src/tokenize.c

/*
** 2001 September 15
**
** The author disclaims copyright to this source code.  In place of
** a legal notice, here is a blessing:
**
**    May you do good and not evil.
**    May you find forgiveness for yourself and forgive others.
**    May you share freely, never taking more than you give.
**
*************************************************************************
** An tokenizer for SQL
**
** This file contains C code that splits an SQL input string up into
** individual tokens and sends those tokens one-by-one over to the
** parser for analysis.
**
** $Id: tokenize.c,v 1.100 2005/01/31 12:42:29 danielk1977 Exp $
*/
#include "sqliteInt.h"
#include "os.h"
#include <ctype.h>
#include <stdlib.h>

/*
** The sqlite3KeywordCode function looks up an identifier to determine if
** it is a keyword.  If it is a keyword, the token code of that keyword is 
** returned.  If the input is not a keyword, TK_ID is returned.
**
** The implementation of this routine was generated by a program,
** mkkeywordhash.h, located in the tool subdirectory of the distribution.
** The output of the mkkeywordhash.c program is written into a file
** named keywordhash.h and then included into this source file by
** the #include below.
*/
#include "keywordhash.h"


/*
** If X is a character that can be used in an identifier and
** X&0x80==0 then isIdChar[X] will be 1.  If X&0x80==0x80 then
** X is always an identifier character.  (Hence all UTF-8
** characters can be part of an identifier).  isIdChar[X] will
** be 0 for every character in the lower 128 ASCII characters
** that cannot be used as part of an identifier.
**
** In this implementation, an identifier can be a string of
** alphabetic characters, digits, and "_" plus any character
** with the high-order bit set.  The latter rule means that
** any sequence of UTF-8 characters or characters taken from
** an extended ISO8859 character set can form an identifier.
**
** 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.
*/
static const char isIdChar[] = {
/* x0 x1 x2 x3 x4 x5 x6 x7 x8 x9 xA xB xC xD xE xF */
    0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,  /* 2x */
    1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 0, 0, 0, 0,  /* 3x */
    0, 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, 0, 0, 0, 0, 1,  /* 5x */
    0, 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, 0, 0, 0, 0, 0,  /* 7x */
};

#define IdChar(C)  (((c=C)&0x80)!=0 || (c>0x1f && isIdChar[c-0x20]))

/*
** Return the length of the token that begins at z[0]. 
** Store the token type in *tokenType before returning.
*/
static int getToken(const unsigned char *z, int *tokenType){
  int i, c;
  switch( *z ){
    case ' ': case '\t': case '\n': case '\f': case '\r': {
      for(i=1; isspace(z[i]); i++){}
      *tokenType = TK_SPACE;
      return i;
    }
    case '-': {
      if( z[1]=='-' ){
        for(i=2; (c=z[i])!=0 && c!='\n'; i++){}
        *tokenType = TK_COMMENT;
        return i;
      }
      *tokenType = TK_MINUS;
      return 1;
    }
    case '(': {
      *tokenType = TK_LP;
      return 1;
    }
    case ')': {
      *tokenType = TK_RP;
      return 1;
    }
    case ';': {
      *tokenType = TK_SEMI;
      return 1;
    }
    case '+': {
      *tokenType = TK_PLUS;
      return 1;
    }
    case '*': {
      *tokenType = TK_STAR;
      return 1;
    }
    case '/': {
      if( z[1]!='*' || z[2]==0 ){
        *tokenType = TK_SLASH;
        return 1;
      }
      for(i=3, c=z[2]; (c!='*' || z[i]!='/') && (c=z[i])!=0; i++){}
      if( c ) i++;
      *tokenType = TK_COMMENT;
      return i;
    }
    case '%': {
      *tokenType = TK_REM;
      return 1;
    }
    case '=': {
      *tokenType = TK_EQ;
      return 1 + (z[1]=='=');
    }
    case '<': {
      if( (c=z[1])=='=' ){
        *tokenType = TK_LE;
        return 2;
      }else if( c=='>' ){
        *tokenType = TK_NE;
        return 2;
      }else if( c=='<' ){
        *tokenType = TK_LSHIFT;
        return 2;
      }else{
        *tokenType = TK_LT;
        return 1;
      }
    }
    case '>': {
      if( (c=z[1])=='=' ){
        *tokenType = TK_GE;
        return 2;
      }else if( c=='>' ){
        *tokenType = TK_RSHIFT;
        return 2;
      }else{
        *tokenType = TK_GT;
        return 1;
      }
    }
    case '!': {
      if( z[1]!='=' ){
        *tokenType = TK_ILLEGAL;
        return 2;
      }else{
        *tokenType = TK_NE;
        return 2;
      }
    }
    case '|': {
      if( z[1]!='|' ){
        *tokenType = TK_BITOR;
        return 1;
      }else{
        *tokenType = TK_CONCAT;
        return 2;
      }
    }
    case ',': {
      *tokenType = TK_COMMA;
      return 1;
    }
    case '&': {
      *tokenType = TK_BITAND;
      return 1;
    }
    case '~': {
      *tokenType = TK_BITNOT;
      return 1;
    }
    case '#': {
      for(i=1; isdigit(z[i]) || (i==1 && z[1]=='-'); i++){}
      *tokenType = TK_REGISTER;
      return i;
    }
    case '\'': case '"': {
      int delim = z[0];
      for(i=1; (c=z[i])!=0; i++){
        if( c==delim ){
          if( z[i+1]==delim ){
            i++;
          }else{
            break;
          }
        }
      }
      if( c ) i++;
      *tokenType = TK_STRING;
      return i;
    }
    case '.': {
      *tokenType = TK_DOT;
      return 1;
    }
    case '0': case '1': case '2': case '3': case '4':
    case '5': case '6': case '7': case '8': case '9': {
      *tokenType = TK_INTEGER;
      for(i=1; isdigit(z[i]); i++){}
#ifndef SQLITE_OMIT_FLOATING_POINT
      if( z[i]=='.' && isdigit(z[i+1]) ){
        i += 2;
        while( isdigit(z[i]) ){ i++; }
        *tokenType = TK_FLOAT;
      }
      if( (z[i]=='e' || z[i]=='E') &&
           ( isdigit(z[i+1]) 
            || ((z[i+1]=='+' || z[i+1]=='-') && isdigit(z[i+2]))
           )
      ){
        i += 2;
        while( isdigit(z[i]) ){ i++; }
        *tokenType = TK_FLOAT;
      }
#endif
      return i;
    }
    case '[': {
      for(i=1, c=z[0]; c!=']' && (c=z[i])!=0; i++){}
      *tokenType = TK_ID;
      return i;
    }
    case '?': {
      *tokenType = TK_VARIABLE;
      for(i=1; isdigit(z[i]); i++){}
      return i;
    }
    case ':': {
      for(i=1; IdChar(z[i]); i++){}
      *tokenType = i>1 ? TK_VARIABLE : TK_ILLEGAL;
      return i;
    }
#ifndef SQLITE_OMIT_TCL_VARIABLE
    case '$': {
      *tokenType = TK_VARIABLE;
      if( z[1]=='{' ){
        int nBrace = 1;
        for(i=2; (c=z[i])!=0 && nBrace; i++){
          if( c=='{' ){
            nBrace++;
          }else if( c=='}' ){
            nBrace--;
          }
        }
        if( c==0 ) *tokenType = TK_ILLEGAL;
      }else{
        int n = 0;
        for(i=1; (c=z[i])!=0; i++){
          if( isalnum(c) || c=='_' ){
            n++;
          }else if( c=='(' && n>0 ){
            do{
              i++;
            }while( (c=z[i])!=0 && !isspace(c) && c!=')' );
            if( c==')' ){
              i++;
            }else{
              *tokenType = TK_ILLEGAL;
            }
            break;
          }else if( c==':' && z[i+1]==':' ){
            i++;
          }else{
            break;
          }
        }
        if( n==0 ) *tokenType = TK_ILLEGAL;
      }
      return i;
    }
#endif
#ifndef SQLITE_OMIT_BLOB_LITERAL
    case 'x': case 'X': {
      if( (c=z[1])=='\'' || c=='"' ){
        int delim = c;
        *tokenType = TK_BLOB;
        for(i=2; (c=z[i])!=0; i++){
          if( c==delim ){
            if( i%2 ) *tokenType = TK_ILLEGAL;
            break;
          }
          if( !isxdigit(c) ){
            *tokenType = TK_ILLEGAL;
            return i;
          }
        }
        if( c ) i++;
        return i;
      }
      /* Otherwise fall through to the next case */
    }
#endif
    default: {
      if( !IdChar(*z) ){
        break;
      }
      for(i=1; IdChar(z[i]); i++){}
      *tokenType = keywordCode((char*)z, i);
      return i;
    }
  }
  *tokenType = TK_ILLEGAL;
  return 1;
}
int sqlite3GetToken(const unsigned char *z, int *tokenType){
  return getToken(z, tokenType);
}

/*
** Run the parser on the given SQL string.  The parser structure is
** passed in.  An SQLITE_ status code is returned.  If an error occurs
** and pzErrMsg!=NULL then an error message might be written into 
** memory obtained from malloc() and *pzErrMsg made to point to that
** error message.  Or maybe not.
*/
int sqlite3RunParser(Parse *pParse, const char *zSql, char **pzErrMsg){
  int nErr = 0;
  int i;
  void *pEngine;
  int tokenType;
  int lastTokenParsed = -1;
  sqlite3 *db = pParse->db;
  extern void *sqlite3ParserAlloc(void*(*)(int));
  extern void sqlite3ParserFree(void*, void(*)(void*));
  extern int sqlite3Parser(void*, int, Token, Parse*);

  db->flags &= ~SQLITE_Interrupt;
  pParse->rc = SQLITE_OK;
  i = 0;
  pEngine = sqlite3ParserAlloc((void*(*)(int))malloc);
  if( pEngine==0 ){
    sqlite3SetString(pzErrMsg, "out of memory", (char*)0);
    return 1;
  }
  assert( pParse->sLastToken.dyn==0 );
  assert( pParse->pNewTable==0 );
  assert( pParse->pNewTrigger==0 );
  assert( pParse->nVar==0 );
  assert( pParse->nVarExpr==0 );
  assert( pParse->nVarExprAlloc==0 );
  assert( pParse->apVarExpr==0 );
  pParse->zTail = pParse->zSql = zSql;
  while( sqlite3_malloc_failed==0 && zSql[i]!=0 ){
    assert( i>=0 );
    pParse->sLastToken.z = &zSql[i];
    assert( pParse->sLastToken.dyn==0 );
    pParse->sLastToken.n = getToken((unsigned char*)&zSql[i],&tokenType);
    i += pParse->sLastToken.n;
    switch( tokenType ){
      case TK_SPACE:
      case TK_COMMENT: {
        if( (db->flags & SQLITE_Interrupt)!=0 ){
          pParse->rc = SQLITE_INTERRUPT;
          sqlite3SetString(pzErrMsg, "interrupt", (char*)0);
          goto abort_parse;
        }
        break;
      }
      case TK_ILLEGAL: {
        if( pzErrMsg ){
          sqliteFree(*pzErrMsg);
          *pzErrMsg = sqlite3MPrintf("unrecognized token: \"%T\"",
                          &pParse->sLastToken);
        }
        nErr++;
        goto abort_parse;
      }
      case TK_SEMI: {
        pParse->zTail = &zSql[i];
        /* Fall thru into the default case */
      }
      default: {
        sqlite3Parser(pEngine, tokenType, pParse->sLastToken, pParse);
        lastTokenParsed = tokenType;
        if( pParse->rc!=SQLITE_OK ){
          goto abort_parse;
        }
        break;
      }
    }
  }
abort_parse:
  if( zSql[i]==0 && nErr==0 && pParse->rc==SQLITE_OK ){
    if( lastTokenParsed!=TK_SEMI ){
      sqlite3Parser(pEngine, TK_SEMI, pParse->sLastToken, pParse);
      pParse->zTail = &zSql[i];
    }
    sqlite3Parser(pEngine, 0, pParse->sLastToken, pParse);
  }
  sqlite3ParserFree(pEngine, free);
  if( sqlite3_malloc_failed ){
    pParse->rc = SQLITE_NOMEM;
  }
  if( pParse->rc!=SQLITE_OK && pParse->rc!=SQLITE_DONE && pParse->zErrMsg==0 ){
    sqlite3SetString(&pParse->zErrMsg, sqlite3ErrStr(pParse->rc),
                    (char*)0);
  }
  if( pParse->zErrMsg ){
    if( pzErrMsg && *pzErrMsg==0 ){
      *pzErrMsg = pParse->zErrMsg;
    }else{
      sqliteFree(pParse->zErrMsg);
    }
    pParse->zErrMsg = 0;
    if( !nErr ) nErr++;
  }
  if( pParse->pVdbe && pParse->nErr>0 && pParse->nested==0 ){
    sqlite3VdbeDelete(pParse->pVdbe);
    pParse->pVdbe = 0;
  }
  sqlite3DeleteTable(pParse->db, pParse->pNewTable);
  sqlite3DeleteTrigger(pParse->pNewTrigger);
  sqliteFree(pParse->apVarExpr);
  if( nErr>0 && (pParse->rc==SQLITE_OK || pParse->rc==SQLITE_DONE) ){
    pParse->rc = SQLITE_ERROR;
  }
  return nErr;
}

/*
** Token types used by the sqlite3_complete() routine.  See the header
** comments on that procedure for additional information.
*/
#define tkSEMI    0
#define tkWS      1
#define tkOTHER   2
#define tkEXPLAIN 3
#define tkCREATE  4
#define tkTEMP    5
#define tkTRIGGER 6
#define tkEND     7

/*
** Return TRUE if the given SQL string ends in a semicolon.
**
** Special handling is require for CREATE TRIGGER statements.
** Whenever the CREATE TRIGGER keywords are seen, the statement
** must end with ";END;".
**
** This implementation uses a state machine with 7 states:
**
**   (0) START     At the beginning or end of an SQL statement.  This routine
**                 returns 1 if it ends in the START state and 0 if it ends
**                 in any other state.
**
**   (1) NORMAL    We are in the middle of statement which ends with a single
**                 semicolon.
**
**   (2) EXPLAIN   The keyword EXPLAIN has been seen at the beginning of 
**                 a statement.
**
**   (3) CREATE    The keyword CREATE has been seen at the beginning of a
**                 statement, possibly preceeded by EXPLAIN and/or followed by
**                 TEMP or TEMPORARY
**
**   (4) TRIGGER   We are in the middle of a trigger definition that must be
**                 ended by a semicolon, the keyword END, and another semicolon.
**
**   (5) SEMI      We've seen the first semicolon in the ";END;" that occurs at
**                 the end of a trigger definition.
**
**   (6) 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.
**   (3) tkEXPLAIN   The "explain" keyword.
**   (4) tkCREATE    The "create" keyword.
**   (5) tkTEMP      The "temp" or "temporary" keyword.
**   (6) tkTRIGGER   The "trigger" keyword.
**   (7) tkEND       The "end" keyword.
**
** Whitespace never causes a state transition and is always ignored.
**
** If we compile with SQLITE_OMIT_TRIGGER, all of the computation needed
** to recognize the end of a trigger can be omitted.  All we have to do
** is look for a semicolon that is not part of an string or comment.
*/
int sqlite3_complete(const char *zSql){
  u8 state = 0;   /* Current state, using numbers defined in header comment */
  u8 token;       /* Value of the next token */

#ifndef SQLITE_OMIT_TRIGGER
  /* A complex statement machine used to detect the end of a CREATE TRIGGER
  ** statement.  This is the normal case.
  */
  static const u8 trans[7][8] = {
                     /* Token:                                                */
     /* State:       **  SEMI  WS  OTHER EXPLAIN  CREATE  TEMP  TRIGGER  END  */
     /* 0   START: */ {    0,  0,     1,      2,      3,    1,       1,   1,  },
     /* 1  NORMAL: */ {    0,  1,     1,      1,      1,    1,       1,   1,  },
     /* 2 EXPLAIN: */ {    0,  2,     1,      1,      3,    1,       1,   1,  },
     /* 3  CREATE: */ {    0,  3,     1,      1,      1,    3,       4,   1,  },
     /* 4 TRIGGER: */ {    5,  4,     4,      4,      4,    4,       4,   4,  },
     /* 5    SEMI: */ {    5,  5,     4,      4,      4,    4,       4,   6,  },
     /* 6     END: */ {    0,  6,     4,      4,      4,    4,       4,   4,  },
  };
#else
  /* If triggers are not suppored by this compile then the statement machine
  ** used to detect the end of a statement is much simplier
  */
  static const u8 trans[2][3] = {
                     /* Token:           */
     /* State:       **  SEMI  WS  OTHER */
     /* 0   START: */ {    0,  0,     1, },
     /* 1  NORMAL: */ {    0,  1,     1, },
  };
#endif /* SQLITE_OMIT_TRIGGER */

  while( *zSql ){
    switch( *zSql ){
      case ';': {  /* A semicolon */
        token = tkSEMI;
        break;
      }
      case ' ':
      case '\r':
      case '\t':
      case '\n':
      case '\f': {  /* White space is ignored */
        token = tkWS;
        break;
      }
      case '/': {   /* C-style comments */
        if( zSql[1]!='*' ){
          token = tkOTHER;
          break;
        }
        zSql += 2;
        while( zSql[0] && (zSql[0]!='*' || zSql[1]!='/') ){ zSql++; }
        if( zSql[0]==0 ) return 0;
        zSql++;
        token = tkWS;
        break;
      }
      case '-': {   /* SQL-style comments from "--" to end of line */
        if( zSql[1]!='-' ){
          token = tkOTHER;
          break;
        }
        while( *zSql && *zSql!='\n' ){ zSql++; }
        if( *zSql==0 ) return state==0;
        token = tkWS;
        break;
      }
      case '[': {   /* Microsoft-style identifiers in [...] */
        zSql++;
        while( *zSql && *zSql!=']' ){ zSql++; }
        if( *zSql==0 ) return 0;
        token = tkOTHER;
        break;
      }
      case '"':     /* single- and double-quoted strings */
      case '\'': {
        int c = *zSql;
        zSql++;
        while( *zSql && *zSql!=c ){ zSql++; }
        if( *zSql==0 ) return 0;
        token = tkOTHER;
        break;
      }
      default: {
        int c;
        if( IdChar((u8)*zSql) ){
          /* Keywords and unquoted identifiers */
          int nId;
          for(nId=1; IdChar(zSql[nId]); nId++){}
#ifdef SQLITE_OMIT_TRIGGER
          token = tkOTHER;
#else
          switch( *zSql ){
            case 'c': case 'C': {
              if( nId==6 && sqlite3StrNICmp(zSql, "create", 6)==0 ){
                token = tkCREATE;
              }else{
                token = tkOTHER;
              }
              break;
            }
            case 't': case 'T': {
              if( nId==7 && sqlite3StrNICmp(zSql, "trigger", 7)==0 ){
                token = tkTRIGGER;
              }else if( nId==4 && sqlite3StrNICmp(zSql, "temp", 4)==0 ){
                token = tkTEMP;
              }else if( nId==9 && sqlite3StrNICmp(zSql, "temporary", 9)==0 ){
                token = tkTEMP;
              }else{
                token = tkOTHER;
              }
              break;
            }
            case 'e':  case 'E': {
              if( nId==3 && sqlite3StrNICmp(zSql, "end", 3)==0 ){
                token = tkEND;
              }else
#ifndef SQLITE_OMIT_EXPLAIN
              if( nId==7 && sqlite3StrNICmp(zSql, "explain", 7)==0 ){
                token = tkEXPLAIN;
              }else
#endif
              {
                token = tkOTHER;
              }
              break;
            }
            default: {
              token = tkOTHER;
              break;
            }
          }
#endif /* SQLITE_OMIT_TRIGGER */
          zSql += nId-1;
        }else{
          /* Operators and special symbols */
          token = tkOTHER;
        }
        break;
      }
    }
    state = trans[state][token];
    zSql++;
  }
  return state==0;
}

#ifndef SQLITE_OMIT_UTF16
/*
** This routine is the same as the sqlite3_complete() routine described
** above, except that the parameter is required to be UTF-16 encoded, not
** UTF-8.
*/
int sqlite3_complete16(const void *zSql){
  sqlite3_value *pVal;
  char const *zSql8;
  int rc = 0;

  pVal = sqlite3ValueNew();
  sqlite3ValueSetStr(pVal, -1, zSql, SQLITE_UTF16NATIVE, SQLITE_STATIC);
  zSql8 = sqlite3ValueText(pVal, SQLITE_UTF8);
  if( zSql8 ){
    rc = sqlite3_complete(zSql8);
  }
  sqlite3ValueFree(pVal);
  return rc;
}
#endif /* SQLITE_OMIT_UTF16 */