
  }

Changes to src/btree.c.

5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
....
1140
1141
1142
1143
1144
1145
1146
1147
1148
1149
1150
1151
1152
1153
1154
....
2597
2598
2599
2600
2601
2602
2603
2604
2605
2606
2607
2608
2609
2610
2611
....
3061
3062
3063
3064
3065
3066
3067
3068
3069
3070
3071
3072
3073
3074
3075
3076
3077
3078
3079
3080
3081
3082
3083
....
3536
3537
3538
3539
3540
3541
3542
3543
3544
3545
3546
3547
3548
3549
3550
....
3770
3771
3772
3773
3774
3775
3776
3777
3778
3779
3780
3781
3782
3783
3784
....
3987
3988
3989
3990
3991
3992
3993
3994
3995
3996
3997
3998
3999
4000
4001
....
4189
4190
4191
4192
4193
4194
4195
4196
4197
4198
4199
4200
4201
4202
4203
....
5885
5886
5887
5888
5889
5890
5891
5892
5893
5894
5895
5896
5897
5898
5899
....
6551
6552
6553
6554
6555
6556
6557
6558
6559
6560
6561
6562
6563
6564
6565
....
6946
6947
6948
6949
6950
6951
6952
6953
6954
6955
6956
6957
6958
6959
6960
....
7023
7024
7025
7026
7027
7028
7029
7030
7031
7032
7033
7034
7035
7036
7037
....
7180
7181
7182
7183
7184
7185
7186
7187
7188
7189
7190
7191
7192
7193
7194
....
7878
7879
7880
7881
7882
7883
7884
7885
7886
7887
7888
7889
7890
7891
7892
** 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.
**
*************************************************************************
** 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"

/*
** The header string that appears at the beginning of every
................................................................................
** Defragment the page given.  All Cells are moved to the
** end of the page and all free space is collected into one
** big FreeBlk that occurs in between the header and cell
** pointer array and the cell content area.
*/
static int defragmentPage(MemPage *pPage){
  int i;                     /* Loop counter */
  int pc;                    /* Address of a i-th cell */
  int hdr;                   /* Offset to the page header */
  int size;                  /* Size of a cell */
  int usableSize;            /* Number of usable bytes on a page */
  int cellOffset;            /* Offset to the cell pointer array */
  int cbrk;                  /* Offset to the cell content area */
  int nCell;                 /* Number of cells on the page */
  unsigned char *data;       /* The page data */
................................................................................
** in assert() expressions, so it is only compiled if NDEBUG is not
** defined.
**
** Only write cursors are counted if wrOnly is true.  If wrOnly is
** false then all cursors are counted.
**
** For the purposes of this routine, a cursor is any cursor that
** is capable of reading or writing to the databse.  Cursors that
** have been tripped into the CURSOR_FAULT state are not counted.
*/
static int countValidCursors(BtShared *pBt, int wrOnly){
  BtCursor *pCur;
  int r = 0;
  for(pCur=pBt->pCursor; pCur; pCur=pCur->pNext){
    if( (wrOnly==0 || (pCur->curFlags & BTCF_WriteFlag)!=0)
................................................................................

/*
** Perform a single step of an incremental-vacuum. If successful, return
** SQLITE_OK. If there is no work to do (and therefore no point in 
** calling this function again), return SQLITE_DONE. Or, if an error 
** occurs, return some other error code.
**
** More specificly, this function attempts to re-organize the database so 
** that the last page of the file currently in use is no longer in use.
**
** Parameter nFin is the number of pages that this database would contain
** were this function called until it returns SQLITE_DONE.
**
** If the bCommit parameter is non-zero, this function assumes that the 
** caller will keep calling incrVacuumStep() until it returns SQLITE_DONE 
** or an error. bCommit is passed true for an auto-vacuum-on-commmit 
** operation, or false for an incremental vacuum.
*/
static int incrVacuumStep(BtShared *pBt, Pgno nFin, Pgno iLastPg, int bCommit){
  Pgno nFreeList;           /* Number of pages still on the free-list */
  int rc;

  assert( sqlite3_mutex_held(pBt->mutex) );
................................................................................

  btreeEndTransaction(p);
  sqlite3BtreeLeave(p);
  return rc;
}

/*
** Start a statement subtransaction. The subtransaction can can be rolled
** back independently of the main transaction. You must start a transaction 
** before starting a subtransaction. The subtransaction is ended automatically 
** if the main transaction commits or rolls back.
**
** Statement subtransactions are used around individual SQL statements
** that are contained within a BEGIN...COMMIT block.  If a constraint
** error occurs within the statement, the effect of that one statement
................................................................................
** BtCursor.info is a cache of the information in the current cell.
** Using this cache reduces the number of calls to btreeParseCell().
**
** 2007-06-25:  There is a bug in some versions of MSVC that cause the
** compiler to crash when getCellInfo() is implemented as a macro.
** But there is a measureable speed advantage to using the macro on gcc
** (when less compiler optimizations like -Os or -O0 are used and the
** compiler is not doing agressive inlining.)  So we use a real function
** for MSVC and a macro for everything else.  Ticket #2457.
*/
#ifndef NDEBUG
  static void assertCellInfo(BtCursor *pCur){
    CellInfo info;
    int iPage = pCur->iPage;
    memset(&info, 0, sizeof(info));
................................................................................
** Data is read to or from the buffer pBuf.
**
** The content being read or written might appear on the main page
** or be scattered out on multiple overflow pages.
**
** If the current cursor entry uses one or more overflow pages and the
** eOp argument is not 2, this function may allocate space for and lazily 
** popluates the overflow page-list cache array (BtCursor.aOverflow). 
** Subsequent calls use this cache to make seeking to the supplied offset 
** more efficient.
**
** Once an overflow page-list cache has been allocated, it may be
** invalidated if some other cursor writes to the same table, or if
** the cursor is moved to a different row. Additionally, in auto-vacuum
** mode, the following events may invalidate an overflow page-list cache.
................................................................................
    return SQLITE_CORRUPT_BKPT;
  }
  return rc;
}

/*
** Read part of the key associated with cursor pCur.  Exactly
** "amt" bytes will be transfered into pBuf[].  The transfer
** begins at "offset".
**
** The caller must ensure that pCur is pointing to a valid row
** in the table.
**
** Return SQLITE_OK on success or an error code if anything goes
** wrong.  An error is returned if "offset+amt" is larger than
................................................................................
}

/*
** Add a list of cells to a page.  The page should be initially empty.
** The cells are guaranteed to fit on the page.
*/
static void assemblePage(
  MemPage *pPage,   /* The page to be assemblied */
  int nCell,        /* The number of cells to add to this page */
  u8 **apCell,      /* Pointers to cell bodies */
  u16 *aSize        /* Sizes of the cells */
){
  int i;            /* Loop counter */
  u8 *pCellptr;     /* Address of next cell pointer */
  int cellbody;     /* Address of next cell body */
................................................................................
    if( rc ) goto balance_cleanup;
    releasePage(apOld[i]);
    apOld[i] = 0;
    i++;
  }

  /*
  ** Put the new pages in accending order.  This helps to
  ** keep entries in the disk file in order so that a scan
  ** of the table is a linear scan through the file.  That
  ** in turn helps the operating system to deliver pages
  ** from the disk more rapidly.
  **
  ** An O(n^2) insertion sort algorithm is used, but since
  ** n is never more than NB (a small constant), that should
................................................................................
         && pPage->aiOvfl[0]==pPage->nCell
         && pParent->pgno!=1
         && pParent->nCell==iIdx
        ){
          /* Call balance_quick() to create a new sibling of pPage on which
          ** to store the overflow cell. balance_quick() inserts a new cell
          ** into pParent, which may cause pParent overflow. If this
          ** happens, the next interation of the do-loop will balance pParent 
          ** use either balance_nonroot() or balance_deeper(). Until this
          ** happens, the overflow cell is stored in the aBalanceQuickSpace[]
          ** buffer. 
          **
          ** The purpose of the following assert() is to check that only a
          ** single call to balance_quick() is made for each call to this
          ** function. If this were not verified, a subtle bug involving reuse
................................................................................
** For an INTKEY table, only the nKey value of the key is used.  pKey is
** ignored.  For a ZERODATA table, the pData and nData are both ignored.
**
** If the seekResult parameter is non-zero, then a successful call to
** MovetoUnpacked() to seek cursor pCur to (pKey, nKey) has already
** been performed. seekResult is the search result returned (a negative
** number if pCur points at an entry that is smaller than (pKey, nKey), or
** a positive value if pCur points at an etry that is larger than 
** (pKey, nKey)). 
**
** If the seekResult parameter is non-zero, then the caller guarantees that
** cursor pCur is pointing at the existing copy of a row that is to be
** overwritten.  If the seekResult parameter is 0, then cursor pCur may
** point to any entry or to no entry at all and so this function has to seek
** the cursor before the new key can be inserted.
................................................................................

end_insert:
  return rc;
}

/*
** Delete the entry that the cursor is pointing to.  The cursor
** is left pointing at a arbitrary location.
*/
int sqlite3BtreeDelete(BtCursor *pCur){
  Btree *p = pCur->pBtree;
  BtShared *pBt = p->pBt;              
  int rc;                              /* Return code */
  MemPage *pPage;                      /* Page to delete cell from */
  unsigned char *pCell;                /* Pointer to cell to delete */
................................................................................
  pCheck->aPgRef[iPg/8] |= (1 << (iPg & 0x07));
}


/*
** Add 1 to the reference count for page iPage.  If this is the second
** reference to the page, add an error message to pCheck->zErrMsg.
** Return 1 if there are 2 ore more references to the page and 0 if
** if this is the first reference to the page.
**
** Also check that the page number is in bounds.
*/
static int checkRef(IntegrityCk *pCheck, Pgno iPage, char *zContext){
  if( iPage==0 ) return 1;
  if( iPage>pCheck->nPage ){







|







 







|







 







|







 







|







|







 







|







 







|







 







|







 







|







 







|







 







|







 







|







 







|







 







|







 







|







5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
....
1140
1141
1142
1143
1144
1145
1146
1147
1148
1149
1150
1151
1152
1153
1154
....
2597
2598
2599
2600
2601
2602
2603
2604
2605
2606
2607
2608
2609
2610
2611
....
3061
3062
3063
3064
3065
3066
3067
3068
3069
3070
3071
3072
3073
3074
3075
3076
3077
3078
3079
3080
3081
3082
3083
....
3536
3537
3538
3539
3540
3541
3542
3543
3544
3545
3546
3547
3548
3549
3550
....
3770
3771
3772
3773
3774
3775
3776
3777
3778
3779
3780
3781
3782
3783
3784
....
3987
3988
3989
3990
3991
3992
3993
3994
3995
3996
3997
3998
3999
4000
4001
....
4189
4190
4191
4192
4193
4194
4195
4196
4197
4198
4199
4200
4201
4202
4203
....
5885
5886
5887
5888
5889
5890
5891
5892
5893
5894
5895
5896
5897
5898
5899
....
6551
6552
6553
6554
6555
6556
6557
6558
6559
6560
6561
6562
6563
6564
6565
....
6946
6947
6948
6949
6950
6951
6952
6953
6954
6955
6956
6957
6958
6959
6960
....
7023
7024
7025
7026
7027
7028
7029
7030
7031
7032
7033
7034
7035
7036
7037
....
7180
7181
7182
7183
7184
7185
7186
7187
7188
7189
7190
7191
7192
7193
7194
....
7878
7879
7880
7881
7882
7883
7884
7885
7886
7887
7888
7889
7890
7891
7892
** 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.
**
*************************************************************************
** This file implements an 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"

/*
** The header string that appears at the beginning of every
................................................................................
** Defragment the page given.  All Cells are moved to the
** end of the page and all free space is collected into one
** big FreeBlk that occurs in between the header and cell
** pointer array and the cell content area.
*/
static int defragmentPage(MemPage *pPage){
  int i;                     /* Loop counter */
  int pc;                    /* Address of the i-th cell */
  int hdr;                   /* Offset to the page header */
  int size;                  /* Size of a cell */
  int usableSize;            /* Number of usable bytes on a page */
  int cellOffset;            /* Offset to the cell pointer array */
  int cbrk;                  /* Offset to the cell content area */
  int nCell;                 /* Number of cells on the page */
  unsigned char *data;       /* The page data */
................................................................................
** in assert() expressions, so it is only compiled if NDEBUG is not
** defined.
**
** Only write cursors are counted if wrOnly is true.  If wrOnly is
** false then all cursors are counted.
**
** For the purposes of this routine, a cursor is any cursor that
** is capable of reading or writing to the database.  Cursors that
** have been tripped into the CURSOR_FAULT state are not counted.
*/
static int countValidCursors(BtShared *pBt, int wrOnly){
  BtCursor *pCur;
  int r = 0;
  for(pCur=pBt->pCursor; pCur; pCur=pCur->pNext){
    if( (wrOnly==0 || (pCur->curFlags & BTCF_WriteFlag)!=0)
................................................................................

/*
** Perform a single step of an incremental-vacuum. If successful, return
** SQLITE_OK. If there is no work to do (and therefore no point in 
** calling this function again), return SQLITE_DONE. Or, if an error 
** occurs, return some other error code.
**
** More specifically, this function attempts to re-organize the database so 
** that the last page of the file currently in use is no longer in use.
**
** Parameter nFin is the number of pages that this database would contain
** were this function called until it returns SQLITE_DONE.
**
** If the bCommit parameter is non-zero, this function assumes that the 
** caller will keep calling incrVacuumStep() until it returns SQLITE_DONE 
** or an error. bCommit is passed true for an auto-vacuum-on-commit 
** operation, or false for an incremental vacuum.
*/
static int incrVacuumStep(BtShared *pBt, Pgno nFin, Pgno iLastPg, int bCommit){
  Pgno nFreeList;           /* Number of pages still on the free-list */
  int rc;

  assert( sqlite3_mutex_held(pBt->mutex) );
................................................................................

  btreeEndTransaction(p);
  sqlite3BtreeLeave(p);
  return rc;
}

/*
** Start a statement subtransaction. The subtransaction can be rolled
** back independently of the main transaction. You must start a transaction 
** before starting a subtransaction. The subtransaction is ended automatically 
** if the main transaction commits or rolls back.
**
** Statement subtransactions are used around individual SQL statements
** that are contained within a BEGIN...COMMIT block.  If a constraint
** error occurs within the statement, the effect of that one statement
................................................................................
** BtCursor.info is a cache of the information in the current cell.
** Using this cache reduces the number of calls to btreeParseCell().
**
** 2007-06-25:  There is a bug in some versions of MSVC that cause the
** compiler to crash when getCellInfo() is implemented as a macro.
** But there is a measureable speed advantage to using the macro on gcc
** (when less compiler optimizations like -Os or -O0 are used and the
** compiler is not doing aggressive inlining.)  So we use a real function
** for MSVC and a macro for everything else.  Ticket #2457.
*/
#ifndef NDEBUG
  static void assertCellInfo(BtCursor *pCur){
    CellInfo info;
    int iPage = pCur->iPage;
    memset(&info, 0, sizeof(info));
................................................................................
** Data is read to or from the buffer pBuf.
**
** The content being read or written might appear on the main page
** or be scattered out on multiple overflow pages.
**
** If the current cursor entry uses one or more overflow pages and the
** eOp argument is not 2, this function may allocate space for and lazily 
** populates the overflow page-list cache array (BtCursor.aOverflow). 
** Subsequent calls use this cache to make seeking to the supplied offset 
** more efficient.
**
** Once an overflow page-list cache has been allocated, it may be
** invalidated if some other cursor writes to the same table, or if
** the cursor is moved to a different row. Additionally, in auto-vacuum
** mode, the following events may invalidate an overflow page-list cache.
................................................................................
    return SQLITE_CORRUPT_BKPT;
  }
  return rc;
}

/*
** Read part of the key associated with cursor pCur.  Exactly
** "amt" bytes will be transferred into pBuf[].  The transfer
** begins at "offset".
**
** The caller must ensure that pCur is pointing to a valid row
** in the table.
**
** Return SQLITE_OK on success or an error code if anything goes
** wrong.  An error is returned if "offset+amt" is larger than
................................................................................
}

/*
** Add a list of cells to a page.  The page should be initially empty.
** The cells are guaranteed to fit on the page.
*/
static void assemblePage(
  MemPage *pPage,   /* The page to be assembled */
  int nCell,        /* The number of cells to add to this page */
  u8 **apCell,      /* Pointers to cell bodies */
  u16 *aSize        /* Sizes of the cells */
){
  int i;            /* Loop counter */
  u8 *pCellptr;     /* Address of next cell pointer */
  int cellbody;     /* Address of next cell body */
................................................................................
    if( rc ) goto balance_cleanup;
    releasePage(apOld[i]);
    apOld[i] = 0;
    i++;
  }

  /*
  ** Put the new pages in ascending order.  This helps to
  ** keep entries in the disk file in order so that a scan
  ** of the table is a linear scan through the file.  That
  ** in turn helps the operating system to deliver pages
  ** from the disk more rapidly.
  **
  ** An O(n^2) insertion sort algorithm is used, but since
  ** n is never more than NB (a small constant), that should
................................................................................
         && pPage->aiOvfl[0]==pPage->nCell
         && pParent->pgno!=1
         && pParent->nCell==iIdx
        ){
          /* Call balance_quick() to create a new sibling of pPage on which
          ** to store the overflow cell. balance_quick() inserts a new cell
          ** into pParent, which may cause pParent overflow. If this
          ** happens, the next iteration of the do-loop will balance pParent 
          ** use either balance_nonroot() or balance_deeper(). Until this
          ** happens, the overflow cell is stored in the aBalanceQuickSpace[]
          ** buffer. 
          **
          ** The purpose of the following assert() is to check that only a
          ** single call to balance_quick() is made for each call to this
          ** function. If this were not verified, a subtle bug involving reuse
................................................................................
** For an INTKEY table, only the nKey value of the key is used.  pKey is
** ignored.  For a ZERODATA table, the pData and nData are both ignored.
**
** If the seekResult parameter is non-zero, then a successful call to
** MovetoUnpacked() to seek cursor pCur to (pKey, nKey) has already
** been performed. seekResult is the search result returned (a negative
** number if pCur points at an entry that is smaller than (pKey, nKey), or
** a positive value if pCur points at an entry that is larger than 
** (pKey, nKey)). 
**
** If the seekResult parameter is non-zero, then the caller guarantees that
** cursor pCur is pointing at the existing copy of a row that is to be
** overwritten.  If the seekResult parameter is 0, then cursor pCur may
** point to any entry or to no entry at all and so this function has to seek
** the cursor before the new key can be inserted.
................................................................................

end_insert:
  return rc;
}

/*
** Delete the entry that the cursor is pointing to.  The cursor
** is left pointing at an arbitrary location.
*/
int sqlite3BtreeDelete(BtCursor *pCur){
  Btree *p = pCur->pBtree;
  BtShared *pBt = p->pBt;              
  int rc;                              /* Return code */
  MemPage *pPage;                      /* Page to delete cell from */
  unsigned char *pCell;                /* Pointer to cell to delete */
................................................................................
  pCheck->aPgRef[iPg/8] |= (1 << (iPg & 0x07));
}


/*
** Add 1 to the reference count for page iPage.  If this is the second
** reference to the page, add an error message to pCheck->zErrMsg.
** Return 1 if there are 2 or more references to the page and 0 if
** if this is the first reference to the page.
**
** Also check that the page number is in bounds.
*/
static int checkRef(IntegrityCk *pCheck, Pgno iPage, char *zContext){
  if( iPage==0 ) return 1;
  if( iPage>pCheck->nPage ){

Changes to src/btreeInt.h.

5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
...
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
...
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
** 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.
**
*************************************************************************
** This file implements a external (disk-based) database using BTrees.
** For a detailed discussion of BTrees, refer to
**
**     Donald E. Knuth, THE ART OF COMPUTER PROGRAMMING, Volume 3:
**     "Sorting And Searching", pages 473-480. Addison-Wesley
**     Publishing Company, Reading, Massachusetts.
**
** The basic idea is that each page of the file contains N database
................................................................................
**      3       2      number of cells on this page
**      5       2      first byte of the cell content area
**      7       1      number of fragmented free bytes
**      8       4      Right child (the Ptr(N) value).  Omitted on leaves.
**
** The flags define the format of this btree page.  The leaf flag means that
** this page has no children.  The zerodata flag means that this page carries
** only keys and no data.  The intkey flag means that the key is a integer
** which is stored in the key size entry of the cell header rather than in
** the payload area.
**
** The cell pointer array begins on the first byte after the page header.
** The cell pointer array contains zero or more 2-byte numbers which are
** offsets from the beginning of the page to the cell content in the cell
** content area.  The cell pointers occur in sorted order.  The system strives
................................................................................
**   The table that this cursor was opened on still exists, but has been 
**   modified since the cursor was last used. The cursor position is saved
**   in variables BtCursor.pKey and BtCursor.nKey. When a cursor is in 
**   this state, restoreCursorPosition() can be called to attempt to
**   seek the cursor to the saved position.
**
** CURSOR_FAULT:
**   A unrecoverable error (an I/O error or a malloc failure) has occurred
**   on a different connection that shares the BtShared cache with this
**   cursor.  The error has left the cache in an inconsistent state.
**   Do nothing else with this cursor.  Any attempt to use the cursor
**   should return the error code stored in BtCursor.skip
*/
#define CURSOR_INVALID           0
#define CURSOR_VALID             1







|







 







|







 







|







5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
...
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
...
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
** 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.
**
*************************************************************************
** This file implements an external (disk-based) database using BTrees.
** For a detailed discussion of BTrees, refer to
**
**     Donald E. Knuth, THE ART OF COMPUTER PROGRAMMING, Volume 3:
**     "Sorting And Searching", pages 473-480. Addison-Wesley
**     Publishing Company, Reading, Massachusetts.
**
** The basic idea is that each page of the file contains N database
................................................................................
**      3       2      number of cells on this page
**      5       2      first byte of the cell content area
**      7       1      number of fragmented free bytes
**      8       4      Right child (the Ptr(N) value).  Omitted on leaves.
**
** The flags define the format of this btree page.  The leaf flag means that
** this page has no children.  The zerodata flag means that this page carries
** only keys and no data.  The intkey flag means that the key is an integer
** which is stored in the key size entry of the cell header rather than in
** the payload area.
**
** The cell pointer array begins on the first byte after the page header.
** The cell pointer array contains zero or more 2-byte numbers which are
** offsets from the beginning of the page to the cell content in the cell
** content area.  The cell pointers occur in sorted order.  The system strives
................................................................................
**   The table that this cursor was opened on still exists, but has been 
**   modified since the cursor was last used. The cursor position is saved
**   in variables BtCursor.pKey and BtCursor.nKey. When a cursor is in 
**   this state, restoreCursorPosition() can be called to attempt to
**   seek the cursor to the saved position.
**
** CURSOR_FAULT:
**   An unrecoverable error (an I/O error or a malloc failure) has occurred
**   on a different connection that shares the BtShared cache with this
**   cursor.  The error has left the cache in an inconsistent state.
**   Do nothing else with this cursor.  Any attempt to use the cursor
**   should return the error code stored in BtCursor.skip
*/
#define CURSOR_INVALID           0
#define CURSOR_VALID             1

Changes to src/build.c.

1615
1616
1617
1618
1619
1620
1621
1622
1623
1624
1625
1626
1627
1628
1629
....
1636
1637
1638
1639
1640
1641
1642
1643
1644
1645
1646
1647
1648
1649
1650
....
2650
2651
2652
2653
2654
2655
2656
2657
2658
2659
2660
2661
2662
2663
2664
....
3250
3251
3252
3253
3254
3255
3256
3257
3258
3259
3260
3261
3262
3263
3264
....
3629
3630
3631
3632
3633
3634
3635
3636
3637
3638
3639
3640
3641
3642
3643
** are appropriate for a WITHOUT ROWID table instead of a rowid table.
** Changes include:
**
**     (1)  Convert the OP_CreateTable into an OP_CreateIndex.  There is
**          no rowid btree for a WITHOUT ROWID.  Instead, the canonical
**          data storage is a covering index btree.
**     (2)  Bypass the creation of the sqlite_master table entry
**          for the PRIMARY KEY as the the primary key index is now
**          identified by the sqlite_master table entry of the table itself.
**     (3)  Set the Index.tnum of the PRIMARY KEY Index object in the
**          schema to the rootpage from the main table.
**     (4)  Set all columns of the PRIMARY KEY schema object to be NOT NULL.
**     (5)  Add all table columns to the PRIMARY KEY Index object
**          so that the PRIMARY KEY is a covering index.  The surplus
**          columns are part of KeyInfo.nXField and are not used for
................................................................................
  Index *pPk;
  int nPk;
  int i, j;
  sqlite3 *db = pParse->db;
  Vdbe *v = pParse->pVdbe;

  /* Convert the OP_CreateTable opcode that would normally create the
  ** root-page for the table into a OP_CreateIndex opcode.  The index
  ** created will become the PRIMARY KEY index.
  */
  if( pParse->addrCrTab ){
    assert( v );
    sqlite3VdbeGetOp(v, pParse->addrCrTab)->opcode = OP_CreateIndex;
  }

................................................................................
  int iSorter;                   /* Cursor opened by OpenSorter (if in use) */
  int addr1;                     /* Address of top of loop */
  int addr2;                     /* Address to jump to for next iteration */
  int tnum;                      /* Root page of index */
  int iPartIdxLabel;             /* Jump to this label to skip a row */
  Vdbe *v;                       /* Generate code into this virtual machine */
  KeyInfo *pKey;                 /* KeyInfo for index */
  int regRecord;                 /* Register holding assemblied index record */
  sqlite3 *db = pParse->db;      /* The database connection */
  int iDb = sqlite3SchemaToIndex(db, pIndex->pSchema);

#ifndef SQLITE_OMIT_AUTHORIZATION
  if( sqlite3AuthCheck(pParse, SQLITE_REINDEX, pIndex->zName, 0,
      db->aDb[iDb].zName ) ){
    return;
................................................................................
  return pRet;
}

/*
** Fill the Index.aiRowEst[] array with default information - information
** to be used when we have not run the ANALYZE command.
**
** aiRowEst[0] is suppose to contain the number of elements in the index.
** Since we do not know, guess 1 million.  aiRowEst[1] is an estimate of the
** number of rows in the table that match any particular value of the
** first column of the index.  aiRowEst[2] is an estimate of the number
** of rows that match any particular combination of the first 2 columns
** of the index.  And so forth.  It must always be the case that
*
**           aiRowEst[N]<=aiRowEst[N-1]
................................................................................
/*
** This routine is called by the parser to add a new term to the
** end of a growing FROM clause.  The "p" parameter is the part of
** the FROM clause that has already been constructed.  "p" is NULL
** if this is the first term of the FROM clause.  pTable and pDatabase
** are the name of the table and database named in the FROM clause term.
** pDatabase is NULL if the database name qualifier is missing - the
** usual case.  If the term has a alias, then pAlias points to the
** alias token.  If the term is a subquery, then pSubquery is the
** SELECT statement that the subquery encodes.  The pTable and
** pDatabase parameters are NULL for subqueries.  The pOn and pUsing
** parameters are the content of the ON and USING clauses.
**
** Return a new SrcList which encodes is the FROM with the new
** term added.







|







 







|







 







|







 







|







 







|







1615
1616
1617
1618
1619
1620
1621
1622
1623
1624
1625
1626
1627
1628
1629
....
1636
1637
1638
1639
1640
1641
1642
1643
1644
1645
1646
1647
1648
1649
1650
....
2650
2651
2652
2653
2654
2655
2656
2657
2658
2659
2660
2661
2662
2663
2664
....
3250
3251
3252
3253
3254
3255
3256
3257
3258
3259
3260
3261
3262
3263
3264
....
3629
3630
3631
3632
3633
3634
3635
3636
3637
3638
3639
3640
3641
3642
3643
** are appropriate for a WITHOUT ROWID table instead of a rowid table.
** Changes include:
**
**     (1)  Convert the OP_CreateTable into an OP_CreateIndex.  There is
**          no rowid btree for a WITHOUT ROWID.  Instead, the canonical
**          data storage is a covering index btree.
**     (2)  Bypass the creation of the sqlite_master table entry
**          for the PRIMARY KEY as the primary key index is now
**          identified by the sqlite_master table entry of the table itself.
**     (3)  Set the Index.tnum of the PRIMARY KEY Index object in the
**          schema to the rootpage from the main table.
**     (4)  Set all columns of the PRIMARY KEY schema object to be NOT NULL.
**     (5)  Add all table columns to the PRIMARY KEY Index object
**          so that the PRIMARY KEY is a covering index.  The surplus
**          columns are part of KeyInfo.nXField and are not used for
................................................................................
  Index *pPk;
  int nPk;
  int i, j;
  sqlite3 *db = pParse->db;
  Vdbe *v = pParse->pVdbe;

  /* Convert the OP_CreateTable opcode that would normally create the
  ** root-page for the table into an OP_CreateIndex opcode.  The index
  ** created will become the PRIMARY KEY index.
  */
  if( pParse->addrCrTab ){
    assert( v );
    sqlite3VdbeGetOp(v, pParse->addrCrTab)->opcode = OP_CreateIndex;
  }

................................................................................
  int iSorter;                   /* Cursor opened by OpenSorter (if in use) */
  int addr1;                     /* Address of top of loop */
  int addr2;                     /* Address to jump to for next iteration */
  int tnum;                      /* Root page of index */
  int iPartIdxLabel;             /* Jump to this label to skip a row */
  Vdbe *v;                       /* Generate code into this virtual machine */
  KeyInfo *pKey;                 /* KeyInfo for index */
  int regRecord;                 /* Register holding assembled index record */
  sqlite3 *db = pParse->db;      /* The database connection */
  int iDb = sqlite3SchemaToIndex(db, pIndex->pSchema);

#ifndef SQLITE_OMIT_AUTHORIZATION
  if( sqlite3AuthCheck(pParse, SQLITE_REINDEX, pIndex->zName, 0,
      db->aDb[iDb].zName ) ){
    return;
................................................................................
  return pRet;
}

/*
** Fill the Index.aiRowEst[] array with default information - information
** to be used when we have not run the ANALYZE command.
**
** aiRowEst[0] is supposed to contain the number of elements in the index.
** Since we do not know, guess 1 million.  aiRowEst[1] is an estimate of the
** number of rows in the table that match any particular value of the
** first column of the index.  aiRowEst[2] is an estimate of the number
** of rows that match any particular combination of the first 2 columns
** of the index.  And so forth.  It must always be the case that
*
**           aiRowEst[N]<=aiRowEst[N-1]
................................................................................
/*
** This routine is called by the parser to add a new term to the
** end of a growing FROM clause.  The "p" parameter is the part of
** the FROM clause that has already been constructed.  "p" is NULL
** if this is the first term of the FROM clause.  pTable and pDatabase
** are the name of the table and database named in the FROM clause term.
** pDatabase is NULL if the database name qualifier is missing - the
** usual case.  If the term has an alias, then pAlias points to the
** alias token.  If the term is a subquery, then pSubquery is the
** SELECT statement that the subquery encodes.  The pTable and
** pDatabase parameters are NULL for subqueries.  The pOn and pUsing
** parameters are the content of the ON and USING clauses.
**
** Return a new SrcList which encodes is the FROM with the new
** term added.

Changes to src/callback.c.

138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
/*
** Locate and return an entry from the db.aCollSeq hash table. If the entry
** specified by zName and nName is not found and parameter 'create' is
** true, then create a new entry. Otherwise return NULL.
**
** Each pointer stored in the sqlite3.aCollSeq hash table contains an
** array of three CollSeq structures. The first is the collation sequence
** prefferred for UTF-8, the second UTF-16le, and the third UTF-16be.
**
** Stored immediately after the three collation sequences is a copy of
** the collation sequence name. A pointer to this string is stored in
** each collation sequence structure.
*/
static CollSeq *findCollSeqEntry(
  sqlite3 *db,          /* Database connection */







|







138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
/*
** Locate and return an entry from the db.aCollSeq hash table. If the entry
** specified by zName and nName is not found and parameter 'create' is
** true, then create a new entry. Otherwise return NULL.
**
** Each pointer stored in the sqlite3.aCollSeq hash table contains an
** array of three CollSeq structures. The first is the collation sequence
** preferred for UTF-8, the second UTF-16le, and the third UTF-16be.
**
** Stored immediately after the three collation sequences is a copy of
** the collation sequence name. A pointer to this string is stored in
** each collation sequence structure.
*/
static CollSeq *findCollSeqEntry(
  sqlite3 *db,          /* Database connection */

Changes to src/complete.c.

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
...
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
**   (2) NORMAL    We are in the middle of statement which ends with a single
**                 semicolon.
**
**   (3) EXPLAIN   The keyword EXPLAIN has been seen at the beginning of 
**                 a statement.
**
**   (4) CREATE    The keyword CREATE has been seen at the beginning of a
**                 statement, possibly preceeded by EXPLAIN and/or followed by
**                 TEMP or TEMPORARY
**
**   (5) TRIGGER   We are in the middle of a trigger definition that must be
**                 ended by a semicolon, the keyword END, and another semicolon.
**
**   (6) SEMI      We've seen the first semicolon in the ";END;" that occurs at
**                 the end of a trigger definition.
**
**   (7) END       We've seen the ";END" of the ";END;" that occurs at the end
**                 of a trigger difinition.
**
** Transitions between states above are determined by tokens extracted
** from the input.  The following tokens are significant:
**
**   (0) tkSEMI      A semicolon.
**   (1) tkWS        Whitespace.
**   (2) tkOTHER     Any other SQL token.
................................................................................
     /* 4  CREATE: */ {    1,  4,     2,       2,      2,    4,       5,   2, },
     /* 5 TRIGGER: */ {    6,  5,     5,       5,      5,    5,       5,   5, },
     /* 6    SEMI: */ {    6,  6,     5,       5,      5,    5,       5,   7, },
     /* 7     END: */ {    1,  7,     5,       5,      5,    5,       5,   5, },
  };
#else
  /* If triggers are not supported by this compile then the statement machine
  ** used to detect the end of a statement is much simplier
  */
  static const u8 trans[3][3] = {
                     /* Token:           */
     /* State:       **  SEMI  WS  OTHER */
     /* 0 INVALID: */ {    1,  0,     2, },
     /* 1   START: */ {    1,  1,     2, },
     /* 2  NORMAL: */ {    1,  2,     2, },







|









|







 







|







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
...
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
**   (2) NORMAL    We are in the middle of statement which ends with a single
**                 semicolon.
**
**   (3) EXPLAIN   The keyword EXPLAIN has been seen at the beginning of 
**                 a statement.
**
**   (4) CREATE    The keyword CREATE has been seen at the beginning of a
**                 statement, possibly preceded by EXPLAIN and/or followed by
**                 TEMP or TEMPORARY
**
**   (5) TRIGGER   We are in the middle of a trigger definition that must be
**                 ended by a semicolon, the keyword END, and another semicolon.
**
**   (6) SEMI      We've seen the first semicolon in the ";END;" that occurs at
**                 the end of a trigger definition.
**
**   (7) END       We've seen the ";END" of the ";END;" that occurs at the end
**                 of a trigger definition.
**
** Transitions between states above are determined by tokens extracted
** from the input.  The following tokens are significant:
**
**   (0) tkSEMI      A semicolon.
**   (1) tkWS        Whitespace.
**   (2) tkOTHER     Any other SQL token.
................................................................................
     /* 4  CREATE: */ {    1,  4,     2,       2,      2,    4,       5,   2, },
     /* 5 TRIGGER: */ {    6,  5,     5,       5,      5,    5,       5,   5, },
     /* 6    SEMI: */ {    6,  6,     5,       5,      5,    5,       5,   7, },
     /* 7     END: */ {    1,  7,     5,       5,      5,    5,       5,   5, },
  };
#else
  /* If triggers are not supported by this compile then the statement machine
  ** used to detect the end of a statement is much simpler
  */
  static const u8 trans[3][3] = {
                     /* Token:           */
     /* State:       **  SEMI  WS  OTHER */
     /* 0 INVALID: */ {    1,  0,     2, },
     /* 1   START: */ {    1,  1,     2, },
     /* 2  NORMAL: */ {    1,  2,     2, },

Changes to src/date.c.

20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
** dates and times are stored as the number of days since noon
** in Greenwich on November 24, 4714 B.C. according to the Gregorian
** calendar system. 
**
** 1970-01-01 00:00:00 is JD 2440587.5
** 2000-01-01 00:00:00 is JD 2451544.5
**
** This implemention requires years to be expressed as a 4-digit number
** which means that only dates between 0000-01-01 and 9999-12-31 can
** be represented, even though julian day numbers allow a much wider
** range of dates.
**
** The Gregorian calendar system is used for all dates and times,
** even those that predate the Gregorian calendar.  Historians usually
** use the Julian calendar for dates prior to 1582-10-15 and for some







|







20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
** dates and times are stored as the number of days since noon
** in Greenwich on November 24, 4714 B.C. according to the Gregorian
** calendar system. 
**
** 1970-01-01 00:00:00 is JD 2440587.5
** 2000-01-01 00:00:00 is JD 2451544.5
**
** This implementation requires years to be expressed as a 4-digit number
** which means that only dates between 0000-01-01 and 9999-12-31 can
** be represented, even though julian day numbers allow a much wider
** range of dates.
**
** The Gregorian calendar system is used for all dates and times,
** even those that predate the Gregorian calendar.  Historians usually
** use the Julian calendar for dates prior to 1582-10-15 and for some

Changes to src/delete.c.

86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
...
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
...
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
...
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
...
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
** pWhere argument is an optional WHERE clause that restricts the
** set of rows in the view that are to be added to the ephemeral table.
*/
void sqlite3MaterializeView(
  Parse *pParse,       /* Parsing context */
  Table *pView,        /* View definition */
  Expr *pWhere,        /* Optional WHERE clause to be added */
  int iCur             /* Cursor number for ephemerial table */
){
  SelectDest dest;
  Select *pSel;
  SrcList *pFrom;
  sqlite3 *db = pParse->db;
  int iDb = sqlite3SchemaToIndex(db, pView->pSchema);
  pWhere = sqlite3ExprDup(db, pWhere, 0);
................................................................................
  int iKey;              /* Memory cell holding key of row to be deleted */
  i16 nKey;              /* Number of memory cells in the row key */
  int iEphCur = 0;       /* Ephemeral table holding all primary key values */
  int iRowSet = 0;       /* Register for rowset of rows to delete */
  int addrBypass = 0;    /* Address of jump over the delete logic */
  int addrLoop = 0;      /* Top of the delete loop */
  int addrDelete = 0;    /* Jump directly to the delete logic */
  int addrEphOpen = 0;   /* Instruction to open the Ephermeral table */
 
#ifndef SQLITE_OMIT_TRIGGER
  int isView;                  /* True if attempting to delete from a view */
  Trigger *pTrigger;           /* List of table triggers, if required */
#endif

  memset(&sContext, 0, sizeof(sContext));
................................................................................
  if( v==0 ){
    goto delete_from_cleanup;
  }
  if( pParse->nested==0 ) sqlite3VdbeCountChanges(v);
  sqlite3BeginWriteOperation(pParse, 1, iDb);

  /* If we are trying to delete from a view, realize that view into
  ** a ephemeral table.
  */
#if !defined(SQLITE_OMIT_VIEW) && !defined(SQLITE_OMIT_TRIGGER)
  if( isView ){
    sqlite3MaterializeView(pParse, pTab, pWhere, iTabCur);
    iDataCur = iIdxCur = iTabCur;
  }
#endif
................................................................................
    if( HasRowid(pTab) ){
      /* For a rowid table, initialize the RowSet to an empty set */
      pPk = 0;
      nPk = 1;
      iRowSet = ++pParse->nMem;
      sqlite3VdbeAddOp2(v, OP_Null, 0, iRowSet);
    }else{
      /* For a WITHOUT ROWID table, create an ephermeral table used to
      ** hold all primary keys for rows to be deleted. */
      pPk = sqlite3PrimaryKeyIndex(pTab);
      assert( pPk!=0 );
      nPk = pPk->nKeyCol;
      iPk = pParse->nMem+1;
      pParse->nMem += nPk;
      iEphCur = pParse->nTab++;
................................................................................
  sqlite3AuthContextPop(&sContext);
  sqlite3SrcListDelete(db, pTabList);
  sqlite3ExprDelete(db, pWhere);
  sqlite3DbFree(db, aToOpen);
  return;
}
/* Make sure "isView" and other macros defined above are undefined. Otherwise
** thely may interfere with compilation of other functions in this file
** (or in another file, if this file becomes part of the amalgamation).  */
#ifdef isView
 #undef isView
#endif
#ifdef pTrigger
 #undef pTrigger
#endif







|







 







|







 







|







 







|







 







|







86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
...
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
...
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
...
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
...
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
** pWhere argument is an optional WHERE clause that restricts the
** set of rows in the view that are to be added to the ephemeral table.
*/
void sqlite3MaterializeView(
  Parse *pParse,       /* Parsing context */
  Table *pView,        /* View definition */
  Expr *pWhere,        /* Optional WHERE clause to be added */
  int iCur             /* Cursor number for ephemeral table */
){
  SelectDest dest;
  Select *pSel;
  SrcList *pFrom;
  sqlite3 *db = pParse->db;
  int iDb = sqlite3SchemaToIndex(db, pView->pSchema);
  pWhere = sqlite3ExprDup(db, pWhere, 0);
................................................................................
  int iKey;              /* Memory cell holding key of row to be deleted */
  i16 nKey;              /* Number of memory cells in the row key */
  int iEphCur = 0;       /* Ephemeral table holding all primary key values */
  int iRowSet = 0;       /* Register for rowset of rows to delete */
  int addrBypass = 0;    /* Address of jump over the delete logic */
  int addrLoop = 0;      /* Top of the delete loop */
  int addrDelete = 0;    /* Jump directly to the delete logic */
  int addrEphOpen = 0;   /* Instruction to open the Ephemeral table */
 
#ifndef SQLITE_OMIT_TRIGGER
  int isView;                  /* True if attempting to delete from a view */
  Trigger *pTrigger;           /* List of table triggers, if required */
#endif

  memset(&sContext, 0, sizeof(sContext));
................................................................................
  if( v==0 ){
    goto delete_from_cleanup;
  }
  if( pParse->nested==0 ) sqlite3VdbeCountChanges(v);
  sqlite3BeginWriteOperation(pParse, 1, iDb);

  /* If we are trying to delete from a view, realize that view into
  ** an ephemeral table.
  */
#if !defined(SQLITE_OMIT_VIEW) && !defined(SQLITE_OMIT_TRIGGER)
  if( isView ){
    sqlite3MaterializeView(pParse, pTab, pWhere, iTabCur);
    iDataCur = iIdxCur = iTabCur;
  }
#endif
................................................................................
    if( HasRowid(pTab) ){
      /* For a rowid table, initialize the RowSet to an empty set */
      pPk = 0;
      nPk = 1;
      iRowSet = ++pParse->nMem;
      sqlite3VdbeAddOp2(v, OP_Null, 0, iRowSet);
    }else{
      /* For a WITHOUT ROWID table, create an ephemeral table used to
      ** hold all primary keys for rows to be deleted. */
      pPk = sqlite3PrimaryKeyIndex(pTab);
      assert( pPk!=0 );
      nPk = pPk->nKeyCol;
      iPk = pParse->nMem+1;
      pParse->nMem += nPk;
      iEphCur = pParse->nTab++;
................................................................................
  sqlite3AuthContextPop(&sContext);
  sqlite3SrcListDelete(db, pTabList);
  sqlite3ExprDelete(db, pWhere);
  sqlite3DbFree(db, aToOpen);
  return;
}
/* Make sure "isView" and other macros defined above are undefined. Otherwise
** they may interfere with compilation of other functions in this file
** (or in another file, if this file becomes part of the amalgamation).  */
#ifdef isView
 #undef isView
#endif
#ifdef pTrigger
 #undef pTrigger
#endif

Changes to src/expr.c.

18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
...
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
...
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
...
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
...
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
....
1537
1538
1539
1540
1541
1542
1543
1544
1545
1546
1547
1548
1549
1550
1551
....
1667
1668
1669
1670
1671
1672
1673
1674
1675
1676
1677
1678
1679
1680
1681
....
2754
2755
2756
2757
2758
2759
2760
2761
2762
2763
2764
2765
2766
2767
2768
....
3193
3194
3195
3196
3197
3198
3199
3200
3201
3202
3203
3204
3205
3206
3207
....
3548
3549
3550
3551
3552
3553
3554
3555
3556
3557
3558
3559
3560
3561
3562
....
4285
4286
4287
4288
4289
4290
4291
4292
4293
4294
4295
4296
4297
4298
4299
** Return the 'affinity' of the expression pExpr if any.
**
** If pExpr is a column, a reference to a column via an 'AS' alias,
** or a sub-select with a column as the return value, then the 
** affinity of that column is returned. Otherwise, 0x00 is returned,
** indicating no affinity for the expression.
**
** i.e. the WHERE clause expresssions in the following statements all
** have an affinity:
**
** CREATE TABLE t1(a);
** SELECT * FROM t1 WHERE a;
** SELECT a AS b FROM t1 WHERE b;
** SELECT * FROM t1 WHERE (select a from t1);
*/
................................................................................
      pRoot->flags |= EP_Collate & pLeft->flags;
    }
    exprSetHeight(pRoot);
  }
}

/*
** Allocate a Expr node which joins as many as two subtrees.
**
** One or both of the subtrees can be NULL.  Return a pointer to the new
** Expr node.  Or, if an OOM error occurs, set pParse->db->mallocFailed,
** free the subtrees and return NULL.
*/
Expr *sqlite3PExpr(
  Parse *pParse,          /* Parsing context */
................................................................................
**
** Wildcards of the form "?nnn" are assigned the number "nnn".  We make
** sure "nnn" is not too be to avoid a denial of service attack when
** the SQL statement comes from an external source.
**
** Wildcards of the form ":aaa", "@aaa", or "$aaa" are assigned the same number
** as the previous instance of the same wildcard.  Or if this is the first
** instance of the wildcard, the next sequenial variable number is
** assigned.
*/
void sqlite3ExprAssignVarNumber(Parse *pParse, Expr *pExpr){
  sqlite3 *db = pParse->db;
  const char *z;

  if( pExpr==0 ) return;
................................................................................
** return value with EP_Reduced|EP_TokenOnly.
**
** Note that with flags==EXPRDUP_REDUCE, this routines works on full-size
** (unreduced) Expr objects as they or originally constructed by the parser.
** During expression analysis, extra information is computed and moved into
** later parts of teh Expr object and that extra information might get chopped
** off if the expression is reduced.  Note also that it does not work to
** make a EXPRDUP_REDUCE copy of a reduced expression.  It is only legal
** to reduce a pristine expression tree from the parser.  The implementation
** of dupedExprStructSize() contain multiple assert() statements that attempt
** to enforce this constraint.
*/
static int dupedExprStructSize(Expr *p, int flags){
  int nSize;
  assert( flags==EXPRDUP_REDUCE || flags==0 ); /* Only one flag value allowed */
................................................................................
}

/*
** This function is similar to sqlite3ExprDup(), except that if pzBuffer 
** is not NULL then *pzBuffer is assumed to point to a buffer large enough 
** to store the copy of expression p, the copies of p->u.zToken
** (if applicable), and the copies of the p->pLeft and p->pRight expressions,
** if any. Before returning, *pzBuffer is set to the first byte passed the
** portion of the buffer copied into by this function.
*/
static Expr *exprDup(sqlite3 *db, Expr *p, int flags, u8 **pzBuffer){
  Expr *pNew = 0;                      /* Value to return */
  if( p ){
    const int isReduced = (flags&EXPRDUP_REDUCE);
    u8 *zAlloc;
................................................................................
** An existing b-tree might be used if the RHS expression pX is a simple
** subquery such as:
**
**     SELECT <column> FROM <table>
**
** If the RHS of the IN operator is a list or a more complex subquery, then
** an ephemeral table might need to be generated from the RHS and then
** pX->iTable made to point to the ephermeral table instead of an
** existing table.
**
** The inFlags parameter must contain exactly one of the bits
** IN_INDEX_MEMBERSHIP or IN_INDEX_LOOP.  If inFlags contains
** IN_INDEX_MEMBERSHIP, then the generated table will be used for a
** fast membership test.  When the IN_INDEX_LOOP bit is set, the
** IN index will be used to loop over all values of the RHS of the
................................................................................
    }
  }

  /* If no preexisting index is available for the IN clause
  ** and IN_INDEX_NOOP is an allowed reply
  ** and the RHS of the IN operator is a list, not a subquery
  ** and the RHS is not contant or has two or fewer terms,
  ** then it is not worth creating an ephermeral table to evaluate
  ** the IN operator so return IN_INDEX_NOOP.
  */
  if( eType==0
   && (inFlags & IN_INDEX_NOOP_OK)
   && !ExprHasProperty(pX, EP_xIsSelect)
   && (!sqlite3InRhsIsConstant(pX) || pX->x.pList->nExpr<=2)
  ){
................................................................................
      pDef = sqlite3FindFunction(db, zId, nId, nFarg, enc, 0);
      if( pDef==0 || pDef->xFunc==0 ){
        sqlite3ErrorMsg(pParse, "unknown function: %.*s()", nId, zId);
        break;
      }

      /* Attempt a direct implementation of the built-in COALESCE() and
      ** IFNULL() functions.  This avoids unnecessary evalation of
      ** arguments past the first non-NULL argument.
      */
      if( pDef->funcFlags & SQLITE_FUNC_COALESCE ){
        int endCoalesce = sqlite3VdbeMakeLabel(v);
        assert( nFarg>=2 );
        sqlite3ExprCode(pParse, pFarg->a[0].pExpr, target);
        for(i=1; i<nFarg; i++){
................................................................................
    sqlite3ExprCodeAtInit(pParse, pExpr, target, 0);
  }else{
    sqlite3ExprCode(pParse, pExpr, target);
  }
}

/*
** Generate code that evalutes the given expression and puts the result
** in register target.
**
** Also make a copy of the expression results into another "cache" register
** and modify the expression so that the next time it is evaluated,
** the result is a copy of the cache register.
**
** This routine is used for expressions that are used multiple 
................................................................................
**    x BETWEEN y AND z
**
** The above is equivalent to 
**
**    x>=y AND x<=z
**
** Code it as such, taking care to do the common subexpression
** elementation of x.
*/
static void exprCodeBetween(
  Parse *pParse,    /* Parsing and code generating context */
  Expr *pExpr,      /* The BETWEEN expression */
  int dest,         /* Jump here if the jump is taken */
  int jumpIfTrue,   /* Take the jump if the BETWEEN is true */
  int jumpIfNull    /* Take the jump if the BETWEEN is NULL */
................................................................................
}

/*
** Deallocate a register, making available for reuse for some other
** purpose.
**
** If a register is currently being used by the column cache, then
** the dallocation is deferred until the column cache line that uses
** the register becomes stale.
*/
void sqlite3ReleaseTempReg(Parse *pParse, int iReg){
  if( iReg && pParse->nTempReg<ArraySize(pParse->aTempReg) ){
    int i;
    struct yColCache *p;
    for(i=0, p=pParse->aColCache; i<SQLITE_N_COLCACHE; i++, p++){







|







 







|







 







|







 







|







 







|







 







|







 







|







 







|







 







|







 







|







 







|







18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
...
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
...
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
...
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
...
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
....
1537
1538
1539
1540
1541
1542
1543
1544
1545
1546
1547
1548
1549
1550
1551
....
1667
1668
1669
1670
1671
1672
1673
1674
1675
1676
1677
1678
1679
1680
1681
....
2754
2755
2756
2757
2758
2759
2760
2761
2762
2763
2764
2765
2766
2767
2768
....
3193
3194
3195
3196
3197
3198
3199
3200
3201
3202
3203
3204
3205
3206
3207
....
3548
3549
3550
3551
3552
3553
3554
3555
3556
3557
3558
3559
3560
3561
3562
....
4285
4286
4287
4288
4289
4290
4291
4292
4293
4294
4295
4296
4297
4298
4299
** Return the 'affinity' of the expression pExpr if any.
**
** If pExpr is a column, a reference to a column via an 'AS' alias,
** or a sub-select with a column as the return value, then the 
** affinity of that column is returned. Otherwise, 0x00 is returned,
** indicating no affinity for the expression.
**
** i.e. the WHERE clause expressions in the following statements all
** have an affinity:
**
** CREATE TABLE t1(a);
** SELECT * FROM t1 WHERE a;
** SELECT a AS b FROM t1 WHERE b;
** SELECT * FROM t1 WHERE (select a from t1);
*/
................................................................................
      pRoot->flags |= EP_Collate & pLeft->flags;
    }
    exprSetHeight(pRoot);
  }
}

/*
** Allocate an Expr node which joins as many as two subtrees.
**
** One or both of the subtrees can be NULL.  Return a pointer to the new
** Expr node.  Or, if an OOM error occurs, set pParse->db->mallocFailed,
** free the subtrees and return NULL.
*/
Expr *sqlite3PExpr(
  Parse *pParse,          /* Parsing context */
................................................................................
**
** Wildcards of the form "?nnn" are assigned the number "nnn".  We make
** sure "nnn" is not too be to avoid a denial of service attack when
** the SQL statement comes from an external source.
**
** Wildcards of the form ":aaa", "@aaa", or "$aaa" are assigned the same number
** as the previous instance of the same wildcard.  Or if this is the first
** instance of the wildcard, the next sequential variable number is
** assigned.
*/
void sqlite3ExprAssignVarNumber(Parse *pParse, Expr *pExpr){
  sqlite3 *db = pParse->db;
  const char *z;

  if( pExpr==0 ) return;
................................................................................
** return value with EP_Reduced|EP_TokenOnly.
**
** Note that with flags==EXPRDUP_REDUCE, this routines works on full-size
** (unreduced) Expr objects as they or originally constructed by the parser.
** During expression analysis, extra information is computed and moved into
** later parts of teh Expr object and that extra information might get chopped
** off if the expression is reduced.  Note also that it does not work to
** make an EXPRDUP_REDUCE copy of a reduced expression.  It is only legal
** to reduce a pristine expression tree from the parser.  The implementation
** of dupedExprStructSize() contain multiple assert() statements that attempt
** to enforce this constraint.
*/
static int dupedExprStructSize(Expr *p, int flags){
  int nSize;
  assert( flags==EXPRDUP_REDUCE || flags==0 ); /* Only one flag value allowed */
................................................................................
}

/*
** This function is similar to sqlite3ExprDup(), except that if pzBuffer 
** is not NULL then *pzBuffer is assumed to point to a buffer large enough 
** to store the copy of expression p, the copies of p->u.zToken
** (if applicable), and the copies of the p->pLeft and p->pRight expressions,
** if any. Before returning, *pzBuffer is set to the first byte past the
** portion of the buffer copied into by this function.
*/
static Expr *exprDup(sqlite3 *db, Expr *p, int flags, u8 **pzBuffer){
  Expr *pNew = 0;                      /* Value to return */
  if( p ){
    const int isReduced = (flags&EXPRDUP_REDUCE);
    u8 *zAlloc;
................................................................................
** An existing b-tree might be used if the RHS expression pX is a simple
** subquery such as:
**
**     SELECT <column> FROM <table>
**
** If the RHS of the IN operator is a list or a more complex subquery, then
** an ephemeral table might need to be generated from the RHS and then
** pX->iTable made to point to the ephemeral table instead of an
** existing table.
**
** The inFlags parameter must contain exactly one of the bits
** IN_INDEX_MEMBERSHIP or IN_INDEX_LOOP.  If inFlags contains
** IN_INDEX_MEMBERSHIP, then the generated table will be used for a
** fast membership test.  When the IN_INDEX_LOOP bit is set, the
** IN index will be used to loop over all values of the RHS of the
................................................................................
    }
  }

  /* If no preexisting index is available for the IN clause
  ** and IN_INDEX_NOOP is an allowed reply
  ** and the RHS of the IN operator is a list, not a subquery
  ** and the RHS is not contant or has two or fewer terms,
  ** then it is not worth creating an ephemeral table to evaluate
  ** the IN operator so return IN_INDEX_NOOP.
  */
  if( eType==0
   && (inFlags & IN_INDEX_NOOP_OK)
   && !ExprHasProperty(pX, EP_xIsSelect)
   && (!sqlite3InRhsIsConstant(pX) || pX->x.pList->nExpr<=2)
  ){
................................................................................
      pDef = sqlite3FindFunction(db, zId, nId, nFarg, enc, 0);
      if( pDef==0 || pDef->xFunc==0 ){
        sqlite3ErrorMsg(pParse, "unknown function: %.*s()", nId, zId);
        break;
      }

      /* Attempt a direct implementation of the built-in COALESCE() and
      ** IFNULL() functions.  This avoids unnecessary evaluation of
      ** arguments past the first non-NULL argument.
      */
      if( pDef->funcFlags & SQLITE_FUNC_COALESCE ){
        int endCoalesce = sqlite3VdbeMakeLabel(v);
        assert( nFarg>=2 );
        sqlite3ExprCode(pParse, pFarg->a[0].pExpr, target);
        for(i=1; i<nFarg; i++){
................................................................................
    sqlite3ExprCodeAtInit(pParse, pExpr, target, 0);
  }else{
    sqlite3ExprCode(pParse, pExpr, target);
  }
}

/*
** Generate code that evaluates the given expression and puts the result
** in register target.
**
** Also make a copy of the expression results into another "cache" register
** and modify the expression so that the next time it is evaluated,
** the result is a copy of the cache register.
**
** This routine is used for expressions that are used multiple 
................................................................................
**    x BETWEEN y AND z
**
** The above is equivalent to 
**
**    x>=y AND x<=z
**
** Code it as such, taking care to do the common subexpression
** elimination of x.
*/
static void exprCodeBetween(
  Parse *pParse,    /* Parsing and code generating context */
  Expr *pExpr,      /* The BETWEEN expression */
  int dest,         /* Jump here if the jump is taken */
  int jumpIfTrue,   /* Take the jump if the BETWEEN is true */
  int jumpIfNull    /* Take the jump if the BETWEEN is NULL */
................................................................................
}

/*
** Deallocate a register, making available for reuse for some other
** purpose.
**
** If a register is currently being used by the column cache, then
** the deallocation is deferred until the column cache line that uses
** the register becomes stale.
*/
void sqlite3ReleaseTempReg(Parse *pParse, int iReg){
  if( iReg && pParse->nTempReg<ArraySize(pParse->aTempReg) ){
    int i;
    struct yColCache *p;
    for(i=0, p=pParse->aColCache; i<SQLITE_N_COLCACHE; i++, p++){

Changes to src/fkey.c.

169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
**
**   3) No parent key columns were provided explicitly as part of the
**      foreign key definition, and the parent table does not have a
**      PRIMARY KEY, or
**
**   4) No parent key columns were provided explicitly as part of the
**      foreign key definition, and the PRIMARY KEY of the parent table 
**      consists of a a different number of columns to the child key in 
**      the child table.
**
** then non-zero is returned, and a "foreign key mismatch" error loaded
** into pParse. If an OOM error occurs, non-zero is returned and the
** pParse->db->mallocFailed flag is set.
*/
int sqlite3FkLocateIndex(







|







169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
**
**   3) No parent key columns were provided explicitly as part of the
**      foreign key definition, and the parent table does not have a
**      PRIMARY KEY, or
**
**   4) No parent key columns were provided explicitly as part of the
**      foreign key definition, and the PRIMARY KEY of the parent table 
**      consists of a different number of columns to the child key in 
**      the child table.
**
** then non-zero is returned, and a "foreign key mismatch" error loaded
** into pParse. If an OOM error occurs, non-zero is returned and the
** pParse->db->mallocFailed flag is set.
*/
int sqlite3FkLocateIndex(

Changes to src/func.c.

5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
....
1634
1635
1636
1637
1638
1639
1640
1641
1642
1643
1644
1645
1646
1647
1648
** 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.
**
*************************************************************************
** This file contains the C-language implementions for many of the SQL
** functions of SQLite.  (Some function, and in particular the date and
** time functions, are implemented separately.)
*/
#include "sqliteInt.h"
#include <stdlib.h>
#include <assert.h>
#include "vdbeInt.h"
................................................................................
  assert( &((char*)&likeInfoAlt)[1] == (char*)&likeInfoAlt.matchOne );
  assert( &((char*)&likeInfoAlt)[2] == (char*)&likeInfoAlt.matchSet );
  *pIsNocase = (pDef->funcFlags & SQLITE_FUNC_CASE)==0;
  return 1;
}

/*
** All all of the FuncDef structures in the aBuiltinFunc[] array above
** to the global function hash table.  This occurs at start-time (as
** a consequence of calling sqlite3_initialize()).
**
** After this routine runs
*/
void sqlite3RegisterGlobalFunctions(void){
  /*







|







 







|







5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
....
1634
1635
1636
1637
1638
1639
1640
1641
1642
1643
1644
1645
1646
1647
1648
** 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.
**
*************************************************************************
** This file contains the C-language implementations for many of the SQL
** functions of SQLite.  (Some function, and in particular the date and
** time functions, are implemented separately.)
*/
#include "sqliteInt.h"
#include <stdlib.h>
#include <assert.h>
#include "vdbeInt.h"
................................................................................
  assert( &((char*)&likeInfoAlt)[1] == (char*)&likeInfoAlt.matchOne );
  assert( &((char*)&likeInfoAlt)[2] == (char*)&likeInfoAlt.matchSet );
  *pIsNocase = (pDef->funcFlags & SQLITE_FUNC_CASE)==0;
  return 1;
}

/*
** All of the FuncDef structures in the aBuiltinFunc[] array above
** to the global function hash table.  This occurs at start-time (as
** a consequence of calling sqlite3_initialize()).
**
** After this routine runs
*/
void sqlite3RegisterGlobalFunctions(void){
  /*

Changes to src/global.c.

6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
**
**    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.
**
*************************************************************************
**
** This file contains definitions of global variables and contants.
*/
#include "sqliteInt.h"

/* An array to map all upper-case characters into their corresponding
** lower-case character. 
**
** SQLite only considers US-ASCII (or EBCDIC) characters.  We do not







|







6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
**
**    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.
**
*************************************************************************
**
** This file contains definitions of global variables and constants.
*/
#include "sqliteInt.h"

/* An array to map all upper-case characters into their corresponding
** lower-case character. 
**
** SQLite only considers US-ASCII (or EBCDIC) characters.  We do not

Changes to src/insert.c.

406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
...
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
....
1023
1024
1025
1026
1027
1028
1029
1030
1031
1032
1033
1034
1035
1036
1037
....
1139
1140
1141
1142
1143
1144
1145
1146
1147
1148
1149
1150
1151
1152
1153
....
1543
1544
1545
1546
1547
1548
1549
1550
1551
1552
1553
1554
1555
1556
1557
....
1668
1669
1670
1671
1672
1673
1674
1675
1676
1677
1678
1679
1680
1681
1682
....
1735
1736
1737
1738
1739
1740
1741
1742
1743
1744
1745
1746
1747
1748
1749
**         insert the select result into <table> from R..R+n
**         goto C
**      D: cleanup
**
** The 4th template is used if the insert statement takes its
** values from a SELECT but the data is being inserted into a table
** that is also read as part of the SELECT.  In the third form,
** we have to use a intermediate table to store the results of
** the select.  The template is like this:
**
**         X <- A
**         goto B
**      A: setup for the SELECT
**         loop over the tables in the SELECT
**           load value into register R..R+n
................................................................................

  /* If this is an AUTOINCREMENT table, look up the sequence number in the
  ** sqlite_sequence table and store it in memory cell regAutoinc.
  */
  regAutoinc = autoIncBegin(pParse, iDb, pTab);

  /* Allocate registers for holding the rowid of the new row,
  ** the content of the new row, and the assemblied row record.
  */
  regRowid = regIns = pParse->nMem+1;
  pParse->nMem += pTab->nCol + 1;
  if( IsVirtual(pTab) ){
    regRowid++;
    pParse->nMem++;
  }
................................................................................
  sqlite3ExprListDelete(db, pList);
  sqlite3SelectDelete(db, pSelect);
  sqlite3IdListDelete(db, pColumn);
  sqlite3DbFree(db, aRegIdx);
}

/* Make sure "isView" and other macros defined above are undefined. Otherwise
** thely may interfere with compilation of other functions in this file
** (or in another file, if this file becomes part of the amalgamation).  */
#ifdef isView
 #undef isView
#endif
#ifdef pTrigger
 #undef pTrigger
#endif
................................................................................
  Index *pIdx;         /* Pointer to one of the indices */
  Index *pPk = 0;      /* The PRIMARY KEY index */
  sqlite3 *db;         /* Database connection */
  int i;               /* loop counter */
  int ix;              /* Index loop counter */
  int nCol;            /* Number of columns */
  int onError;         /* Conflict resolution strategy */
  int j1;              /* Addresss of jump instruction */
  int seenReplace = 0; /* True if REPLACE is used to resolve INT PK conflict */
  int nPkField;        /* Number of fields in PRIMARY KEY. 1 for ROWID tables */
  int ipkTop = 0;      /* Top of the rowid change constraint check */
  int ipkBottom = 0;   /* Bottom of the rowid change constraint check */
  u8 isUpdate;         /* True if this is an UPDATE operation */
  u8 bAffinityDone = 0;  /* True if the OP_Affinity operation has been run */
  int regRowid = -1;   /* Register holding ROWID value */
................................................................................
  int appendBias,     /* True if this is likely to be an append */
  int useSeekResult   /* True to set the USESEEKRESULT flag on OP_[Idx]Insert */
){
  Vdbe *v;            /* Prepared statements under construction */
  Index *pIdx;        /* An index being inserted or updated */
  u8 pik_flags;       /* flag values passed to the btree insert */
  int regData;        /* Content registers (after the rowid) */
  int regRec;         /* Register holding assemblied record for the table */
  int i;              /* Loop counter */
  u8 bAffinityDone = 0; /* True if OP_Affinity has been run already */

  v = sqlite3GetVdbe(pParse);
  assert( v!=0 );
  assert( pTab->pSelect==0 );  /* This table is not a VIEW */
  for(i=0, pIdx=pTab->pIndex; pIdx; pIdx=pIdx->pNext, i++){
................................................................................


#ifdef SQLITE_TEST
/*
** The following global variable is incremented whenever the
** transfer optimization is used.  This is used for testing
** purposes only - to make sure the transfer optimization really
** is happening when it is suppose to.
*/
int sqlite3_xferopt_count;
#endif /* SQLITE_TEST */


#ifndef SQLITE_OMIT_XFER_OPT
/*
................................................................................

/*
** Attempt the transfer optimization on INSERTs of the form
**
**     INSERT INTO tab1 SELECT * FROM tab2;
**
** The xfer optimization transfers raw records from tab2 over to tab1.  
** Columns are not decoded and reassemblied, which greatly improves
** performance.  Raw index records are transferred in the same way.
**
** The xfer optimization is only attempted if tab1 and tab2 are compatible.
** There are lots of rules for determining compatibility - see comments
** embedded in the code for details.
**
** This routine returns TRUE if the optimization is guaranteed to be used.







|







 







|







 







|







 







|







 







|







 







|







 







|







406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
...
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
....
1023
1024
1025
1026
1027
1028
1029
1030
1031
1032
1033
1034
1035
1036
1037
....
1139
1140
1141
1142
1143
1144
1145
1146
1147
1148
1149
1150
1151
1152
1153
....
1543
1544
1545
1546
1547
1548
1549
1550
1551
1552
1553
1554
1555
1556
1557
....
1668
1669
1670
1671
1672
1673
1674
1675
1676
1677
1678
1679
1680
1681
1682
....
1735
1736
1737
1738
1739
1740
1741
1742
1743
1744
1745
1746
1747
1748
1749
**         insert the select result into <table> from R..R+n
**         goto C
**      D: cleanup
**
** The 4th template is used if the insert statement takes its
** values from a SELECT but the data is being inserted into a table
** that is also read as part of the SELECT.  In the third form,
** we have to use an intermediate table to store the results of
** the select.  The template is like this:
**
**         X <- A
**         goto B
**      A: setup for the SELECT
**         loop over the tables in the SELECT
**           load value into register R..R+n
................................................................................

  /* If this is an AUTOINCREMENT table, look up the sequence number in the
  ** sqlite_sequence table and store it in memory cell regAutoinc.
  */
  regAutoinc = autoIncBegin(pParse, iDb, pTab);

  /* Allocate registers for holding the rowid of the new row,
  ** the content of the new row, and the assembled row record.
  */
  regRowid = regIns = pParse->nMem+1;
  pParse->nMem += pTab->nCol + 1;
  if( IsVirtual(pTab) ){
    regRowid++;
    pParse->nMem++;
  }
................................................................................
  sqlite3ExprListDelete(db, pList);
  sqlite3SelectDelete(db, pSelect);
  sqlite3IdListDelete(db, pColumn);
  sqlite3DbFree(db, aRegIdx);
}

/* Make sure "isView" and other macros defined above are undefined. Otherwise
** they may interfere with compilation of other functions in this file
** (or in another file, if this file becomes part of the amalgamation).  */
#ifdef isView
 #undef isView
#endif
#ifdef pTrigger
 #undef pTrigger
#endif
................................................................................
  Index *pIdx;         /* Pointer to one of the indices */
  Index *pPk = 0;      /* The PRIMARY KEY index */
  sqlite3 *db;         /* Database connection */
  int i;               /* loop counter */
  int ix;              /* Index loop counter */
  int nCol;            /* Number of columns */
  int onError;         /* Conflict resolution strategy */
  int j1;              /* Address of jump instruction */
  int seenReplace = 0; /* True if REPLACE is used to resolve INT PK conflict */
  int nPkField;        /* Number of fields in PRIMARY KEY. 1 for ROWID tables */
  int ipkTop = 0;      /* Top of the rowid change constraint check */
  int ipkBottom = 0;   /* Bottom of the rowid change constraint check */
  u8 isUpdate;         /* True if this is an UPDATE operation */
  u8 bAffinityDone = 0;  /* True if the OP_Affinity operation has been run */
  int regRowid = -1;   /* Register holding ROWID value */
................................................................................
  int appendBias,     /* True if this is likely to be an append */
  int useSeekResult   /* True to set the USESEEKRESULT flag on OP_[Idx]Insert */
){
  Vdbe *v;            /* Prepared statements under construction */
  Index *pIdx;        /* An index being inserted or updated */
  u8 pik_flags;       /* flag values passed to the btree insert */
  int regData;        /* Content registers (after the rowid) */
  int regRec;         /* Register holding assembled record for the table */
  int i;              /* Loop counter */
  u8 bAffinityDone = 0; /* True if OP_Affinity has been run already */

  v = sqlite3GetVdbe(pParse);
  assert( v!=0 );
  assert( pTab->pSelect==0 );  /* This table is not a VIEW */
  for(i=0, pIdx=pTab->pIndex; pIdx; pIdx=pIdx->pNext, i++){
................................................................................


#ifdef SQLITE_TEST
/*
** The following global variable is incremented whenever the
** transfer optimization is used.  This is used for testing
** purposes only - to make sure the transfer optimization really
** is happening when it is supposed to.
*/
int sqlite3_xferopt_count;
#endif /* SQLITE_TEST */


#ifndef SQLITE_OMIT_XFER_OPT
/*
................................................................................

/*
** Attempt the transfer optimization on INSERTs of the form
**
**     INSERT INTO tab1 SELECT * FROM tab2;
**
** The xfer optimization transfers raw records from tab2 over to tab1.  
** Columns are not decoded and reassembled, which greatly improves
** performance.  Raw index records are transferred in the same way.
**
** The xfer optimization is only attempted if tab1 and tab2 are compatible.
** There are lots of rules for determining compatibility - see comments
** embedded in the code for details.
**
** This routine returns TRUE if the optimization is guaranteed to be used.

Changes to src/main.c.

2849
2850
2851
2852
2853
2854
2855
2856
2857
2858
2859
2860
2861
2862
2863
2864
2865
....
3165
3166
3167
3168
3169
3170
3171
3172
3173
3174
3175
3176
3177
3178
3179
** by the next COMMIT or ROLLBACK.
*/
int sqlite3_get_autocommit(sqlite3 *db){
  return db->autoCommit;
}

/*
** The following routines are subtitutes for constants SQLITE_CORRUPT,
** SQLITE_MISUSE, SQLITE_CANTOPEN, SQLITE_IOERR and possibly other error
** constants.  They server two purposes:
**
**   1.  Serve as a convenient place to set a breakpoint in a debugger
**       to detect when version error conditions occurs.
**
**   2.  Invoke sqlite3_log() to provide the source code location where
**       a low-level error is first detected.
*/
................................................................................
    ** Set the PENDING byte to the value in the argument, if X>0.
    ** Make no changes if X==0.  Return the value of the pending byte
    ** as it existing before this routine was called.
    **
    ** IMPORTANT:  Changing the PENDING byte from 0x40000000 results in
    ** an incompatible database file format.  Changing the PENDING byte
    ** while any database connection is open results in undefined and
    ** dileterious behavior.
    */
    case SQLITE_TESTCTRL_PENDING_BYTE: {
      rc = PENDING_BYTE;
#ifndef SQLITE_OMIT_WSD
      {
        unsigned int newVal = va_arg(ap, unsigned int);
        if( newVal ) sqlite3PendingByte = newVal;







|

|







 







|







2849
2850
2851
2852
2853
2854
2855
2856
2857
2858
2859
2860
2861
2862
2863
2864
2865
....
3165
3166
3167
3168
3169
3170
3171
3172
3173
3174
3175
3176
3177
3178
3179
** by the next COMMIT or ROLLBACK.
*/
int sqlite3_get_autocommit(sqlite3 *db){
  return db->autoCommit;
}

/*
** The following routines are substitutes for constants SQLITE_CORRUPT,
** SQLITE_MISUSE, SQLITE_CANTOPEN, SQLITE_IOERR and possibly other error
** constants.  They serve two purposes:
**
**   1.  Serve as a convenient place to set a breakpoint in a debugger
**       to detect when version error conditions occurs.
**
**   2.  Invoke sqlite3_log() to provide the source code location where
**       a low-level error is first detected.
*/
................................................................................
    ** Set the PENDING byte to the value in the argument, if X>0.
    ** Make no changes if X==0.  Return the value of the pending byte
    ** as it existing before this routine was called.
    **
    ** IMPORTANT:  Changing the PENDING byte from 0x40000000 results in
    ** an incompatible database file format.  Changing the PENDING byte
    ** while any database connection is open results in undefined and
    ** deleterious behavior.
    */
    case SQLITE_TESTCTRL_PENDING_BYTE: {
      rc = PENDING_BYTE;
#ifndef SQLITE_OMIT_WSD
      {
        unsigned int newVal = va_arg(ap, unsigned int);
        if( newVal ) sqlite3PendingByte = newVal;

Changes to src/mem1.c.

184
185
186
187
188
189
190
191
192
193
194
195
196
197
198

/*
** Like realloc().  Resize an allocation previously obtained from
** sqlite3MemMalloc().
**
** For this low-level interface, we know that pPrior!=0.  Cases where
** pPrior==0 while have been intercepted by higher-level routine and
** redirected to xMalloc.  Similarly, we know that nByte>0 becauses
** cases where nByte<=0 will have been intercepted by higher-level
** routines and redirected to xFree.
*/
static void *sqlite3MemRealloc(void *pPrior, int nByte){
#ifdef SQLITE_MALLOCSIZE
  void *p = SQLITE_REALLOC(pPrior, nByte);
  if( p==0 ){







|







184
185
186
187
188
189
190
191
192
193
194
195
196
197
198

/*
** Like realloc().  Resize an allocation previously obtained from
** sqlite3MemMalloc().
**
** For this low-level interface, we know that pPrior!=0.  Cases where
** pPrior==0 while have been intercepted by higher-level routine and
** redirected to xMalloc.  Similarly, we know that nByte>0 because
** cases where nByte<=0 will have been intercepted by higher-level
** routines and redirected to xFree.
*/
static void *sqlite3MemRealloc(void *pPrior, int nByte){
#ifdef SQLITE_MALLOCSIZE
  void *p = SQLITE_REALLOC(pPrior, nByte);
  if( p==0 ){

Changes to src/mem5.c.

24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
** in the build only if SQLITE_ENABLE_MEMSYS5 is defined.
**
** This memory allocator uses the following algorithm:
**
**   1.  All memory allocations sizes are rounded up to a power of 2.
**
**   2.  If two adjacent free blocks are the halves of a larger block,
**       then the two blocks are coalesed into the single larger block.
**
**   3.  New memory is allocated from the first available free block.
**
** This algorithm is described in: J. M. Robson. "Bounds for Some Functions
** Concerning Dynamic Storage Allocation". Journal of the Association for
** Computing Machinery, Volume 21, Number 8, July 1974, pages 491-499.
** 







|







24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
** in the build only if SQLITE_ENABLE_MEMSYS5 is defined.
**
** This memory allocator uses the following algorithm:
**
**   1.  All memory allocations sizes are rounded up to a power of 2.
**
**   2.  If two adjacent free blocks are the halves of a larger block,
**       then the two blocks are coalesced into the single larger block.
**
**   3.  New memory is allocated from the first available free block.
**
** This algorithm is described in: J. M. Robson. "Bounds for Some Functions
** Concerning Dynamic Storage Allocation". Journal of the Association for
** Computing Machinery, Volume 21, Number 8, July 1974, pages 491-499.
** 

Changes to src/memjournal.c.

22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
typedef struct FileChunk FileChunk;

/* Space to hold the rollback journal is allocated in increments of
** this many bytes.
**
** The size chosen is a little less than a power of two.  That way,
** the FileChunk object will have a size that almost exactly fills
** a power-of-two allocation.  This mimimizes wasted space in power-of-two
** memory allocators.
*/
#define JOURNAL_CHUNKSIZE ((int)(1024-sizeof(FileChunk*)))

/*
** The rollback journal is composed of a linked list of these structures.
*/







|







22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
typedef struct FileChunk FileChunk;

/* Space to hold the rollback journal is allocated in increments of
** this many bytes.
**
** The size chosen is a little less than a power of two.  That way,
** the FileChunk object will have a size that almost exactly fills
** a power-of-two allocation.  This minimizes wasted space in power-of-two
** memory allocators.
*/
#define JOURNAL_CHUNKSIZE ((int)(1024-sizeof(FileChunk*)))

/*
** The rollback journal is composed of a linked list of these structures.
*/

Changes to src/mutex.h.

21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
*/


/*
** Figure out what version of the code to use.  The choices are
**
**   SQLITE_MUTEX_OMIT         No mutex logic.  Not even stubs.  The
**                             mutexes implemention cannot be overridden
**                             at start-time.
**
**   SQLITE_MUTEX_NOOP         For single-threaded applications.  No
**                             mutual exclusion is provided.  But this
**                             implementation can be overridden at
**                             start-time.
**







|







21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
*/


/*
** Figure out what version of the code to use.  The choices are
**
**   SQLITE_MUTEX_OMIT         No mutex logic.  Not even stubs.  The
**                             mutexes implementation cannot be overridden
**                             at start-time.
**
**   SQLITE_MUTEX_NOOP         For single-threaded applications.  No
**                             mutual exclusion is provided.  But this
**                             implementation can be overridden at
**                             start-time.
**

Changes to src/os.h.

116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
**
** The following #defines specify the range of bytes used for locking.
** SHARED_SIZE is the number of bytes available in the pool from which
** a random byte is selected for a shared lock.  The pool of bytes for
** shared locks begins at SHARED_FIRST. 
**
** The same locking strategy and
** byte ranges are used for Unix.  This leaves open the possiblity of having
** clients on win95, winNT, and unix all talking to the same shared file
** and all locking correctly.  To do so would require that samba (or whatever
** tool is being used for file sharing) implements locks correctly between
** windows and unix.  I'm guessing that isn't likely to happen, but by
** using the same locking range we are at least open to the possibility.
**
** Locking in windows is manditory.  For this reason, we cannot store







|







116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
**
** The following #defines specify the range of bytes used for locking.
** SHARED_SIZE is the number of bytes available in the pool from which
** a random byte is selected for a shared lock.  The pool of bytes for
** shared locks begins at SHARED_FIRST. 
**
** The same locking strategy and
** byte ranges are used for Unix.  This leaves open the possibility of having
** clients on win95, winNT, and unix all talking to the same shared file
** and all locking correctly.  To do so would require that samba (or whatever
** tool is being used for file sharing) implements locks correctly between
** windows and unix.  I'm guessing that isn't likely to happen, but by
** using the same locking range we are at least open to the possibility.
**
** Locking in windows is manditory.  For this reason, we cannot store

Changes to src/os_unix.c.

627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
....
3094
3095
3096
3097
3098
3099
3100
3101
3102
3103
3104
3105
3106
3107
3108
....
3726
3727
3728
3729
3730
3731
3732
3733
3734
3735
3736
3737
3738
3739
3740
....
3933
3934
3935
3936
3937
3938
3939
3940
3941
3942
3943
3944
3945
3946
3947
....
4905
4906
4907
4908
4909
4910
4911
4912
4913
4914
4915
4916
4917
4918
4919
....
5167
5168
5169
5170
5171
5172
5173
5174
5175
5176
5177
5178
5179
5180
5181
....
5481
5482
5483
5484
5485
5486
5487
5488
5489
5490
5491
5492
5493
5494
5495
....
5512
5513
5514
5515
5516
5517
5518
5519
5520
5521
5522
5523
5524
5525
5526
....
6304
6305
6306
6307
6308
6309
6310
6311
6312
6313
6314
6315
6316
6317
6318
....
6356
6357
6358
6359
6360
6361
6362
6363
6364
6365
6366
6367
6368
6369
6370
}
#endif


#if defined(SQLITE_TEST) && defined(SQLITE_DEBUG)
/*
** Helper function for printing out trace information from debugging
** binaries. This returns the string represetation of the supplied
** integer lock-type.
*/
static const char *azFileLock(int eFileLock){
  switch( eFileLock ){
    case NO_LOCK: return "NONE";
    case SHARED_LOCK: return "SHARED";
    case RESERVED_LOCK: return "RESERVED";
................................................................................
/*
** Seek to the offset passed as the second argument, then read cnt 
** bytes into pBuf. Return the number of bytes actually read.
**
** NB:  If you define USE_PREAD or USE_PREAD64, then it might also
** be necessary to define _XOPEN_SOURCE to be 500.  This varies from
** one system to another.  Since SQLite does not define USE_PREAD
** any any form by default, we will not attempt to define _XOPEN_SOURCE.
** See tickets #2741 and #2681.
**
** To avoid stomping the errno value on a failed read the lastErrno value
** is set before returning.
*/
static int seekAndRead(unixFile *id, sqlite3_int64 offset, void *pBuf, int cnt){
  int got;
................................................................................
  }
#endif

  return SQLITE_OK;
}

/*
** If *pArg is inititially negative then this is a query.  Set *pArg to
** 1 or 0 depending on whether or not bit mask of pFile->ctrlFlags is set.
**
** If *pArg is 0 or 1, then clear or set the mask bit of pFile->ctrlFlags.
*/
static void unixModeBit(unixFile *pFile, unsigned char mask, int *pArg){
  if( *pArg<0 ){
    *pArg = (pFile->ctrlFlags & mask)!=0;
................................................................................
}
#endif /* __QNXNTO__ */

/*
** Return the device characteristics for the file.
**
** This VFS is set up to return SQLITE_IOCAP_POWERSAFE_OVERWRITE by default.
** However, that choice is contraversial since technically the underlying
** file system does not always provide powersafe overwrites.  (In other
** words, after a power-loss event, parts of the file that were never
** written might end up being altered.)  However, non-PSOW behavior is very,
** very rare.  And asserting PSOW makes a large reduction in the amount
** of required I/O for journaling, since a lot of padding is eliminated.
**  Hence, while POWERSAFE_OVERWRITE is on by default, there is a file-control
** available to turn it off and URI query parameter available to turn it off.
................................................................................
** the correct finder-function for that VFS.
**
** Most finder functions return a pointer to a fixed sqlite3_io_methods
** object.  The only interesting finder-function is autolockIoFinder, which
** looks at the filesystem type and tries to guess the best locking
** strategy from that.
**
** For finder-funtion F, two objects are created:
**
**    (1) The real finder-function named "FImpt()".
**
**    (2) A constant pointer to this function named just "F".
**
**
** A pointer to the F pointer is used as the pAppData value for VFS
................................................................................
}
static const sqlite3_io_methods 
  *(*const autolockIoFinder)(const char*,unixFile*) = autolockIoFinderImpl;

#endif /* OS_VXWORKS && SQLITE_ENABLE_LOCKING_STYLE */

/*
** An abstract type for a pointer to a IO method finder function:
*/
typedef const sqlite3_io_methods *(*finder_type)(const char*,unixFile*);


/****************************************************************************
**************************** sqlite3_vfs methods ****************************
**
................................................................................
  /* A stat() call may fail for various reasons. If this happens, it is
  ** almost certain that an open() call on the same path will also fail.
  ** For this reason, if an error occurs in the stat() call here, it is
  ** ignored and -1 is returned. The caller will try to open a new file
  ** descriptor on the same path, fail, and return an error to SQLite.
  **
  ** Even if a subsequent open() call does succeed, the consequences of
  ** not searching for a resusable file descriptor are not dire.  */
  if( 0==osStat(zPath, &sStat) ){
    unixInodeInfo *pInode;

    unixEnterMutex();
    pInode = inodeList;
    while( pInode && (pInode->fileId.dev!=sStat.st_dev
                     || pInode->fileId.ino!=sStat.st_ino) ){
................................................................................
/*
** This function is called by unixOpen() to determine the unix permissions
** to create new files with. If no error occurs, then SQLITE_OK is returned
** and a value suitable for passing as the third argument to open(2) is
** written to *pMode. If an IO error occurs, an SQLite error code is 
** returned and the value of *pMode is not modified.
**
** In most cases cases, this routine sets *pMode to 0, which will become
** an indication to robust_open() to create the file using
** SQLITE_DEFAULT_FILE_PERMISSIONS adjusted by the umask.
** But if the file being opened is a WAL or regular journal file, then 
** this function queries the file-system for the permissions on the 
** corresponding database file and sets *pMode to this value. Whenever 
** possible, WAL and journal files are created using the same permissions 
** as the associated database file.
................................................................................
**
** The conch file - to use a proxy file, sqlite must first "hold the conch"
** by taking an sqlite-style shared lock on the conch file, reading the
** contents and comparing the host's unique host ID (see below) and lock
** proxy path against the values stored in the conch.  The conch file is
** stored in the same directory as the database file and the file name
** is patterned after the database file name as ".<databasename>-conch".
** If the conch file does not exist, or it's contents do not match the
** host ID and/or proxy path, then the lock is escalated to an exclusive
** lock and the conch file contents is updated with the host ID and proxy
** path and the lock is downgraded to a shared lock again.  If the conch
** is held by another process (with a shared lock), the exclusive lock
** will fail and SQLITE_BUSY is returned.
**
** The proxy file - a single-byte file used for all advisory file locks
................................................................................
**       lock proxy files, only used when LOCKPROXYDIR is not set.
**    
**    
** As mentioned above, when compiled with SQLITE_PREFER_PROXY_LOCKING,
** setting the environment variable SQLITE_FORCE_PROXY_LOCKING to 1 will
** force proxy locking to be used for every database file opened, and 0
** will force automatic proxy locking to be disabled for all database
** files (explicity calling the SQLITE_SET_LOCKPROXYFILE pragma or
** sqlite_file_control API is not affected by SQLITE_FORCE_PROXY_LOCKING).
*/

/*
** Proxy locking is only available on MacOSX 
*/
#if defined(__APPLE__) && SQLITE_ENABLE_LOCKING_STYLE







|







 







|







 







|







 







|







 







|







 







|







 







|







 







|







 







|







 







|







627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
....
3094
3095
3096
3097
3098
3099
3100
3101
3102
3103
3104
3105
3106
3107
3108
....
3726
3727
3728
3729
3730
3731
3732
3733
3734
3735
3736
3737
3738
3739
3740
....
3933
3934
3935
3936
3937
3938
3939
3940
3941
3942
3943
3944
3945
3946
3947
....
4905
4906
4907
4908
4909
4910
4911
4912
4913
4914
4915
4916
4917
4918
4919
....
5167
5168
5169
5170
5171
5172
5173
5174
5175
5176
5177
5178
5179
5180
5181
....
5481
5482
5483
5484
5485
5486
5487
5488
5489
5490
5491
5492
5493
5494
5495
....
5512
5513
5514
5515
5516
5517
5518
5519
5520
5521
5522
5523
5524
5525
5526
....
6304
6305
6306
6307
6308
6309
6310
6311
6312
6313
6314
6315
6316
6317
6318
....
6356
6357
6358
6359
6360
6361
6362
6363
6364
6365
6366
6367
6368
6369
6370
}
#endif


#if defined(SQLITE_TEST) && defined(SQLITE_DEBUG)
/*
** Helper function for printing out trace information from debugging
** binaries. This returns the string representation of the supplied
** integer lock-type.
*/
static const char *azFileLock(int eFileLock){
  switch( eFileLock ){
    case NO_LOCK: return "NONE";
    case SHARED_LOCK: return "SHARED";
    case RESERVED_LOCK: return "RESERVED";
................................................................................
/*
** Seek to the offset passed as the second argument, then read cnt 
** bytes into pBuf. Return the number of bytes actually read.
**
** NB:  If you define USE_PREAD or USE_PREAD64, then it might also
** be necessary to define _XOPEN_SOURCE to be 500.  This varies from
** one system to another.  Since SQLite does not define USE_PREAD
** in any form by default, we will not attempt to define _XOPEN_SOURCE.
** See tickets #2741 and #2681.
**
** To avoid stomping the errno value on a failed read the lastErrno value
** is set before returning.
*/
static int seekAndRead(unixFile *id, sqlite3_int64 offset, void *pBuf, int cnt){
  int got;
................................................................................
  }
#endif

  return SQLITE_OK;
}

/*
** If *pArg is initially negative then this is a query.  Set *pArg to
** 1 or 0 depending on whether or not bit mask of pFile->ctrlFlags is set.
**
** If *pArg is 0 or 1, then clear or set the mask bit of pFile->ctrlFlags.
*/
static void unixModeBit(unixFile *pFile, unsigned char mask, int *pArg){
  if( *pArg<0 ){
    *pArg = (pFile->ctrlFlags & mask)!=0;
................................................................................
}
#endif /* __QNXNTO__ */

/*
** Return the device characteristics for the file.
**
** This VFS is set up to return SQLITE_IOCAP_POWERSAFE_OVERWRITE by default.
** However, that choice is controversial since technically the underlying
** file system does not always provide powersafe overwrites.  (In other
** words, after a power-loss event, parts of the file that were never
** written might end up being altered.)  However, non-PSOW behavior is very,
** very rare.  And asserting PSOW makes a large reduction in the amount
** of required I/O for journaling, since a lot of padding is eliminated.
**  Hence, while POWERSAFE_OVERWRITE is on by default, there is a file-control
** available to turn it off and URI query parameter available to turn it off.
................................................................................
** the correct finder-function for that VFS.
**
** Most finder functions return a pointer to a fixed sqlite3_io_methods
** object.  The only interesting finder-function is autolockIoFinder, which
** looks at the filesystem type and tries to guess the best locking
** strategy from that.
**
** For finder-function F, two objects are created:
**
**    (1) The real finder-function named "FImpt()".
**
**    (2) A constant pointer to this function named just "F".
**
**
** A pointer to the F pointer is used as the pAppData value for VFS
................................................................................
}
static const sqlite3_io_methods 
  *(*const autolockIoFinder)(const char*,unixFile*) = autolockIoFinderImpl;

#endif /* OS_VXWORKS && SQLITE_ENABLE_LOCKING_STYLE */

/*
** An abstract type for a pointer to an IO method finder function:
*/
typedef const sqlite3_io_methods *(*finder_type)(const char*,unixFile*);


/****************************************************************************
**************************** sqlite3_vfs methods ****************************
**
................................................................................
  /* A stat() call may fail for various reasons. If this happens, it is
  ** almost certain that an open() call on the same path will also fail.
  ** For this reason, if an error occurs in the stat() call here, it is
  ** ignored and -1 is returned. The caller will try to open a new file
  ** descriptor on the same path, fail, and return an error to SQLite.
  **
  ** Even if a subsequent open() call does succeed, the consequences of
  ** not searching for a reusable file descriptor are not dire.  */
  if( 0==osStat(zPath, &sStat) ){
    unixInodeInfo *pInode;

    unixEnterMutex();
    pInode = inodeList;
    while( pInode && (pInode->fileId.dev!=sStat.st_dev
                     || pInode->fileId.ino!=sStat.st_ino) ){
................................................................................
/*
** This function is called by unixOpen() to determine the unix permissions
** to create new files with. If no error occurs, then SQLITE_OK is returned
** and a value suitable for passing as the third argument to open(2) is
** written to *pMode. If an IO error occurs, an SQLite error code is 
** returned and the value of *pMode is not modified.
**
** In most cases, this routine sets *pMode to 0, which will become
** an indication to robust_open() to create the file using
** SQLITE_DEFAULT_FILE_PERMISSIONS adjusted by the umask.
** But if the file being opened is a WAL or regular journal file, then 
** this function queries the file-system for the permissions on the 
** corresponding database file and sets *pMode to this value. Whenever 
** possible, WAL and journal files are created using the same permissions 
** as the associated database file.
................................................................................
**
** The conch file - to use a proxy file, sqlite must first "hold the conch"
** by taking an sqlite-style shared lock on the conch file, reading the
** contents and comparing the host's unique host ID (see below) and lock
** proxy path against the values stored in the conch.  The conch file is
** stored in the same directory as the database file and the file name
** is patterned after the database file name as ".<databasename>-conch".
** If the conch file does not exist, or its contents do not match the
** host ID and/or proxy path, then the lock is escalated to an exclusive
** lock and the conch file contents is updated with the host ID and proxy
** path and the lock is downgraded to a shared lock again.  If the conch
** is held by another process (with a shared lock), the exclusive lock
** will fail and SQLITE_BUSY is returned.
**
** The proxy file - a single-byte file used for all advisory file locks
................................................................................
**       lock proxy files, only used when LOCKPROXYDIR is not set.
**    
**    
** As mentioned above, when compiled with SQLITE_PREFER_PROXY_LOCKING,
** setting the environment variable SQLITE_FORCE_PROXY_LOCKING to 1 will
** force proxy locking to be used for every database file opened, and 0
** will force automatic proxy locking to be disabled for all database
** files (explicitly calling the SQLITE_SET_LOCKPROXYFILE pragma or
** sqlite_file_control API is not affected by SQLITE_FORCE_PROXY_LOCKING).
*/

/*
** Proxy locking is only available on MacOSX 
*/
#if defined(__APPLE__) && SQLITE_ENABLE_LOCKING_STYLE

Changes to src/os_win.c.

3124
3125
3126
3127
3128
3129
3130
3131
3132
3133
3134
3135
3136
3137
3138
....
4138
4139
4140
4141
4142
4143
4144
4145
4146
4147
4148
4149
4150
4151
4152
  pFile->locktype = (u8)locktype;
  OSTRACE(("UNLOCK file=%p, lock=%d, rc=%s\n",
           pFile->h, pFile->locktype, sqlite3ErrName(rc)));
  return rc;
}

/*
** If *pArg is inititially negative then this is a query.  Set *pArg to
** 1 or 0 depending on whether or not bit mask of pFile->ctrlFlags is set.
**
** If *pArg is 0 or 1, then clear or set the mask bit of pFile->ctrlFlags.
*/
static void winModeBit(winFile *pFile, unsigned char mask, int *pArg){
  if( *pArg<0 ){
    *pArg = (pFile->ctrlFlags & mask)!=0;
................................................................................
           osGetCurrentProcessId(), pFd, iOff, p));

  if( p ){
    pFd->nFetchOut--;
  }else{
    /* FIXME:  If Windows truly always prevents truncating or deleting a
    ** file while a mapping is held, then the following winUnmapfile() call
    ** is unnecessary can can be omitted - potentially improving
    ** performance.  */
    winUnmapfile(pFd);
  }

  assert( pFd->nFetchOut>=0 );
#endif








|







 







|







3124
3125
3126
3127
3128
3129
3130
3131
3132
3133
3134
3135
3136
3137
3138
....
4138
4139
4140
4141
4142
4143
4144
4145
4146
4147
4148
4149
4150
4151
4152
  pFile->locktype = (u8)locktype;
  OSTRACE(("UNLOCK file=%p, lock=%d, rc=%s\n",
           pFile->h, pFile->locktype, sqlite3ErrName(rc)));
  return rc;
}

/*
** If *pArg is initially negative then this is a query.  Set *pArg to
** 1 or 0 depending on whether or not bit mask of pFile->ctrlFlags is set.
**
** If *pArg is 0 or 1, then clear or set the mask bit of pFile->ctrlFlags.
*/
static void winModeBit(winFile *pFile, unsigned char mask, int *pArg){
  if( *pArg<0 ){
    *pArg = (pFile->ctrlFlags & mask)!=0;
................................................................................
           osGetCurrentProcessId(), pFd, iOff, p));

  if( p ){
    pFd->nFetchOut--;
  }else{
    /* FIXME:  If Windows truly always prevents truncating or deleting a
    ** file while a mapping is held, then the following winUnmapfile() call
    ** is unnecessary can be omitted - potentially improving
    ** performance.  */
    winUnmapfile(pFd);
  }

  assert( pFd->nFetchOut>=0 );
#endif

Changes to src/pager.c.

72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
...
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
...
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
...
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
....
2493
2494
2495
2496
2497
2498
2499
2500
2501
2502
2503
2504
2505
2506
2507
....
2552
2553
2554
2555
2556
2557
2558
2559
2560
2561
2562
2563
2564
2565
2566
....
3754
3755
3756
3757
3758
3759
3760
3761
3762
3763
3764
3765
3766
3767
3768
....
4382
4383
4384
4385
4386
4387
4388
4389
4390
4391
4392
4393
4394
4395
4396
....
4922
4923
4924
4925
4926
4927
4928
4929
4930
4931
4932
4933
4934
4935
4936
** 
** (6) If a master journal file is used, then all writes to the database file
**     are synced prior to the master journal being deleted.
** 
** Definition: Two databases (or the same database at two points it time)
** are said to be "logically equivalent" if they give the same answer to
** all queries.  Note in particular the content of freelist leaf
** pages can be changed arbitarily without effecting the logical equivalence
** of the database.
** 
** (7) At any time, if any subset, including the empty set and the total set,
**     of the unsynced changes to a rollback journal are removed and the 
**     journal is rolled back, the resulting database file will be logical
**     equivalent to the database file at the beginning of the transaction.
** 
** (8) When a transaction is rolled back, the xTruncate method of the VFS
**     is called to restore the database file to the same size it was at
**     the beginning of the transaction.  (In some VFSes, the xTruncate
**     method is a no-op, but that does not change the fact the SQLite will
**     invoke it.)
................................................................................
**
** This is usually safe. If an xUnlock fails or appears to fail, there may 
** be a few redundant xLock() calls or a lock may be held for longer than
** required, but nothing really goes wrong.
**
** The exception is when the database file is unlocked as the pager moves
** from ERROR to OPEN state. At this point there may be a hot-journal file 
** in the file-system that needs to be rolled back (as part of a OPEN->SHARED
** transition, by the same pager or any other). If the call to xUnlock()
** fails at this point and the pager is left holding an EXCLUSIVE lock, this
** can confuse the call to xCheckReservedLock() call made later as part
** of hot-journal detection.
**
** xCheckReservedLock() is defined as returning true "if there is a RESERVED 
** lock held by this process or any others". So xCheckReservedLock may 
................................................................................
** Bits of the Pager.doNotSpill flag.  See further description below.
*/
#define SPILLFLAG_OFF         0x01      /* Never spill cache.  Set via pragma */
#define SPILLFLAG_ROLLBACK    0x02      /* Current rolling back, so do not spill */
#define SPILLFLAG_NOSYNC      0x04      /* Spill is ok, but do not sync */

/*
** A open page cache is an instance of struct Pager. A description of
** some of the more important member variables follows:
**
** eState
**
**   The current 'state' of the pager object. See the comment and state
**   diagram above for a description of the pager state.
**
................................................................................
  u8 tempFile;                /* zFilename is a temporary or immutable file */
  u8 noLock;                  /* Do not lock (except in WAL mode) */
  u8 readOnly;                /* True for a read-only database */
  u8 memDb;                   /* True to inhibit all file I/O */

  /**************************************************************************
  ** The following block contains those class members that change during
  ** routine opertion.  Class members not in this block are either fixed
  ** when the pager is first created or else only change when there is a
  ** significant mode change (such as changing the page_size, locking_mode,
  ** or the journal_mode).  From another view, these class members describe
  ** the "state" of the pager, while other class members describe the
  ** "configuration" of the pager.
  */
  u8 eState;                  /* Pager state (OPEN, READER, WRITER_LOCKED..) */
................................................................................
**
** If the main database file is not open, or the pager is not in either
** DBMOD or OPEN state, this function is a no-op. Otherwise, the size 
** of the file is changed to nPage pages (nPage*pPager->pageSize bytes). 
** If the file on disk is currently larger than nPage pages, then use the VFS
** xTruncate() method to truncate it.
**
** Or, it might might be the case that the file on disk is smaller than 
** nPage pages. Some operating system implementations can get confused if 
** you try to truncate a file to some size that is larger than it 
** currently is, so detect this case and write a single zero byte to 
** the end of the new file instead.
**
** If successful, return SQLITE_OK. If an IO error occurs while modifying
** the database file, return the error code to the caller.
................................................................................
  }
  return iRet;
}

/*
** Set the value of the Pager.sectorSize variable for the given
** pager based on the value returned by the xSectorSize method
** of the open database file. The sector size will be used used 
** to determine the size and alignment of journal header and 
** master journal pointers within created journal files.
**
** For temporary files the effective sector size is always 512 bytes.
**
** Otherwise, for non-temporary files, the effective sector size is
** the value returned by the xSectorSize() method rounded up to 32 if
................................................................................
** the lock. If the lock is obtained successfully, set the Pager.state 
** variable to locktype before returning.
*/
static int pager_wait_on_lock(Pager *pPager, int locktype){
  int rc;                              /* Return code */

  /* Check that this is either a no-op (because the requested lock is 
  ** already held, or one of the transistions that the busy-handler
  ** may be invoked during, according to the comment above
  ** sqlite3PagerSetBusyhandler().
  */
  assert( (pPager->eLock>=locktype)
       || (pPager->eLock==NO_LOCK && locktype==SHARED_LOCK)
       || (pPager->eLock==RESERVED_LOCK && locktype==EXCLUSIVE_LOCK)
  );
................................................................................
  ** pages belonging to the same sector.
  **
  ** The doNotSpill ROLLBACK and OFF bits inhibits all cache spilling
  ** regardless of whether or not a sync is required.  This is set during
  ** a rollback or by user request, respectively.
  **
  ** Spilling is also prohibited when in an error state since that could
  ** lead to database corruption.   In the current implementaton it 
  ** is impossible for sqlite3PcacheFetch() to be called with createFlag==3
  ** while in the error state, hence it is impossible for this routine to
  ** be called in the error state.  Nevertheless, we include a NEVER()
  ** test for the error state as a safeguard against future changes.
  */
  if( NEVER(pPager->errCode) ) return SQLITE_OK;
  testcase( pPager->doNotSpill & SPILLFLAG_ROLLBACK );
................................................................................
            }
            if( !jrnlOpen ){
              sqlite3OsClose(pPager->jfd);
            }
            *pExists = (first!=0);
          }else if( rc==SQLITE_CANTOPEN ){
            /* If we cannot open the rollback journal file in order to see if
            ** its has a zero header, that might be due to an I/O error, or
            ** it might be due to the race condition described above and in
            ** ticket #3883.  Either way, assume that the journal is hot.
            ** This might be a false positive.  But if it is, then the
            ** automatic journal playback and recovery mechanism will deal
            ** with it under an EXCLUSIVE lock where we do not need to
            ** worry so much with race conditions.
            */







|




|







 







|







 







|







 







|







 







|







 







|







 







|







 







|







 







|







72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
...
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
...
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
...
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
....
2493
2494
2495
2496
2497
2498
2499
2500
2501
2502
2503
2504
2505
2506
2507
....
2552
2553
2554
2555
2556
2557
2558
2559
2560
2561
2562
2563
2564
2565
2566
....
3754
3755
3756
3757
3758
3759
3760
3761
3762
3763
3764
3765
3766
3767
3768
....
4382
4383
4384
4385
4386
4387
4388
4389
4390
4391
4392
4393
4394
4395
4396
....
4922
4923
4924
4925
4926
4927
4928
4929
4930
4931
4932
4933
4934
4935
4936
** 
** (6) If a master journal file is used, then all writes to the database file
**     are synced prior to the master journal being deleted.
** 
** Definition: Two databases (or the same database at two points it time)
** are said to be "logically equivalent" if they give the same answer to
** all queries.  Note in particular the content of freelist leaf
** pages can be changed arbitrarily without affecting the logical equivalence
** of the database.
** 
** (7) At any time, if any subset, including the empty set and the total set,
**     of the unsynced changes to a rollback journal are removed and the 
**     journal is rolled back, the resulting database file will be logically
**     equivalent to the database file at the beginning of the transaction.
** 
** (8) When a transaction is rolled back, the xTruncate method of the VFS
**     is called to restore the database file to the same size it was at
**     the beginning of the transaction.  (In some VFSes, the xTruncate
**     method is a no-op, but that does not change the fact the SQLite will
**     invoke it.)
................................................................................
**
** This is usually safe. If an xUnlock fails or appears to fail, there may 
** be a few redundant xLock() calls or a lock may be held for longer than
** required, but nothing really goes wrong.
**
** The exception is when the database file is unlocked as the pager moves
** from ERROR to OPEN state. At this point there may be a hot-journal file 
** in the file-system that needs to be rolled back (as part of an OPEN->SHARED
** transition, by the same pager or any other). If the call to xUnlock()
** fails at this point and the pager is left holding an EXCLUSIVE lock, this
** can confuse the call to xCheckReservedLock() call made later as part
** of hot-journal detection.
**
** xCheckReservedLock() is defined as returning true "if there is a RESERVED 
** lock held by this process or any others". So xCheckReservedLock may 
................................................................................
** Bits of the Pager.doNotSpill flag.  See further description below.
*/
#define SPILLFLAG_OFF         0x01      /* Never spill cache.  Set via pragma */
#define SPILLFLAG_ROLLBACK    0x02      /* Current rolling back, so do not spill */
#define SPILLFLAG_NOSYNC      0x04      /* Spill is ok, but do not sync */

/*
** An open page cache is an instance of struct Pager. A description of
** some of the more important member variables follows:
**
** eState
**
**   The current 'state' of the pager object. See the comment and state
**   diagram above for a description of the pager state.
**
................................................................................
  u8 tempFile;                /* zFilename is a temporary or immutable file */
  u8 noLock;                  /* Do not lock (except in WAL mode) */
  u8 readOnly;                /* True for a read-only database */
  u8 memDb;                   /* True to inhibit all file I/O */

  /**************************************************************************
  ** The following block contains those class members that change during
  ** routine operation.  Class members not in this block are either fixed
  ** when the pager is first created or else only change when there is a
  ** significant mode change (such as changing the page_size, locking_mode,
  ** or the journal_mode).  From another view, these class members describe
  ** the "state" of the pager, while other class members describe the
  ** "configuration" of the pager.
  */
  u8 eState;                  /* Pager state (OPEN, READER, WRITER_LOCKED..) */
................................................................................
**
** If the main database file is not open, or the pager is not in either
** DBMOD or OPEN state, this function is a no-op. Otherwise, the size 
** of the file is changed to nPage pages (nPage*pPager->pageSize bytes). 
** If the file on disk is currently larger than nPage pages, then use the VFS
** xTruncate() method to truncate it.
**
** Or, it might be the case that the file on disk is smaller than 
** nPage pages. Some operating system implementations can get confused if 
** you try to truncate a file to some size that is larger than it 
** currently is, so detect this case and write a single zero byte to 
** the end of the new file instead.
**
** If successful, return SQLITE_OK. If an IO error occurs while modifying
** the database file, return the error code to the caller.
................................................................................
  }
  return iRet;
}

/*
** Set the value of the Pager.sectorSize variable for the given
** pager based on the value returned by the xSectorSize method
** of the open database file. The sector size will be used 
** to determine the size and alignment of journal header and 
** master journal pointers within created journal files.
**
** For temporary files the effective sector size is always 512 bytes.
**
** Otherwise, for non-temporary files, the effective sector size is
** the value returned by the xSectorSize() method rounded up to 32 if
................................................................................
** the lock. If the lock is obtained successfully, set the Pager.state 
** variable to locktype before returning.
*/
static int pager_wait_on_lock(Pager *pPager, int locktype){
  int rc;                              /* Return code */

  /* Check that this is either a no-op (because the requested lock is 
  ** already held), or one of the transitions that the busy-handler
  ** may be invoked during, according to the comment above
  ** sqlite3PagerSetBusyhandler().
  */
  assert( (pPager->eLock>=locktype)
       || (pPager->eLock==NO_LOCK && locktype==SHARED_LOCK)
       || (pPager->eLock==RESERVED_LOCK && locktype==EXCLUSIVE_LOCK)
  );
................................................................................
  ** pages belonging to the same sector.
  **
  ** The doNotSpill ROLLBACK and OFF bits inhibits all cache spilling
  ** regardless of whether or not a sync is required.  This is set during
  ** a rollback or by user request, respectively.
  **
  ** Spilling is also prohibited when in an error state since that could
  ** lead to database corruption.   In the current implementation it 
  ** is impossible for sqlite3PcacheFetch() to be called with createFlag==3
  ** while in the error state, hence it is impossible for this routine to
  ** be called in the error state.  Nevertheless, we include a NEVER()
  ** test for the error state as a safeguard against future changes.
  */
  if( NEVER(pPager->errCode) ) return SQLITE_OK;
  testcase( pPager->doNotSpill & SPILLFLAG_ROLLBACK );
................................................................................
            }
            if( !jrnlOpen ){
              sqlite3OsClose(pPager->jfd);
            }
            *pExists = (first!=0);
          }else if( rc==SQLITE_CANTOPEN ){
            /* If we cannot open the rollback journal file in order to see if
            ** it has a zero header, that might be due to an I/O error, or
            ** it might be due to the race condition described above and in
            ** ticket #3883.  Either way, assume that the journal is hot.
            ** This might be a false positive.  But if it is, then the
            ** automatic journal playback and recovery mechanism will deal
            ** with it under an EXCLUSIVE lock where we do not need to
            ** worry so much with race conditions.
            */

Changes to src/pcache.c.

386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
    pCache->pPage1 = pPgHdr;
  }
  return pPgHdr;
}

/*
** Decrement the reference count on a page. If the page is clean and the
** reference count drops to 0, then it is made elible for recycling.
*/
void SQLITE_NOINLINE sqlite3PcacheRelease(PgHdr *p){
  assert( p->nRef>0 );
  p->nRef--;
  if( p->nRef==0 ){
    p->pCache->nRef--;
    if( (p->flags&PGHDR_DIRTY)==0 ){







|







386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
    pCache->pPage1 = pPgHdr;
  }
  return pPgHdr;
}

/*
** Decrement the reference count on a page. If the page is clean and the
** reference count drops to 0, then it is made eligible for recycling.
*/
void SQLITE_NOINLINE sqlite3PcacheRelease(PgHdr *p){
  assert( p->nRef>0 );
  p->nRef--;
  if( p->nRef==0 ){
    p->pCache->nRef--;
    if( (p->flags&PGHDR_DIRTY)==0 ){

Changes to src/pcache1.c.

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
**    May you share freely, never taking more than you give.
**
*************************************************************************
**
** This file implements the default page cache implementation (the
** sqlite3_pcache interface). It also contains part of the implementation
** of the SQLITE_CONFIG_PAGECACHE and sqlite3_release_memory() features.
** If the default page cache implementation is overriden, then neither of
** these two features are available.
*/

#include "sqliteInt.h"

typedef struct PCache1 PCache1;
typedef struct PgHdr1 PgHdr1;
typedef struct PgFreeslot PgFreeslot;
typedef struct PGroup PGroup;

/* Each page cache (or PCache) belongs to a PGroup.  A PGroup is a set 
** of one or more PCaches that are able to recycle each others unpinned
** pages when they are under memory pressure.  A PGroup is an instance of
** the following object.
**
** This page cache implementation works in one of two modes:
**
**   (1)  Every PCache is the sole member of its own PGroup.  There is
**        one PGroup per PCache.







|











|







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
**    May you share freely, never taking more than you give.
**
*************************************************************************
**
** This file implements the default page cache implementation (the
** sqlite3_pcache interface). It also contains part of the implementation
** of the SQLITE_CONFIG_PAGECACHE and sqlite3_release_memory() features.
** If the default page cache implementation is overridden, then neither of
** these two features are available.
*/

#include "sqliteInt.h"

typedef struct PCache1 PCache1;
typedef struct PgHdr1 PgHdr1;
typedef struct PgFreeslot PgFreeslot;
typedef struct PGroup PGroup;

/* Each page cache (or PCache) belongs to a PGroup.  A PGroup is a set 
** of one or more PCaches that are able to recycle each other's unpinned
** pages when they are under memory pressure.  A PGroup is an instance of
** the following object.
**
** This page cache implementation works in one of two modes:
**
**   (1)  Every PCache is the sole member of its own PGroup.  There is
**        one PGroup per PCache.

Changes to src/printf.c.

900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
  z = sqlite3VMPrintf(db, zFormat, ap);
  va_end(ap);
  return z;
}

/*
** Like sqlite3MPrintf(), but call sqlite3DbFree() on zStr after formatting
** the string and before returnning.  This routine is intended to be used
** to modify an existing string.  For example:
**
**       x = sqlite3MPrintf(db, x, "prefix %s suffix", x);
**
*/
char *sqlite3MAppendf(sqlite3 *db, char *zStr, const char *zFormat, ...){
  va_list ap;







|







900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
  z = sqlite3VMPrintf(db, zFormat, ap);
  va_end(ap);
  return z;
}

/*
** Like sqlite3MPrintf(), but call sqlite3DbFree() on zStr after formatting
** the string and before returning.  This routine is intended to be used
** to modify an existing string.  For example:
**
**       x = sqlite3MPrintf(db, x, "prefix %s suffix", x);
**
*/
char *sqlite3MAppendf(sqlite3 *db, char *zStr, const char *zFormat, ...){
  va_list ap;

Changes to src/resolve.c.

1114
1115
1116
1117
1118
1119
1120
1121
1122
1123
1124
1125
1126
1127
1128
      }
    }
  }
  return sqlite3ResolveOrderGroupBy(pParse, pSelect, pOrderBy, zType);
}

/*
** Resolve names in the SELECT statement p and all of its descendents.
*/
static int resolveSelectStep(Walker *pWalker, Select *p){
  NameContext *pOuterNC;  /* Context that contains this SELECT */
  NameContext sNC;        /* Name context of this SELECT */
  int isCompound;         /* True if p is a compound select */
  int nCompound;          /* Number of compound terms processed so far */
  Parse *pParse;          /* Parsing context */







|







1114
1115
1116
1117
1118
1119
1120
1121
1122
1123
1124
1125
1126
1127
1128
      }
    }
  }
  return sqlite3ResolveOrderGroupBy(pParse, pSelect, pOrderBy, zType);
}

/*
** Resolve names in the SELECT statement p and all of its descendants.
*/
static int resolveSelectStep(Walker *pWalker, Select *p){
  NameContext *pOuterNC;  /* Context that contains this SELECT */
  NameContext sNC;        /* Name context of this SELECT */
  int isCompound;         /* True if p is a compound select */
  int nCompound;          /* Number of compound terms processed so far */
  Parse *pParse;          /* Parsing context */

Changes to src/rowset.c.

46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
...
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
** value added by the INSERT will not be visible to the second TEST.
** The initial batch number is zero, so if the very first TEST contains
** a non-zero batch number, it will see all prior INSERTs.
**
** No INSERTs may occurs after a SMALLEST.  An assertion will fail if
** that is attempted.
**
** The cost of an INSERT is roughly constant.  (Sometime new memory
** has to be allocated on an INSERT.)  The cost of a TEST with a new
** batch number is O(NlogN) where N is the number of elements in the RowSet.
** The cost of a TEST using the same batch number is O(logN).  The cost
** of the first SMALLEST is O(NlogN).  Second and subsequent SMALLEST
** primitives are constant time.  The cost of DESTROY is O(N).
**
** There is an added cost of O(N) when switching between TEST and
................................................................................
  }
}

/*
** Check to see if element iRowid was inserted into the rowset as
** part of any insert batch prior to iBatch.  Return 1 or 0.
**
** If this is the first test of a new batch and if there exist entires
** on pRowSet->pEntry, then sort those entires into the forest at
** pRowSet->pForest so that they can be tested.
*/
int sqlite3RowSetTest(RowSet *pRowSet, int iBatch, sqlite3_int64 iRowid){
  struct RowSetEntry *p, *pTree;

  /* This routine is never called after sqlite3RowSetNext() */
  assert( pRowSet!=0 && (pRowSet->rsFlags & ROWSET_NEXT)==0 );







|







 







|
|







46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
...
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
** value added by the INSERT will not be visible to the second TEST.
** The initial batch number is zero, so if the very first TEST contains
** a non-zero batch number, it will see all prior INSERTs.
**
** No INSERTs may occurs after a SMALLEST.  An assertion will fail if
** that is attempted.
**
** The cost of an INSERT is roughly constant.  (Sometimes new memory
** has to be allocated on an INSERT.)  The cost of a TEST with a new
** batch number is O(NlogN) where N is the number of elements in the RowSet.
** The cost of a TEST using the same batch number is O(logN).  The cost
** of the first SMALLEST is O(NlogN).  Second and subsequent SMALLEST
** primitives are constant time.  The cost of DESTROY is O(N).
**
** There is an added cost of O(N) when switching between TEST and
................................................................................
  }
}

/*
** Check to see if element iRowid was inserted into the rowset as
** part of any insert batch prior to iBatch.  Return 1 or 0.
**
** If this is the first test of a new batch and if there exist entries
** on pRowSet->pEntry, then sort those entries into the forest at
** pRowSet->pForest so that they can be tested.
*/
int sqlite3RowSetTest(RowSet *pRowSet, int iBatch, sqlite3_int64 iRowid){
  struct RowSetEntry *p, *pTree;

  /* This routine is never called after sqlite3RowSetNext() */
  assert( pRowSet!=0 && (pRowSet->rsFlags & ROWSET_NEXT)==0 );

Changes to src/select.c.

1006
1007
1008
1009
1010
1011
1012
1013
1014
1015
1016
1017
1018
1019
1020
....
1537
1538
1539
1540
1541
1542
1543
1544
1545
1546
1547
1548
1549
1550
1551
....
1610
1611
1612
1613
1614
1615
1616
1617
1618
1619
1620
1621
1622
1623
1624
....
3094
3095
3096
3097
3098
3099
3100
3101
3102
3103
3104
3105
3106
3107
3108
....
3238
3239
3240
3241
3242
3243
3244
3245
3246
3247
3248
3249
3250
3251
3252
....
3624
3625
3626
3627
3628
3629
3630
3631
3632
3633
3634
3635
3636
3637
3638
....
3954
3955
3956
3957
3958
3959
3960
3961
3962
3963
3964
3965
3966
3967
3968
3969
3970
3971
**
** If the ExprList is an ORDER BY or GROUP BY clause then the resulting
** KeyInfo structure is appropriate for initializing a virtual index to
** implement that clause.  If the ExprList is the result set of a SELECT
** then the KeyInfo structure is appropriate for initializing a virtual
** index to implement a DISTINCT test.
**
** Space to hold the KeyInfo structure is obtain from malloc.  The calling
** function is responsible for seeing that this structure is eventually
** freed.
*/
static KeyInfo *keyInfoFromExprList(
  Parse *pParse,       /* Parsing context */
  ExprList *pList,     /* Form the KeyInfo object from this ExprList */
  int iStart,          /* Begin with this column of pList */
................................................................................
      sqlite3VdbeSetColName(v, i, COLNAME_NAME, z, SQLITE_DYNAMIC);
    }
  }
  generateColumnTypes(pParse, pTabList, pEList);
}

/*
** Given a an expression list (which is really the list of expressions
** that form the result set of a SELECT statement) compute appropriate
** column names for a table that would hold the expression list.
**
** All column names will be unique.
**
** Only the column names are computed.  Column.zType, Column.zColl,
** and other fields of Column are zeroed.
................................................................................
    }
    if( db->mallocFailed ){
      sqlite3DbFree(db, zName);
      break;
    }

    /* Make sure the column name is unique.  If the name is not unique,
    ** append a integer to the name so that it becomes unique.
    */
    nName = sqlite3Strlen30(zName);
    for(j=cnt=0; j<i; j++){
      if( sqlite3StrICmp(aCol[j].zName, zName)==0 ){
        char *zNewName;
        int k;
        for(k=nName-1; k>1 && sqlite3Isdigit(zName[k]); k--){}
................................................................................
** optimized.
**
** This routine attempts to rewrite queries such as the above into
** a single flat select, like this:
**
**     SELECT x+y AS a FROM t1 WHERE z<100 AND a>5
**
** The code generated for this simpification gives the same result
** but only has to scan the data once.  And because indices might 
** exist on the table t1, a complete scan of the data might be
** avoided.
**
** Flattening is only attempted if all of the following are true:
**
**   (1)  The subquery and the outer query do not both use aggregates.
................................................................................
  pSub = pSubitem->pSelect;
  assert( pSub!=0 );
  if( isAgg && subqueryIsAgg ) return 0;                 /* Restriction (1)  */
  if( subqueryIsAgg && pSrc->nSrc>1 ) return 0;          /* Restriction (2)  */
  pSubSrc = pSub->pSrc;
  assert( pSubSrc );
  /* Prior to version 3.1.2, when LIMIT and OFFSET had to be simple constants,
  ** not arbitrary expresssions, we allowed some combining of LIMIT and OFFSET
  ** because they could be computed at compile-time.  But when LIMIT and OFFSET
  ** became arbitrary expressions, we were forced to add restrictions (13)
  ** and (14). */
  if( pSub->pLimit && p->pLimit ) return 0;              /* Restriction (13) */
  if( pSub->pOffset ) return 0;                          /* Restriction (14) */
  if( (p->selFlags & SF_Compound)!=0 && pSub->pLimit ){
    return 0;                                            /* Restriction (15) */
................................................................................

  assert( *ppMinMax==0 || (*ppMinMax)->nExpr==1 );
  return eRet;
}

/*
** The select statement passed as the first argument is an aggregate query.
** The second argment is the associated aggregate-info object. This 
** function tests if the SELECT is of the form:
**
**   SELECT count(*) FROM <tbl>
**
** where table is a database table, not a sub-select or view. If the query
** does match this pattern, then a pointer to the Table object representing
** <tbl> is returned. Otherwise, 0 is returned.
................................................................................
**         element of the FROM clause.
**
**    (2)  Fill in the pTabList->a[].pTab fields in the SrcList that 
**         defines FROM clause.  When views appear in the FROM clause,
**         fill pTabList->a[].pSelect with a copy of the SELECT statement
**         that implements the view.  A copy is made of the view's SELECT
**         statement so that we can freely modify or delete that statement
**         without worrying about messing up the presistent representation
**         of the view.
**
**    (3)  Add terms to the WHERE clause to accomodate the NATURAL keyword
**         on joins and the ON and USING clause of joins.
**
**    (4)  Scan the list of columns in the result set (pEList) looking
**         for instances of the "*" operator or the TABLE.* operator.
**         If found, expand each "*" to be every column in every table
**         and TABLE.* to be every column in TABLE.
**







|







 







|







 







|







 







|







 







|







 







|







 







|


|







1006
1007
1008
1009
1010
1011
1012
1013
1014
1015
1016
1017
1018
1019
1020
....
1537
1538
1539
1540
1541
1542
1543
1544
1545
1546
1547
1548
1549
1550
1551
....
1610
1611
1612
1613
1614
1615
1616
1617
1618
1619
1620
1621
1622
1623
1624
....
3094
3095
3096
3097
3098
3099
3100
3101
3102
3103
3104
3105
3106
3107
3108
....
3238
3239
3240
3241
3242
3243
3244
3245
3246
3247
3248
3249
3250
3251
3252
....
3624
3625
3626
3627
3628
3629
3630
3631
3632
3633
3634
3635
3636
3637
3638
....
3954
3955
3956
3957
3958
3959
3960
3961
3962
3963
3964
3965
3966
3967
3968
3969
3970
3971
**
** If the ExprList is an ORDER BY or GROUP BY clause then the resulting
** KeyInfo structure is appropriate for initializing a virtual index to
** implement that clause.  If the ExprList is the result set of a SELECT
** then the KeyInfo structure is appropriate for initializing a virtual
** index to implement a DISTINCT test.
**
** Space to hold the KeyInfo structure is obtained from malloc.  The calling
** function is responsible for seeing that this structure is eventually
** freed.
*/
static KeyInfo *keyInfoFromExprList(
  Parse *pParse,       /* Parsing context */
  ExprList *pList,     /* Form the KeyInfo object from this ExprList */
  int iStart,          /* Begin with this column of pList */
................................................................................
      sqlite3VdbeSetColName(v, i, COLNAME_NAME, z, SQLITE_DYNAMIC);
    }
  }
  generateColumnTypes(pParse, pTabList, pEList);
}

/*
** Given an expression list (which is really the list of expressions
** that form the result set of a SELECT statement) compute appropriate
** column names for a table that would hold the expression list.
**
** All column names will be unique.
**
** Only the column names are computed.  Column.zType, Column.zColl,
** and other fields of Column are zeroed.
................................................................................
    }
    if( db->mallocFailed ){
      sqlite3DbFree(db, zName);
      break;
    }

    /* Make sure the column name is unique.  If the name is not unique,
    ** append an integer to the name so that it becomes unique.
    */
    nName = sqlite3Strlen30(zName);
    for(j=cnt=0; j<i; j++){
      if( sqlite3StrICmp(aCol[j].zName, zName)==0 ){
        char *zNewName;
        int k;
        for(k=nName-1; k>1 && sqlite3Isdigit(zName[k]); k--){}
................................................................................
** optimized.
**
** This routine attempts to rewrite queries such as the above into
** a single flat select, like this:
**
**     SELECT x+y AS a FROM t1 WHERE z<100 AND a>5
**
** The code generated for this simplification gives the same result
** but only has to scan the data once.  And because indices might 
** exist on the table t1, a complete scan of the data might be
** avoided.
**
** Flattening is only attempted if all of the following are true:
**
**   (1)  The subquery and the outer query do not both use aggregates.
................................................................................
  pSub = pSubitem->pSelect;
  assert( pSub!=0 );
  if( isAgg && subqueryIsAgg ) return 0;                 /* Restriction (1)  */
  if( subqueryIsAgg && pSrc->nSrc>1 ) return 0;          /* Restriction (2)  */
  pSubSrc = pSub->pSrc;
  assert( pSubSrc );
  /* Prior to version 3.1.2, when LIMIT and OFFSET had to be simple constants,
  ** not arbitrary expressions, we allowed some combining of LIMIT and OFFSET
  ** because they could be computed at compile-time.  But when LIMIT and OFFSET
  ** became arbitrary expressions, we were forced to add restrictions (13)
  ** and (14). */
  if( pSub->pLimit && p->pLimit ) return 0;              /* Restriction (13) */
  if( pSub->pOffset ) return 0;                          /* Restriction (14) */
  if( (p->selFlags & SF_Compound)!=0 && pSub->pLimit ){
    return 0;                                            /* Restriction (15) */
................................................................................

  assert( *ppMinMax==0 || (*ppMinMax)->nExpr==1 );
  return eRet;
}

/*
** The select statement passed as the first argument is an aggregate query.
** The second argument is the associated aggregate-info object. This 
** function tests if the SELECT is of the form:
**
**   SELECT count(*) FROM <tbl>
**
** where table is a database table, not a sub-select or view. If the query
** does match this pattern, then a pointer to the Table object representing
** <tbl> is returned. Otherwise, 0 is returned.
................................................................................
**         element of the FROM clause.
**
**    (2)  Fill in the pTabList->a[].pTab fields in the SrcList that 
**         defines FROM clause.  When views appear in the FROM clause,
**         fill pTabList->a[].pSelect with a copy of the SELECT statement
**         that implements the view.  A copy is made of the view's SELECT
**         statement so that we can freely modify or delete that statement
**         without worrying about messing up the persistent representation
**         of the view.
**
**    (3)  Add terms to the WHERE clause to accommodate the NATURAL keyword
**         on joins and the ON and USING clause of joins.
**
**    (4)  Scan the list of columns in the result set (pEList) looking
**         for instances of the "*" operator or the TABLE.* operator.
**         If found, expand each "*" to be every column in every table
**         and TABLE.* to be every column in TABLE.
**

Changes to src/sqlite.h.in.

2090
2091
2092
2093
2094
2095
2096
2097
2098
2099
2100
2101
2102
2103
2104
....
4160
4161
4162
4163
4164
4165
4166
4167
4168
4169
4170
4171
4172
4173
4174
....
6353
6354
6355
6356
6357
6358
6359
6360
6361
6362
6363
6364
6365
6366
6367
6368
6369
6370
6371
6372
6373
6374
6375
6376
6377
6378
6379
6380
6381
** the handler returns 0 which causes [sqlite3_step()] to return
** [SQLITE_BUSY].
**
** ^Calling this routine with an argument less than or equal to zero
** turns off all busy handlers.
**
** ^(There can only be a single busy handler for a particular
** [database connection] any any given moment.  If another busy handler
** was defined  (using [sqlite3_busy_handler()]) prior to calling
** this routine, that other busy handler is cleared.)^
**
** See also:  [PRAGMA busy_timeout]
*/
int sqlite3_busy_timeout(sqlite3*, int ms);

................................................................................
** extract values from the [sqlite3_value] objects.
**
** These routines work only with [protected sqlite3_value] objects.
** Any attempt to use these routines on an [unprotected sqlite3_value]
** object results in undefined behavior.
**
** ^These routines work just like the corresponding [column access functions]
** except that  these routines take a single [protected sqlite3_value] object
** pointer instead of a [sqlite3_stmt*] pointer and an integer column number.
**
** ^The sqlite3_value_text16() interface extracts a UTF-16 string
** in the native byte-order of the host machine.  ^The
** sqlite3_value_text16be() and sqlite3_value_text16le() interfaces
** extract UTF-16 strings as big-endian and little-endian respectively.
**
................................................................................
** <dd>This parameter returns the number malloc attempts that might have
** been satisfied using lookaside memory but failed due to all lookaside
** memory already being in use.
** Only the high-water value is meaningful;
** the current value is always zero.)^
**
** [[SQLITE_DBSTATUS_CACHE_USED]] ^(<dt>SQLITE_DBSTATUS_CACHE_USED</dt>
** <dd>This parameter returns the approximate number of of bytes of heap
** memory used by all pager caches associated with the database connection.)^
** ^The highwater mark associated with SQLITE_DBSTATUS_CACHE_USED is always 0.
**
** [[SQLITE_DBSTATUS_SCHEMA_USED]] ^(<dt>SQLITE_DBSTATUS_SCHEMA_USED</dt>
** <dd>This parameter returns the approximate number of of bytes of heap
** memory used to store the schema for all databases associated
** with the connection - main, temp, and any [ATTACH]-ed databases.)^ 
** ^The full amount of memory used by the schemas is reported, even if the
** schema memory is shared with other database connections due to
** [shared cache mode] being enabled.
** ^The highwater mark associated with SQLITE_DBSTATUS_SCHEMA_USED is always 0.
**
** [[SQLITE_DBSTATUS_STMT_USED]] ^(<dt>SQLITE_DBSTATUS_STMT_USED</dt>
** <dd>This parameter returns the approximate number of of bytes of heap
** and lookaside memory used by all prepared statements associated with
** the database connection.)^
** ^The highwater mark associated with SQLITE_DBSTATUS_STMT_USED is always 0.
** </dd>
**
** [[SQLITE_DBSTATUS_CACHE_HIT]] ^(<dt>SQLITE_DBSTATUS_CACHE_HIT</dt>
** <dd>This parameter returns the number of pager cache hits that have







|







 







|







 







|




|








|







2090
2091
2092
2093
2094
2095
2096
2097
2098
2099
2100
2101
2102
2103
2104
....
4160
4161
4162
4163
4164
4165
4166
4167
4168
4169
4170
4171
4172
4173
4174
....
6353
6354
6355
6356
6357
6358
6359
6360
6361
6362
6363
6364
6365
6366
6367
6368
6369
6370
6371
6372
6373
6374
6375
6376
6377
6378
6379
6380
6381
** the handler returns 0 which causes [sqlite3_step()] to return
** [SQLITE_BUSY].
**
** ^Calling this routine with an argument less than or equal to zero
** turns off all busy handlers.
**
** ^(There can only be a single busy handler for a particular
** [database connection] at any given moment.  If another busy handler
** was defined  (using [sqlite3_busy_handler()]) prior to calling
** this routine, that other busy handler is cleared.)^
**
** See also:  [PRAGMA busy_timeout]
*/
int sqlite3_busy_timeout(sqlite3*, int ms);

................................................................................
** extract values from the [sqlite3_value] objects.
**
** These routines work only with [protected sqlite3_value] objects.
** Any attempt to use these routines on an [unprotected sqlite3_value]
** object results in undefined behavior.
**
** ^These routines work just like the corresponding [column access functions]
** except that these routines take a single [protected sqlite3_value] object
** pointer instead of a [sqlite3_stmt*] pointer and an integer column number.
**
** ^The sqlite3_value_text16() interface extracts a UTF-16 string
** in the native byte-order of the host machine.  ^The
** sqlite3_value_text16be() and sqlite3_value_text16le() interfaces
** extract UTF-16 strings as big-endian and little-endian respectively.
**
................................................................................
** <dd>This parameter returns the number malloc attempts that might have
** been satisfied using lookaside memory but failed due to all lookaside
** memory already being in use.
** Only the high-water value is meaningful;
** the current value is always zero.)^
**
** [[SQLITE_DBSTATUS_CACHE_USED]] ^(<dt>SQLITE_DBSTATUS_CACHE_USED</dt>
** <dd>This parameter returns the approximate number of bytes of heap
** memory used by all pager caches associated with the database connection.)^
** ^The highwater mark associated with SQLITE_DBSTATUS_CACHE_USED is always 0.
**
** [[SQLITE_DBSTATUS_SCHEMA_USED]] ^(<dt>SQLITE_DBSTATUS_SCHEMA_USED</dt>
** <dd>This parameter returns the approximate number of bytes of heap
** memory used to store the schema for all databases associated
** with the connection - main, temp, and any [ATTACH]-ed databases.)^ 
** ^The full amount of memory used by the schemas is reported, even if the
** schema memory is shared with other database connections due to
** [shared cache mode] being enabled.
** ^The highwater mark associated with SQLITE_DBSTATUS_SCHEMA_USED is always 0.
**
** [[SQLITE_DBSTATUS_STMT_USED]] ^(<dt>SQLITE_DBSTATUS_STMT_USED</dt>
** <dd>This parameter returns the approximate number of bytes of heap
** and lookaside memory used by all prepared statements associated with
** the database connection.)^
** ^The highwater mark associated with SQLITE_DBSTATUS_STMT_USED is always 0.
** </dd>
**
** [[SQLITE_DBSTATUS_CACHE_HIT]] ^(<dt>SQLITE_DBSTATUS_CACHE_HIT</dt>
** <dd>This parameter returns the number of pager cache hits that have

Changes to src/sqlite3ext.h.

24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
...
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
/*
** The following structure holds pointers to all of the SQLite API
** routines.
**
** WARNING:  In order to maintain backwards compatibility, add new
** interfaces to the end of this structure only.  If you insert new
** interfaces in the middle of this structure, then older different
** versions of SQLite will not be able to load each others' shared
** libraries!
*/
struct sqlite3_api_routines {
  void * (*aggregate_context)(sqlite3_context*,int nBytes);
  int  (*aggregate_count)(sqlite3_context*);
  int  (*bind_blob)(sqlite3_stmt*,int,const void*,int n,void(*)(void*));
  int  (*bind_double)(sqlite3_stmt*,int,double);
................................................................................
  const char *(*uri_parameter)(const char*,const char*);
  char *(*vsnprintf)(int,char*,const char*,va_list);
  int (*wal_checkpoint_v2)(sqlite3*,const char*,int,int*,int*);
};

/*
** The following macros redefine the API routines so that they are
** redirected throught the global sqlite3_api structure.
**
** This header file is also used by the loadext.c source file
** (part of the main SQLite library - not an extension) so that
** it can get access to the sqlite3_api_routines structure
** definition.  But the main library does not want to redefine
** the API.  So the redefinition macros are only valid if the
** SQLITE_CORE macros is undefined.







|







 







|







24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
...
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
/*
** The following structure holds pointers to all of the SQLite API
** routines.
**
** WARNING:  In order to maintain backwards compatibility, add new
** interfaces to the end of this structure only.  If you insert new
** interfaces in the middle of this structure, then older different
** versions of SQLite will not be able to load each other's shared
** libraries!
*/
struct sqlite3_api_routines {
  void * (*aggregate_context)(sqlite3_context*,int nBytes);
  int  (*aggregate_count)(sqlite3_context*);
  int  (*bind_blob)(sqlite3_stmt*,int,const void*,int n,void(*)(void*));
  int  (*bind_double)(sqlite3_stmt*,int,double);
................................................................................
  const char *(*uri_parameter)(const char*,const char*);
  char *(*vsnprintf)(int,char*,const char*,va_list);
  int (*wal_checkpoint_v2)(sqlite3*,const char*,int,int*,int*);
};

/*
** The following macros redefine the API routines so that they are
** redirected through the global sqlite3_api structure.
**
** This header file is also used by the loadext.c source file
** (part of the main SQLite library - not an extension) so that
** it can get access to the sqlite3_api_routines structure
** definition.  But the main library does not want to redefine
** the API.  So the redefinition macros are only valid if the
** SQLITE_CORE macros is undefined.

Changes to src/sqliteInt.h.

347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
...
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
....
2879
2880
2881
2882
2883
2884
2885
2886
2887
2888
2889
2890
2891
2892
2893
2894
# define NEVER(X)       ((X)?(assert(0),1):0)
#else
# define ALWAYS(X)      (X)
# define NEVER(X)       (X)
#endif

/*
** Return true (non-zero) if the input is a integer that is too large
** to fit in 32-bits.  This macro is used inside of various testcase()
** macros to verify that we have tested SQLite for large-file support.
*/
#define IS_BIG_INT(X)  (((X)&~(i64)0xffffffff)!=0)

/*
** The macro unlikely() is a hint that surrounds a boolean
................................................................................

/*
** Assert that the pointer X is aligned to an 8-byte boundary.  This
** macro is used only within assert() to verify that the code gets
** all alignment restrictions correct.
**
** Except, if SQLITE_4_BYTE_ALIGNED_MALLOC is defined, then the
** underlying malloc() implemention might return us 4-byte aligned
** pointers.  In that case, only verify 4-byte alignment.
*/
#ifdef SQLITE_4_BYTE_ALIGNED_MALLOC
# define EIGHT_BYTE_ALIGNMENT(X)   ((((char*)(X) - (char*)0)&3)==0)
#else
# define EIGHT_BYTE_ALIGNMENT(X)   ((((char*)(X) - (char*)0)&7)==0)
#endif
................................................................................
#define SQLITE_CORRUPT_BKPT sqlite3CorruptError(__LINE__)
#define SQLITE_MISUSE_BKPT sqlite3MisuseError(__LINE__)
#define SQLITE_CANTOPEN_BKPT sqlite3CantopenError(__LINE__)


/*
** FTS4 is really an extension for FTS3.  It is enabled using the
** SQLITE_ENABLE_FTS3 macro.  But to avoid confusion we also all
** the SQLITE_ENABLE_FTS4 macro to serve as an alisse for SQLITE_ENABLE_FTS3.
*/
#if defined(SQLITE_ENABLE_FTS4) && !defined(SQLITE_ENABLE_FTS3)
# define SQLITE_ENABLE_FTS3
#endif

/*
** The ctype.h header is needed for non-ASCII systems.  It is also







|







 







|







 







|
|







347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
...
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
....
2879
2880
2881
2882
2883
2884
2885
2886
2887
2888
2889
2890
2891
2892
2893
2894
# define NEVER(X)       ((X)?(assert(0),1):0)
#else
# define ALWAYS(X)      (X)
# define NEVER(X)       (X)
#endif

/*
** Return true (non-zero) if the input is an integer that is too large
** to fit in 32-bits.  This macro is used inside of various testcase()
** macros to verify that we have tested SQLite for large-file support.
*/
#define IS_BIG_INT(X)  (((X)&~(i64)0xffffffff)!=0)

/*
** The macro unlikely() is a hint that surrounds a boolean
................................................................................

/*
** Assert that the pointer X is aligned to an 8-byte boundary.  This
** macro is used only within assert() to verify that the code gets
** all alignment restrictions correct.
**
** Except, if SQLITE_4_BYTE_ALIGNED_MALLOC is defined, then the
** underlying malloc() implementation might return us 4-byte aligned
** pointers.  In that case, only verify 4-byte alignment.
*/
#ifdef SQLITE_4_BYTE_ALIGNED_MALLOC
# define EIGHT_BYTE_ALIGNMENT(X)   ((((char*)(X) - (char*)0)&3)==0)
#else
# define EIGHT_BYTE_ALIGNMENT(X)   ((((char*)(X) - (char*)0)&7)==0)
#endif
................................................................................
#define SQLITE_CORRUPT_BKPT sqlite3CorruptError(__LINE__)
#define SQLITE_MISUSE_BKPT sqlite3MisuseError(__LINE__)
#define SQLITE_CANTOPEN_BKPT sqlite3CantopenError(__LINE__)


/*
** FTS4 is really an extension for FTS3.  It is enabled using the
** SQLITE_ENABLE_FTS3 macro.  But to avoid confusion we also call
** the SQLITE_ENABLE_FTS4 macro to serve as an alias for SQLITE_ENABLE_FTS3.
*/
#if defined(SQLITE_ENABLE_FTS4) && !defined(SQLITE_ENABLE_FTS3)
# define SQLITE_ENABLE_FTS3
#endif

/*
** The ctype.h header is needed for non-ASCII systems.  It is also

Changes to src/table.c.

178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
  return rc;
}

/*
** This routine frees the space the sqlite3_get_table() malloced.
*/
void sqlite3_free_table(
  char **azResult            /* Result returned from from sqlite3_get_table() */
){
  if( azResult ){
    int i, n;
    azResult--;
    assert( azResult!=0 );
    n = SQLITE_PTR_TO_INT(azResult[0]);
    for(i=1; i<n; i++){ if( azResult[i] ) sqlite3_free(azResult[i]); }
    sqlite3_free(azResult);
  }
}

#endif /* SQLITE_OMIT_GET_TABLE */







|












178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
  return rc;
}

/*
** This routine frees the space the sqlite3_get_table() malloced.
*/
void sqlite3_free_table(
  char **azResult            /* Result returned from sqlite3_get_table() */
){
  if( azResult ){
    int i, n;
    azResult--;
    assert( azResult!=0 );
    n = SQLITE_PTR_TO_INT(azResult[0]);
    for(i=1; i<n; i++){ if( azResult[i] ) sqlite3_free(azResult[i]); }
    sqlite3_free(azResult);
  }
}

#endif /* SQLITE_OMIT_GET_TABLE */

Changes to src/tclsqlite.c.

756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
    Tcl_IncrRefCount(pCmd);
    rc = Tcl_EvalObjEx(p->interp, pCmd, 0);
    Tcl_DecrRefCount(pCmd);
  }else{
    /* If there are arguments to the function, make a shallow copy of the
    ** script object, lappend the arguments, then evaluate the copy.
    **
    ** By "shallow" copy, we mean a only the outer list Tcl_Obj is duplicated.
    ** The new Tcl_Obj contains pointers to the original list elements. 
    ** That way, when Tcl_EvalObjv() is run and shimmers the first element
    ** of the list to tclCmdNameType, that alternate representation will
    ** be preserved and reused on the next invocation.
    */
    Tcl_Obj **aArg;
    int nArg;







|







756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
    Tcl_IncrRefCount(pCmd);
    rc = Tcl_EvalObjEx(p->interp, pCmd, 0);
    Tcl_DecrRefCount(pCmd);
  }else{
    /* If there are arguments to the function, make a shallow copy of the
    ** script object, lappend the arguments, then evaluate the copy.
    **
    ** By "shallow" copy, we mean only the outer list Tcl_Obj is duplicated.
    ** The new Tcl_Obj contains pointers to the original list elements. 
    ** That way, when Tcl_EvalObjv() is run and shimmers the first element
    ** of the list to tclCmdNameType, that alternate representation will
    ** be preserved and reused on the next invocation.
    */
    Tcl_Obj **aArg;
    int nArg;

Changes to src/test1.c.

2601
2602
2603
2604
2605
2606
2607
2608
2609
2610
2611
2612
2613
2614
2615
....
3754
3755
3756
3757
3758
3759
3760
3761
3762
3763
3764
3765
3766
3767
3768
**   "test_collate <enc> <lhs> <rhs>"
**
** The <lhs> and <rhs> are the two values being compared, encoded in UTF-8.
** The <enc> parameter is the encoding of the collation function that
** SQLite selected to call. The TCL test script implements the
** "test_collate" proc.
**
** Note that this will only work with one intepreter at a time, as the
** interp pointer to use when evaluating the TCL script is stored in
** pTestCollateInterp.
*/
static Tcl_Interp* pTestCollateInterp;
static int test_collate_func(
  void *pCtx, 
  int nA, const void *zA,
................................................................................
  return TCL_OK;
}

/*
** Usage: sqlite3_prepare_tkt3134 DB
**
** Generate a prepared statement for a zero-byte string as a test
** for ticket #3134.  The string should be preceeded by a zero byte.
*/
static int test_prepare_tkt3134(
  void * clientData,
  Tcl_Interp *interp,
  int objc,
  Tcl_Obj *CONST objv[]
){







|







 







|







2601
2602
2603
2604
2605
2606
2607
2608
2609
2610
2611
2612
2613
2614
2615
....
3754
3755
3756
3757
3758
3759
3760
3761
3762
3763
3764
3765
3766
3767
3768
**   "test_collate <enc> <lhs> <rhs>"
**
** The <lhs> and <rhs> are the two values being compared, encoded in UTF-8.
** The <enc> parameter is the encoding of the collation function that
** SQLite selected to call. The TCL test script implements the
** "test_collate" proc.
**
** Note that this will only work with one interpreter at a time, as the
** interp pointer to use when evaluating the TCL script is stored in
** pTestCollateInterp.
*/
static Tcl_Interp* pTestCollateInterp;
static int test_collate_func(
  void *pCtx, 
  int nA, const void *zA,
................................................................................
  return TCL_OK;
}

/*
** Usage: sqlite3_prepare_tkt3134 DB
**
** Generate a prepared statement for a zero-byte string as a test
** for ticket #3134.  The string should be preceded by a zero byte.
*/
static int test_prepare_tkt3134(
  void * clientData,
  Tcl_Interp *interp,
  int objc,
  Tcl_Obj *CONST objv[]
){

Changes to src/test_intarray.c.

33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
  void (*xFree)(void*);     /* Function used to free a[] */
};

/* Objects used internally by the virtual table implementation */
typedef struct intarray_vtab intarray_vtab;
typedef struct intarray_cursor intarray_cursor;

/* A intarray table object */
struct intarray_vtab {
  sqlite3_vtab base;            /* Base class */
  sqlite3_intarray *pContent;   /* Content of the integer array */
};

/* A intarray cursor object */
struct intarray_cursor {
  sqlite3_vtab_cursor base;    /* Base class */
  int i;                       /* Current cursor position */
};

/*
** None of this works unless we have virtual tables.







|





|







33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
  void (*xFree)(void*);     /* Function used to free a[] */
};

/* Objects used internally by the virtual table implementation */
typedef struct intarray_vtab intarray_vtab;
typedef struct intarray_cursor intarray_cursor;

/* An intarray table object */
struct intarray_vtab {
  sqlite3_vtab base;            /* Base class */
  sqlite3_intarray *pContent;   /* Content of the integer array */
};

/* An intarray cursor object */
struct intarray_cursor {
  sqlite3_vtab_cursor base;    /* Base class */
  int i;                       /* Current cursor position */
};

/*
** None of this works unless we have virtual tables.

Changes to src/test_malloc.c.

883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
...
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
}

/*
** Usage:    sqlite3_config_scratch SIZE N
**
** Set the scratch memory buffer using SQLITE_CONFIG_SCRATCH.
** The buffer is static and is of limited size.  N might be
** adjusted downward as needed to accomodate the requested size.
** The revised value of N is returned.
**
** A negative SIZE causes the buffer pointer to be NULL.
*/
static int test_config_scratch(
  void * clientData,
  Tcl_Interp *interp,
................................................................................
}

/*
** Usage:    sqlite3_config_pagecache SIZE N
**
** Set the page-cache memory buffer using SQLITE_CONFIG_PAGECACHE.
** The buffer is static and is of limited size.  N might be
** adjusted downward as needed to accomodate the requested size.
** The revised value of N is returned.
**
** A negative SIZE causes the buffer pointer to be NULL.
*/
static int test_config_pagecache(
  void * clientData,
  Tcl_Interp *interp,







|







 







|







883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
...
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
}

/*
** Usage:    sqlite3_config_scratch SIZE N
**
** Set the scratch memory buffer using SQLITE_CONFIG_SCRATCH.
** The buffer is static and is of limited size.  N might be
** adjusted downward as needed to accommodate the requested size.
** The revised value of N is returned.
**
** A negative SIZE causes the buffer pointer to be NULL.
*/
static int test_config_scratch(
  void * clientData,
  Tcl_Interp *interp,
................................................................................
}

/*
** Usage:    sqlite3_config_pagecache SIZE N
**
** Set the page-cache memory buffer using SQLITE_CONFIG_PAGECACHE.
** The buffer is static and is of limited size.  N might be
** adjusted downward as needed to accommodate the requested size.
** The revised value of N is returned.
**
** A negative SIZE causes the buffer pointer to be NULL.
*/
static int test_config_pagecache(
  void * clientData,
  Tcl_Interp *interp,

Changes to src/test_schema.c.

185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
      while( SQLITE_ROW!=sqlite3_step(pCur->pDbList) ){
        rc = finalize(&pCur->pDbList);
        goto next_exit;
      }

      /* Set zSql to the SQL to pull the list of tables from the 
      ** sqlite_master (or sqlite_temp_master) table of the database
      ** identfied by the row pointed to by the SQL statement pCur->pDbList
      ** (iterating through a "PRAGMA database_list;" statement).
      */
      if( sqlite3_column_int(pCur->pDbList, 0)==1 ){
        zSql = sqlite3_mprintf(
            "SELECT name FROM sqlite_temp_master WHERE type='table'"
        );
      }else{







|







185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
      while( SQLITE_ROW!=sqlite3_step(pCur->pDbList) ){
        rc = finalize(&pCur->pDbList);
        goto next_exit;
      }

      /* Set zSql to the SQL to pull the list of tables from the 
      ** sqlite_master (or sqlite_temp_master) table of the database
      ** identified by the row pointed to by the SQL statement pCur->pDbList
      ** (iterating through a "PRAGMA database_list;" statement).
      */
      if( sqlite3_column_int(pCur->pDbList, 0)==1 ){
        zSql = sqlite3_mprintf(
            "SELECT name FROM sqlite_temp_master WHERE type='table'"
        );
      }else{

Changes to src/tokenize.c.

73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
** allowed in an identifier.  For 7-bit characters, 
** sqlite3IsIdChar[X] must be 1.
**
** For EBCDIC, the rules are more complex but have the same
** end result.
**
** Ticket #1066.  the SQL standard does not allow '$' in the
** middle of identfiers.  But many SQL implementations do. 
** SQLite will allow '$' in identifiers for compatibility.
** But the feature is undocumented.
*/
#ifdef SQLITE_ASCII
#define IdChar(C)  ((sqlite3CtypeMap[(unsigned char)C]&0x46)!=0)
#endif
#ifdef SQLITE_EBCDIC







|







73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
** allowed in an identifier.  For 7-bit characters, 
** sqlite3IsIdChar[X] must be 1.
**
** For EBCDIC, the rules are more complex but have the same
** end result.
**
** Ticket #1066.  the SQL standard does not allow '$' in the
** middle of identifiers.  But many SQL implementations do. 
** SQLite will allow '$' in identifiers for compatibility.
** But the feature is undocumented.
*/
#ifdef SQLITE_ASCII
#define IdChar(C)  ((sqlite3CtypeMap[(unsigned char)C]&0x46)!=0)
#endif
#ifdef SQLITE_EBCDIC

Changes to src/trigger.c.

123
124
125
126
127
128
129
130
131
132
133
134
135
136
137

  /* A long-standing parser bug is that this syntax was allowed:
  **
  **    CREATE TRIGGER attached.demo AFTER INSERT ON attached.tab ....
  **                                                 ^^^^^^^^
  **
  ** To maintain backwards compatibility, ignore the database
  ** name on pTableName if we are reparsing our of SQLITE_MASTER.
  */
  if( db->init.busy && iDb!=1 ){
    sqlite3DbFree(db, pTableName->a[0].zDatabase);
    pTableName->a[0].zDatabase = 0;
  }

  /* If the trigger name was unqualified, and the table is a temp table,







|







123
124
125
126
127
128
129
130
131
132
133
134
135
136
137

  /* A long-standing parser bug is that this syntax was allowed:
  **
  **    CREATE TRIGGER attached.demo AFTER INSERT ON attached.tab ....
  **                                                 ^^^^^^^^
  **
  ** To maintain backwards compatibility, ignore the database
  ** name on pTableName if we are reparsing out of SQLITE_MASTER.
  */
  if( db->init.busy && iDb!=1 ){
    sqlite3DbFree(db, pTableName->a[0].zDatabase);
    pTableName->a[0].zDatabase = 0;
  }

  /* If the trigger name was unqualified, and the table is a temp table,

Changes to src/update.c.

323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
...
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
...
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

  /* Start the view context. */
  if( isView ){
    sqlite3AuthContextPush(pParse, &sContext, pTab->zName);
  }

  /* If we are trying to update a view, realize that view into
  ** a ephemeral table.
  */
#if !defined(SQLITE_OMIT_VIEW) && !defined(SQLITE_OMIT_TRIGGER)
  if( isView ){
    sqlite3MaterializeView(pParse, pTab, pWhere, iDataCur);
  }
#endif

................................................................................
    }
    if( chngRowid==0 && pPk==0 ){
      sqlite3VdbeAddOp2(v, OP_Copy, regOldRowid, regNewRowid);
    }
  }

  /* Populate the array of registers beginning at regNew with the new
  ** row data. This array is used to check constaints, create the new
  ** table and index records, and as the values for any new.* references
  ** made by triggers.
  **
  ** If there are one or more BEFORE triggers, then do not populate the
  ** registers associated with columns that are (a) not modified by
  ** this UPDATE statement and (b) not accessed by new.* references. The
  ** values for registers not modified by the UPDATE must be reloaded from 
................................................................................
  sqlite3DbFree(db, aXRef); /* Also frees aRegIdx[] and aToOpen[] */
  sqlite3SrcListDelete(db, pTabList);
  sqlite3ExprListDelete(db, pChanges);
  sqlite3ExprDelete(db, pWhere);
  return;
}
/* Make sure "isView" and other macros defined above are undefined. Otherwise
** thely may interfere with compilation of other functions in this file
** (or in another file, if this file becomes part of the amalgamation).  */
#ifdef isView
 #undef isView
#endif
#ifdef pTrigger
 #undef pTrigger
#endif

#ifndef SQLITE_OMIT_VIRTUALTABLE
/*
** Generate code for an UPDATE of a virtual table.
**
** The strategy is that we create an ephemerial table that contains
** for each row to be changed:
**
**   (A)  The original rowid of that row.
**   (B)  The revised rowid for the row. (note1)
**   (C)  The content of every column in the row.
**
** Then we loop over this ephemeral table and for each row in
** the ephermeral table call VUpdate.
**
** When finished, drop the ephemeral table.
**
** (note1) Actually, if we know in advance that (A) is always the same
** as (B) we only store (A), then duplicate (A) when pulling
** it out of the ephemeral table before calling VUpdate.
*/







|







 







|







 







|












|







|







323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
...
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
...
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

  /* Start the view context. */
  if( isView ){
    sqlite3AuthContextPush(pParse, &sContext, pTab->zName);
  }

  /* If we are trying to update a view, realize that view into
  ** an ephemeral table.
  */
#if !defined(SQLITE_OMIT_VIEW) && !defined(SQLITE_OMIT_TRIGGER)
  if( isView ){
    sqlite3MaterializeView(pParse, pTab, pWhere, iDataCur);
  }
#endif

................................................................................
    }
    if( chngRowid==0 && pPk==0 ){
      sqlite3VdbeAddOp2(v, OP_Copy, regOldRowid, regNewRowid);
    }
  }

  /* Populate the array of registers beginning at regNew with the new
  ** row data. This array is used to check constants, create the new
  ** table and index records, and as the values for any new.* references
  ** made by triggers.
  **
  ** If there are one or more BEFORE triggers, then do not populate the
  ** registers associated with columns that are (a) not modified by
  ** this UPDATE statement and (b) not accessed by new.* references. The
  ** values for registers not modified by the UPDATE must be reloaded from 
................................................................................
  sqlite3DbFree(db, aXRef); /* Also frees aRegIdx[] and aToOpen[] */
  sqlite3SrcListDelete(db, pTabList);
  sqlite3ExprListDelete(db, pChanges);
  sqlite3ExprDelete(db, pWhere);
  return;
}
/* Make sure "isView" and other macros defined above are undefined. Otherwise
** they may interfere with compilation of other functions in this file
** (or in another file, if this file becomes part of the amalgamation).  */
#ifdef isView
 #undef isView
#endif
#ifdef pTrigger
 #undef pTrigger
#endif

#ifndef SQLITE_OMIT_VIRTUALTABLE
/*
** Generate code for an UPDATE of a virtual table.
**
** The strategy is that we create an ephemeral table that contains
** for each row to be changed:
**
**   (A)  The original rowid of that row.
**   (B)  The revised rowid for the row. (note1)
**   (C)  The content of every column in the row.
**
** Then we loop over this ephemeral table and for each row in
** the ephemeral table call VUpdate.
**
** When finished, drop the ephemeral table.
**
** (note1) Actually, if we know in advance that (A) is always the same
** as (B) we only store (A), then duplicate (A) when pulling
** it out of the ephemeral table before calling VUpdate.
*/

Changes to src/util.c.

200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
** is added to the dequoted string.
**
** The return value is -1 if no dequoting occurs or the length of the
** dequoted string, exclusive of the zero terminator, if dequoting does
** occur.
**
** 2002-Feb-14: This routine is extended to remove MS-Access style
** brackets from around identifers.  For example:  "[a-b-c]" becomes
** "a-b-c".
*/
int sqlite3Dequote(char *z){
  char quote;
  int i, j;
  if( z==0 ) return -1;
  quote = z[0];







|







200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
** is added to the dequoted string.
**
** The return value is -1 if no dequoting occurs or the length of the
** dequoted string, exclusive of the zero terminator, if dequoting does
** occur.
**
** 2002-Feb-14: This routine is extended to remove MS-Access style
** brackets from around identifiers.  For example:  "[a-b-c]" becomes
** "a-b-c".
*/
int sqlite3Dequote(char *z){
  char quote;
  int i, j;
  if( z==0 ) return -1;
  quote = z[0];

Changes to src/vacuum.c.

83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
**        original database.
**
** The transient database requires temporary disk space approximately
** equal to the size of the original database.  The copy operation of
** step (3) requires additional temporary disk space approximately equal
** to the size of the original database for the rollback journal.
** Hence, temporary disk space that is approximately 2x the size of the
** orginal database is required.  Every page of the database is written
** approximately 3 times:  Once for step (2) and twice for step (3).
** Two writes per page are required in step (3) because the original
** database content must be written into the rollback journal prior to
** overwriting the database with the vacuumed content.
**
** Only 1x temporary space and only 1x writes would be required if
** the copy of step (3) were replace by deleting the original database







|







83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
**        original database.
**
** The transient database requires temporary disk space approximately
** equal to the size of the original database.  The copy operation of
** step (3) requires additional temporary disk space approximately equal
** to the size of the original database for the rollback journal.
** Hence, temporary disk space that is approximately 2x the size of the
** original database is required.  Every page of the database is written
** approximately 3 times:  Once for step (2) and twice for step (3).
** Two writes per page are required in step (3) because the original
** database content must be written into the rollback journal prior to
** overwriting the database with the vacuumed content.
**
** Only 1x temporary space and only 1x writes would be required if
** the copy of step (3) were replace by deleting the original database

Changes to src/vdbe.c.

2556
2557
2558
2559
2560
2561
2562
2563
2564
2565
2566
2567
2568
2569
2570
....
3543
3544
3545
3546
3547
3548
3549
3550
3551
3552
3553
3554
3555
3556
3557
  ** like this:
  **
  ** ------------------------------------------------------------------------
  ** | hdr-size | type 0 | type 1 | ... | type N-1 | data0 | ... | data N-1 | 
  ** ------------------------------------------------------------------------
  **
  ** Data(0) is taken from register P1.  Data(1) comes from register P1+1
  ** and so froth.
  **
  ** Each type field is a varint representing the serial type of the 
  ** corresponding data element (see sqlite3VdbeSerialType()). The
  ** hdr-size field is also a varint which is the offset from the beginning
  ** of the record to data0.
  */
  nData = 0;         /* Number of bytes of data space */
................................................................................
  pC->nullRow = 0;
#ifdef SQLITE_DEBUG
  pC->seekOp = pOp->opcode;
#endif
  if( pC->isTable ){
    /* The input value in P3 might be of any type: integer, real, string,
    ** blob, or NULL.  But it needs to be an integer before we can do
    ** the seek, so covert it. */
    pIn3 = &aMem[pOp->p3];
    if( (pIn3->flags & (MEM_Int|MEM_Real))==0 ){
      applyNumericAffinity(pIn3, 0);
    }
    iKey = sqlite3VdbeIntValue(pIn3);
    pC->rowidIsValid = 0;








|







 







|







2556
2557
2558
2559
2560
2561
2562
2563
2564
2565
2566
2567
2568
2569
2570
....
3543
3544
3545
3546
3547
3548
3549
3550
3551
3552
3553
3554
3555
3556
3557
  ** like this:
  **
  ** ------------------------------------------------------------------------
  ** | hdr-size | type 0 | type 1 | ... | type N-1 | data0 | ... | data N-1 | 
  ** ------------------------------------------------------------------------
  **
  ** Data(0) is taken from register P1.  Data(1) comes from register P1+1
  ** and so forth.
  **
  ** Each type field is a varint representing the serial type of the 
  ** corresponding data element (see sqlite3VdbeSerialType()). The
  ** hdr-size field is also a varint which is the offset from the beginning
  ** of the record to data0.
  */
  nData = 0;         /* Number of bytes of data space */
................................................................................
  pC->nullRow = 0;
#ifdef SQLITE_DEBUG
  pC->seekOp = pOp->opcode;
#endif
  if( pC->isTable ){
    /* The input value in P3 might be of any type: integer, real, string,
    ** blob, or NULL.  But it needs to be an integer before we can do
    ** the seek, so convert it. */
    pIn3 = &aMem[pOp->p3];
    if( (pIn3->flags & (MEM_Int|MEM_Real))==0 ){
      applyNumericAffinity(pIn3, 0);
    }
    iKey = sqlite3VdbeIntValue(pIn3);
    pC->rowidIsValid = 0;

Changes to src/vdbeInt.h.

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
** is for use inside assert() statements only.
*/
#ifdef SQLITE_DEBUG
#define memIsValid(M)  ((M)->flags & MEM_Undefined)==0
#endif

/*
** Each auxilliary data pointer stored by a user defined function 
** implementation calling sqlite3_set_auxdata() is stored in an instance
** of this structure. All such structures associated with a single VM
** are stored in a linked list headed at Vdbe.pAuxData. All are destroyed
** when the VM is halted (if not before).
*/
struct AuxData {
  int iOp;                        /* Instruction number of OP_Function opcode */
................................................................................
  int iArg;                       /* Index of function argument. */
  void *pAux;                     /* Aux data pointer */
  void (*xDelete)(void *);        /* Destructor for the aux data */
  AuxData *pNext;                 /* Next element in list */
};

/*
** The "context" argument for a installable function.  A pointer to an
** instance of this structure is the first argument to the routines used
** implement the SQL functions.
**
** There is a typedef for this structure in sqlite.h.  So all routines,
** even the public interface to SQLite, can use a pointer to this structure.
** But this file is the only place where the internal details of this
** structure are known.







|







 







|







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
** is for use inside assert() statements only.
*/
#ifdef SQLITE_DEBUG
#define memIsValid(M)  ((M)->flags & MEM_Undefined)==0
#endif

/*
** Each auxiliary data pointer stored by a user defined function 
** implementation calling sqlite3_set_auxdata() is stored in an instance
** of this structure. All such structures associated with a single VM
** are stored in a linked list headed at Vdbe.pAuxData. All are destroyed
** when the VM is halted (if not before).
*/
struct AuxData {
  int iOp;                        /* Instruction number of OP_Function opcode */
................................................................................
  int iArg;                       /* Index of function argument. */
  void *pAux;                     /* Aux data pointer */
  void (*xDelete)(void *);        /* Destructor for the aux data */
  AuxData *pNext;                 /* Next element in list */
};

/*
** The "context" argument for an installable function.  A pointer to an
** instance of this structure is the first argument to the routines used
** implement the SQL functions.
**
** There is a typedef for this structure in sqlite.h.  So all routines,
** even the public interface to SQLite, can use a pointer to this structure.
** But this file is the only place where the internal details of this
** structure are known.

Changes to src/vdbeapi.c.

208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
...
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
...
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
...
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
...
972
973
974
975
976
977
978
979
980
981
982
983
984
985
986
...
988
989
990
991
992
993
994
995
996
997
998
999
1000
1001
1002
....
1004
1005
1006
1007
1008
1009
1010
1011
1012
1013
1014
1015
1016
1017
1018
....
1284
1285
1286
1287
1288
1289
1290
1291
1292
1293
1294
1295
1296
1297
1298
  return aType[pVal->flags&MEM_AffMask];
}

/**************************** sqlite3_result_  *******************************
** The following routines are used by user-defined functions to specify
** the function result.
**
** The setStrOrError() funtion calls sqlite3VdbeMemSetStr() to store the
** result as a string or blob but if the string or blob is too large, it
** then sets the error code to SQLITE_TOOBIG
*/
static void setResultStrOrError(
  sqlite3_context *pCtx,  /* Function context */
  const char *z,          /* String pointer */
  int n,                  /* Bytes in string, or negative */
................................................................................
    return createAggContext(p, nByte);
  }else{
    return (void*)p->pMem->z;
  }
}

/*
** Return the auxilary data pointer, if any, for the iArg'th argument to
** the user-function defined by pCtx.
*/
void *sqlite3_get_auxdata(sqlite3_context *pCtx, int iArg){
  AuxData *pAuxData;

  assert( sqlite3_mutex_held(pCtx->pOut->db->mutex) );
  for(pAuxData=pCtx->pVdbe->pAuxData; pAuxData; pAuxData=pAuxData->pNext){
................................................................................
    if( pAuxData->iOp==pCtx->iOp && pAuxData->iArg==iArg ) break;
  }

  return (pAuxData ? pAuxData->pAux : 0);
}

/*
** Set the auxilary data pointer and delete function, for the iArg'th
** argument to the user-function defined by pCtx. Any previous value is
** deleted by calling the delete function specified when it was set.
*/
void sqlite3_set_auxdata(
  sqlite3_context *pCtx, 
  int iArg, 
  void *pAux, 
................................................................................
  if( xDelete ){
    xDelete(pAux);
  }
}

#ifndef SQLITE_OMIT_DEPRECATED
/*
** Return the number of times the Step function of a aggregate has been 
** called.
**
** This function is deprecated.  Do not use it for new code.  It is
** provide only to avoid breaking legacy code.  New aggregate function
** implementations should keep their own counts within their aggregate
** context.
*/
................................................................................
#endif /* SQLITE_OMIT_UTF16 */
#endif /* SQLITE_OMIT_DECLTYPE */

#ifdef SQLITE_ENABLE_COLUMN_METADATA
/*
** Return the name of the database from which a result column derives.
** NULL is returned if the result column is an expression or constant or
** anything else which is not an unabiguous reference to a database column.
*/
const char *sqlite3_column_database_name(sqlite3_stmt *pStmt, int N){
  return columnName(
      pStmt, N, (const void*(*)(Mem*))sqlite3_value_text, COLNAME_DATABASE);
}
#ifndef SQLITE_OMIT_UTF16
const void *sqlite3_column_database_name16(sqlite3_stmt *pStmt, int N){
................................................................................
      pStmt, N, (const void*(*)(Mem*))sqlite3_value_text16, COLNAME_DATABASE);
}
#endif /* SQLITE_OMIT_UTF16 */

/*
** Return the name of the table from which a result column derives.
** NULL is returned if the result column is an expression or constant or
** anything else which is not an unabiguous reference to a database column.
*/
const char *sqlite3_column_table_name(sqlite3_stmt *pStmt, int N){
  return columnName(
      pStmt, N, (const void*(*)(Mem*))sqlite3_value_text, COLNAME_TABLE);
}
#ifndef SQLITE_OMIT_UTF16
const void *sqlite3_column_table_name16(sqlite3_stmt *pStmt, int N){
................................................................................
      pStmt, N, (const void*(*)(Mem*))sqlite3_value_text16, COLNAME_TABLE);
}
#endif /* SQLITE_OMIT_UTF16 */

/*
** Return the name of the table column from which a result column derives.
** NULL is returned if the result column is an expression or constant or
** anything else which is not an unabiguous reference to a database column.
*/
const char *sqlite3_column_origin_name(sqlite3_stmt *pStmt, int N){
  return columnName(
      pStmt, N, (const void*(*)(Mem*))sqlite3_value_text, COLNAME_COLUMN);
}
#ifndef SQLITE_OMIT_UTF16
const void *sqlite3_column_origin_name16(sqlite3_stmt *pStmt, int N){
................................................................................
}

#ifndef SQLITE_OMIT_DEPRECATED
/*
** Deprecated external interface.  Internal/core SQLite code
** should call sqlite3TransferBindings.
**
** Is is misuse to call this routine with statements from different
** database connections.  But as this is a deprecated interface, we
** will not bother to check for that condition.
**
** If the two statements contain a different number of bindings, then
** an SQLITE_ERROR is returned.  Nothing else can go wrong, so otherwise
** SQLITE_OK is returned.
*/







|







 







|







 







|







 







|







 







|







 







|







 







|







 







|







208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
...
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
...
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
...
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
...
972
973
974
975
976
977
978
979
980
981
982
983
984
985
986
...
988
989
990
991
992
993
994
995
996
997
998
999
1000
1001
1002
....
1004
1005
1006
1007
1008
1009
1010
1011
1012
1013
1014
1015
1016
1017
1018
....
1284
1285
1286
1287
1288
1289
1290
1291
1292
1293
1294
1295
1296
1297
1298
  return aType[pVal->flags&MEM_AffMask];
}

/**************************** sqlite3_result_  *******************************
** The following routines are used by user-defined functions to specify
** the function result.
**
** The setStrOrError() function calls sqlite3VdbeMemSetStr() to store the
** result as a string or blob but if the string or blob is too large, it
** then sets the error code to SQLITE_TOOBIG
*/
static void setResultStrOrError(
  sqlite3_context *pCtx,  /* Function context */
  const char *z,          /* String pointer */
  int n,                  /* Bytes in string, or negative */
................................................................................
    return createAggContext(p, nByte);
  }else{
    return (void*)p->pMem->z;
  }
}

/*
** Return the auxiliary data pointer, if any, for the iArg'th argument to
** the user-function defined by pCtx.
*/
void *sqlite3_get_auxdata(sqlite3_context *pCtx, int iArg){
  AuxData *pAuxData;

  assert( sqlite3_mutex_held(pCtx->pOut->db->mutex) );
  for(pAuxData=pCtx->pVdbe->pAuxData; pAuxData; pAuxData=pAuxData->pNext){
................................................................................
    if( pAuxData->iOp==pCtx->iOp && pAuxData->iArg==iArg ) break;
  }

  return (pAuxData ? pAuxData->pAux : 0);
}

/*
** Set the auxiliary data pointer and delete function, for the iArg'th
** argument to the user-function defined by pCtx. Any previous value is
** deleted by calling the delete function specified when it was set.
*/
void sqlite3_set_auxdata(
  sqlite3_context *pCtx, 
  int iArg, 
  void *pAux, 
................................................................................
  if( xDelete ){
    xDelete(pAux);
  }
}

#ifndef SQLITE_OMIT_DEPRECATED
/*
** Return the number of times the Step function of an aggregate has been 
** called.
**
** This function is deprecated.  Do not use it for new code.  It is
** provide only to avoid breaking legacy code.  New aggregate function
** implementations should keep their own counts within their aggregate
** context.
*/
................................................................................
#endif /* SQLITE_OMIT_UTF16 */
#endif /* SQLITE_OMIT_DECLTYPE */

#ifdef SQLITE_ENABLE_COLUMN_METADATA
/*
** Return the name of the database from which a result column derives.
** NULL is returned if the result column is an expression or constant or
** anything else which is not an unambiguous reference to a database column.
*/
const char *sqlite3_column_database_name(sqlite3_stmt *pStmt, int N){
  return columnName(
      pStmt, N, (const void*(*)(Mem*))sqlite3_value_text, COLNAME_DATABASE);
}
#ifndef SQLITE_OMIT_UTF16
const void *sqlite3_column_database_name16(sqlite3_stmt *pStmt, int N){
................................................................................
      pStmt, N, (const void*(*)(Mem*))sqlite3_value_text16, COLNAME_DATABASE);
}
#endif /* SQLITE_OMIT_UTF16 */

/*
** Return the name of the table from which a result column derives.
** NULL is returned if the result column is an expression or constant or
** anything else which is not an unambiguous reference to a database column.
*/
const char *sqlite3_column_table_name(sqlite3_stmt *pStmt, int N){
  return columnName(
      pStmt, N, (const void*(*)(Mem*))sqlite3_value_text, COLNAME_TABLE);
}
#ifndef SQLITE_OMIT_UTF16
const void *sqlite3_column_table_name16(sqlite3_stmt *pStmt, int N){
................................................................................
      pStmt, N, (const void*(*)(Mem*))sqlite3_value_text16, COLNAME_TABLE);
}
#endif /* SQLITE_OMIT_UTF16 */

/*
** Return the name of the table column from which a result column derives.
** NULL is returned if the result column is an expression or constant or
** anything else which is not an unambiguous reference to a database column.
*/
const char *sqlite3_column_origin_name(sqlite3_stmt *pStmt, int N){
  return columnName(
      pStmt, N, (const void*(*)(Mem*))sqlite3_value_text, COLNAME_COLUMN);
}
#ifndef SQLITE_OMIT_UTF16
const void *sqlite3_column_origin_name16(sqlite3_stmt *pStmt, int N){
................................................................................
}

#ifndef SQLITE_OMIT_DEPRECATED
/*
** Deprecated external interface.  Internal/core SQLite code
** should call sqlite3TransferBindings.
**
** It is misuse to call this routine with statements from different
** database connections.  But as this is a deprecated interface, we
** will not bother to check for that condition.
**
** If the two statements contain a different number of bindings, then
** an SQLITE_ERROR is returned.  Nothing else can go wrong, so otherwise
** SQLITE_OK is returned.
*/

Changes to src/vdbeaux.c.

891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
....
1606
1607
1608
1609
1610
1611
1612
1613
1614
1615
1616
1617
1618
1619
1620
1621
1622
....
1986
1987
1988
1989
1990
1991
1992
1993
1994
1995
1996
1997
1998
1999
2000
....
2634
2635
2636
2637
2638
2639
2640
2641
2642
2643
2644
2645
2646
2647
2648
....
2738
2739
2740
2741
2742
2743
2744
2745
2746
2747
2748
2749
2750
2751
2752
....
3263
3264
3265
3266
3267
3268
3269
3270
3271
3272
3273
3274
3275
3276
3277
....
3454
3455
3456
3457
3458
3459
3460
3461
3462
3463
3464
3465
3466
3467
3468
....
3644
3645
3646
3647
3648
3649
3650
3651
3652
3653
3654
3655
3656
3657
3658
** Return the opcode for a given address.  If the address is -1, then
** return the most recently inserted opcode.
**
** If a memory allocation error has occurred prior to the calling of this
** routine, then a pointer to a dummy VdbeOp will be returned.  That opcode
** is readable but not writable, though it is cast to a writable value.
** The return of a dummy opcode allows the call to continue functioning
** after a OOM fault without having to check to see if the return from 
** this routine is a valid pointer.  But because the dummy.opcode is 0,
** dummy will never be written to.  This is verified by code inspection and
** by running with Valgrind.
*/
VdbeOp *sqlite3VdbeGetOp(Vdbe *p, int addr){
  /* C89 specifies that the constant "dummy" will be initialized to all
  ** zeros, which is correct.  MSVC generates a warning, nevertheless. */
................................................................................
/*
** Prepare a virtual machine for execution for the first time after
** creating the virtual machine.  This involves things such
** as allocating stack space and initializing the program counter.
** After the VDBE has be prepped, it can be executed by one or more
** calls to sqlite3VdbeExec().  
**
** This function may be called exact once on a each virtual machine.
** After this routine is called the VM has been "packaged" and is ready
** to run.  After this routine is called, futher calls to 
** sqlite3VdbeAddOp() functions are prohibited.  This routine disconnects
** the Vdbe from the Parse object that helped generate it so that the
** the Vdbe becomes an independent entity and the Parse object can be
** destroyed.
**
** Use the sqlite3VdbeRewind() procedure to restore a virtual machine back
** to its initial state after it has been run.
................................................................................
    if( rc==SQLITE_OK ){
      sqlite3VtabCommit(db);
    }
  }

  /* The complex case - There is a multi-file write-transaction active.
  ** This requires a master journal file to ensure the transaction is
  ** committed atomicly.
  */
#ifndef SQLITE_OMIT_DISKIO
  else{
    sqlite3_vfs *pVfs = db->pVfs;
    int needSync = 0;
    char *zMaster = 0;   /* File-name for the master journal */
    char const *zMainFile = sqlite3BtreeGetFilename(db->aDb[0].pBt);
................................................................................
** function invoked by the OP_Function opcode at instruction iOp of 
** VM pVdbe, and only then if:
**
**    * the associated function parameter is the 32nd or later (counting
**      from left to right), or
**
**    * the corresponding bit in argument mask is clear (where the first
**      function parameter corrsponds to bit 0 etc.).
*/
void sqlite3VdbeDeleteAuxData(Vdbe *pVdbe, int iOp, int mask){
  AuxData **pp = &pVdbe->pAuxData;
  while( *pp ){
    AuxData *pAux = *pp;
    if( (iOp<0)
     || (pAux->iOp==iOp && (pAux->iArg>31 || !(mask & MASKBIT32(pAux->iArg))))
................................................................................
  return SQLITE_OK;
}

/*
** Something has moved cursor "p" out of place.  Maybe the row it was
** pointed to was deleted out from under it.  Or maybe the btree was
** rebalanced.  Whatever the cause, try to restore "p" to the place it
** is suppose to be pointing.  If the row was deleted out from under the
** cursor, set the cursor to point to a NULL row.
*/
static int SQLITE_NOINLINE handleMovedCursor(VdbeCursor *p){
  int isDifferentRow, rc;
  assert( p->pCursor!=0 );
  assert( sqlite3BtreeCursorHasMoved(p->pCursor) );
  rc = sqlite3BtreeCursorRestore(p->pCursor, &isDifferentRow);
................................................................................
  /* No memory allocation is ever used on mem1.  Prove this using
  ** the following assert().  If the assert() fails, it indicates a
  ** memory leak and a need to call sqlite3VdbeMemRelease(&mem1).
  */
  assert( mem1.zMalloc==0 );

  /* rc==0 here means that one of the keys ran out of fields and
  ** all the fields up to that point were equal. Return the the default_rc
  ** value.  */
  rc = pPKey2->default_rc;

debugCompareEnd:
  if( desiredResult==0 && rc==0 ) return 1;
  if( desiredResult<0 && rc<0 ) return 1;
  if( desiredResult>0 && rc>0 ) return 1;
................................................................................
}

/*
** This function compares the two table rows or index records
** specified by {nKey1, pKey1} and pPKey2.  It returns a negative, zero
** or positive integer if key1 is less than, equal to or 
** greater than key2.  The {nKey1, pKey1} key must be a blob
** created by th OP_MakeRecord opcode of the VDBE.  The pPKey2
** key must be a parsed key such as obtained from
** sqlite3VdbeParseRecord.
**
** If argument bSkip is non-zero, it is assumed that the caller has already
** determined that the first fields of the keys are equal.
**
** Key1 and Key2 do not have to contain the same number of fields. If all 
................................................................................

  /* No memory allocation is ever used on mem1.  Prove this using
  ** the following assert().  If the assert() fails, it indicates a
  ** memory leak and a need to call sqlite3VdbeMemRelease(&mem1).  */
  assert( mem1.zMalloc==0 );

  /* rc==0 here means that one or both of the keys ran out of fields and
  ** all the fields up to that point were equal. Return the the default_rc
  ** value.  */
  assert( CORRUPT_DB 
       || vdbeRecordCompareDebug(nKey1, pKey1, pPKey2, pPKey2->default_rc) 
       || pKeyInfo->db->mallocFailed
  );
  return pPKey2->default_rc;
}







|







 







|

|







 







|







 







|







 







|







 







|







 







|







 







|







891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
....
1606
1607
1608
1609
1610
1611
1612
1613
1614
1615
1616
1617
1618
1619
1620
1621
1622
....
1986
1987
1988
1989
1990
1991
1992
1993
1994
1995
1996
1997
1998
1999
2000
....
2634
2635
2636
2637
2638
2639
2640
2641
2642
2643
2644
2645
2646
2647
2648
....
2738
2739
2740
2741
2742
2743
2744
2745
2746
2747
2748
2749
2750
2751
2752
....
3263
3264
3265
3266
3267
3268
3269
3270
3271
3272
3273
3274
3275
3276
3277
....
3454
3455
3456
3457
3458
3459
3460
3461
3462
3463
3464
3465
3466
3467
3468
....
3644
3645
3646
3647
3648
3649
3650
3651
3652
3653
3654
3655
3656
3657
3658
** Return the opcode for a given address.  If the address is -1, then
** return the most recently inserted opcode.
**
** If a memory allocation error has occurred prior to the calling of this
** routine, then a pointer to a dummy VdbeOp will be returned.  That opcode
** is readable but not writable, though it is cast to a writable value.
** The return of a dummy opcode allows the call to continue functioning
** after an OOM fault without having to check to see if the return from 
** this routine is a valid pointer.  But because the dummy.opcode is 0,
** dummy will never be written to.  This is verified by code inspection and
** by running with Valgrind.
*/
VdbeOp *sqlite3VdbeGetOp(Vdbe *p, int addr){
  /* C89 specifies that the constant "dummy" will be initialized to all
  ** zeros, which is correct.  MSVC generates a warning, nevertheless. */
................................................................................
/*
** Prepare a virtual machine for execution for the first time after
** creating the virtual machine.  This involves things such
** as allocating stack space and initializing the program counter.
** After the VDBE has be prepped, it can be executed by one or more
** calls to sqlite3VdbeExec().  
**
** This function may be called exactly once on each virtual machine.
** After this routine is called the VM has been "packaged" and is ready
** to run.  After this routine is called, further calls to 
** sqlite3VdbeAddOp() functions are prohibited.  This routine disconnects
** the Vdbe from the Parse object that helped generate it so that the
** the Vdbe becomes an independent entity and the Parse object can be
** destroyed.
**
** Use the sqlite3VdbeRewind() procedure to restore a virtual machine back
** to its initial state after it has been run.
................................................................................
    if( rc==SQLITE_OK ){
      sqlite3VtabCommit(db);
    }
  }

  /* The complex case - There is a multi-file write-transaction active.
  ** This requires a master journal file to ensure the transaction is
  ** committed atomically.
  */
#ifndef SQLITE_OMIT_DISKIO
  else{
    sqlite3_vfs *pVfs = db->pVfs;
    int needSync = 0;
    char *zMaster = 0;   /* File-name for the master journal */
    char const *zMainFile = sqlite3BtreeGetFilename(db->aDb[0].pBt);
................................................................................
** function invoked by the OP_Function opcode at instruction iOp of 
** VM pVdbe, and only then if:
**
**    * the associated function parameter is the 32nd or later (counting
**      from left to right), or
**
**    * the corresponding bit in argument mask is clear (where the first
**      function parameter corresponds to bit 0 etc.).
*/
void sqlite3VdbeDeleteAuxData(Vdbe *pVdbe, int iOp, int mask){
  AuxData **pp = &pVdbe->pAuxData;
  while( *pp ){
    AuxData *pAux = *pp;
    if( (iOp<0)
     || (pAux->iOp==iOp && (pAux->iArg>31 || !(mask & MASKBIT32(pAux->iArg))))
................................................................................
  return SQLITE_OK;
}

/*
** Something has moved cursor "p" out of place.  Maybe the row it was
** pointed to was deleted out from under it.  Or maybe the btree was
** rebalanced.  Whatever the cause, try to restore "p" to the place it
** is supposed to be pointing.  If the row was deleted out from under the
** cursor, set the cursor to point to a NULL row.
*/
static int SQLITE_NOINLINE handleMovedCursor(VdbeCursor *p){
  int isDifferentRow, rc;
  assert( p->pCursor!=0 );
  assert( sqlite3BtreeCursorHasMoved(p->pCursor) );
  rc = sqlite3BtreeCursorRestore(p->pCursor, &isDifferentRow);
................................................................................
  /* No memory allocation is ever used on mem1.  Prove this using
  ** the following assert().  If the assert() fails, it indicates a
  ** memory leak and a need to call sqlite3VdbeMemRelease(&mem1).
  */
  assert( mem1.zMalloc==0 );

  /* rc==0 here means that one of the keys ran out of fields and
  ** all the fields up to that point were equal. Return the default_rc
  ** value.  */
  rc = pPKey2->default_rc;

debugCompareEnd:
  if( desiredResult==0 && rc==0 ) return 1;
  if( desiredResult<0 && rc<0 ) return 1;
  if( desiredResult>0 && rc>0 ) return 1;
................................................................................
}

/*
** This function compares the two table rows or index records
** specified by {nKey1, pKey1} and pPKey2.  It returns a negative, zero
** or positive integer if key1 is less than, equal to or 
** greater than key2.  The {nKey1, pKey1} key must be a blob
** created by the OP_MakeRecord opcode of the VDBE.  The pPKey2
** key must be a parsed key such as obtained from
** sqlite3VdbeParseRecord.
**
** If argument bSkip is non-zero, it is assumed that the caller has already
** determined that the first fields of the keys are equal.
**
** Key1 and Key2 do not have to contain the same number of fields. If all 
................................................................................

  /* No memory allocation is ever used on mem1.  Prove this using
  ** the following assert().  If the assert() fails, it indicates a
  ** memory leak and a need to call sqlite3VdbeMemRelease(&mem1).  */
  assert( mem1.zMalloc==0 );

  /* rc==0 here means that one or both of the keys ran out of fields and
  ** all the fields up to that point were equal. Return the default_rc
  ** value.  */
  assert( CORRUPT_DB 
       || vdbeRecordCompareDebug(nKey1, pKey1, pPKey2, pPKey2->default_rc) 
       || pKeyInfo->db->mallocFailed
  );
  return pPKey2->default_rc;
}

Changes to src/vdbemem.c.

33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
...
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
...
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
...
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
  assert( (p->flags & MEM_Dyn)!=0 || p->xDel==0 );

  /* If p holds a string or blob, the Mem.z must point to exactly
  ** one of the following:
  **
  **   (1) Memory in Mem.zMalloc and managed by the Mem object
  **   (2) Memory to be freed using Mem.xDel
  **   (3) An ephermal string or blob
  **   (4) A static string or blob
  */
  if( (p->flags & (MEM_Str|MEM_Blob)) && p->z!=0 ){
    assert( 
      ((p->z==p->zMalloc)? 1 : 0) +
      ((p->flags&MEM_Dyn)!=0 ? 1 : 0) +
      ((p->flags&MEM_Ephem)!=0 ? 1 : 0) +
................................................................................
** Existing representations MEM_Int and MEM_Real are invalidated if
** bForce is true but are retained if bForce is false.
**
** A MEM_Null value will never be passed to this function. This function is
** used for converting values to text for returning to the user (i.e. via
** sqlite3_value_text()), or for ensuring that values to be used as btree
** keys are strings. In the former case a NULL pointer is returned the
** user and the later is an internal programming error.
*/
int sqlite3VdbeMemStringify(Mem *pMem, u8 enc, u8 bForce){
  int fg = pMem->flags;
  const int nByte = 32;

  assert( pMem->db==0 || sqlite3_mutex_held(pMem->db->mutex) );
  assert( !(fg&MEM_Zero) );
................................................................................

/*
** Return some kind of integer value which is the best we can do
** at representing the value that *pMem describes as an integer.
** If pMem is an integer, then the value is exact.  If pMem is
** a floating-point then the value returned is the integer part.
** If pMem is a string or blob, then we make an attempt to convert
** it into a integer and return that.  If pMem represents an
** an SQL-NULL value, return 0.
**
** If pMem represents a string value, its encoding might be changed.
*/
i64 sqlite3VdbeIntValue(Mem *pMem){
  int flags;
  assert( pMem->db==0 || sqlite3_mutex_held(pMem->db->mutex) );
................................................................................
    return n>p->db->aLimit[SQLITE_LIMIT_LENGTH];
  }
  return 0; 
}

#ifdef SQLITE_DEBUG
/*
** This routine prepares a memory cell for modication by breaking
** its link to a shallow copy and by marking any current shallow
** copies of this cell as invalid.
**
** This is used for testing and debugging only - to make sure shallow
** copies are not misused.
*/
void sqlite3VdbeMemAboutToChange(Vdbe *pVdbe, Mem *pMem){







|







 







|







 







|







 







|







33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
...
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
...
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
...
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
  assert( (p->flags & MEM_Dyn)!=0 || p->xDel==0 );

  /* If p holds a string or blob, the Mem.z must point to exactly
  ** one of the following:
  **
  **   (1) Memory in Mem.zMalloc and managed by the Mem object
  **   (2) Memory to be freed using Mem.xDel
  **   (3) An ephemeral string or blob
  **   (4) A static string or blob
  */
  if( (p->flags & (MEM_Str|MEM_Blob)) && p->z!=0 ){
    assert( 
      ((p->z==p->zMalloc)? 1 : 0) +
      ((p->flags&MEM_Dyn)!=0 ? 1 : 0) +
      ((p->flags&MEM_Ephem)!=0 ? 1 : 0) +
................................................................................
** Existing representations MEM_Int and MEM_Real are invalidated if
** bForce is true but are retained if bForce is false.
**
** A MEM_Null value will never be passed to this function. This function is
** used for converting values to text for returning to the user (i.e. via
** sqlite3_value_text()), or for ensuring that values to be used as btree
** keys are strings. In the former case a NULL pointer is returned the
** user and the latter is an internal programming error.
*/
int sqlite3VdbeMemStringify(Mem *pMem, u8 enc, u8 bForce){
  int fg = pMem->flags;
  const int nByte = 32;

  assert( pMem->db==0 || sqlite3_mutex_held(pMem->db->mutex) );
  assert( !(fg&MEM_Zero) );
................................................................................

/*
** Return some kind of integer value which is the best we can do
** at representing the value that *pMem describes as an integer.
** If pMem is an integer, then the value is exact.  If pMem is
** a floating-point then the value returned is the integer part.
** If pMem is a string or blob, then we make an attempt to convert
** it into an integer and return that.  If pMem represents an
** an SQL-NULL value, return 0.
**
** If pMem represents a string value, its encoding might be changed.
*/
i64 sqlite3VdbeIntValue(Mem *pMem){
  int flags;
  assert( pMem->db==0 || sqlite3_mutex_held(pMem->db->mutex) );
................................................................................
    return n>p->db->aLimit[SQLITE_LIMIT_LENGTH];
  }
  return 0; 
}

#ifdef SQLITE_DEBUG
/*
** This routine prepares a memory cell for modification by breaking
** its link to a shallow copy and by marking any current shallow
** copies of this cell as invalid.
**
** This is used for testing and debugging only - to make sure shallow
** copies are not misused.
*/
void sqlite3VdbeMemAboutToChange(Vdbe *pVdbe, Mem *pMem){

Changes to src/vdbesort.c.

2055
2056
2057
2058
2059
2060
2061
2062
2063
2064
2065
2066
2067
2068
2069
               );
  pReader->pIncr->pTask->bDone = 1;
  return pRet;
}

/*
** Use a background thread to invoke vdbePmaReaderIncrMergeInit(INCRINIT_TASK) 
** on the the PmaReader object passed as the first argument.
**
** This call will initialize the various fields of the pReadr->pIncr 
** structure and, if it is a multi-threaded IncrMerger, launch a 
** background thread to populate aFile[1].
*/
static int vdbePmaReaderBgIncrInit(PmaReader *pReadr){
  void *pCtx = (void*)pReadr;







|







2055
2056
2057
2058
2059
2060
2061
2062
2063
2064
2065
2066
2067
2068
2069
               );
  pReader->pIncr->pTask->bDone = 1;
  return pRet;
}

/*
** Use a background thread to invoke vdbePmaReaderIncrMergeInit(INCRINIT_TASK) 
** on the PmaReader object passed as the first argument.
**
** This call will initialize the various fields of the pReadr->pIncr 
** structure and, if it is a multi-threaded IncrMerger, launch a 
** background thread to populate aFile[1].
*/
static int vdbePmaReaderBgIncrInit(PmaReader *pReadr){
  void *pCtx = (void*)pReadr;

Changes to src/vdbetrace.c.

60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
**
** The calling function is responsible for making sure the memory returned
** is eventually freed.
**
** ALGORITHM:  Scan the input string looking for host parameters in any of
** these forms:  ?, ?N, $A, @A, :A.  Take care to avoid text within
** string literals, quoted identifier names, and comments.  For text forms,
** the host parameter index is found by scanning the perpared
** statement for the corresponding OP_Variable opcode.  Once the host
** parameter index is known, locate the value in p->aVar[].  Then render
** the value as a literal in place of the host parameter name.
*/
char *sqlite3VdbeExpandSql(
  Vdbe *p,                 /* The prepared statement being evaluated */
  const char *zRawSql      /* Raw text of the SQL statement */







|







60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
**
** The calling function is responsible for making sure the memory returned
** is eventually freed.
**
** ALGORITHM:  Scan the input string looking for host parameters in any of
** these forms:  ?, ?N, $A, @A, :A.  Take care to avoid text within
** string literals, quoted identifier names, and comments.  For text forms,
** the host parameter index is found by scanning the prepared
** statement for the corresponding OP_Variable opcode.  Once the host
** parameter index is known, locate the value in p->aVar[].  Then render
** the value as a literal in place of the host parameter name.
*/
char *sqlite3VdbeExpandSql(
  Vdbe *p,                 /* The prepared statement being evaluated */
  const char *zRawSql      /* Raw text of the SQL statement */

Changes to src/wal.c.

570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
...
984
985
986
987
988
989
990
991
992
993
994
995
996
997
998
....
1642
1643
1644
1645
1646
1647
1648
1649
1650
1651
1652
1653
1654
1655
1656
....
1946
1947
1948
1949
1950
1951
1952
1953
1954
1955
1956
1957
1958
1959
1960
....
2150
2151
2152
2153
2154
2155
2156
2157
2158
2159
2160
2161
2162
2163
2164
....
2818
2819
2820
2821
2822
2823
2824
2825
2826
2827
2828
2829
2830
2831
2832
  return (volatile WalIndexHdr*)pWal->apWiData[0];
}

/*
** The argument to this macro must be of type u32. On a little-endian
** architecture, it returns the u32 value that results from interpreting
** the 4 bytes as a big-endian value. On a big-endian architecture, it
** returns the value that would be produced by intepreting the 4 bytes
** of the input value as a little-endian integer.
*/
#define BYTESWAP32(x) ( \
    (((x)&0x000000FF)<<24) + (((x)&0x0000FF00)<<8)  \
  + (((x)&0x00FF0000)>>8)  + (((x)&0xFF000000)>>24) \
)

................................................................................
    int idx;                      /* Value to write to hash-table slot */
    int nCollide;                 /* Number of hash collisions */

    idx = iFrame - iZero;
    assert( idx <= HASHTABLE_NSLOT/2 + 1 );
    
    /* If this is the first entry to be added to this hash-table, zero the
    ** entire hash table and aPgno[] array before proceding. 
    */
    if( idx==1 ){
      int nByte = (int)((u8 *)&aHash[HASHTABLE_NSLOT] - (u8 *)&aPgno[1]);
      memset((void*)&aPgno[1], 0, nByte);
    }

    /* If the entry in aPgno[] is already set, then the previous writer
................................................................................
**
** Fsync is also called on the database file if (and only if) the entire
** WAL content is copied into the database file.  This second fsync makes
** it safe to delete the WAL since the new content will persist in the
** database file.
**
** This routine uses and updates the nBackfill field of the wal-index header.
** This is the only routine tha will increase the value of nBackfill.  
** (A WAL reset or recovery will revert nBackfill to zero, but not increase
** its value.)
**
** The caller must be holding sufficient locks to ensure that no other
** checkpoint is running (in any other thread or process) at the same
** time.
*/
................................................................................

/*
** Read the wal-index header from the wal-index and into pWal->hdr.
** If the wal-header appears to be corrupt, try to reconstruct the
** wal-index from the WAL before returning.
**
** Set *pChanged to 1 if the wal-index header value in pWal->hdr is
** changed by this opertion.  If pWal->hdr is unchanged, set *pChanged
** to 0.
**
** If the wal-index header is successfully read, return SQLITE_OK. 
** Otherwise an SQLite error code.
*/
static int walIndexReadHdr(Wal *pWal, int *pChanged){
  int rc;                         /* Return code */
................................................................................
    walShmBarrier(pWal);
    if( rc==SQLITE_OK ){
      if( memcmp((void *)walIndexHdr(pWal), &pWal->hdr, sizeof(WalIndexHdr)) ){
        /* It is not safe to allow the reader to continue here if frames
        ** may have been appended to the log before READ_LOCK(0) was obtained.
        ** When holding READ_LOCK(0), the reader ignores the entire log file,
        ** which implies that the database file contains a trustworthy
        ** snapshoT. Since holding READ_LOCK(0) prevents a checkpoint from
        ** happening, this is usually correct.
        **
        ** However, if frames have been appended to the log (or if the log 
        ** is wrapped and written for that matter) before the READ_LOCK(0)
        ** is obtained, that is not necessarily true. A checkpointer may
        ** have started to backfill the appended frames but crashed before
        ** it finished. Leaving a corrupt image in the database file.
................................................................................
  }

  /* If this is the end of a transaction, then we might need to pad
  ** the transaction and/or sync the WAL file.
  **
  ** Padding and syncing only occur if this set of frames complete a
  ** transaction and if PRAGMA synchronous=FULL.  If synchronous==NORMAL
  ** or synchonous==OFF, then no padding or syncing are needed.
  **
  ** If SQLITE_IOCAP_POWERSAFE_OVERWRITE is defined, then padding is not
  ** needed and only the sync is done.  If padding is needed, then the
  ** final frame is repeated (with its commit mark) until the next sector
  ** boundary is crossed.  Only the part of the WAL prior to the last
  ** sector boundary is synced; the part of the last frame that extends
  ** past the sector boundary is written after the sync.







|







 







|







 







|







 







|







 







|







 







|







570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
...
984
985
986
987
988
989
990
991
992
993
994
995
996
997
998
....
1642
1643
1644
1645
1646
1647
1648
1649
1650
1651
1652
1653
1654
1655
1656
....
1946
1947
1948
1949
1950
1951
1952
1953
1954
1955
1956
1957
1958
1959
1960
....
2150
2151
2152
2153
2154
2155
2156
2157
2158
2159
2160
2161
2162
2163
2164
....
2818
2819
2820
2821
2822
2823
2824
2825
2826
2827
2828
2829
2830
2831
2832
  return (volatile WalIndexHdr*)pWal->apWiData[0];
}

/*
** The argument to this macro must be of type u32. On a little-endian
** architecture, it returns the u32 value that results from interpreting
** the 4 bytes as a big-endian value. On a big-endian architecture, it
** returns the value that would be produced by interpreting the 4 bytes
** of the input value as a little-endian integer.
*/
#define BYTESWAP32(x) ( \
    (((x)&0x000000FF)<<24) + (((x)&0x0000FF00)<<8)  \
  + (((x)&0x00FF0000)>>8)  + (((x)&0xFF000000)>>24) \
)

................................................................................
    int idx;                      /* Value to write to hash-table slot */
    int nCollide;                 /* Number of hash collisions */

    idx = iFrame - iZero;
    assert( idx <= HASHTABLE_NSLOT/2 + 1 );
    
    /* If this is the first entry to be added to this hash-table, zero the
    ** entire hash table and aPgno[] array before proceeding. 
    */
    if( idx==1 ){
      int nByte = (int)((u8 *)&aHash[HASHTABLE_NSLOT] - (u8 *)&aPgno[1]);
      memset((void*)&aPgno[1], 0, nByte);
    }

    /* If the entry in aPgno[] is already set, then the previous writer
................................................................................
**
** Fsync is also called on the database file if (and only if) the entire
** WAL content is copied into the database file.  This second fsync makes
** it safe to delete the WAL since the new content will persist in the
** database file.
**
** This routine uses and updates the nBackfill field of the wal-index header.
** This is the only routine that will increase the value of nBackfill.  
** (A WAL reset or recovery will revert nBackfill to zero, but not increase
** its value.)
**
** The caller must be holding sufficient locks to ensure that no other
** checkpoint is running (in any other thread or process) at the same
** time.
*/
................................................................................

/*
** Read the wal-index header from the wal-index and into pWal->hdr.
** If the wal-header appears to be corrupt, try to reconstruct the
** wal-index from the WAL before returning.
**
** Set *pChanged to 1 if the wal-index header value in pWal->hdr is
** changed by this operation.  If pWal->hdr is unchanged, set *pChanged
** to 0.
**
** If the wal-index header is successfully read, return SQLITE_OK. 
** Otherwise an SQLite error code.
*/
static int walIndexReadHdr(Wal *pWal, int *pChanged){
  int rc;                         /* Return code */
................................................................................
    walShmBarrier(pWal);
    if( rc==SQLITE_OK ){
      if( memcmp((void *)walIndexHdr(pWal), &pWal->hdr, sizeof(WalIndexHdr)) ){
        /* It is not safe to allow the reader to continue here if frames
        ** may have been appended to the log before READ_LOCK(0) was obtained.
        ** When holding READ_LOCK(0), the reader ignores the entire log file,
        ** which implies that the database file contains a trustworthy
        ** snapshot. Since holding READ_LOCK(0) prevents a checkpoint from
        ** happening, this is usually correct.
        **
        ** However, if frames have been appended to the log (or if the log 
        ** is wrapped and written for that matter) before the READ_LOCK(0)
        ** is obtained, that is not necessarily true. A checkpointer may
        ** have started to backfill the appended frames but crashed before
        ** it finished. Leaving a corrupt image in the database file.
................................................................................
  }

  /* If this is the end of a transaction, then we might need to pad
  ** the transaction and/or sync the WAL file.
  **
  ** Padding and syncing only occur if this set of frames complete a
  ** transaction and if PRAGMA synchronous=FULL.  If synchronous==NORMAL
  ** or synchronous==OFF, then no padding or syncing are needed.
  **
  ** If SQLITE_IOCAP_POWERSAFE_OVERWRITE is defined, then padding is not
  ** needed and only the sync is done.  If padding is needed, then the
  ** final frame is repeated (with its commit mark) until the next sector
  ** boundary is crossed.  Only the part of the WAL prior to the last
  ** sector boundary is synced; the part of the last frame that extends
  ** past the sector boundary is written after the sync.

Changes to src/walker.c.

15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
#include "sqliteInt.h"
#include <stdlib.h>
#include <string.h>


/*
** Walk an expression tree.  Invoke the callback once for each node
** of the expression, while decending.  (In other words, the callback
** is invoked before visiting children.)
**
** The return value from the callback should be one of the WRC_*
** constants to specify how to proceed with the walk.
**
**    WRC_Continue      Continue descending down the tree.
**







|







15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
#include "sqliteInt.h"
#include <stdlib.h>
#include <string.h>


/*
** Walk an expression tree.  Invoke the callback once for each node
** of the expression, while descending.  (In other words, the callback
** is invoked before visiting children.)
**
** The return value from the callback should be one of the WRC_*
** constants to specify how to proceed with the walk.
**
**    WRC_Continue      Continue descending down the tree.
**

Changes to src/where.c.

697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
...
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
...
975
976
977
978
979
980
981
982
983
984
985
986
987
988
989
....
1386
1387
1388
1389
1390
1391
1392
1393
1394
1395
1396
1397
1398
1399
1400
....
2136
2137
2138
2139
2140
2141
2142
2143
2144
2145
2146
2147
2148
2149
2150
....
3413
3414
3415
3416
3417
3418
3419
3420
3421
3422
3423
3424
3425
3426
3427
....
3464
3465
3466
3467
3468
3469
3470
3471
3472
3473
3474
3475
3476
3477
3478
....
4719
4720
4721
4722
4723
4724
4725
4726
4727
4728
4729
4730
4731
4732
4733
....
5149
5150
5151
5152
5153
5154
5155
5156
5157
5158
5159
5160
5161
5162
5163
        Vdbe *v = pParse->pVdbe;
        sqlite3VdbeSetVarmask(v, pRight->iColumn);
        if( *pisComplete && pRight->u.zToken[1] ){
          /* If the rhs of the LIKE expression is a variable, and the current
          ** value of the variable means there is no need to invoke the LIKE
          ** function, then no OP_Variable will be added to the program.
          ** This causes problems for the sqlite3_bind_parameter_name()
          ** API. To workaround them, add a dummy OP_Variable here.
          */ 
          int r1 = sqlite3GetTempReg(pParse);
          sqlite3ExprCodeTarget(pParse, pRight, r1);
          sqlite3VdbeChangeP3(v, sqlite3VdbeCurrentAddr(v)-1, 0);
          sqlite3ReleaseTempReg(pParse, r1);
        }
      }
................................................................................
** From another point of view, "indexable" means that the subterm could
** potentially be used with an index if an appropriate index exists.
** This analysis does not consider whether or not the index exists; that
** is decided elsewhere.  This analysis only looks at whether subterms
** appropriate for indexing exist.
**
** All examples A through E above satisfy case 2.  But if a term
** also statisfies case 1 (such as B) we know that the optimizer will
** always prefer case 1, so in that case we pretend that case 2 is not
** satisfied.
**
** It might be the case that multiple tables are indexable.  For example,
** (E) above is indexable on tables P, Q, and R.
**
** Terms that satisfy case 2 are candidates for lookup by using
................................................................................
          /* This is the 2-bit case and we are on the second iteration and
          ** current term is from the first iteration.  So skip this term. */
          assert( j==1 );
          continue;
        }
        if( (chngToIN & getMask(&pWInfo->sMaskSet, pOrTerm->leftCursor))==0 ){
          /* This term must be of the form t1.a==t2.b where t2 is in the
          ** chngToIN set but t1 is not.  This term will be either preceeded
          ** or follwed by an inverted copy (t2.b==t1.a).  Skip this term 
          ** and use its inversion. */
          testcase( pOrTerm->wtFlags & TERM_COPIED );
          testcase( pOrTerm->wtFlags & TERM_VIRTUAL );
          assert( pOrTerm->wtFlags & (TERM_COPIED|TERM_VIRTUAL) );
          continue;
        }
................................................................................
  /* Prevent ON clause terms of a LEFT JOIN from being used to drive
  ** an index for tables to the left of the join.
  */
  pTerm->prereqRight |= extraRight;
}

/*
** This function searches pList for a entry that matches the iCol-th column
** of index pIdx.
**
** If such an expression is found, its index in pList->a[] is returned. If
** no expression is found, -1 is returned.
*/
static int findIndexCol(
  Parse *pParse,                  /* Parse context */
................................................................................
**
** then nEq is set to 0.
**
** When this function is called, *pnOut is set to the sqlite3LogEst() of the
** number of rows that the index scan is expected to visit without 
** considering the range constraints. If nEq is 0, this is the number of 
** rows in the index. Assuming no error occurs, *pnOut is adjusted (reduced)
** to account for the range contraints pLower and pUpper.
** 
** In the absence of sqlite_stat4 ANALYZE data, or if such data cannot be
** used, a single range inequality reduces the search space by a factor of 4. 
** and a pair of constraints (x>? AND x<?) reduces the expected number of
** rows visited by a factor of 64.
*/
static int whereRangeScanEst(
................................................................................
    **       A: <loop body>                 # Return data, whatever.
    **
    **          Return     2                # Jump back to the Gosub
    **
    **       B: <after the loop>
    **
    ** Added 2014-05-26: If the table is a WITHOUT ROWID table, then
    ** use an ephermeral index instead of a RowSet to record the primary
    ** keys of the rows we have already seen.
    **
    */
    WhereClause *pOrWc;    /* The OR-clause broken out into subterms */
    SrcList *pOrTab;       /* Shortened table list or OR-clause generation */
    Index *pCov = 0;             /* Potential covering index (or NULL) */
    int iCovCur = pParse->nTab++;  /* Cursor used for index scans (if any) */
................................................................................
        memcpy(&pOrTab->a[k], &origSrc[pLevel[k].iFrom], sizeof(pOrTab->a[k]));
      }
    }else{
      pOrTab = pWInfo->pTabList;
    }

    /* Initialize the rowset register to contain NULL. An SQL NULL is 
    ** equivalent to an empty rowset.  Or, create an ephermeral index
    ** capable of holding primary keys in the case of a WITHOUT ROWID.
    **
    ** Also initialize regReturn to contain the address of the instruction 
    ** immediately following the OP_Return at the bottom of the loop. This
    ** is required in a few obscure LEFT JOIN cases where control jumps
    ** over the top of the loop into the body of it. In this case the 
    ** correct response for the end-of-loop code (the OP_Return) is to 
................................................................................
        /* TUNING: One-time cost for computing the automatic index is
        ** approximately 7*N*log2(N) where N is the number of rows in
        ** the table being indexed. */
        pNew->rSetup = rLogSize + rSize + 28;  assert( 28==sqlite3LogEst(7) );
        ApplyCostMultiplier(pNew->rSetup, pTab->costMult);
        /* TUNING: Each index lookup yields 20 rows in the table.  This
        ** is more than the usual guess of 10 rows, since we have no way
        ** of knowning how selective the index will ultimately be.  It would
        ** not be unreasonable to make this value much larger. */
        pNew->nOut = 43;  assert( 43==sqlite3LogEst(20) );
        pNew->rRun = sqlite3LogEstAdd(rLogSize,pNew->nOut);
        pNew->wsFlags = WHERE_AUTO_INDEX;
        pNew->prereq = mExtra | pTerm->prereqRight;
        rc = whereLoopInsert(pBuilder, pNew);
      }
................................................................................
**   N==0:  No terms of the ORDER BY clause are satisfied
**   N<0:   Unknown yet how many terms of ORDER BY might be satisfied.   
**
** Note that processing for WHERE_GROUPBY and WHERE_DISTINCTBY is not as
** strict.  With GROUP BY and DISTINCT the only requirement is that
** equivalent rows appear immediately adjacent to one another.  GROUP BY
** and DISTINCT do not require rows to appear in any particular order as long
** as equivelent rows are grouped together.  Thus for GROUP BY and DISTINCT
** the pOrderBy terms can be matched in any order.  With ORDER BY, the 
** pOrderBy terms must be matched in strict left-to-right order.
*/
static i8 wherePathSatisfiesOrderBy(
  WhereInfo *pWInfo,    /* The WHERE clause */
  ExprList *pOrderBy,   /* ORDER BY or GROUP BY or DISTINCT clause to check */
  WherePath *pPath,     /* The WherePath to check */







|







 







|







 







|







 







|







 







|







 







|







 







|







 







|







 







|







697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
...
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
...
975
976
977
978
979
980
981
982
983
984
985
986
987
988
989
....
1386
1387
1388
1389
1390
1391
1392
1393
1394
1395
1396
1397
1398
1399
1400
....
2136
2137
2138
2139
2140
2141
2142
2143
2144
2145
2146
2147
2148
2149
2150
....
3413
3414
3415
3416
3417
3418
3419
3420
3421
3422
3423
3424
3425
3426
3427
....
3464
3465
3466
3467
3468
3469
3470
3471
3472
3473
3474
3475
3476
3477
3478
....
4719
4720
4721
4722
4723
4724
4725
4726
4727
4728
4729
4730
4731
4732
4733
....
5149
5150
5151
5152
5153
5154
5155
5156
5157
5158
5159
5160
5161
5162
5163
        Vdbe *v = pParse->pVdbe;
        sqlite3VdbeSetVarmask(v, pRight->iColumn);
        if( *pisComplete && pRight->u.zToken[1] ){
          /* If the rhs of the LIKE expression is a variable, and the current
          ** value of the variable means there is no need to invoke the LIKE
          ** function, then no OP_Variable will be added to the program.
          ** This causes problems for the sqlite3_bind_parameter_name()
          ** API. To work around them, add a dummy OP_Variable here.
          */ 
          int r1 = sqlite3GetTempReg(pParse);
          sqlite3ExprCodeTarget(pParse, pRight, r1);
          sqlite3VdbeChangeP3(v, sqlite3VdbeCurrentAddr(v)-1, 0);
          sqlite3ReleaseTempReg(pParse, r1);
        }
      }
................................................................................
** From another point of view, "indexable" means that the subterm could
** potentially be used with an index if an appropriate index exists.
** This analysis does not consider whether or not the index exists; that
** is decided elsewhere.  This analysis only looks at whether subterms
** appropriate for indexing exist.
**
** All examples A through E above satisfy case 2.  But if a term
** also satisfies case 1 (such as B) we know that the optimizer will
** always prefer case 1, so in that case we pretend that case 2 is not
** satisfied.
**
** It might be the case that multiple tables are indexable.  For example,
** (E) above is indexable on tables P, Q, and R.
**
** Terms that satisfy case 2 are candidates for lookup by using
................................................................................
          /* This is the 2-bit case and we are on the second iteration and
          ** current term is from the first iteration.  So skip this term. */
          assert( j==1 );
          continue;
        }
        if( (chngToIN & getMask(&pWInfo->sMaskSet, pOrTerm->leftCursor))==0 ){
          /* This term must be of the form t1.a==t2.b where t2 is in the
          ** chngToIN set but t1 is not.  This term will be either preceded
          ** or follwed by an inverted copy (t2.b==t1.a).  Skip this term 
          ** and use its inversion. */
          testcase( pOrTerm->wtFlags & TERM_COPIED );
          testcase( pOrTerm->wtFlags & TERM_VIRTUAL );
          assert( pOrTerm->wtFlags & (TERM_COPIED|TERM_VIRTUAL) );
          continue;
        }
................................................................................
  /* Prevent ON clause terms of a LEFT JOIN from being used to drive
  ** an index for tables to the left of the join.
  */
  pTerm->prereqRight |= extraRight;
}

/*
** This function searches pList for an entry that matches the iCol-th column
** of index pIdx.
**
** If such an expression is found, its index in pList->a[] is returned. If
** no expression is found, -1 is returned.
*/
static int findIndexCol(
  Parse *pParse,                  /* Parse context */
................................................................................
**
** then nEq is set to 0.
**
** When this function is called, *pnOut is set to the sqlite3LogEst() of the
** number of rows that the index scan is expected to visit without 
** considering the range constraints. If nEq is 0, this is the number of 
** rows in the index. Assuming no error occurs, *pnOut is adjusted (reduced)
** to account for the range constraints pLower and pUpper.
** 
** In the absence of sqlite_stat4 ANALYZE data, or if such data cannot be
** used, a single range inequality reduces the search space by a factor of 4. 
** and a pair of constraints (x>? AND x<?) reduces the expected number of
** rows visited by a factor of 64.
*/
static int whereRangeScanEst(
................................................................................
    **       A: <loop body>                 # Return data, whatever.
    **
    **          Return     2                # Jump back to the Gosub
    **
    **       B: <after the loop>
    **
    ** Added 2014-05-26: If the table is a WITHOUT ROWID table, then
    ** use an ephemeral index instead of a RowSet to record the primary
    ** keys of the rows we have already seen.
    **
    */
    WhereClause *pOrWc;    /* The OR-clause broken out into subterms */
    SrcList *pOrTab;       /* Shortened table list or OR-clause generation */
    Index *pCov = 0;             /* Potential covering index (or NULL) */
    int iCovCur = pParse->nTab++;  /* Cursor used for index scans (if any) */
................................................................................
        memcpy(&pOrTab->a[k], &origSrc[pLevel[k].iFrom], sizeof(pOrTab->a[k]));
      }
    }else{
      pOrTab = pWInfo->pTabList;
    }

    /* Initialize the rowset register to contain NULL. An SQL NULL is 
    ** equivalent to an empty rowset.  Or, create an ephemeral index
    ** capable of holding primary keys in the case of a WITHOUT ROWID.
    **
    ** Also initialize regReturn to contain the address of the instruction 
    ** immediately following the OP_Return at the bottom of the loop. This
    ** is required in a few obscure LEFT JOIN cases where control jumps
    ** over the top of the loop into the body of it. In this case the 
    ** correct response for the end-of-loop code (the OP_Return) is to 
................................................................................
        /* TUNING: One-time cost for computing the automatic index is
        ** approximately 7*N*log2(N) where N is the number of rows in
        ** the table being indexed. */
        pNew->rSetup = rLogSize + rSize + 28;  assert( 28==sqlite3LogEst(7) );
        ApplyCostMultiplier(pNew->rSetup, pTab->costMult);
        /* TUNING: Each index lookup yields 20 rows in the table.  This
        ** is more than the usual guess of 10 rows, since we have no way
        ** of knowing how selective the index will ultimately be.  It would
        ** not be unreasonable to make this value much larger. */
        pNew->nOut = 43;  assert( 43==sqlite3LogEst(20) );
        pNew->rRun = sqlite3LogEstAdd(rLogSize,pNew->nOut);
        pNew->wsFlags = WHERE_AUTO_INDEX;
        pNew->prereq = mExtra | pTerm->prereqRight;
        rc = whereLoopInsert(pBuilder, pNew);
      }
................................................................................
**   N==0:  No terms of the ORDER BY clause are satisfied
**   N<0:   Unknown yet how many terms of ORDER BY might be satisfied.   
**
** Note that processing for WHERE_GROUPBY and WHERE_DISTINCTBY is not as
** strict.  With GROUP BY and DISTINCT the only requirement is that
** equivalent rows appear immediately adjacent to one another.  GROUP BY
** and DISTINCT do not require rows to appear in any particular order as long
** as equivalent rows are grouped together.  Thus for GROUP BY and DISTINCT
** the pOrderBy terms can be matched in any order.  With ORDER BY, the 
** pOrderBy terms must be matched in strict left-to-right order.
*/
static i8 wherePathSatisfiesOrderBy(
  WhereInfo *pWInfo,    /* The WHERE clause */
  ExprList *pOrderBy,   /* ORDER BY or GROUP BY or DISTINCT clause to check */
  WherePath *pPath,     /* The WherePath to check */

Changes to src/whereInt.h.

172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
** Then a WherePath object is a path through the graph that visits some
** or all of the WhereLoop objects once.
**
** The "solver" works by creating the N best WherePath objects of length
** 1.  Then using those as a basis to compute the N best WherePath objects
** of length 2.  And so forth until the length of WherePaths equals the
** number of nodes in the FROM clause.  The best (lowest cost) WherePath
** at the end is the choosen query plan.
*/
struct WherePath {
  Bitmask maskLoop;     /* Bitmask of all WhereLoop objects in this path */
  Bitmask revLoop;      /* aLoop[]s that should be reversed for ORDER BY */
  LogEst nRow;          /* Estimated number of rows generated by this path */
  LogEst rCost;         /* Total cost of this path */
  LogEst rUnsorted;     /* Total cost of this path ignoring sorting costs */







|







172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
** Then a WherePath object is a path through the graph that visits some
** or all of the WhereLoop objects once.
**
** The "solver" works by creating the N best WherePath objects of length
** 1.  Then using those as a basis to compute the N best WherePath objects
** of length 2.  And so forth until the length of WherePaths equals the
** number of nodes in the FROM clause.  The best (lowest cost) WherePath
** at the end is the chosen query plan.
*/
struct WherePath {
  Bitmask maskLoop;     /* Bitmask of all WhereLoop objects in this path */
  Bitmask revLoop;      /* aLoop[]s that should be reversed for ORDER BY */
  LogEst nRow;          /* Estimated number of rows generated by this path */
  LogEst rCost;         /* Total cost of this path */
  LogEst rUnsorted;     /* Total cost of this path ignoring sorting costs */