xref: /sqlite-3.40.0/src/tokenize.c (revision d5578433)
1 /*
2 ** 2001 September 15
3 **
4 ** The author disclaims copyright to this source code.  In place of
5 ** a legal notice, here is a blessing:
6 **
7 **    May you do good and not evil.
8 **    May you find forgiveness for yourself and forgive others.
9 **    May you share freely, never taking more than you give.
10 **
11 *************************************************************************
12 ** An tokenizer for SQL
13 **
14 ** This file contains C code that splits an SQL input string up into
15 ** individual tokens and sends those tokens one-by-one over to the
16 ** parser for analysis.
17 */
18 #include "sqliteInt.h"
19 #include <stdlib.h>
20 
21 /*
22 ** The charMap() macro maps alphabetic characters into their
23 ** lower-case ASCII equivalent.  On ASCII machines, this is just
24 ** an upper-to-lower case map.  On EBCDIC machines we also need
25 ** to adjust the encoding.  Only alphabetic characters and underscores
26 ** need to be translated.
27 */
28 #ifdef SQLITE_ASCII
29 # define charMap(X) sqlite3UpperToLower[(unsigned char)X]
30 #endif
31 #ifdef SQLITE_EBCDIC
32 # define charMap(X) ebcdicToAscii[(unsigned char)X]
33 const unsigned char ebcdicToAscii[] = {
34 /* 0   1   2   3   4   5   6   7   8   9   A   B   C   D   E   F */
35    0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  /* 0x */
36    0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  /* 1x */
37    0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  /* 2x */
38    0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  /* 3x */
39    0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  /* 4x */
40    0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  /* 5x */
41    0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0, 95,  0,  0,  /* 6x */
42    0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  /* 7x */
43    0, 97, 98, 99,100,101,102,103,104,105,  0,  0,  0,  0,  0,  0,  /* 8x */
44    0,106,107,108,109,110,111,112,113,114,  0,  0,  0,  0,  0,  0,  /* 9x */
45    0,  0,115,116,117,118,119,120,121,122,  0,  0,  0,  0,  0,  0,  /* Ax */
46    0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  /* Bx */
47    0, 97, 98, 99,100,101,102,103,104,105,  0,  0,  0,  0,  0,  0,  /* Cx */
48    0,106,107,108,109,110,111,112,113,114,  0,  0,  0,  0,  0,  0,  /* Dx */
49    0,  0,115,116,117,118,119,120,121,122,  0,  0,  0,  0,  0,  0,  /* Ex */
50    0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  /* Fx */
51 };
52 #endif
53 
54 /*
55 ** The sqlite3KeywordCode function looks up an identifier to determine if
56 ** it is a keyword.  If it is a keyword, the token code of that keyword is
57 ** returned.  If the input is not a keyword, TK_ID is returned.
58 **
59 ** The implementation of this routine was generated by a program,
60 ** mkkeywordhash.h, located in the tool subdirectory of the distribution.
61 ** The output of the mkkeywordhash.c program is written into a file
62 ** named keywordhash.h and then included into this source file by
63 ** the #include below.
64 */
65 #include "keywordhash.h"
66 
67 
68 /*
69 ** If X is a character that can be used in an identifier then
70 ** IdChar(X) will be true.  Otherwise it is false.
71 **
72 ** For ASCII, any character with the high-order bit set is
73 ** allowed in an identifier.  For 7-bit characters,
74 ** sqlite3IsIdChar[X] must be 1.
75 **
76 ** For EBCDIC, the rules are more complex but have the same
77 ** end result.
78 **
79 ** Ticket #1066.  the SQL standard does not allow '$' in the
80 ** middle of identfiers.  But many SQL implementations do.
81 ** SQLite will allow '$' in identifiers for compatibility.
82 ** But the feature is undocumented.
83 */
84 #ifdef SQLITE_ASCII
85 #define IdChar(C)  ((sqlite3CtypeMap[(unsigned char)C]&0x46)!=0)
86 #endif
87 #ifdef SQLITE_EBCDIC
88 const char sqlite3IsEbcdicIdChar[] = {
89 /* x0 x1 x2 x3 x4 x5 x6 x7 x8 x9 xA xB xC xD xE xF */
90     0, 0, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 0, 0, 0, 0,  /* 4x */
91     0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 1, 0, 0, 0, 0,  /* 5x */
92     0, 0, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 0, 1, 0, 0,  /* 6x */
93     0, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 0, 0, 0, 0, 0,  /* 7x */
94     0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 1, 1, 1, 0,  /* 8x */
95     0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 1, 0, 1, 0,  /* 9x */
96     1, 0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 1, 1, 1, 0,  /* Ax */
97     0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,  /* Bx */
98     0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 1, 1, 1, 1, 1,  /* Cx */
99     0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 1, 1, 1, 1, 1,  /* Dx */
100     0, 0, 1, 1, 1, 1, 1, 1, 1, 1, 0, 1, 1, 1, 1, 1,  /* Ex */
101     1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 1, 1, 1, 1, 0,  /* Fx */
102 };
103 #define IdChar(C)  (((c=C)>=0x42 && sqlite3IsEbcdicIdChar[c-0x40]))
104 #endif
105 
106 
107 /*
108 ** Return the length of the token that begins at z[0].
109 ** Store the token type in *tokenType before returning.
110 */
111 int sqlite3GetToken(const unsigned char *z, int *tokenType){
112   int i, c;
113   switch( *z ){
114     case ' ': case '\t': case '\n': case '\f': case '\r': {
115       testcase( z[0]==' ' );
116       testcase( z[0]=='\t' );
117       testcase( z[0]=='\n' );
118       testcase( z[0]=='\f' );
119       testcase( z[0]=='\r' );
120       for(i=1; sqlite3Isspace(z[i]); i++){}
121       *tokenType = TK_SPACE;
122       return i;
123     }
124     case '-': {
125       if( z[1]=='-' ){
126         /* IMP: R-50417-27976 -- syntax diagram for comments */
127         for(i=2; (c=z[i])!=0 && c!='\n'; i++){}
128         *tokenType = TK_SPACE;   /* IMP: R-22934-25134 */
129         return i;
130       }
131       *tokenType = TK_MINUS;
132       return 1;
133     }
134     case '(': {
135       *tokenType = TK_LP;
136       return 1;
137     }
138     case ')': {
139       *tokenType = TK_RP;
140       return 1;
141     }
142     case ';': {
143       *tokenType = TK_SEMI;
144       return 1;
145     }
146     case '+': {
147       *tokenType = TK_PLUS;
148       return 1;
149     }
150     case '*': {
151       *tokenType = TK_STAR;
152       return 1;
153     }
154     case '/': {
155       if( z[1]!='*' || z[2]==0 ){
156         *tokenType = TK_SLASH;
157         return 1;
158       }
159       /* IMP: R-50417-27976 -- syntax diagram for comments */
160       for(i=3, c=z[2]; (c!='*' || z[i]!='/') && (c=z[i])!=0; i++){}
161       if( c ) i++;
162       *tokenType = TK_SPACE;   /* IMP: R-22934-25134 */
163       return i;
164     }
165     case '%': {
166       *tokenType = TK_REM;
167       return 1;
168     }
169     case '=': {
170       *tokenType = TK_EQ;
171       return 1 + (z[1]=='=');
172     }
173     case '<': {
174       if( (c=z[1])=='=' ){
175         *tokenType = TK_LE;
176         return 2;
177       }else if( c=='>' ){
178         *tokenType = TK_NE;
179         return 2;
180       }else if( c=='<' ){
181         *tokenType = TK_LSHIFT;
182         return 2;
183       }else{
184         *tokenType = TK_LT;
185         return 1;
186       }
187     }
188     case '>': {
189       if( (c=z[1])=='=' ){
190         *tokenType = TK_GE;
191         return 2;
192       }else if( c=='>' ){
193         *tokenType = TK_RSHIFT;
194         return 2;
195       }else{
196         *tokenType = TK_GT;
197         return 1;
198       }
199     }
200     case '!': {
201       if( z[1]!='=' ){
202         *tokenType = TK_ILLEGAL;
203         return 2;
204       }else{
205         *tokenType = TK_NE;
206         return 2;
207       }
208     }
209     case '|': {
210       if( z[1]!='|' ){
211         *tokenType = TK_BITOR;
212         return 1;
213       }else{
214         *tokenType = TK_CONCAT;
215         return 2;
216       }
217     }
218     case ',': {
219       *tokenType = TK_COMMA;
220       return 1;
221     }
222     case '&': {
223       *tokenType = TK_BITAND;
224       return 1;
225     }
226     case '~': {
227       *tokenType = TK_BITNOT;
228       return 1;
229     }
230     case '`':
231     case '\'':
232     case '"': {
233       int delim = z[0];
234       testcase( delim=='`' );
235       testcase( delim=='\'' );
236       testcase( delim=='"' );
237       for(i=1; (c=z[i])!=0; i++){
238         if( c==delim ){
239           if( z[i+1]==delim ){
240             i++;
241           }else{
242             break;
243           }
244         }
245       }
246       if( c=='\'' ){
247         *tokenType = TK_STRING;
248         return i+1;
249       }else if( c!=0 ){
250         *tokenType = TK_ID;
251         return i+1;
252       }else{
253         *tokenType = TK_ILLEGAL;
254         return i;
255       }
256     }
257     case '.': {
258 #ifndef SQLITE_OMIT_FLOATING_POINT
259       if( !sqlite3Isdigit(z[1]) )
260 #endif
261       {
262         *tokenType = TK_DOT;
263         return 1;
264       }
265       /* If the next character is a digit, this is a floating point
266       ** number that begins with ".".  Fall thru into the next case */
267     }
268     case '0': case '1': case '2': case '3': case '4':
269     case '5': case '6': case '7': case '8': case '9': {
270       testcase( z[0]=='0' );  testcase( z[0]=='1' );  testcase( z[0]=='2' );
271       testcase( z[0]=='3' );  testcase( z[0]=='4' );  testcase( z[0]=='5' );
272       testcase( z[0]=='6' );  testcase( z[0]=='7' );  testcase( z[0]=='8' );
273       testcase( z[0]=='9' );
274       *tokenType = TK_INTEGER;
275       for(i=0; sqlite3Isdigit(z[i]); i++){}
276 #ifndef SQLITE_OMIT_FLOATING_POINT
277       if( z[i]=='.' ){
278         i++;
279         while( sqlite3Isdigit(z[i]) ){ i++; }
280         *tokenType = TK_FLOAT;
281       }
282       if( (z[i]=='e' || z[i]=='E') &&
283            ( sqlite3Isdigit(z[i+1])
284             || ((z[i+1]=='+' || z[i+1]=='-') && sqlite3Isdigit(z[i+2]))
285            )
286       ){
287         i += 2;
288         while( sqlite3Isdigit(z[i]) ){ i++; }
289         *tokenType = TK_FLOAT;
290       }
291 #endif
292       while( IdChar(z[i]) ){
293         *tokenType = TK_ILLEGAL;
294         i++;
295       }
296       return i;
297     }
298     case '[': {
299       for(i=1, c=z[0]; c!=']' && (c=z[i])!=0; i++){}
300       *tokenType = c==']' ? TK_ID : TK_ILLEGAL;
301       return i;
302     }
303     case '?': {
304       *tokenType = TK_VARIABLE;
305       for(i=1; sqlite3Isdigit(z[i]); i++){}
306       return i;
307     }
308     case '#': {
309       for(i=1; sqlite3Isdigit(z[i]); i++){}
310       if( i>1 ){
311         /* Parameters of the form #NNN (where NNN is a number) are used
312         ** internally by sqlite3NestedParse.  */
313         *tokenType = TK_REGISTER;
314         return i;
315       }
316       /* Fall through into the next case if the '#' is not followed by
317       ** a digit. Try to match #AAAA where AAAA is a parameter name. */
318     }
319 #ifndef SQLITE_OMIT_TCL_VARIABLE
320     case '$':
321 #endif
322     case '@':  /* For compatibility with MS SQL Server */
323     case ':': {
324       int n = 0;
325       testcase( z[0]=='$' );  testcase( z[0]=='@' );  testcase( z[0]==':' );
326       *tokenType = TK_VARIABLE;
327       for(i=1; (c=z[i])!=0; i++){
328         if( IdChar(c) ){
329           n++;
330 #ifndef SQLITE_OMIT_TCL_VARIABLE
331         }else if( c=='(' && n>0 ){
332           do{
333             i++;
334           }while( (c=z[i])!=0 && !sqlite3Isspace(c) && c!=')' );
335           if( c==')' ){
336             i++;
337           }else{
338             *tokenType = TK_ILLEGAL;
339           }
340           break;
341         }else if( c==':' && z[i+1]==':' ){
342           i++;
343 #endif
344         }else{
345           break;
346         }
347       }
348       if( n==0 ) *tokenType = TK_ILLEGAL;
349       return i;
350     }
351 #ifndef SQLITE_OMIT_BLOB_LITERAL
352     case 'x': case 'X': {
353       testcase( z[0]=='x' ); testcase( z[0]=='X' );
354       if( z[1]=='\'' ){
355         *tokenType = TK_BLOB;
356         for(i=2; sqlite3Isxdigit(z[i]); i++){}
357         if( z[i]!='\'' || i%2 ){
358           *tokenType = TK_ILLEGAL;
359           while( z[i] && z[i]!='\'' ){ i++; }
360         }
361         if( z[i] ) i++;
362         return i;
363       }
364       /* Otherwise fall through to the next case */
365     }
366 #endif
367     default: {
368       if( !IdChar(*z) ){
369         break;
370       }
371       for(i=1; IdChar(z[i]); i++){}
372       *tokenType = keywordCode((char*)z, i);
373       return i;
374     }
375   }
376   *tokenType = TK_ILLEGAL;
377   return 1;
378 }
379 
380 /*
381 ** Run the parser on the given SQL string.  The parser structure is
382 ** passed in.  An SQLITE_ status code is returned.  If an error occurs
383 ** then an and attempt is made to write an error message into
384 ** memory obtained from sqlite3_malloc() and to make *pzErrMsg point to that
385 ** error message.
386 */
387 int sqlite3RunParser(Parse *pParse, const char *zSql, char **pzErrMsg){
388   int nErr = 0;                   /* Number of errors encountered */
389   int i;                          /* Loop counter */
390   void *pEngine;                  /* The LEMON-generated LALR(1) parser */
391   int tokenType;                  /* type of the next token */
392   int lastTokenParsed = -1;       /* type of the previous token */
393   u8 enableLookaside;             /* Saved value of db->lookaside.bEnabled */
394   sqlite3 *db = pParse->db;       /* The database connection */
395   int mxSqlLen;                   /* Max length of an SQL string */
396 
397 
398   mxSqlLen = db->aLimit[SQLITE_LIMIT_SQL_LENGTH];
399   if( db->activeVdbeCnt==0 ){
400     db->u1.isInterrupted = 0;
401   }
402   pParse->rc = SQLITE_OK;
403   pParse->zTail = zSql;
404   i = 0;
405   assert( pzErrMsg!=0 );
406   pEngine = sqlite3ParserAlloc((void*(*)(size_t))sqlite3Malloc);
407   if( pEngine==0 ){
408     db->mallocFailed = 1;
409     return SQLITE_NOMEM;
410   }
411   assert( pParse->pNewTable==0 );
412   assert( pParse->pNewTrigger==0 );
413   assert( pParse->nVar==0 );
414   assert( pParse->nzVar==0 );
415   assert( pParse->azVar==0 );
416   enableLookaside = db->lookaside.bEnabled;
417   if( db->lookaside.pStart ) db->lookaside.bEnabled = 1;
418   while( !db->mallocFailed && zSql[i]!=0 ){
419     assert( i>=0 );
420     pParse->sLastToken.z = &zSql[i];
421     pParse->sLastToken.n = sqlite3GetToken((unsigned char*)&zSql[i],&tokenType);
422     i += pParse->sLastToken.n;
423     if( i>mxSqlLen ){
424       pParse->rc = SQLITE_TOOBIG;
425       break;
426     }
427     switch( tokenType ){
428       case TK_SPACE: {
429         if( db->u1.isInterrupted ){
430           sqlite3ErrorMsg(pParse, "interrupt");
431           pParse->rc = SQLITE_INTERRUPT;
432           goto abort_parse;
433         }
434         break;
435       }
436       case TK_ILLEGAL: {
437         sqlite3DbFree(db, *pzErrMsg);
438         *pzErrMsg = sqlite3MPrintf(db, "unrecognized token: \"%T\"",
439                         &pParse->sLastToken);
440         nErr++;
441         goto abort_parse;
442       }
443       case TK_SEMI: {
444         pParse->zTail = &zSql[i];
445         /* Fall thru into the default case */
446       }
447       default: {
448         sqlite3Parser(pEngine, tokenType, pParse->sLastToken, pParse);
449         lastTokenParsed = tokenType;
450         if( pParse->rc!=SQLITE_OK ){
451           goto abort_parse;
452         }
453         break;
454       }
455     }
456   }
457 abort_parse:
458   if( zSql[i]==0 && nErr==0 && pParse->rc==SQLITE_OK ){
459     if( lastTokenParsed!=TK_SEMI ){
460       sqlite3Parser(pEngine, TK_SEMI, pParse->sLastToken, pParse);
461       pParse->zTail = &zSql[i];
462     }
463     sqlite3Parser(pEngine, 0, pParse->sLastToken, pParse);
464   }
465 #ifdef YYTRACKMAXSTACKDEPTH
466   sqlite3StatusSet(SQLITE_STATUS_PARSER_STACK,
467       sqlite3ParserStackPeak(pEngine)
468   );
469 #endif /* YYDEBUG */
470   sqlite3ParserFree(pEngine, sqlite3_free);
471   db->lookaside.bEnabled = enableLookaside;
472   if( db->mallocFailed ){
473     pParse->rc = SQLITE_NOMEM;
474   }
475   if( pParse->rc!=SQLITE_OK && pParse->rc!=SQLITE_DONE && pParse->zErrMsg==0 ){
476     sqlite3SetString(&pParse->zErrMsg, db, "%s", sqlite3ErrStr(pParse->rc));
477   }
478   assert( pzErrMsg!=0 );
479   if( pParse->zErrMsg ){
480     *pzErrMsg = pParse->zErrMsg;
481     sqlite3_log(pParse->rc, "%s", *pzErrMsg);
482     pParse->zErrMsg = 0;
483     nErr++;
484   }
485   if( pParse->pVdbe && pParse->nErr>0 && pParse->nested==0 ){
486     sqlite3VdbeDelete(pParse->pVdbe);
487     pParse->pVdbe = 0;
488   }
489 #ifndef SQLITE_OMIT_SHARED_CACHE
490   if( pParse->nested==0 ){
491     sqlite3DbFree(db, pParse->aTableLock);
492     pParse->aTableLock = 0;
493     pParse->nTableLock = 0;
494   }
495 #endif
496 #ifndef SQLITE_OMIT_VIRTUALTABLE
497   sqlite3_free(pParse->apVtabLock);
498 #endif
499 
500   if( !IN_DECLARE_VTAB ){
501     /* If the pParse->declareVtab flag is set, do not delete any table
502     ** structure built up in pParse->pNewTable. The calling code (see vtab.c)
503     ** will take responsibility for freeing the Table structure.
504     */
505     sqlite3DeleteTable(db, pParse->pNewTable);
506   }
507 
508   sqlite3DeleteTrigger(db, pParse->pNewTrigger);
509   for(i=pParse->nzVar-1; i>=0; i--) sqlite3DbFree(db, pParse->azVar[i]);
510   sqlite3DbFree(db, pParse->azVar);
511   sqlite3DbFree(db, pParse->aAlias);
512   while( pParse->pAinc ){
513     AutoincInfo *p = pParse->pAinc;
514     pParse->pAinc = p->pNext;
515     sqlite3DbFree(db, p);
516   }
517   while( pParse->pZombieTab ){
518     Table *p = pParse->pZombieTab;
519     pParse->pZombieTab = p->pNextZombie;
520     sqlite3DeleteTable(db, p);
521   }
522   if( nErr>0 && pParse->rc==SQLITE_OK ){
523     pParse->rc = SQLITE_ERROR;
524   }
525   return nErr;
526 }
527