SQLite

1.0.29

** Author contact information:
**   drh@hwaci.com
**   http://www.hwaci.com/drh/
**
*************************************************************************
** Internal interface definitions for SQLite.
**
** @(#) $Id: sqliteInt.h,v 1.39 2001/04/04 21:10:19 drh Exp $
*/
#include "sqlite.h"
#include "dbbe.h"
#include "vdbe.h"
#include "parse.h"
#include <gdbm.h>
#include <stdio.h>

void sqliteWhereEnd(WhereInfo*);
void sqliteExprCode(Parse*, Expr*);
void sqliteExprIfTrue(Parse*, Expr*, int);
void sqliteExprIfFalse(Parse*, Expr*, int);
Table *sqliteFindTable(sqlite*,char*);
void sqliteCopy(Parse*, Token*, Token*, Token*);
void sqliteVacuum(Parse*, Token*);
int sqliteGlobCompare(const unsigned char*,const unsigned char*);
int sqliteLikeCompare(const unsigned char*,const unsigned char*);
char *sqliteTableNameFromToken(Token*);
int sqliteExprCheck(Parse*, Expr*, int, int*);
int sqliteExprCompare(Expr*, Expr*);
int sqliteFuncId(Token*);
int sqliteExprResolveIds(Parse*, IdList*, Expr*);
void sqliteExprResolveInSelect(Parse*, Expr*);

**
*************************************************************************
** Utility functions used throughout sqlite.
**
** This file contains functions for allocating memory, comparing
** strings, and stuff like that.
**
** $Id: util.c,v 1.19 2001/04/04 21:10:19 drh Exp $
*/
#include "sqliteInt.h"
#include <stdarg.h>
#include <ctype.h>

/*
** If MEMORY_DEBUG is defined, then use versions of malloc() and

  fprintf(stderr,"string at 0x%x is %s\n", (int)*pz, *pz);
#endif
#endif
}

/*
** Works like sqliteSetString, but each string is now followed by
** a length integer which specifies how much of the source string 
** to copy (in bytes).  -1 means use the whole string.
*/
void sqliteSetNString(char **pz, ...){
  va_list ap;
  int nByte;
  const char *z;
  char *zResult;
  int n;

    }
  }
  if( *a=='-' ) res = -res;
  return res;
}

/*
** When the first byte of a UTF-8 character is used as the
** index of the following array, then the value is the number
** of bytes in the whole UTF-8 character.  This matrix assumes
** a well-formed UTF-8 string.  All bets are off if the input
** is not well-formed.
*/
static const unsigned char utf8_width[] = {
        /* x0 x1 x2 x3 x4 x5 x6 x7 x8 x9 xA xB xC xD xE xF */
/* 0x */   0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,
/* 1x */   1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,
/* 2x */   1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,
/* 3x */   1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,
/* 4x */   1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,
/* 5x */   1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,
/* 6x */   1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,
/* 7x */   1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,
/* 8x */   1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,
/* 9x */   1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,
/* Ax */   1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,
/* Bx */   1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,
/* Cx */   2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2,
/* Dx */   2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2,
/* Ex */   3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3,
/* Fx */   4, 4, 4, 4, 4, 4, 4, 4, 5, 5, 5, 5, 6, 6, 1, 1,
};

/*
** This routine computes the number of bytes to the start of the
** next UTF-8 character.  We could just do 
**
**      z += utf8_width[*z]
**
** accomplish the same thing, if we know that z was a well-formed
** UTF-8 string.  If it is not, then z might be incremented past
** its null terminator.  This function, though slower, will never
** increment z past its terminator.
*/
static int utf8_char_size(const unsigned char *z){
  int i, n = utf8_width[*z];
  for(i=1; i<n && z[i]!=0; i++){}
  return i;
}

/*
** Convert the UTF-8 character pointed to by the input parameter
** into a 31-bit UCS character and return an integer holding the
** 31-bit UCS character.
*/
static int utf8_to_int(const unsigned char *z){
  int n = utf8_width[*z];
  int c;
  switch( n ){
    case 0: {
      return 0;
    }
    case 1: {
      return *z;
    }
    case 2: {
      c = 0x1f & *(z++);
      break;
    }
    case 3: {
      c = 0x0f & *(z++);
      break;
    }
    case 4: {
      c = 0x07 & *(z++);
      break;
    }
    case 5: {
      c = 0x03 & *(z++);
      break;
    }
    case 6: {
      c = 0x01 & *(z++);
      break;
    }
  }
  while( (--n) > 0 ){
    c = (c<<6) | (0x3f & *(z++));
  }
  return c;
}

/*
** Compare two UTF-8 strings for equality where the first string can
** potentially be a "glob" expression.  Return true (1) if they
** are the same and false (0) if they are different.
**
** Globbing rules:
**
**      '*'       Matches any sequence of zero or more characters.
**

**
** This routine is usually quick, but can be N**2 in the worst case.
**
** Hints: to match '*' or '?', put them in "[]".  Like this:
**
**         abc[*]xyz        Matches "abc*xyz" only
*/
int 
sqliteGlobCompare(const unsigned char *zPattern, const unsigned char *zString){
  register int c;
  int invert;
  int seen;
  int c2;

  while( (c = *zPattern)!=0 ){
    switch( c ){
      case '*':
        while( (c=zPattern[1]) == '*' || c == '?' ){
          if( c=='?' ){
            if( *zString==0 ) return 0;
            zString += utf8_char_size(zString);
          }
          zPattern++;
        }
        if( c==0 ) return 1;
        c = UpperToLower[c];
        if( c=='[' ){
          while( *zString && sqliteGlobCompare(&zPattern[1],zString)==0 ){
            zString += utf8_char_size(zString);
          }
          return *zString!=0;
        }else{
          while( (c2 = *zString)!=0 ){
            while( c2 != 0 && c2 != c ){ c2 = *++zString; }
            if( c2==0 ) return 0;
            if( sqliteGlobCompare(&zPattern[1],zString) ) return 1;
            zString += utf8_char_size(zString);
          }
          return 0;
        }
      case '?': {
        if( *zString==0 ) return 0;
        zString += utf8_char_size(zString);
        zPattern++;
        break;
      }
      case '[': {
        int prior_c = 0;
        seen = 0;
        invert = 0;
        c = utf8_to_int(zString);
        if( c==0 ) return 0;
        c2 = *++zPattern;
        if( c2=='^' ){ invert = 1; c2 = *++zPattern; }
        if( c2==']' ){
          if( c==']' ) seen = 1;
          c2 = *++zPattern;
        }
        while( (c2 = utf8_to_int(zPattern))!=0 && c2!=']' ){
          if( c2=='-' && zPattern[1]!=']' && zPattern[1]!=0 && prior_c>0 ){
            zPattern++;
            c2 = utf8_to_int(zPattern);
            if( c>=prior_c && c<=c2 ) seen = 1;
            prior_c = 0;
          }else if( c==c2 ){
            seen = 1;
            prior_c = c2;
          }else{
            prior_c = c2;
          }
          zPattern += utf8_char_size(zPattern);
        }
        if( c2==0 || (seen ^ invert)==0 ) return 0;
        zString += utf8_char_size(zString);
        zPattern++;
        break;
      }
      default: {
        if( c != *zString ) return 0;


        zPattern++;
        zString++;
        break;
      }
    }
  }
  return *zString==0;
}

/*
** Compare two UTF-8 strings for equality using the "LIKE" operator of
** SQL.  The '%' character matches any sequence of 0 or more
** characters and '_' matches any single character.  Case is
** not significant.
**
** This routine is just an adaptation of the sqliteGlobCompare()
** routine above.
*/
int 
sqliteLikeCompare(const unsigned char *zPattern, const unsigned char *zString){
  register int c;
  int c2;

  while( (c = UpperToLower[*zPattern])!=0 ){
    switch( c ){
      case '%': {

        while( (c=zPattern[1]) == '%' || c == '_' ){

          if( c=='_' ){
            if( *zString==0 ) return 0;
            zString += utf8_char_size(zString);
          }
          zPattern++;
        }
        if( c==0 ) return 1;
        c = UpperToLower[c];
        while( (c2=UpperToLower[*zString])!=0 ){
          while( c2 != 0 && c2 != c ){ c2 = UpperToLower[*++zString]; }
          if( c2==0 ) return 0;
          if( sqliteLikeCompare(&zPattern[1],zString) ) return 1;
          zString += utf8_char_size(zString);
        }
        return 0;
      }
      case '_': {
        if( *zString==0 ) return 0;
        zString += utf8_char_size(zString);
        zPattern++;
        break;
      }
      default: {
        if( c != UpperToLower[*zString] ) return 0;


        zPattern++;
        zString++;
        break;
      }
    }
  }
  return *zString==0;
}

#   drh@hwaci.com
#   http://www.hwaci.com/drh/
#
#***********************************************************************
# This file implements regression tests for SQLite library.  The
# focus of this file is testing expressions.
#
# $Id: expr.test,v 1.11 2001/04/04 21:10:19 drh Exp $

set testdir [file dirname $argv0]
source $testdir/tester.tcl

# Create a table to work with.
#
execsql {CREATE TABLE test1(i1 int, i2 int, r1 real, r2 real, t1 text, t2 text)}

test_expr expr-5.8b {t1='abxyzzycy', t2='A%CX'} {t1 LIKE t2} 0
test_expr expr-5.9 {t1='abc', t2='A%_C'} {t1 LIKE t2} 1
test_expr expr-5.9b {t1='ac', t2='A%_C'} {t1 LIKE t2} 0
test_expr expr-5.10 {t1='abxyzzyc', t2='A%_C'} {t1 LIKE t2} 1
test_expr expr-5.11 {t1='abc', t2='xyz'} {t1 NOT LIKE t2} 1
test_expr expr-5.12 {t1='abc', t2='ABC'} {t1 NOT LIKE t2} 0

# The following tests only work on versions of TCL that support
# Unicode.
#
test_expr expr-5.13 "t1='a\u0080c', t2='A_C'" {t1 LIKE t2} 1
test_expr expr-5.14 "t1='a\u07FFc', t2='A_C'" {t1 LIKE t2} 1
test_expr expr-5.15 "t1='a\u0800c', t2='A_C'" {t1 LIKE t2} 1
test_expr expr-5.16 "t1='a\uFFFFc', t2='A_C'" {t1 LIKE t2} 1
test_expr expr-5.17 "t1='a\u0080', t2='A__'" {t1 LIKE t2} 0
test_expr expr-5.18 "t1='a\u07FF', t2='A__'" {t1 LIKE t2} 0
test_expr expr-5.19 "t1='a\u0800', t2='A__'" {t1 LIKE t2} 0
test_expr expr-5.20 "t1='a\uFFFF', t2='A__'" {t1 LIKE t2} 0
test_expr expr-5.21 "t1='ax\uABCD', t2='A_\uABCD'" {t1 LIKE t2} 1
test_expr expr-5.22 "t1='ax\u1234', t2='A%\u1234'" {t1 LIKE t2} 1
test_expr expr-5.23 "t1='ax\uFEDC', t2='A_%'" {t1 LIKE t2} 1
test_expr expr-5.24 "t1='ax\uFEDCy\uFEDC', t2='A%\uFEDC'" {t1 LIKE t2} 1

test_expr expr-6.1 {t1='abc', t2='xyz'} {t1 GLOB t2} 0
test_expr expr-6.2 {t1='abc', t2='ABC'} {t1 GLOB t2} 0
test_expr expr-6.3 {t1='abc', t2='A?C'} {t1 GLOB t2} 0
test_expr expr-6.4 {t1='abc', t2='a?c'} {t1 GLOB t2} 1
test_expr expr-6.5 {t1='abc', t2='abc?'} {t1 GLOB t2} 0
test_expr expr-6.6 {t1='abc', t2='A*C'} {t1 GLOB t2} 0
test_expr expr-6.7 {t1='abc', t2='a*c'} {t1 GLOB t2} 1

test_expr expr-6.19 {t1='abc', t2='a[]b]c'} {t1 GLOB t2} 1
test_expr expr-6.20 {t1='abc', t2='a[^]b]c'} {t1 GLOB t2} 0
test_expr expr-6.21 {t1='abcdefg', t2='a*[de]g'} {t1 GLOB t2} 0
test_expr expr-6.22 {t1='abcdefg', t2='a*[^de]g'} {t1 GLOB t2} 1
test_expr expr-6.23 {t1='abcdefg', t2='a*?g'} {t1 GLOB t2} 1
test_expr expr-6.24 {t1='ac', t2='a*c'} {t1 GLOB t2} 1
test_expr expr-6.25 {t1='ac', t2='a*?c'} {t1 GLOB t2} 0

# These tests only work on versions of TCL that support Unicode
#
test_expr expr-6.26 "t1='a\u0080c', t2='a?c'" {t1 GLOB t2} 1
test_expr expr-6.27 "t1='a\u07ffc', t2='a?c'" {t1 GLOB t2} 1
test_expr expr-6.28 "t1='a\u0800c', t2='a?c'" {t1 GLOB t2} 1
test_expr expr-6.29 "t1='a\uffffc', t2='a?c'" {t1 GLOB t2} 1
test_expr expr-6.30 "t1='a\u1234', t2='a?'" {t1 GLOB t2} 1
test_expr expr-6.31 "t1='a\u1234', t2='a??'" {t1 GLOB t2} 0
test_expr expr-6.32 "t1='ax\u1234', t2='a?\u1234'" {t1 GLOB t2} 1
test_expr expr-6.33 "t1='ax\u1234', t2='a*\u1234'" {t1 GLOB t2} 1
test_expr expr-6.34 "t1='ax\u1234y\u1234', t2='a*\u1234'" {t1 GLOB t2} 1
test_expr expr-6.35 "t1='a\u1234b', t2='a\[x\u1234y\]b'" {t1 GLOB t2} 1
test_expr expr-6.36 "t1='a\u1234b', t2='a\[\u1233-\u1235\]b'" {t1 GLOB t2} 1
test_expr expr-6.37 "t1='a\u1234b', t2='a\[\u1234-\u124f\]b'" {t1 GLOB t2} 1
test_expr expr-6.38 "t1='a\u1234b', t2='a\[\u1235-\u124f\]b'" {t1 GLOB t2} 0
test_expr expr-6.39 "t1='a\u1234b', t2='a\[a-\u1235\]b'" {t1 GLOB t2} 1
test_expr expr-6.40 "t1='a\u1234b', t2='a\[a-\u1234\]b'" {t1 GLOB t2} 1
test_expr expr-6.41 "t1='a\u1234b', t2='a\[a-\u1233\]b'" {t1 GLOB t2} 0


# The sqliteExprIfFalse and sqliteExprIfTrue routines are only
# executed as part of a WHERE clause.  Create a table suitable
# for testing these functions.
#
execsql {DROP TABLE test1}
execsql {CREATE TABLE test1(a int, b int);}

}


proc chng {date desc} {
  puts "<DT><B>$date</B></DT>"
  puts "<DD><P><UL>$desc</UL></P></DD>"
}

chng {2001 Apr 5 (1.0.29)} {
<li>The LIKE and GLOB operators now assume both operands are
    UTF-8 strings.
    <font color="red">** This change could potentially
    break existing code **</font></li>
}

chng {2001 Apr 4 (1.0.28)} {
<li>Added limited support for transactions.  At this point, transactions
    will do table locking on the GDBM backend.  There is no support (yet)
    for rollback or atomic commit.</li>
<li>Added special column names ROWID, OID, and _ROWID_ that refer to the
    unique random integer key associated with every row of every table.</li>

#
# Run this TCL script to generate HTML for the index.html file.
#
set rcsid {$Id: index.tcl,v 1.34 2001/04/04 21:10:19 drh Exp $}

puts {<html>
<head><title>SQLite: An SQL Database Library Built Atop GDBM</title></head>
<body bgcolor=white>
<h1 align=center>SQLite: An SQL Database Library Built Atop
<a href="http://www.gnu.org/software/gdbm/gdbm.html">GDBM</a></h1>
<p align=center>}

three functions and one opaque structure.</li>
<li>A <a href="tclsqlite.html">Tcl</a> interface is
included.</li>
<li>Command-line access program <a href="sqlite.html">sqlite</a> uses
the <a href="http://www.google.com/search?q=gnu+readline+library">GNU
Readline library</a></li>
<li>A Tcl-based test suite provides near 100% code coverage</li>
<li>Approximately 9500 lines of C code.  No external dependencies other 
than GDBM.</li>
<li>Built and tested under Linux, HPUX, and WinNT.</li>
</ul>
</p>

<h2>Current Status</h2>

<p>A <a href="changes.html">change history</a> is available online.
There are currently no <em>known</em> bugs or memory leaks
in the library.  <a href="http://gcc.gnu.org/onlinedocs/gcov_1.html">Gcov</a>
is used to verify test coverage.  The test suite currently exercises
all code except for a few areas which are unreachable or which are
only reached when <tt>malloc()</tt> fails.  The code has been tested
for memory leaks and is found to be clean.</p>

<p><b>Important Note:</b>  Beginning with version 1.0.29, the LIKE and
GLOB operators assume both operands are UTF-8 strings.  Prior to that,
both operators assumed plain ASCII strings.  Users of earlier versions
of SQLite that invoke LIKE or GLOB to compare strings containing
characters greater than 127 may have problems when they upgrade to
version 1.0.29 or later.</p>



<p><b>Important Note:</b>  Serious bugs have been found in versions
1.0.22 on Unix and 1.0.26 on Windows.  Users of these or earlier




versions of SQLite should upgrade.</p>

<h2>Documentation</h2>

<p>The following documentation is currently available:</p>

<p><ul>
<li>Information on the <a href="sqlite.html">sqlite</a>