xref: /sqlite-3.40.0/src/tokenize.c (revision 8a29dfde)
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 ** $Id: tokenize.c,v 1.141 2008/04/05 18:41:43 drh Exp $
19 */
20 #include "sqliteInt.h"
21 #include <ctype.h>
22 #include <stdlib.h>
23 
24 /*
25 ** The charMap() macro maps alphabetic characters into their
26 ** lower-case ASCII equivalent.  On ASCII machines, this is just
27 ** an upper-to-lower case map.  On EBCDIC machines we also need
28 ** to adjust the encoding.  Only alphabetic characters and underscores
29 ** need to be translated.
30 */
31 #ifdef SQLITE_ASCII
32 # define charMap(X) sqlite3UpperToLower[(unsigned char)X]
33 #endif
34 #ifdef SQLITE_EBCDIC
35 # define charMap(X) ebcdicToAscii[(unsigned char)X]
36 const unsigned char ebcdicToAscii[] = {
37 /* 0   1   2   3   4   5   6   7   8   9   A   B   C   D   E   F */
38    0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  /* 0x */
39    0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  /* 1x */
40    0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  /* 2x */
41    0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  /* 3x */
42    0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  /* 4x */
43    0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  /* 5x */
44    0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0, 95,  0,  0,  /* 6x */
45    0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  /* 7x */
46    0, 97, 98, 99,100,101,102,103,104,105,  0,  0,  0,  0,  0,  0,  /* 8x */
47    0,106,107,108,109,110,111,112,113,114,  0,  0,  0,  0,  0,  0,  /* 9x */
48    0,  0,115,116,117,118,119,120,121,122,  0,  0,  0,  0,  0,  0,  /* Ax */
49    0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  /* Bx */
50    0, 97, 98, 99,100,101,102,103,104,105,  0,  0,  0,  0,  0,  0,  /* Cx */
51    0,106,107,108,109,110,111,112,113,114,  0,  0,  0,  0,  0,  0,  /* Dx */
52    0,  0,115,116,117,118,119,120,121,122,  0,  0,  0,  0,  0,  0,  /* Ex */
53    0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  0,  /* Fx */
54 };
55 #endif
56 
57 /*
58 ** The sqlite3KeywordCode function looks up an identifier to determine if
59 ** it is a keyword.  If it is a keyword, the token code of that keyword is
60 ** returned.  If the input is not a keyword, TK_ID is returned.
61 **
62 ** The implementation of this routine was generated by a program,
63 ** mkkeywordhash.h, located in the tool subdirectory of the distribution.
64 ** The output of the mkkeywordhash.c program is written into a file
65 ** named keywordhash.h and then included into this source file by
66 ** the #include below.
67 */
68 #include "keywordhash.h"
69 
70 
71 /*
72 ** If X is a character that can be used in an identifier then
73 ** IdChar(X) will be true.  Otherwise it is false.
74 **
75 ** For ASCII, any character with the high-order bit set is
76 ** allowed in an identifier.  For 7-bit characters,
77 ** sqlite3IsIdChar[X] must be 1.
78 **
79 ** For EBCDIC, the rules are more complex but have the same
80 ** end result.
81 **
82 ** Ticket #1066.  the SQL standard does not allow '$' in the
83 ** middle of identfiers.  But many SQL implementations do.
84 ** SQLite will allow '$' in identifiers for compatibility.
85 ** But the feature is undocumented.
86 */
87 #ifdef SQLITE_ASCII
88 const char sqlite3IsAsciiIdChar[] = {
89 /* x0 x1 x2 x3 x4 x5 x6 x7 x8 x9 xA xB xC xD xE xF */
90     0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,  /* 2x */
91     1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 0, 0, 0, 0,  /* 3x */
92     0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,  /* 4x */
93     1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 0, 0, 1,  /* 5x */
94     0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,  /* 6x */
95     1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 0, 0, 0,  /* 7x */
96 };
97 #define IdChar(C)  (((c=C)&0x80)!=0 || (c>0x1f && sqlite3IsAsciiIdChar[c-0x20]))
98 #endif
99 #ifdef SQLITE_EBCDIC
100 const char sqlite3IsEbcdicIdChar[] = {
101 /* x0 x1 x2 x3 x4 x5 x6 x7 x8 x9 xA xB xC xD xE xF */
102     0, 0, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 0, 0, 0, 0,  /* 4x */
103     0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 1, 0, 0, 0, 0,  /* 5x */
104     0, 0, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 0, 1, 0, 0,  /* 6x */
105     0, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 0, 0, 0, 0, 0,  /* 7x */
106     0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 1, 1, 1, 0,  /* 8x */
107     0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 1, 0, 1, 0,  /* 9x */
108     1, 0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 1, 1, 1, 0,  /* Ax */
109     0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,  /* Bx */
110     0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 1, 1, 1, 1, 1,  /* Cx */
111     0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 1, 1, 1, 1, 1,  /* Dx */
112     0, 0, 1, 1, 1, 1, 1, 1, 1, 1, 0, 1, 1, 1, 1, 1,  /* Ex */
113     1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 1, 1, 1, 1, 0,  /* Fx */
114 };
115 #define IdChar(C)  (((c=C)>=0x42 && sqlite3IsEbcdicIdChar[c-0x40]))
116 #endif
117 
118 
119 /*
120 ** Return the length of the token that begins at z[0].
121 ** Store the token type in *tokenType before returning.
122 */
123 int sqlite3GetToken(const unsigned char *z, int *tokenType){
124   int i, c;
125   switch( *z ){
126     case ' ': case '\t': case '\n': case '\f': case '\r': {
127       for(i=1; isspace(z[i]); i++){}
128       *tokenType = TK_SPACE;
129       return i;
130     }
131     case '-': {
132       if( z[1]=='-' ){
133         for(i=2; (c=z[i])!=0 && c!='\n'; i++){}
134         *tokenType = TK_COMMENT;
135         return i;
136       }
137       *tokenType = TK_MINUS;
138       return 1;
139     }
140     case '(': {
141       *tokenType = TK_LP;
142       return 1;
143     }
144     case ')': {
145       *tokenType = TK_RP;
146       return 1;
147     }
148     case ';': {
149       *tokenType = TK_SEMI;
150       return 1;
151     }
152     case '+': {
153       *tokenType = TK_PLUS;
154       return 1;
155     }
156     case '*': {
157       *tokenType = TK_STAR;
158       return 1;
159     }
160     case '/': {
161       if( z[1]!='*' || z[2]==0 ){
162         *tokenType = TK_SLASH;
163         return 1;
164       }
165       for(i=3, c=z[2]; (c!='*' || z[i]!='/') && (c=z[i])!=0; i++){}
166       if( c ) i++;
167       *tokenType = TK_COMMENT;
168       return i;
169     }
170     case '%': {
171       *tokenType = TK_REM;
172       return 1;
173     }
174     case '=': {
175       *tokenType = TK_EQ;
176       return 1 + (z[1]=='=');
177     }
178     case '<': {
179       if( (c=z[1])=='=' ){
180         *tokenType = TK_LE;
181         return 2;
182       }else if( c=='>' ){
183         *tokenType = TK_NE;
184         return 2;
185       }else if( c=='<' ){
186         *tokenType = TK_LSHIFT;
187         return 2;
188       }else{
189         *tokenType = TK_LT;
190         return 1;
191       }
192     }
193     case '>': {
194       if( (c=z[1])=='=' ){
195         *tokenType = TK_GE;
196         return 2;
197       }else if( c=='>' ){
198         *tokenType = TK_RSHIFT;
199         return 2;
200       }else{
201         *tokenType = TK_GT;
202         return 1;
203       }
204     }
205     case '!': {
206       if( z[1]!='=' ){
207         *tokenType = TK_ILLEGAL;
208         return 2;
209       }else{
210         *tokenType = TK_NE;
211         return 2;
212       }
213     }
214     case '|': {
215       if( z[1]!='|' ){
216         *tokenType = TK_BITOR;
217         return 1;
218       }else{
219         *tokenType = TK_CONCAT;
220         return 2;
221       }
222     }
223     case ',': {
224       *tokenType = TK_COMMA;
225       return 1;
226     }
227     case '&': {
228       *tokenType = TK_BITAND;
229       return 1;
230     }
231     case '~': {
232       *tokenType = TK_BITNOT;
233       return 1;
234     }
235     case '`':
236     case '\'':
237     case '"': {
238       int delim = z[0];
239       for(i=1; (c=z[i])!=0; i++){
240         if( c==delim ){
241           if( z[i+1]==delim ){
242             i++;
243           }else{
244             break;
245           }
246         }
247       }
248       if( c ){
249         *tokenType = TK_STRING;
250         return i+1;
251       }else{
252         *tokenType = TK_ILLEGAL;
253         return i;
254       }
255     }
256     case '.': {
257 #ifndef SQLITE_OMIT_FLOATING_POINT
258       if( !isdigit(z[1]) )
259 #endif
260       {
261         *tokenType = TK_DOT;
262         return 1;
263       }
264       /* If the next character is a digit, this is a floating point
265       ** number that begins with ".".  Fall thru into the next case */
266     }
267     case '0': case '1': case '2': case '3': case '4':
268     case '5': case '6': case '7': case '8': case '9': {
269       *tokenType = TK_INTEGER;
270       for(i=0; isdigit(z[i]); i++){}
271 #ifndef SQLITE_OMIT_FLOATING_POINT
272       if( z[i]=='.' ){
273         i++;
274         while( isdigit(z[i]) ){ i++; }
275         *tokenType = TK_FLOAT;
276       }
277       if( (z[i]=='e' || z[i]=='E') &&
278            ( isdigit(z[i+1])
279             || ((z[i+1]=='+' || z[i+1]=='-') && isdigit(z[i+2]))
280            )
281       ){
282         i += 2;
283         while( isdigit(z[i]) ){ i++; }
284         *tokenType = TK_FLOAT;
285       }
286 #endif
287       while( IdChar(z[i]) ){
288         *tokenType = TK_ILLEGAL;
289         i++;
290       }
291       return i;
292     }
293     case '[': {
294       for(i=1, c=z[0]; c!=']' && (c=z[i])!=0; i++){}
295       *tokenType = c==']' ? TK_ID : TK_ILLEGAL;
296       return i;
297     }
298     case '?': {
299       *tokenType = TK_VARIABLE;
300       for(i=1; isdigit(z[i]); i++){}
301       return i;
302     }
303     case '#': {
304       for(i=1; isdigit(z[i]); i++){}
305       if( i>1 ){
306         /* Parameters of the form #NNN (where NNN is a number) are used
307         ** internally by sqlite3NestedParse.  */
308         *tokenType = TK_REGISTER;
309         return i;
310       }
311       /* Fall through into the next case if the '#' is not followed by
312       ** a digit. Try to match #AAAA where AAAA is a parameter name. */
313     }
314 #ifndef SQLITE_OMIT_TCL_VARIABLE
315     case '$':
316 #endif
317     case '@':  /* For compatibility with MS SQL Server */
318     case ':': {
319       int n = 0;
320       *tokenType = TK_VARIABLE;
321       for(i=1; (c=z[i])!=0; i++){
322         if( IdChar(c) ){
323           n++;
324 #ifndef SQLITE_OMIT_TCL_VARIABLE
325         }else if( c=='(' && n>0 ){
326           do{
327             i++;
328           }while( (c=z[i])!=0 && !isspace(c) && c!=')' );
329           if( c==')' ){
330             i++;
331           }else{
332             *tokenType = TK_ILLEGAL;
333           }
334           break;
335         }else if( c==':' && z[i+1]==':' ){
336           i++;
337 #endif
338         }else{
339           break;
340         }
341       }
342       if( n==0 ) *tokenType = TK_ILLEGAL;
343       return i;
344     }
345 #ifndef SQLITE_OMIT_BLOB_LITERAL
346     case 'x': case 'X': {
347       if( z[1]=='\'' ){
348         *tokenType = TK_BLOB;
349         for(i=2; (c=z[i])!=0 && c!='\''; i++){
350           if( !isxdigit(c) ){
351             *tokenType = TK_ILLEGAL;
352           }
353         }
354         if( i%2 || !c ) *tokenType = TK_ILLEGAL;
355         if( c ) i++;
356         return i;
357       }
358       /* Otherwise fall through to the next case */
359     }
360 #endif
361     default: {
362       if( !IdChar(*z) ){
363         break;
364       }
365       for(i=1; IdChar(z[i]); i++){}
366       *tokenType = keywordCode((char*)z, i);
367       return i;
368     }
369   }
370   *tokenType = TK_ILLEGAL;
371   return 1;
372 }
373 
374 /*
375 ** Run the parser on the given SQL string.  The parser structure is
376 ** passed in.  An SQLITE_ status code is returned.  If an error occurs
377 ** and pzErrMsg!=NULL then an error message might be written into
378 ** memory obtained from sqlite3_malloc() and *pzErrMsg made to point to that
379 ** error message.  Or maybe not.
380 */
381 int sqlite3RunParser(Parse *pParse, const char *zSql, char **pzErrMsg){
382   int nErr = 0;
383   int i;
384   void *pEngine;
385   int tokenType;
386   int lastTokenParsed = -1;
387   sqlite3 *db = pParse->db;
388   int mxSqlLen = db->aLimit[SQLITE_LIMIT_SQL_LENGTH];
389 
390   if( db->activeVdbeCnt==0 ){
391     db->u1.isInterrupted = 0;
392   }
393   pParse->rc = SQLITE_OK;
394   pParse->zTail = pParse->zSql = zSql;
395   i = 0;
396   pEngine = sqlite3ParserAlloc((void*(*)(size_t))sqlite3_malloc);
397   if( pEngine==0 ){
398     db->mallocFailed = 1;
399     return SQLITE_NOMEM;
400   }
401   assert( pParse->sLastToken.dyn==0 );
402   assert( pParse->pNewTable==0 );
403   assert( pParse->pNewTrigger==0 );
404   assert( pParse->nVar==0 );
405   assert( pParse->nVarExpr==0 );
406   assert( pParse->nVarExprAlloc==0 );
407   assert( pParse->apVarExpr==0 );
408   while( !db->mallocFailed && zSql[i]!=0 ){
409     assert( i>=0 );
410     pParse->sLastToken.z = (u8*)&zSql[i];
411     assert( pParse->sLastToken.dyn==0 );
412     pParse->sLastToken.n = sqlite3GetToken((unsigned char*)&zSql[i],&tokenType);
413     i += pParse->sLastToken.n;
414     if( i>mxSqlLen ){
415       pParse->rc = SQLITE_TOOBIG;
416       break;
417     }
418     switch( tokenType ){
419       case TK_SPACE:
420       case TK_COMMENT: {
421         if( db->u1.isInterrupted ){
422           pParse->rc = SQLITE_INTERRUPT;
423           sqlite3SetString(pzErrMsg, "interrupt", (char*)0);
424           goto abort_parse;
425         }
426         break;
427       }
428       case TK_ILLEGAL: {
429         if( pzErrMsg ){
430           sqlite3_free(*pzErrMsg);
431           *pzErrMsg = sqlite3MPrintf(db, "unrecognized token: \"%T\"",
432                           &pParse->sLastToken);
433         }
434         nErr++;
435         goto abort_parse;
436       }
437       case TK_SEMI: {
438         pParse->zTail = &zSql[i];
439         /* Fall thru into the default case */
440       }
441       default: {
442         sqlite3Parser(pEngine, tokenType, pParse->sLastToken, pParse);
443         lastTokenParsed = tokenType;
444         if( pParse->rc!=SQLITE_OK ){
445           goto abort_parse;
446         }
447         break;
448       }
449     }
450   }
451 abort_parse:
452   if( zSql[i]==0 && nErr==0 && pParse->rc==SQLITE_OK ){
453     if( lastTokenParsed!=TK_SEMI ){
454       sqlite3Parser(pEngine, TK_SEMI, pParse->sLastToken, pParse);
455       pParse->zTail = &zSql[i];
456     }
457     sqlite3Parser(pEngine, 0, pParse->sLastToken, pParse);
458   }
459   sqlite3ParserFree(pEngine, sqlite3_free);
460   if( db->mallocFailed ){
461     pParse->rc = SQLITE_NOMEM;
462   }
463   if( pParse->rc!=SQLITE_OK && pParse->rc!=SQLITE_DONE && pParse->zErrMsg==0 ){
464     sqlite3SetString(&pParse->zErrMsg, sqlite3ErrStr(pParse->rc), (char*)0);
465   }
466   if( pParse->zErrMsg ){
467     if( pzErrMsg && *pzErrMsg==0 ){
468       *pzErrMsg = pParse->zErrMsg;
469     }else{
470       sqlite3_free(pParse->zErrMsg);
471     }
472     pParse->zErrMsg = 0;
473     nErr++;
474   }
475   if( pParse->pVdbe && pParse->nErr>0 && pParse->nested==0 ){
476     sqlite3VdbeDelete(pParse->pVdbe);
477     pParse->pVdbe = 0;
478   }
479 #ifndef SQLITE_OMIT_SHARED_CACHE
480   if( pParse->nested==0 ){
481     sqlite3_free(pParse->aTableLock);
482     pParse->aTableLock = 0;
483     pParse->nTableLock = 0;
484   }
485 #endif
486 
487   if( !IN_DECLARE_VTAB ){
488     /* If the pParse->declareVtab flag is set, do not delete any table
489     ** structure built up in pParse->pNewTable. The calling code (see vtab.c)
490     ** will take responsibility for freeing the Table structure.
491     */
492     sqlite3DeleteTable(pParse->pNewTable);
493   }
494 
495   sqlite3DeleteTrigger(pParse->pNewTrigger);
496   sqlite3_free(pParse->apVarExpr);
497   if( nErr>0 && (pParse->rc==SQLITE_OK || pParse->rc==SQLITE_DONE) ){
498     pParse->rc = SQLITE_ERROR;
499   }
500   return nErr;
501 }
502