xref: /core/basic/source/comp/scanner.cxx (revision 7ddedd25) 
1 /* -*- Mode: C++; tab-width: 4; indent-tabs-mode: nil; c-basic-offset: 4 -*- */
2 /*
3  * This file is part of the LibreOffice project.
4  *
5  * This Source Code Form is subject to the terms of the Mozilla Public
6  * License, v. 2.0. If a copy of the MPL was not distributed with this
7  * file, You can obtain one at http://mozilla.org/MPL/2.0/.
8  *
9  * This file incorporates work covered by the following license notice:
10  *
11  *   Licensed to the Apache Software Foundation (ASF) under one or more
12  *   contributor license agreements. See the NOTICE file distributed
13  *   with this work for additional information regarding copyright
14  *   ownership. The ASF licenses this file to you under the Apache
15  *   License, Version 2.0 (the "License"); you may not use this file
16  *   except in compliance with the License. You may obtain a copy of
17  *   the License at http://www.apache.org/licenses/LICENSE-2.0 .
18  */
19 
20 #include <basiccharclass.hxx>
21 #include <scanner.hxx>
22 #include <sbintern.hxx>
23 #include <runtime.hxx>
24 
25 #include <basic/sberrors.hxx>
26 #include <i18nlangtag/lang.h>
27 #include <svl/zforlist.hxx>
28 #include <rtl/character.hxx>
29 
30 SbiScanner::SbiScanner( const OUString& rBuf, StarBASIC* p ) : aBuf( rBuf )
31 {
32     pBasic   = p;
33     nLineIdx = -1;
34     nVal     = 0;
35     eScanType = SbxVARIANT;
36     nErrors  = 0;
37     nBufPos  = 0;
38     nSavedCol1 = 0;
39     nColLock = 0;
40     nLine    = 0;
41     nCol1    = 0;
42     nCol2    = 0;
43     nCol     = 0;
44     bError   =
45     bAbort   =
46     bSpaces  =
47     bNumber  =
48     bSymbol  =
49     bCompatible =
50     bVBASupportOn =
51     bInStatement =
52     bPrevLineExtentsComment = false;
53     bHash    = true;
54     nSaveLineIdx = -1;
55 }
56 
57 void SbiScanner::LockColumn()
58 {
59     if( !nColLock++ )
60         nSavedCol1 = nCol1;
61 }
62 
63 void SbiScanner::UnlockColumn()
64 {
65     if( nColLock )
66         nColLock--;
67 }
68 
69 void SbiScanner::GenError( ErrCode code )
70 {
71     if( GetSbData()->bBlockCompilerError )
72     {
73         bAbort = true;
74         return;
75     }
76     if( !bError )
77     {
78         bool bRes = true;
79         // report only one error per statement
80         bError = true;
81         if( pBasic )
82         {
83             // in case of EXPECTED or UNEXPECTED it always refers
84             // to the last token, so take the Col1 over
85             sal_Int32 nc = nColLock ? nSavedCol1 : nCol1;
86             if ( code.anyOf(
87                     ERRCODE_BASIC_EXPECTED,
88                     ERRCODE_BASIC_UNEXPECTED,
89                     ERRCODE_BASIC_SYMBOL_EXPECTED,
90                     ERRCODE_BASIC_LABEL_EXPECTED) )
91             {
92                     nc = nCol1;
93                     if( nc > nCol2 ) nCol2 = nc;
94             }
95             bRes = pBasic->CError( code, aError, nLine, nc, nCol2 );
96         }
97         bAbort = bAbort || !bRes  || ( code == ERRCODE_BASIC_NO_MEMORY || code == ERRCODE_BASIC_PROG_TOO_LARGE );
98     }
99     nErrors++;
100 }
101 
102 
103 // used by SbiTokenizer::MayBeLabel() to detect a label
104 bool SbiScanner::DoesColonFollow()
105 {
106     if(nCol < aLine.getLength() && aLine[nCol] == ':')
107     {
108         ++nLineIdx; ++nCol;
109         return true;
110     }
111     else
112         return false;
113 }
114 
115 // test for legal suffix
116 static SbxDataType GetSuffixType( sal_Unicode c )
117 {
118     switch (c)
119     {
120     case '%':
121         return SbxINTEGER;
122     case '&':
123         return SbxLONG;
124     case '!':
125         return SbxSINGLE;
126     case '#':
127         return SbxDOUBLE;
128     case '@':
129         return SbxCURRENCY;
130     case '$':
131         return SbxSTRING;
132     default:
133         return SbxVARIANT;
134     }
135 }
136 
137 // reading the next symbol into the variables aSym, nVal and eType
138 // return value is sal_False at EOF or errors
139 #define BUF_SIZE 80
140 
141 void SbiScanner::scanAlphanumeric()
142 {
143     sal_Int32 n = nCol;
144     while(nCol < aLine.getLength() && (BasicCharClass::isAlphaNumeric(aLine[nCol], bCompatible) || aLine[nCol] == '_'))
145     {
146         ++nLineIdx;
147         ++nCol;
148     }
149     aSym = aLine.copy(n, nCol - n);
150 }
151 
152 void SbiScanner::scanGoto()
153 {
154     sal_Int32 n = nCol;
155     while(n < aLine.getLength() && BasicCharClass::isWhitespace(aLine[n]))
156         ++n;
157 
158     if(n + 1 < aLine.getLength())
159     {
160         OUString aTemp = aLine.copy(n, 2);
161         if(aTemp.equalsIgnoreAsciiCase("to"))
162         {
163             aSym = "goto";
164             nLineIdx += n + 2 - nCol;
165             nCol = n + 2;
166         }
167     }
168 }
169 
170 bool SbiScanner::readLine()
171 {
172     if(nBufPos >= aBuf.getLength())
173         return false;
174 
175     sal_Int32 n = nBufPos;
176     sal_Int32 nLen = aBuf.getLength();
177 
178     while(n < nLen && aBuf[n] != '\r' && aBuf[n] != '\n')
179         ++n;
180 
181     // Trim trailing whitespace
182     sal_Int32 nEnd = n;
183     while(nBufPos < nEnd && BasicCharClass::isWhitespace(aBuf[nEnd - 1]))
184         --nEnd;
185 
186     aLine = aBuf.copy(nBufPos, nEnd - nBufPos);
187 
188     // Fast-forward past the line ending
189     if(n + 1 < nLen && aBuf[n] == '\r' && aBuf[n + 1] == '\n')
190         n += 2;
191     else if(n < nLen)
192         ++n;
193 
194     nBufPos = n;
195     nLineIdx = 0;
196 
197     ++nLine;
198     nCol = nCol1 = nCol2 = 0;
199     nColLock = 0;
200 
201     return true;
202 }
203 
204 bool SbiScanner::NextSym()
205 {
206     // memorize for the EOLN-case
207     sal_Int32 nOldLine = nLine;
208     sal_Int32 nOldCol1 = nCol1;
209     sal_Int32 nOldCol2 = nCol2;
210     sal_Unicode buf[ BUF_SIZE ], *p = buf;
211 
212     eScanType = SbxVARIANT;
213     aSym.clear();
214     bHash = bSymbol = bNumber = bSpaces = false;
215     bool bCompilerDirective = false;
216 
217     // read in line?
218     if (nLineIdx == -1)
219     {
220         if(!readLine())
221             return false;
222 
223         nOldLine = nLine;
224         nOldCol1 = nOldCol2 = 0;
225     }
226 
227     const sal_Int32 nLineIdxScanStart = nLineIdx;
228 
229     if(nCol < aLine.getLength() && BasicCharClass::isWhitespace(aLine[nCol]))
230     {
231         bSpaces = true;
232         while(nCol < aLine.getLength() && BasicCharClass::isWhitespace(aLine[nCol]))
233         {
234             ++nLineIdx;
235             ++nCol;
236         }
237     }
238 
239     nCol1 = nCol;
240 
241     // only blank line?
242     if(nCol >= aLine.getLength())
243         goto eoln;
244 
245     if( bPrevLineExtentsComment )
246         goto PrevLineCommentLbl;
247 
248     if(nCol < aLine.getLength() && aLine[nCol] == '#')
249     {
250         sal_Int32 nLineTempIdx = nLineIdx;
251         do
252         {
253             nLineTempIdx++;
254         } while (nLineTempIdx < aLine.getLength() && !BasicCharClass::isWhitespace(aLine[nLineTempIdx]) && aLine[nLineTempIdx] != '#');
255         // leave it if it is a date literal - it will be handled later
256         if (nLineTempIdx >= aLine.getLength() || aLine[nLineTempIdx] != '#')
257         {
258             ++nLineIdx;
259             ++nCol;
260             //ignore compiler directives (# is first non-space character)
261             if (nOldCol2 == 0)
262                 bCompilerDirective = true;
263             else
264                 bHash = true;
265         }
266     }
267 
268     // copy character if symbol
269     if(nCol < aLine.getLength() && (BasicCharClass::isAlpha(aLine[nCol], bCompatible) || aLine[nCol] == '_'))
270     {
271         // if there's nothing behind '_' , it's the end of a line!
272         if(nCol + 1 == aLine.getLength() && aLine[nCol] == '_')
273         {
274             // Note that nCol is not incremented here...
275             ++nLineIdx;
276             goto eoln;
277         }
278 
279         bSymbol = true;
280 
281         scanAlphanumeric();
282 
283         // Special handling for "go to"
284         if(nCol < aLine.getLength() && bCompatible && aSym.equalsIgnoreAsciiCase("go"))
285             scanGoto();
286 
287         // replace closing '_' by space when end of line is following
288         // (wrong line continuation otherwise)
289         if (nCol == aLine.getLength() && aLine[nCol - 1] == '_')
290         {
291             // We are going to modify a potentially shared string, so force
292             // a copy, so that aSym is not modified by the following operation
293             OUString aSymCopy( aSym.getStr(), aSym.getLength() );
294             aSym = aSymCopy;
295 
296             // HACK: modifying a potentially shared string here!
297             const_cast<sal_Unicode*>(aLine.getStr())[nLineIdx - 1] = ' ';
298         }
299 
300         // type recognition?
301         // don't test the exclamation mark
302         // if there's a symbol behind it
303         else if((nCol >= aLine.getLength() || aLine[nCol] != '!') ||
304                 (nCol + 1 >= aLine.getLength() || !BasicCharClass::isAlpha(aLine[nCol + 1], bCompatible)))
305         {
306             if(nCol < aLine.getLength())
307             {
308                 SbxDataType t(GetSuffixType(aLine[nCol]));
309                 if( t != SbxVARIANT )
310                 {
311                     eScanType = t;
312                     ++nLineIdx;
313                     ++nCol;
314                 }
315             }
316         }
317     }
318 
319     // read in and convert if number
320     else if((nCol < aLine.getLength() && rtl::isAsciiDigit(aLine[nCol])) ||
321             (nCol + 1 < aLine.getLength() && aLine[nCol] == '.' && rtl::isAsciiDigit(aLine[nCol + 1])))
322     {
323         short exp = 0;
324         short dec = 0;
325         eScanType = SbxDOUBLE;
326         bool bScanError = false;
327         bool bBufOverflow = false;
328         // All this because of 'D' or 'd' floating point type, sigh...
329         while(!bScanError && nCol < aLine.getLength() && strchr("0123456789.DEde", aLine[nCol]))
330         {
331             // from 4.1.1996: buffer full? -> go on scanning empty
332             if( (p-buf) == (BUF_SIZE-1) )
333             {
334                 bBufOverflow = true;
335                 ++nLineIdx;
336                 ++nCol;
337                 continue;
338             }
339             // point or exponent?
340             if(aLine[nCol] == '.')
341             {
342                 if( ++dec > 1 )
343                     bScanError = true;
344                 else
345                     *p++ = '.';
346             }
347             else if(strchr("DdEe", aLine[nCol]))
348             {
349                 if (++exp > 1)
350                     bScanError = true;
351                 else
352                 {
353                     *p++ = 'E';
354                     if (nCol + 1 < aLine.getLength() && (aLine[nCol+1] == '+' || aLine[nCol+1] == '-'))
355                     {
356                         ++nLineIdx;
357                         ++nCol;
358                         if( (p-buf) == (BUF_SIZE-1) )
359                         {
360                             bBufOverflow = true;
361                             continue;
362                         }
363                         *p++ = aLine[nCol];
364                     }
365                 }
366             }
367             else
368             {
369                 *p++ = aLine[nCol];
370             }
371             ++nLineIdx;
372             ++nCol;
373         }
374         *p = 0;
375         aSym = p; bNumber = true;
376 
377         // For bad characters, scan and parse errors generate only one error.
378         ErrCode nError = ERRCODE_NONE;
379         if (bScanError)
380         {
381             --nLineIdx;
382             --nCol;
383             aError = OUString( aLine[nCol]);
384             nError = ERRCODE_BASIC_BAD_CHAR_IN_NUMBER;
385         }
386 
387         rtl_math_ConversionStatus eStatus = rtl_math_ConversionStatus_Ok;
388         const sal_Unicode* pParseEnd = buf;
389         nVal = rtl_math_uStringToDouble( buf, buf+(p-buf), '.', ',', &eStatus, &pParseEnd );
390         if (pParseEnd != buf+(p-buf))
391         {
392             // e.g. "12e" or "12e+", or with bScanError "12d"+"E".
393             sal_Int32 nChars = buf+(p-buf) - pParseEnd;
394             nLineIdx -= nChars;
395             nCol -= nChars;
396             // For bScanError, nLineIdx and nCol were already decremented, just
397             // add that character to the parse end.
398             if (bScanError)
399                 ++nChars;
400             // Copy error position from original string, not the buffer
401             // replacement where "12dE" => "12EE".
402             aError = aLine.copy( nCol, nChars);
403             nError = ERRCODE_BASIC_BAD_CHAR_IN_NUMBER;
404         }
405         else if (eStatus != rtl_math_ConversionStatus_Ok)
406         {
407             // Keep the scan error and character at position, if any.
408             if (!nError)
409                 nError = ERRCODE_BASIC_MATH_OVERFLOW;
410         }
411 
412         if (nError)
413             GenError( nError );
414 
415         if( !dec && !exp )
416         {
417             if( nVal >= SbxMININT && nVal <= SbxMAXINT )
418                 eScanType = SbxINTEGER;
419             else if( nVal >= SbxMINLNG && nVal <= SbxMAXLNG )
420                     eScanType = SbxLONG;
421         }
422 
423         if( bBufOverflow )
424             GenError( ERRCODE_BASIC_MATH_OVERFLOW );
425 
426         // type recognition?
427         if( nCol < aLine.getLength() )
428         {
429             SbxDataType t(GetSuffixType(aLine[nCol]));
430             if( t != SbxVARIANT )
431             {
432                 eScanType = t;
433                 ++nLineIdx;
434                 ++nCol;
435             }
436        }
437     }
438 
439     // Hex/octal number? Read in and convert:
440     else if(aLine.getLength() - nCol > 1 && aLine[nCol] == '&')
441     {
442         ++nLineIdx; ++nCol;
443         sal_Unicode base = 16;
444         sal_Unicode xch  = aLine[nCol];
445         ++nLineIdx; ++nCol;
446         switch( rtl::toAsciiUpperCase( xch ) )
447         {
448             case 'O':
449                 base = 8;
450                 break;
451             case 'H':
452                 break;
453             default :
454                 // treated as an operator
455                 --nLineIdx; --nCol; nCol1 = nCol-1;
456                 aSym = "&";
457                 return true;
458         }
459         bNumber = true;
460         // Hex literals are signed Integers ( as defined by basic
461         // e.g. -2,147,483,648 through 2,147,483,647 (signed)
462         sal_uInt64 lu = 0;
463         bool bOverflow = false;
464         while(nCol < aLine.getLength() && BasicCharClass::isAlphaNumeric(aLine[nCol], false))
465         {
466             sal_Unicode ch = rtl::toAsciiUpperCase(aLine[nCol]);
467             ++nLineIdx; ++nCol;
468             if( ((base == 16 ) && rtl::isAsciiHexDigit( ch ) ) ||
469                      ((base == 8) && rtl::isAsciiOctalDigit( ch )))
470             {
471                 int i = ch  - '0';
472                 if( i > 9 ) i -= 7;
473                 lu = ( lu * base ) + i;
474                 if( lu > SAL_MAX_UINT32 )
475                 {
476                     bOverflow = true;
477                 }
478             }
479             else
480             {
481                 aError = OUString(ch);
482                 GenError( ERRCODE_BASIC_BAD_CHAR_IN_NUMBER );
483             }
484         }
485         if(nCol < aLine.getLength() && aLine[nCol] == '&')
486         {
487             ++nLineIdx;
488             ++nCol;
489         }
490         // tdf#62326 - If the value of the hex string lies within the range of 0x8000 (SbxMAXINT + 1)
491         // and 0xFFFF (SbxMAXUINT) inclusive, cast the value to 16 bit in order to get
492         // signed integers, e.g., SbxMININT through SbxMAXINT
493         sal_Int32 ls = (lu > SbxMAXINT && lu <= SbxMAXUINT) ? static_cast<sal_Int16>(lu) : static_cast<sal_Int32>(lu);
494         nVal = static_cast<double>(ls);
495         eScanType = ( ls >= SbxMININT && ls <= SbxMAXINT ) ? SbxINTEGER : SbxLONG;
496         if( bOverflow )
497             GenError( ERRCODE_BASIC_MATH_OVERFLOW );
498     }
499 
500     // Strings:
501     else if (nLineIdx < aLine.getLength() && (aLine[nLineIdx] == '"' || aLine[nLineIdx] == '['))
502     {
503         sal_Unicode cSep = aLine[nLineIdx];
504         if( cSep == '[' )
505         {
506             bSymbol = true;
507             cSep = ']';
508         }
509         sal_Int32 n = nCol + 1;
510         while (nLineIdx < aLine.getLength())
511         {
512             do
513             {
514                 nLineIdx++;
515                 nCol++;
516             }
517             while (nLineIdx < aLine.getLength() && (aLine[nLineIdx] != cSep));
518             if (nLineIdx < aLine.getLength() && aLine[nLineIdx] == cSep)
519             {
520                 nLineIdx++; nCol++;
521                 if (nLineIdx >= aLine.getLength() || aLine[nLineIdx] != cSep || cSep == ']')
522                 {
523                     // If VBA Interop then doesn't eat the [] chars
524                     if ( cSep == ']' && bVBASupportOn )
525                         aSym = aLine.copy( n - 1, nCol - n  + 1);
526                     else
527                         aSym = aLine.copy( n, nCol - n - 1 );
528                     // get out duplicate string delimiters
529                     OUStringBuffer aSymBuf(aSym.getLength());
530                     for ( sal_Int32 i = 0, len = aSym.getLength(); i < len; ++i )
531                     {
532                         aSymBuf.append( aSym[i] );
533                         if ( aSym[i] == cSep && ( i+1 < len ) && aSym[i+1] == cSep )
534                             ++i;
535                     }
536                     aSym = aSymBuf.makeStringAndClear();
537                     if( cSep != ']' )
538                         eScanType = SbxSTRING;
539                     break;
540                 }
541             }
542             else
543             {
544                 aError = OUString(cSep);
545                 GenError( ERRCODE_BASIC_EXPECTED );
546             }
547         }
548     }
549 
550     // Date:
551     else if (nLineIdx < aLine.getLength() && aLine[nLineIdx] == '#')
552     {
553         sal_Int32 n = nCol + 1;
554         do
555         {
556             nLineIdx++;
557             nCol++;
558         }
559         while (nLineIdx < aLine.getLength() && (aLine[nLineIdx] != '#'));
560         if (nLineIdx < aLine.getLength() && aLine[nLineIdx] == '#')
561         {
562             nLineIdx++; nCol++;
563             aSym = aLine.copy( n, nCol - n - 1 );
564 
565             // parse date literal
566             std::shared_ptr<SvNumberFormatter> pFormatter;
567             if (GetSbData()->pInst)
568             {
569                 pFormatter = GetSbData()->pInst->GetNumberFormatter();
570             }
571             else
572             {
573                 sal_uInt32 nDummy;
574                 pFormatter = SbiInstance::PrepareNumberFormatter( nDummy, nDummy, nDummy );
575             }
576             sal_uInt32 nIndex = pFormatter->GetStandardIndex( LANGUAGE_ENGLISH_US);
577             bool bSuccess = pFormatter->IsNumberFormat(aSym, nIndex, nVal);
578             if( bSuccess )
579             {
580                 SvNumFormatType nType_ = pFormatter->GetType(nIndex);
581                 if( !(nType_ & SvNumFormatType::DATE) )
582                     bSuccess = false;
583             }
584 
585             if (!bSuccess)
586                 GenError( ERRCODE_BASIC_CONVERSION );
587 
588             bNumber = true;
589             eScanType = SbxDOUBLE;
590         }
591         else
592         {
593             aError = OUString('#');
594             GenError( ERRCODE_BASIC_EXPECTED );
595         }
596     }
597     // invalid characters:
598     else if (nLineIdx < aLine.getLength() && aLine[nLineIdx] >= 0x7F)
599     {
600         GenError( ERRCODE_BASIC_SYNTAX ); nLineIdx++; nCol++;
601     }
602     // other groups:
603     else
604     {
605         sal_Int32 n = 1;
606         auto nChar = nLineIdx < aLine.getLength() ? aLine[nLineIdx] : 0;
607         ++nLineIdx;
608         if (nLineIdx < aLine.getLength())
609         {
610             switch (nChar)
611             {
612                 case '<': if( aLine[nLineIdx] == '>' || aLine[nLineIdx] == '=' ) n = 2; break;
613                 case '>': if( aLine[nLineIdx] == '=' ) n = 2; break;
614                 case ':': if( aLine[nLineIdx] == '=' ) n = 2; break;
615             }
616         }
617         aSym = aLine.copy(nCol, std::min(n, aLine.getLength() - nCol));
618         nLineIdx += n-1; nCol = nCol + n;
619     }
620 
621     nCol2 = nCol-1;
622 
623 PrevLineCommentLbl:
624 
625     if( bPrevLineExtentsComment || (eScanType != SbxSTRING &&
626                                     ( bCompilerDirective ||
627                                       aSym.startsWith("'") ||
628                                       aSym.equalsIgnoreAsciiCase( "REM" ) ) ) )
629     {
630         bPrevLineExtentsComment = false;
631         aSym = "REM";
632         sal_Int32 nLen = aLine.getLength() - nLineIdx;
633         if( bCompatible && aLine[nLineIdx + nLen - 1] == '_' && aLine[nLineIdx + nLen - 2] == ' ' )
634             bPrevLineExtentsComment = true;
635         nCol2 = nCol2 + nLen;
636         nLineIdx = -1;
637     }
638 
639     if (nLineIdx == nLineIdxScanStart)
640     {
641         GenError( ERRCODE_BASIC_SYMBOL_EXPECTED );
642         return false;
643     }
644 
645     return true;
646 
647 
648 eoln:
649     if( nCol && aLine[--nLineIdx] == '_' )
650     {
651         nLineIdx = -1;
652         bool bRes = NextSym();
653         if( aSym.startsWith(".") )
654         {
655             // object _
656             //    .Method
657             // ^^^  <- spaces is legal in MSO VBA
658             bSpaces = false;
659         }
660         return bRes;
661     }
662     else
663     {
664         nLineIdx = -1;
665         nLine = nOldLine;
666         nCol1 = nOldCol1;
667         nCol2 = nOldCol2;
668         aSym = "\n";
669         nColLock = 0;
670         return true;
671     }
672 }
673 
674 /* vim:set shiftwidth=4 softtabstop=4 expandtab: */
675