xref: /trunk/main/basic/source/inc/token.hxx (revision cdf0e10c4e3984b49a9502b011690b615761d4a3)
1*cdf0e10cSrcweir /*************************************************************************
2*cdf0e10cSrcweir  *
3*cdf0e10cSrcweir  * DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER.
4*cdf0e10cSrcweir  *
5*cdf0e10cSrcweir  * Copyright 2000, 2010 Oracle and/or its affiliates.
6*cdf0e10cSrcweir  *
7*cdf0e10cSrcweir  * OpenOffice.org - a multi-platform office productivity suite
8*cdf0e10cSrcweir  *
9*cdf0e10cSrcweir  * This file is part of OpenOffice.org.
10*cdf0e10cSrcweir  *
11*cdf0e10cSrcweir  * OpenOffice.org is free software: you can redistribute it and/or modify
12*cdf0e10cSrcweir  * it under the terms of the GNU Lesser General Public License version 3
13*cdf0e10cSrcweir  * only, as published by the Free Software Foundation.
14*cdf0e10cSrcweir  *
15*cdf0e10cSrcweir  * OpenOffice.org is distributed in the hope that it will be useful,
16*cdf0e10cSrcweir  * but WITHOUT ANY WARRANTY; without even the implied warranty of
17*cdf0e10cSrcweir  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
18*cdf0e10cSrcweir  * GNU Lesser General Public License version 3 for more details
19*cdf0e10cSrcweir  * (a copy is included in the LICENSE file that accompanied this code).
20*cdf0e10cSrcweir  *
21*cdf0e10cSrcweir  * You should have received a copy of the GNU Lesser General Public License
22*cdf0e10cSrcweir  * version 3 along with OpenOffice.org.  If not, see
23*cdf0e10cSrcweir  * <http://www.openoffice.org/license.html>
24*cdf0e10cSrcweir  * for a copy of the LGPLv3 License.
25*cdf0e10cSrcweir  *
26*cdf0e10cSrcweir  ************************************************************************/
27*cdf0e10cSrcweir 
28*cdf0e10cSrcweir #ifndef _TOKEN_HXX
29*cdf0e10cSrcweir #define _TOKEN_HXX
30*cdf0e10cSrcweir 
31*cdf0e10cSrcweir #include "scanner.hxx"
32*cdf0e10cSrcweir #ifndef _SBDEF_HXX
33*cdf0e10cSrcweir #include <basic/sbdef.hxx>
34*cdf0e10cSrcweir #endif
35*cdf0e10cSrcweir 
36*cdf0e10cSrcweir #if defined( SHARED )
37*cdf0e10cSrcweir #define SbiTokenSHAREDTMPUNDEF
38*cdf0e10cSrcweir #undef SHARED
39*cdf0e10cSrcweir #endif
40*cdf0e10cSrcweir 
41*cdf0e10cSrcweir // Der Tokenizer ist stand-alone, d.h. er kann von ueberallher verwendet
42*cdf0e10cSrcweir // werden. Eine BASIC-Instanz ist fuer Fehlermeldungen notwendig. Ohne
43*cdf0e10cSrcweir // BASIC werden die Fehler nur gezaehlt. Auch ist Basic notwendig, wenn
44*cdf0e10cSrcweir // eine erweiterte SBX-Variable zur Erkennung von Datentypen etc. verwendet
45*cdf0e10cSrcweir // werden soll.
46*cdf0e10cSrcweir 
47*cdf0e10cSrcweir enum SbiToken {
48*cdf0e10cSrcweir     NIL = 0,
49*cdf0e10cSrcweir     // Token zwischen 0x20 und 0x3F sind Literale:
50*cdf0e10cSrcweir     LPAREN = '(', RPAREN = ')', COMMA = ',', DOT = '.', EXCLAM = '!',
51*cdf0e10cSrcweir     HASH = '#', SEMICOLON = ';',
52*cdf0e10cSrcweir 
53*cdf0e10cSrcweir     // Anweisungen:
54*cdf0e10cSrcweir     FIRSTKWD = 0x40,
55*cdf0e10cSrcweir     AS = FIRSTKWD, ALIAS, ASSIGN,
56*cdf0e10cSrcweir     CALL, CASE, CLOSE, COMPARE, _CONST_,
57*cdf0e10cSrcweir     DECLARE, DIM, DO,
58*cdf0e10cSrcweir 
59*cdf0e10cSrcweir     // in der Reihenfolge der Datentyp-Enums!
60*cdf0e10cSrcweir     DEFINT, DEFLNG, DEFSNG, DEFDBL, DEFCUR, DEFDATE, DEFSTR, DEFOBJ,
61*cdf0e10cSrcweir     DEFERR, DEFBOOL, DEFVAR,
62*cdf0e10cSrcweir     // in der Reihenfolge der Datentyp-Enums!
63*cdf0e10cSrcweir     DATATYPE1,
64*cdf0e10cSrcweir     TINTEGER = DATATYPE1,
65*cdf0e10cSrcweir     TLONG, TSINGLE, TDOUBLE, TCURRENCY, TDATE, TSTRING, TOBJECT,
66*cdf0e10cSrcweir     _ERROR_, TBOOLEAN, TVARIANT, TBYTE,
67*cdf0e10cSrcweir     DATATYPE2 = TBYTE,
68*cdf0e10cSrcweir 
69*cdf0e10cSrcweir     EACH, ELSE, ELSEIF, END, ERASE, EXIT,
70*cdf0e10cSrcweir     FOR, FUNCTION,
71*cdf0e10cSrcweir     GET, GLOBAL, GOSUB, GOTO,
72*cdf0e10cSrcweir     IF, _IN_, INPUT,
73*cdf0e10cSrcweir     LET, LINE, LINEINPUT, LOCAL, LOOP, LPRINT, LSET,
74*cdf0e10cSrcweir     NAME, NEW, NEXT,
75*cdf0e10cSrcweir     ON, OPEN, OPTION, IMPLEMENTS,
76*cdf0e10cSrcweir     PRINT, PRIVATE, PROPERTY, PUBLIC,
77*cdf0e10cSrcweir     REDIM, REM, RESUME, RETURN, RSET,
78*cdf0e10cSrcweir     SELECT, SET, SHARED, STATIC, STEP, STOP, SUB,
79*cdf0e10cSrcweir     TEXT, THEN, TO, TYPE, ENUM,
80*cdf0e10cSrcweir     UNTIL,
81*cdf0e10cSrcweir     WEND, WHILE, WITH, WRITE,
82*cdf0e10cSrcweir     ENDENUM, ENDIF, ENDFUNC, ENDPROPERTY, ENDSUB, ENDTYPE, ENDSELECT, ENDWITH,
83*cdf0e10cSrcweir     // Ende aller Keywords
84*cdf0e10cSrcweir     LASTKWD = ENDWITH,
85*cdf0e10cSrcweir     // Statement-Ende
86*cdf0e10cSrcweir     EOS, EOLN,
87*cdf0e10cSrcweir     // Operatoren:
88*cdf0e10cSrcweir     EXPON, NEG, MUL,
89*cdf0e10cSrcweir     DIV, IDIV, MOD, PLUS, MINUS,
90*cdf0e10cSrcweir     EQ, NE, LT, GT, LE, GE,
91*cdf0e10cSrcweir     NOT, AND, OR, XOR, EQV,
92*cdf0e10cSrcweir     IMP, CAT, LIKE, IS, TYPEOF,
93*cdf0e10cSrcweir     // Sonstiges:
94*cdf0e10cSrcweir     FIRSTEXTRA,
95*cdf0e10cSrcweir     NUMBER=FIRSTEXTRA, FIXSTRING, SYMBOL, _CDECL_, BYVAL, BYREF,
96*cdf0e10cSrcweir     OUTPUT, RANDOM, APPEND, BINARY, ACCESS,
97*cdf0e10cSrcweir     LOCK, READ, PRESERVE, BASE, ANY, LIB, _OPTIONAL_,
98*cdf0e10cSrcweir     EXPLICIT, COMPATIBLE, CLASSMODULE, PARAMARRAY, WITHEVENTS,
99*cdf0e10cSrcweir 
100*cdf0e10cSrcweir     // Ab hier kommen JavaScript-Tokens (gleiches enum, damit gleicher Typ)
101*cdf0e10cSrcweir     FIRSTJAVA,
102*cdf0e10cSrcweir     JS_BREAK=FIRSTJAVA, JS_CONTINUE, JS_FOR, JS_FUNCTION, JS_IF, JS_NEW,
103*cdf0e10cSrcweir     JS_RETURN, JS_THIS, JS_VAR, JS_WHILE, JS_WITH,
104*cdf0e10cSrcweir 
105*cdf0e10cSrcweir     // JavaScript-Operatoren
106*cdf0e10cSrcweir     // _ASS_ = Assignment
107*cdf0e10cSrcweir     JS_COMMA, JS_ASSIGNMENT, JS_ASS_PLUS, JS_ASS_MINUS, JS_ASS_MUL,
108*cdf0e10cSrcweir     JS_ASS_DIV, JS_ASS_MOD, JS_ASS_LSHIFT, JS_ASS_RSHIFT, JS_ASS_RSHIFT_Z,
109*cdf0e10cSrcweir     JS_ASS_AND, JS_ASS_XOR, JS_ASS_OR,
110*cdf0e10cSrcweir     JS_COND_QUEST, JS_COND_SEL, JS_LOG_OR, JS_LOG_AND, JS_BIT_OR,
111*cdf0e10cSrcweir     JS_BIT_XOR, JS_BIT_AND, JS_EQ, JS_NE, JS_LT, JS_LE,
112*cdf0e10cSrcweir     JS_GT, JS_GE, JS_LSHIFT, JS_RSHIFT, JS_RSHIFT_Z,
113*cdf0e10cSrcweir     JS_PLUS, JS_MINUS, JS_MUL, JS_DIV, JS_MOD, JS_LOG_NOT, JS_BIT_NOT,
114*cdf0e10cSrcweir     JS_INC, JS_DEC, JS_LPAREN, JS_RPAREN, JS_LINDEX, JS_RINDEX
115*cdf0e10cSrcweir     , VBASUPPORT
116*cdf0e10cSrcweir };
117*cdf0e10cSrcweir 
118*cdf0e10cSrcweir #ifdef SbiTokenSHAREDTMPUNDEF
119*cdf0e10cSrcweir #define SHARED
120*cdf0e10cSrcweir #undef SbiTokenSHAREDTMPUNDEF
121*cdf0e10cSrcweir #endif
122*cdf0e10cSrcweir 
123*cdf0e10cSrcweir // #i109076
124*cdf0e10cSrcweir class TokenLabelInfo
125*cdf0e10cSrcweir {
126*cdf0e10cSrcweir     bool* m_pTokenCanBeLabelTab;
127*cdf0e10cSrcweir 
128*cdf0e10cSrcweir public:
129*cdf0e10cSrcweir     TokenLabelInfo( void );
130*cdf0e10cSrcweir     TokenLabelInfo( const TokenLabelInfo& rInfo )
131*cdf0e10cSrcweir         : m_pTokenCanBeLabelTab( NULL )
132*cdf0e10cSrcweir             { (void)rInfo; }
133*cdf0e10cSrcweir     ~TokenLabelInfo();
134*cdf0e10cSrcweir 
135*cdf0e10cSrcweir     bool canTokenBeLabel( SbiToken eTok )
136*cdf0e10cSrcweir         { return m_pTokenCanBeLabelTab[eTok]; }
137*cdf0e10cSrcweir };
138*cdf0e10cSrcweir 
139*cdf0e10cSrcweir class SbiTokenizer : public SbiScanner {
140*cdf0e10cSrcweir     TokenLabelInfo  m_aTokenLabelInfo;
141*cdf0e10cSrcweir 
142*cdf0e10cSrcweir protected:
143*cdf0e10cSrcweir     SbiToken eCurTok;               // aktuelles Token
144*cdf0e10cSrcweir     SbiToken ePush;                 // Pushback-Token
145*cdf0e10cSrcweir     sal_uInt16  nPLine, nPCol1, nPCol2; // Pushback-Location
146*cdf0e10cSrcweir     sal_Bool bEof;                      // sal_True bei Dateiende
147*cdf0e10cSrcweir     sal_Bool bEos;                      // sal_True bei Statement-Ende
148*cdf0e10cSrcweir     sal_Bool bKeywords;                 // sal_True, falls Keywords geparst werden
149*cdf0e10cSrcweir     sal_Bool bAs;                       // letztes Keyword war AS
150*cdf0e10cSrcweir     sal_Bool bErrorIsSymbol;            // Handle Error token as Symbol, not keyword
151*cdf0e10cSrcweir public:
152*cdf0e10cSrcweir     SbiTokenizer( const ::rtl::OUString&, StarBASIC* = NULL );
153*cdf0e10cSrcweir    ~SbiTokenizer();
154*cdf0e10cSrcweir 
155*cdf0e10cSrcweir     inline sal_Bool IsEof()             { return bEof; }
156*cdf0e10cSrcweir     inline sal_Bool IsEos()             { return bEos; }
157*cdf0e10cSrcweir 
158*cdf0e10cSrcweir     void  Push( SbiToken );             // Pushback eines Tokens
159*cdf0e10cSrcweir     const String& Symbol( SbiToken );// Rueckumwandlung
160*cdf0e10cSrcweir 
161*cdf0e10cSrcweir     SbiToken Peek();                    // das naechste Token lesen
162*cdf0e10cSrcweir     SbiToken Next();                    // Ein Token lesen
163*cdf0e10cSrcweir     sal_Bool MayBeLabel( sal_Bool= sal_False ); // Kann es ein Label sein?
164*cdf0e10cSrcweir 
165*cdf0e10cSrcweir     void Hilite( SbTextPortions& ); // Syntax-Highlighting
166*cdf0e10cSrcweir 
167*cdf0e10cSrcweir     void Error( SbError c ) { GenError( c ); }
168*cdf0e10cSrcweir     void Error( SbError, SbiToken );
169*cdf0e10cSrcweir     void Error( SbError, const char* );
170*cdf0e10cSrcweir     void Error( SbError, String );
171*cdf0e10cSrcweir 
172*cdf0e10cSrcweir     void Keywords( sal_Bool b ) { bKeywords = b; }
173*cdf0e10cSrcweir 
174*cdf0e10cSrcweir     static sal_Bool IsEoln( SbiToken t )
175*cdf0e10cSrcweir         { return sal_Bool( t == EOS || t == EOLN || t == REM ); }
176*cdf0e10cSrcweir     static sal_Bool IsKwd( SbiToken t )
177*cdf0e10cSrcweir         { return sal_Bool( t >= FIRSTKWD && t <= LASTKWD ); }
178*cdf0e10cSrcweir     static sal_Bool IsExtra( SbiToken t )
179*cdf0e10cSrcweir         { return sal_Bool( t >= FIRSTEXTRA ); }
180*cdf0e10cSrcweir };
181*cdf0e10cSrcweir 
182*cdf0e10cSrcweir 
183*cdf0e10cSrcweir #endif
184