don't use wxScopedPtr<> in wxDocTemplate::CreateDocument() as the document is implici...
[wxWidgets.git] / src / common / string.cpp
1 /////////////////////////////////////////////////////////////////////////////
2 // Name: src/common/string.cpp
3 // Purpose: wxString class
4 // Author: Vadim Zeitlin, Ryan Norton
5 // Modified by:
6 // Created: 29/01/98
7 // RCS-ID: $Id$
8 // Copyright: (c) 1998 Vadim Zeitlin <zeitlin@dptmaths.ens-cachan.fr>
9 // (c) 2004 Ryan Norton <wxprojects@comcast.net>
10 // Licence: wxWindows licence
11 /////////////////////////////////////////////////////////////////////////////
12
13 // ===========================================================================
14 // headers, declarations, constants
15 // ===========================================================================
16
17 // For compilers that support precompilation, includes "wx.h".
18 #include "wx/wxprec.h"
19
20 #ifdef __BORLANDC__
21 #pragma hdrstop
22 #endif
23
24 #ifndef WX_PRECOMP
25 #include "wx/string.h"
26 #include "wx/wxcrtvararg.h"
27 #endif
28
29 #include <ctype.h>
30
31 #ifndef __WXWINCE__
32 #include <errno.h>
33 #endif
34
35 #include <string.h>
36 #include <stdlib.h>
37
38 #include "wx/hashmap.h"
39 #include "wx/vector.h"
40
41 // string handling functions used by wxString:
42 #if wxUSE_UNICODE_UTF8
43 #define wxStringMemcpy memcpy
44 #define wxStringMemcmp memcmp
45 #define wxStringMemchr memchr
46 #define wxStringStrlen strlen
47 #else
48 #define wxStringMemcpy wxTmemcpy
49 #define wxStringMemcmp wxTmemcmp
50 #define wxStringMemchr wxTmemchr
51 #define wxStringStrlen wxStrlen
52 #endif
53
54 // ----------------------------------------------------------------------------
55 // global variables
56 // ----------------------------------------------------------------------------
57
58 namespace wxPrivate
59 {
60
61 static UntypedBufferData s_untypedNullData(NULL);
62
63 UntypedBufferData * const untypedNullDataPtr = &s_untypedNullData;
64
65 } // namespace wxPrivate
66
67 // ---------------------------------------------------------------------------
68 // static class variables definition
69 // ---------------------------------------------------------------------------
70
71 //According to STL _must_ be a -1 size_t
72 const size_t wxString::npos = (size_t) -1;
73
74 #if wxUSE_STRING_POS_CACHE
75
76 #ifdef wxHAS_COMPILER_TLS
77
78 wxTLS_TYPE(wxString::Cache) wxString::ms_cache;
79
80 #else // !wxHAS_COMPILER_TLS
81
82 struct wxStrCacheInitializer
83 {
84 wxStrCacheInitializer()
85 {
86 // calling this function triggers s_cache initialization in it, and
87 // from now on it becomes safe to call from multiple threads
88 wxString::GetCache();
89 }
90 };
91
92 /*
93 wxString::Cache& wxString::GetCache()
94 {
95 static wxTLS_TYPE(Cache) s_cache;
96
97 return wxTLS_VALUE(s_cache);
98 }
99 */
100
101 static wxStrCacheInitializer gs_stringCacheInit;
102
103 #endif // wxHAS_COMPILER_TLS/!wxHAS_COMPILER_TLS
104
105 // gdb seems to be unable to display thread-local variables correctly, at least
106 // not my 6.4.98 version under amd64, so provide this debugging helper to do it
107 #ifdef __WXDEBUG__
108
109 struct wxStrCacheDumper
110 {
111 static void ShowAll()
112 {
113 puts("*** wxString cache dump:");
114 for ( unsigned n = 0; n < wxString::Cache::SIZE; n++ )
115 {
116 const wxString::Cache::Element&
117 c = wxString::GetCacheBegin()[n];
118
119 printf("\t%u%s\t%p: pos=(%lu, %lu), len=%ld\n",
120 n,
121 n == wxString::LastUsedCacheElement() ? " [*]" : "",
122 c.str,
123 (unsigned long)c.pos,
124 (unsigned long)c.impl,
125 (long)c.len);
126 }
127 }
128 };
129
130 void wxDumpStrCache() { wxStrCacheDumper::ShowAll(); }
131
132 #endif // __WXDEBUG__
133
134 #ifdef wxPROFILE_STRING_CACHE
135
136 wxString::CacheStats wxString::ms_cacheStats;
137
138 struct wxStrCacheStatsDumper
139 {
140 ~wxStrCacheStatsDumper()
141 {
142 const wxString::CacheStats& stats = wxString::ms_cacheStats;
143
144 if ( stats.postot )
145 {
146 puts("*** wxString cache statistics:");
147 printf("\tTotal non-trivial calls to PosToImpl(): %u\n",
148 stats.postot);
149 printf("\tHits %u (of which %u not used) or %.2f%%\n",
150 stats.poshits,
151 stats.mishits,
152 100.*float(stats.poshits - stats.mishits)/stats.postot);
153 printf("\tAverage position requested: %.2f\n",
154 float(stats.sumpos) / stats.postot);
155 printf("\tAverage offset after cached hint: %.2f\n",
156 float(stats.sumofs) / stats.postot);
157 }
158
159 if ( stats.lentot )
160 {
161 printf("\tNumber of calls to length(): %u, hits=%.2f%%\n",
162 stats.lentot, 100.*float(stats.lenhits)/stats.lentot);
163 }
164 }
165 };
166
167 static wxStrCacheStatsDumper s_showCacheStats;
168
169 #endif // wxPROFILE_STRING_CACHE
170
171 #endif // wxUSE_STRING_POS_CACHE
172
173 // ----------------------------------------------------------------------------
174 // global functions
175 // ----------------------------------------------------------------------------
176
177 #if wxUSE_STD_IOSTREAM
178
179 #include <iostream>
180
181 wxSTD ostream& operator<<(wxSTD ostream& os, const wxCStrData& str)
182 {
183 #if wxUSE_UNICODE && !wxUSE_UNICODE_UTF8
184 const wxCharBuffer buf(str.AsCharBuf());
185 if ( !buf )
186 os.clear(wxSTD ios_base::failbit);
187 else
188 os << buf.data();
189
190 return os;
191 #else
192 return os << str.AsInternal();
193 #endif
194 }
195
196 wxSTD ostream& operator<<(wxSTD ostream& os, const wxString& str)
197 {
198 return os << str.c_str();
199 }
200
201 wxSTD ostream& operator<<(wxSTD ostream& os, const wxCharBuffer& str)
202 {
203 return os << str.data();
204 }
205
206 #ifndef __BORLANDC__
207 wxSTD ostream& operator<<(wxSTD ostream& os, const wxWCharBuffer& str)
208 {
209 return os << str.data();
210 }
211 #endif
212
213 #if wxUSE_UNICODE && defined(HAVE_WOSTREAM)
214
215 wxSTD wostream& operator<<(wxSTD wostream& wos, const wxString& str)
216 {
217 return wos << str.wc_str();
218 }
219
220 wxSTD wostream& operator<<(wxSTD wostream& wos, const wxCStrData& str)
221 {
222 return wos << str.AsWChar();
223 }
224
225 wxSTD wostream& operator<<(wxSTD wostream& wos, const wxWCharBuffer& str)
226 {
227 return wos << str.data();
228 }
229
230 #endif // wxUSE_UNICODE && defined(HAVE_WOSTREAM)
231
232 #endif // wxUSE_STD_IOSTREAM
233
234 // ===========================================================================
235 // wxString class core
236 // ===========================================================================
237
238 #if wxUSE_UNICODE_UTF8
239
240 void wxString::PosLenToImpl(size_t pos, size_t len,
241 size_t *implPos, size_t *implLen) const
242 {
243 if ( pos == npos )
244 {
245 *implPos = npos;
246 }
247 else // have valid start position
248 {
249 const const_iterator b = GetIterForNthChar(pos);
250 *implPos = wxStringImpl::const_iterator(b.impl()) - m_impl.begin();
251 if ( len == npos )
252 {
253 *implLen = npos;
254 }
255 else // have valid length too
256 {
257 // we need to handle the case of length specifying a substring
258 // going beyond the end of the string, just as std::string does
259 const const_iterator e(end());
260 const_iterator i(b);
261 while ( len && i <= e )
262 {
263 ++i;
264 --len;
265 }
266
267 *implLen = i.impl() - b.impl();
268 }
269 }
270 }
271
272 #endif // wxUSE_UNICODE_UTF8
273
274 // ----------------------------------------------------------------------------
275 // wxCStrData converted strings caching
276 // ----------------------------------------------------------------------------
277
278 // FIXME-UTF8: temporarily disabled because it doesn't work with global
279 // string objects; re-enable after fixing this bug and benchmarking
280 // performance to see if using a hash is a good idea at all
281 #if 0
282
283 // For backward compatibility reasons, it must be possible to assign the value
284 // returned by wxString::c_str() to a char* or wchar_t* variable and work with
285 // it. Returning wxCharBuffer from (const char*)c_str() wouldn't do the trick,
286 // because the memory would be freed immediately, but it has to be valid as long
287 // as the string is not modified, so that code like this still works:
288 //
289 // const wxChar *s = str.c_str();
290 // while ( s ) { ... }
291
292 // FIXME-UTF8: not thread safe!
293 // FIXME-UTF8: we currently clear the cached conversion only when the string is
294 // destroyed, but we should do it when the string is modified, to
295 // keep memory usage down
296 // FIXME-UTF8: we do the conversion every time As[W]Char() is called, but if we
297 // invalidated the cache on every change, we could keep the previous
298 // conversion
299 // FIXME-UTF8: add tracing of usage of these two methods - new code is supposed
300 // to use mb_str() or wc_str() instead of (const [w]char*)c_str()
301
302 template<typename T>
303 static inline void DeleteStringFromConversionCache(T& hash, const wxString *s)
304 {
305 typename T::iterator i = hash.find(wxConstCast(s, wxString));
306 if ( i != hash.end() )
307 {
308 free(i->second);
309 hash.erase(i);
310 }
311 }
312
313 #if wxUSE_UNICODE
314 // NB: non-STL implementation doesn't compile with "const wxString*" key type,
315 // so we have to use wxString* here and const-cast when used
316 WX_DECLARE_HASH_MAP(wxString*, char*, wxPointerHash, wxPointerEqual,
317 wxStringCharConversionCache);
318 static wxStringCharConversionCache gs_stringsCharCache;
319
320 const char* wxCStrData::AsChar() const
321 {
322 // remove previously cache value, if any (see FIXMEs above):
323 DeleteStringFromConversionCache(gs_stringsCharCache, m_str);
324
325 // convert the string and keep it:
326 const char *s = gs_stringsCharCache[wxConstCast(m_str, wxString)] =
327 m_str->mb_str().release();
328
329 return s + m_offset;
330 }
331 #endif // wxUSE_UNICODE
332
333 #if !wxUSE_UNICODE_WCHAR
334 WX_DECLARE_HASH_MAP(wxString*, wchar_t*, wxPointerHash, wxPointerEqual,
335 wxStringWCharConversionCache);
336 static wxStringWCharConversionCache gs_stringsWCharCache;
337
338 const wchar_t* wxCStrData::AsWChar() const
339 {
340 // remove previously cache value, if any (see FIXMEs above):
341 DeleteStringFromConversionCache(gs_stringsWCharCache, m_str);
342
343 // convert the string and keep it:
344 const wchar_t *s = gs_stringsWCharCache[wxConstCast(m_str, wxString)] =
345 m_str->wc_str().release();
346
347 return s + m_offset;
348 }
349 #endif // !wxUSE_UNICODE_WCHAR
350
351 wxString::~wxString()
352 {
353 #if wxUSE_UNICODE
354 // FIXME-UTF8: do this only if locale is not UTF8 if wxUSE_UNICODE_UTF8
355 DeleteStringFromConversionCache(gs_stringsCharCache, this);
356 #endif
357 #if !wxUSE_UNICODE_WCHAR
358 DeleteStringFromConversionCache(gs_stringsWCharCache, this);
359 #endif
360 }
361 #endif
362
363 #if wxUSE_UNICODE && !wxUSE_UTF8_LOCALE_ONLY
364 const char* wxCStrData::AsChar() const
365 {
366 #if wxUSE_UNICODE_UTF8
367 if ( wxLocaleIsUtf8 )
368 return AsInternal();
369 #endif
370 // under non-UTF8 locales, we have to convert the internal UTF-8
371 // representation using wxConvLibc and cache the result
372
373 wxString *str = wxConstCast(m_str, wxString);
374
375 // convert the string:
376 //
377 // FIXME-UTF8: we'd like to do the conversion in the existing buffer (if we
378 // have it) but it's unfortunately not obvious to implement
379 // because we don't know how big buffer do we need for the
380 // given string length (in case of multibyte encodings, e.g.
381 // ISO-2022-JP or UTF-8 when internal representation is wchar_t)
382 //
383 // One idea would be to store more than just m_convertedToChar
384 // in wxString: then we could record the length of the string
385 // which was converted the last time and try to reuse the same
386 // buffer if the current length is not greater than it (this
387 // could still fail because string could have been modified in
388 // place but it would work most of the time, so we'd do it and
389 // only allocate the new buffer if in-place conversion returned
390 // an error). We could also store a bit saying if the string
391 // was modified since the last conversion (and update it in all
392 // operation modifying the string, of course) to avoid unneeded
393 // consequential conversions. But both of these ideas require
394 // adding more fields to wxString and require profiling results
395 // to be sure that we really gain enough from them to justify
396 // doing it.
397 wxCharBuffer buf(str->mb_str());
398
399 // if it failed, return empty string and not NULL to avoid crashes in code
400 // written with either wxWidgets 2 wxString or std::string behaviour in
401 // mind: neither of them ever returns NULL and so we shouldn't neither
402 if ( !buf )
403 return "";
404
405 if ( str->m_convertedToChar &&
406 strlen(buf) == strlen(str->m_convertedToChar) )
407 {
408 // keep the same buffer for as long as possible, so that several calls
409 // to c_str() in a row still work:
410 strcpy(str->m_convertedToChar, buf);
411 }
412 else
413 {
414 str->m_convertedToChar = buf.release();
415 }
416
417 // and keep it:
418 return str->m_convertedToChar + m_offset;
419 }
420 #endif // wxUSE_UNICODE && !wxUSE_UTF8_LOCALE_ONLY
421
422 #if !wxUSE_UNICODE_WCHAR
423 const wchar_t* wxCStrData::AsWChar() const
424 {
425 wxString *str = wxConstCast(m_str, wxString);
426
427 // convert the string:
428 wxWCharBuffer buf(str->wc_str());
429
430 // notice that here, unlike above in AsChar(), conversion can't fail as our
431 // internal UTF-8 is always well-formed -- or the string was corrupted and
432 // all bets are off anyhow
433
434 // FIXME-UTF8: do the conversion in-place in the existing buffer
435 if ( str->m_convertedToWChar &&
436 wxWcslen(buf) == wxWcslen(str->m_convertedToWChar) )
437 {
438 // keep the same buffer for as long as possible, so that several calls
439 // to c_str() in a row still work:
440 memcpy(str->m_convertedToWChar, buf, sizeof(wchar_t) * wxWcslen(buf));
441 }
442 else
443 {
444 str->m_convertedToWChar = buf.release();
445 }
446
447 // and keep it:
448 return str->m_convertedToWChar + m_offset;
449 }
450 #endif // !wxUSE_UNICODE_WCHAR
451
452 // ===========================================================================
453 // wxString class core
454 // ===========================================================================
455
456 // ---------------------------------------------------------------------------
457 // construction and conversion
458 // ---------------------------------------------------------------------------
459
460 #if wxUSE_UNICODE_WCHAR
461 /* static */
462 wxString::SubstrBufFromMB wxString::ConvertStr(const char *psz, size_t nLength,
463 const wxMBConv& conv)
464 {
465 // anything to do?
466 if ( !psz || nLength == 0 )
467 return SubstrBufFromMB(L"", 0);
468
469 if ( nLength == npos )
470 nLength = wxNO_LEN;
471
472 size_t wcLen;
473 wxWCharBuffer wcBuf(conv.cMB2WC(psz, nLength, &wcLen));
474 if ( !wcLen )
475 return SubstrBufFromMB(_T(""), 0);
476 else
477 return SubstrBufFromMB(wcBuf, wcLen);
478 }
479 #endif // wxUSE_UNICODE_WCHAR
480
481 #if wxUSE_UNICODE_UTF8
482 /* static */
483 wxString::SubstrBufFromMB wxString::ConvertStr(const char *psz, size_t nLength,
484 const wxMBConv& conv)
485 {
486 // anything to do?
487 if ( !psz || nLength == 0 )
488 return SubstrBufFromMB("", 0);
489
490 // if psz is already in UTF-8, we don't have to do the roundtrip to
491 // wchar_t* and back:
492 if ( conv.IsUTF8() )
493 {
494 // we need to validate the input because UTF8 iterators assume valid
495 // UTF-8 sequence and psz may be invalid:
496 if ( wxStringOperations::IsValidUtf8String(psz, nLength) )
497 {
498 // we must pass the real string length to SubstrBufFromMB ctor
499 if ( nLength == npos )
500 nLength = psz ? strlen(psz) : 0;
501 return SubstrBufFromMB(wxCharBuffer::CreateNonOwned(psz), nLength);
502 }
503 // else: do the roundtrip through wchar_t*
504 }
505
506 if ( nLength == npos )
507 nLength = wxNO_LEN;
508
509 // first convert to wide string:
510 size_t wcLen;
511 wxWCharBuffer wcBuf(conv.cMB2WC(psz, nLength, &wcLen));
512 if ( !wcLen )
513 return SubstrBufFromMB("", 0);
514
515 // and then to UTF-8:
516 SubstrBufFromMB buf(ConvertStr(wcBuf, wcLen, wxMBConvStrictUTF8()));
517 // widechar -> UTF-8 conversion isn't supposed to ever fail:
518 wxASSERT_MSG( buf.data, _T("conversion to UTF-8 failed") );
519
520 return buf;
521 }
522 #endif // wxUSE_UNICODE_UTF8
523
524 #if wxUSE_UNICODE_UTF8 || !wxUSE_UNICODE
525 /* static */
526 wxString::SubstrBufFromWC wxString::ConvertStr(const wchar_t *pwz, size_t nLength,
527 const wxMBConv& conv)
528 {
529 // anything to do?
530 if ( !pwz || nLength == 0 )
531 return SubstrBufFromWC("", 0);
532
533 if ( nLength == npos )
534 nLength = wxNO_LEN;
535
536 size_t mbLen;
537 wxCharBuffer mbBuf(conv.cWC2MB(pwz, nLength, &mbLen));
538 if ( !mbLen )
539 return SubstrBufFromWC("", 0);
540 else
541 return SubstrBufFromWC(mbBuf, mbLen);
542 }
543 #endif // wxUSE_UNICODE_UTF8 || !wxUSE_UNICODE
544
545
546 #if wxUSE_UNICODE_WCHAR
547
548 //Convert wxString in Unicode mode to a multi-byte string
549 const wxCharBuffer wxString::mb_str(const wxMBConv& conv) const
550 {
551 return conv.cWC2MB(wx_str(), length() + 1 /* size, not length */, NULL);
552 }
553
554 #elif wxUSE_UNICODE_UTF8
555
556 const wxWCharBuffer wxString::wc_str() const
557 {
558 return wxMBConvStrictUTF8().cMB2WC
559 (
560 m_impl.c_str(),
561 m_impl.length() + 1, // size, not length
562 NULL
563 );
564 }
565
566 const wxCharBuffer wxString::mb_str(const wxMBConv& conv) const
567 {
568 if ( conv.IsUTF8() )
569 return wxCharBuffer::CreateNonOwned(m_impl.c_str());
570
571 // FIXME-UTF8: use wc_str() here once we have buffers with length
572
573 size_t wcLen;
574 wxWCharBuffer wcBuf(wxMBConvStrictUTF8().cMB2WC
575 (
576 m_impl.c_str(),
577 m_impl.length() + 1, // size
578 &wcLen
579 ));
580 if ( !wcLen )
581 return wxCharBuffer("");
582
583 return conv.cWC2MB(wcBuf, wcLen+1, NULL);
584 }
585
586 #else // ANSI
587
588 //Converts this string to a wide character string if unicode
589 //mode is not enabled and wxUSE_WCHAR_T is enabled
590 const wxWCharBuffer wxString::wc_str(const wxMBConv& conv) const
591 {
592 return conv.cMB2WC(wx_str(), length() + 1 /* size, not length */, NULL);
593 }
594
595 #endif // Unicode/ANSI
596
597 // shrink to minimal size (releasing extra memory)
598 bool wxString::Shrink()
599 {
600 wxString tmp(begin(), end());
601 swap(tmp);
602 return tmp.length() == length();
603 }
604
605 // deprecated compatibility code:
606 #if WXWIN_COMPATIBILITY_2_8 && !wxUSE_STL_BASED_WXSTRING && !wxUSE_UNICODE_UTF8
607 wxStringCharType *wxString::GetWriteBuf(size_t nLen)
608 {
609 return DoGetWriteBuf(nLen);
610 }
611
612 void wxString::UngetWriteBuf()
613 {
614 DoUngetWriteBuf();
615 }
616
617 void wxString::UngetWriteBuf(size_t nLen)
618 {
619 DoUngetWriteBuf(nLen);
620 }
621 #endif // WXWIN_COMPATIBILITY_2_8 && !wxUSE_STL_BASED_WXSTRING && !wxUSE_UNICODE_UTF8
622
623
624 // ---------------------------------------------------------------------------
625 // data access
626 // ---------------------------------------------------------------------------
627
628 // all functions are inline in string.h
629
630 // ---------------------------------------------------------------------------
631 // concatenation operators
632 // ---------------------------------------------------------------------------
633
634 /*
635 * concatenation functions come in 5 flavours:
636 * string + string
637 * char + string and string + char
638 * C str + string and string + C str
639 */
640
641 wxString operator+(const wxString& str1, const wxString& str2)
642 {
643 #if !wxUSE_STL_BASED_WXSTRING
644 wxASSERT( str1.IsValid() );
645 wxASSERT( str2.IsValid() );
646 #endif
647
648 wxString s = str1;
649 s += str2;
650
651 return s;
652 }
653
654 wxString operator+(const wxString& str, wxUniChar ch)
655 {
656 #if !wxUSE_STL_BASED_WXSTRING
657 wxASSERT( str.IsValid() );
658 #endif
659
660 wxString s = str;
661 s += ch;
662
663 return s;
664 }
665
666 wxString operator+(wxUniChar ch, const wxString& str)
667 {
668 #if !wxUSE_STL_BASED_WXSTRING
669 wxASSERT( str.IsValid() );
670 #endif
671
672 wxString s = ch;
673 s += str;
674
675 return s;
676 }
677
678 wxString operator+(const wxString& str, const char *psz)
679 {
680 #if !wxUSE_STL_BASED_WXSTRING
681 wxASSERT( str.IsValid() );
682 #endif
683
684 wxString s;
685 if ( !s.Alloc(strlen(psz) + str.length()) ) {
686 wxFAIL_MSG( _T("out of memory in wxString::operator+") );
687 }
688 s += str;
689 s += psz;
690
691 return s;
692 }
693
694 wxString operator+(const wxString& str, const wchar_t *pwz)
695 {
696 #if !wxUSE_STL_BASED_WXSTRING
697 wxASSERT( str.IsValid() );
698 #endif
699
700 wxString s;
701 if ( !s.Alloc(wxWcslen(pwz) + str.length()) ) {
702 wxFAIL_MSG( _T("out of memory in wxString::operator+") );
703 }
704 s += str;
705 s += pwz;
706
707 return s;
708 }
709
710 wxString operator+(const char *psz, const wxString& str)
711 {
712 #if !wxUSE_STL_BASED_WXSTRING
713 wxASSERT( str.IsValid() );
714 #endif
715
716 wxString s;
717 if ( !s.Alloc(strlen(psz) + str.length()) ) {
718 wxFAIL_MSG( _T("out of memory in wxString::operator+") );
719 }
720 s = psz;
721 s += str;
722
723 return s;
724 }
725
726 wxString operator+(const wchar_t *pwz, const wxString& str)
727 {
728 #if !wxUSE_STL_BASED_WXSTRING
729 wxASSERT( str.IsValid() );
730 #endif
731
732 wxString s;
733 if ( !s.Alloc(wxWcslen(pwz) + str.length()) ) {
734 wxFAIL_MSG( _T("out of memory in wxString::operator+") );
735 }
736 s = pwz;
737 s += str;
738
739 return s;
740 }
741
742 // ---------------------------------------------------------------------------
743 // string comparison
744 // ---------------------------------------------------------------------------
745
746 bool wxString::IsSameAs(wxUniChar c, bool compareWithCase) const
747 {
748 return (length() == 1) && (compareWithCase ? GetChar(0u) == c
749 : wxToupper(GetChar(0u)) == wxToupper(c));
750 }
751
752 #ifdef HAVE_STD_STRING_COMPARE
753
754 // NB: Comparison code (both if HAVE_STD_STRING_COMPARE and if not) works with
755 // UTF-8 encoded strings too, thanks to UTF-8's design which allows us to
756 // sort strings in characters code point order by sorting the byte sequence
757 // in byte values order (i.e. what strcmp() and memcmp() do).
758
759 int wxString::compare(const wxString& str) const
760 {
761 return m_impl.compare(str.m_impl);
762 }
763
764 int wxString::compare(size_t nStart, size_t nLen,
765 const wxString& str) const
766 {
767 size_t pos, len;
768 PosLenToImpl(nStart, nLen, &pos, &len);
769 return m_impl.compare(pos, len, str.m_impl);
770 }
771
772 int wxString::compare(size_t nStart, size_t nLen,
773 const wxString& str,
774 size_t nStart2, size_t nLen2) const
775 {
776 size_t pos, len;
777 PosLenToImpl(nStart, nLen, &pos, &len);
778
779 size_t pos2, len2;
780 str.PosLenToImpl(nStart2, nLen2, &pos2, &len2);
781
782 return m_impl.compare(pos, len, str.m_impl, pos2, len2);
783 }
784
785 int wxString::compare(const char* sz) const
786 {
787 return m_impl.compare(ImplStr(sz));
788 }
789
790 int wxString::compare(const wchar_t* sz) const
791 {
792 return m_impl.compare(ImplStr(sz));
793 }
794
795 int wxString::compare(size_t nStart, size_t nLen,
796 const char* sz, size_t nCount) const
797 {
798 size_t pos, len;
799 PosLenToImpl(nStart, nLen, &pos, &len);
800
801 SubstrBufFromMB str(ImplStr(sz, nCount));
802
803 return m_impl.compare(pos, len, str.data, str.len);
804 }
805
806 int wxString::compare(size_t nStart, size_t nLen,
807 const wchar_t* sz, size_t nCount) const
808 {
809 size_t pos, len;
810 PosLenToImpl(nStart, nLen, &pos, &len);
811
812 SubstrBufFromWC str(ImplStr(sz, nCount));
813
814 return m_impl.compare(pos, len, str.data, str.len);
815 }
816
817 #else // !HAVE_STD_STRING_COMPARE
818
819 static inline int wxDoCmp(const wxStringCharType* s1, size_t l1,
820 const wxStringCharType* s2, size_t l2)
821 {
822 if( l1 == l2 )
823 return wxStringMemcmp(s1, s2, l1);
824 else if( l1 < l2 )
825 {
826 int ret = wxStringMemcmp(s1, s2, l1);
827 return ret == 0 ? -1 : ret;
828 }
829 else
830 {
831 int ret = wxStringMemcmp(s1, s2, l2);
832 return ret == 0 ? +1 : ret;
833 }
834 }
835
836 int wxString::compare(const wxString& str) const
837 {
838 return ::wxDoCmp(m_impl.data(), m_impl.length(),
839 str.m_impl.data(), str.m_impl.length());
840 }
841
842 int wxString::compare(size_t nStart, size_t nLen,
843 const wxString& str) const
844 {
845 wxASSERT(nStart <= length());
846 size_type strLen = length() - nStart;
847 nLen = strLen < nLen ? strLen : nLen;
848
849 size_t pos, len;
850 PosLenToImpl(nStart, nLen, &pos, &len);
851
852 return ::wxDoCmp(m_impl.data() + pos, len,
853 str.m_impl.data(), str.m_impl.length());
854 }
855
856 int wxString::compare(size_t nStart, size_t nLen,
857 const wxString& str,
858 size_t nStart2, size_t nLen2) const
859 {
860 wxASSERT(nStart <= length());
861 wxASSERT(nStart2 <= str.length());
862 size_type strLen = length() - nStart,
863 strLen2 = str.length() - nStart2;
864 nLen = strLen < nLen ? strLen : nLen;
865 nLen2 = strLen2 < nLen2 ? strLen2 : nLen2;
866
867 size_t pos, len;
868 PosLenToImpl(nStart, nLen, &pos, &len);
869 size_t pos2, len2;
870 str.PosLenToImpl(nStart2, nLen2, &pos2, &len2);
871
872 return ::wxDoCmp(m_impl.data() + pos, len,
873 str.m_impl.data() + pos2, len2);
874 }
875
876 int wxString::compare(const char* sz) const
877 {
878 SubstrBufFromMB str(ImplStr(sz, npos));
879 if ( str.len == npos )
880 str.len = wxStringStrlen(str.data);
881 return ::wxDoCmp(m_impl.data(), m_impl.length(), str.data, str.len);
882 }
883
884 int wxString::compare(const wchar_t* sz) const
885 {
886 SubstrBufFromWC str(ImplStr(sz, npos));
887 if ( str.len == npos )
888 str.len = wxStringStrlen(str.data);
889 return ::wxDoCmp(m_impl.data(), m_impl.length(), str.data, str.len);
890 }
891
892 int wxString::compare(size_t nStart, size_t nLen,
893 const char* sz, size_t nCount) const
894 {
895 wxASSERT(nStart <= length());
896 size_type strLen = length() - nStart;
897 nLen = strLen < nLen ? strLen : nLen;
898
899 size_t pos, len;
900 PosLenToImpl(nStart, nLen, &pos, &len);
901
902 SubstrBufFromMB str(ImplStr(sz, nCount));
903 if ( str.len == npos )
904 str.len = wxStringStrlen(str.data);
905
906 return ::wxDoCmp(m_impl.data() + pos, len, str.data, str.len);
907 }
908
909 int wxString::compare(size_t nStart, size_t nLen,
910 const wchar_t* sz, size_t nCount) const
911 {
912 wxASSERT(nStart <= length());
913 size_type strLen = length() - nStart;
914 nLen = strLen < nLen ? strLen : nLen;
915
916 size_t pos, len;
917 PosLenToImpl(nStart, nLen, &pos, &len);
918
919 SubstrBufFromWC str(ImplStr(sz, nCount));
920 if ( str.len == npos )
921 str.len = wxStringStrlen(str.data);
922
923 return ::wxDoCmp(m_impl.data() + pos, len, str.data, str.len);
924 }
925
926 #endif // HAVE_STD_STRING_COMPARE/!HAVE_STD_STRING_COMPARE
927
928
929 // ---------------------------------------------------------------------------
930 // find_{first,last}_[not]_of functions
931 // ---------------------------------------------------------------------------
932
933 #if !wxUSE_STL_BASED_WXSTRING || wxUSE_UNICODE_UTF8
934
935 // NB: All these functions are implemented with the argument being wxChar*,
936 // i.e. widechar string in any Unicode build, even though native string
937 // representation is char* in the UTF-8 build. This is because we couldn't
938 // use memchr() to determine if a character is in a set encoded as UTF-8.
939
940 size_t wxString::find_first_of(const wxChar* sz, size_t nStart) const
941 {
942 return find_first_of(sz, nStart, wxStrlen(sz));
943 }
944
945 size_t wxString::find_first_not_of(const wxChar* sz, size_t nStart) const
946 {
947 return find_first_not_of(sz, nStart, wxStrlen(sz));
948 }
949
950 size_t wxString::find_first_of(const wxChar* sz, size_t nStart, size_t n) const
951 {
952 wxASSERT_MSG( nStart <= length(), _T("invalid index") );
953
954 size_t idx = nStart;
955 for ( const_iterator i = begin() + nStart; i != end(); ++idx, ++i )
956 {
957 if ( wxTmemchr(sz, *i, n) )
958 return idx;
959 }
960
961 return npos;
962 }
963
964 size_t wxString::find_first_not_of(const wxChar* sz, size_t nStart, size_t n) const
965 {
966 wxASSERT_MSG( nStart <= length(), _T("invalid index") );
967
968 size_t idx = nStart;
969 for ( const_iterator i = begin() + nStart; i != end(); ++idx, ++i )
970 {
971 if ( !wxTmemchr(sz, *i, n) )
972 return idx;
973 }
974
975 return npos;
976 }
977
978
979 size_t wxString::find_last_of(const wxChar* sz, size_t nStart) const
980 {
981 return find_last_of(sz, nStart, wxStrlen(sz));
982 }
983
984 size_t wxString::find_last_not_of(const wxChar* sz, size_t nStart) const
985 {
986 return find_last_not_of(sz, nStart, wxStrlen(sz));
987 }
988
989 size_t wxString::find_last_of(const wxChar* sz, size_t nStart, size_t n) const
990 {
991 size_t len = length();
992
993 if ( nStart == npos )
994 {
995 nStart = len - 1;
996 }
997 else
998 {
999 wxASSERT_MSG( nStart <= len, _T("invalid index") );
1000 }
1001
1002 size_t idx = nStart;
1003 for ( const_reverse_iterator i = rbegin() + (len - nStart - 1);
1004 i != rend(); --idx, ++i )
1005 {
1006 if ( wxTmemchr(sz, *i, n) )
1007 return idx;
1008 }
1009
1010 return npos;
1011 }
1012
1013 size_t wxString::find_last_not_of(const wxChar* sz, size_t nStart, size_t n) const
1014 {
1015 size_t len = length();
1016
1017 if ( nStart == npos )
1018 {
1019 nStart = len - 1;
1020 }
1021 else
1022 {
1023 wxASSERT_MSG( nStart <= len, _T("invalid index") );
1024 }
1025
1026 size_t idx = nStart;
1027 for ( const_reverse_iterator i = rbegin() + (len - nStart - 1);
1028 i != rend(); --idx, ++i )
1029 {
1030 if ( !wxTmemchr(sz, *i, n) )
1031 return idx;
1032 }
1033
1034 return npos;
1035 }
1036
1037 size_t wxString::find_first_not_of(wxUniChar ch, size_t nStart) const
1038 {
1039 wxASSERT_MSG( nStart <= length(), _T("invalid index") );
1040
1041 size_t idx = nStart;
1042 for ( const_iterator i = begin() + nStart; i != end(); ++idx, ++i )
1043 {
1044 if ( *i != ch )
1045 return idx;
1046 }
1047
1048 return npos;
1049 }
1050
1051 size_t wxString::find_last_not_of(wxUniChar ch, size_t nStart) const
1052 {
1053 size_t len = length();
1054
1055 if ( nStart == npos )
1056 {
1057 nStart = len - 1;
1058 }
1059 else
1060 {
1061 wxASSERT_MSG( nStart <= len, _T("invalid index") );
1062 }
1063
1064 size_t idx = nStart;
1065 for ( const_reverse_iterator i = rbegin() + (len - nStart - 1);
1066 i != rend(); --idx, ++i )
1067 {
1068 if ( *i != ch )
1069 return idx;
1070 }
1071
1072 return npos;
1073 }
1074
1075 // the functions above were implemented for wchar_t* arguments in Unicode
1076 // build and char* in ANSI build; below are implementations for the other
1077 // version:
1078 #if wxUSE_UNICODE
1079 #define wxOtherCharType char
1080 #define STRCONV (const wxChar*)wxConvLibc.cMB2WC
1081 #else
1082 #define wxOtherCharType wchar_t
1083 #define STRCONV (const wxChar*)wxConvLibc.cWC2MB
1084 #endif
1085
1086 size_t wxString::find_first_of(const wxOtherCharType* sz, size_t nStart) const
1087 { return find_first_of(STRCONV(sz), nStart); }
1088
1089 size_t wxString::find_first_of(const wxOtherCharType* sz, size_t nStart,
1090 size_t n) const
1091 { return find_first_of(STRCONV(sz, n, NULL), nStart, n); }
1092 size_t wxString::find_last_of(const wxOtherCharType* sz, size_t nStart) const
1093 { return find_last_of(STRCONV(sz), nStart); }
1094 size_t wxString::find_last_of(const wxOtherCharType* sz, size_t nStart,
1095 size_t n) const
1096 { return find_last_of(STRCONV(sz, n, NULL), nStart, n); }
1097 size_t wxString::find_first_not_of(const wxOtherCharType* sz, size_t nStart) const
1098 { return find_first_not_of(STRCONV(sz), nStart); }
1099 size_t wxString::find_first_not_of(const wxOtherCharType* sz, size_t nStart,
1100 size_t n) const
1101 { return find_first_not_of(STRCONV(sz, n, NULL), nStart, n); }
1102 size_t wxString::find_last_not_of(const wxOtherCharType* sz, size_t nStart) const
1103 { return find_last_not_of(STRCONV(sz), nStart); }
1104 size_t wxString::find_last_not_of(const wxOtherCharType* sz, size_t nStart,
1105 size_t n) const
1106 { return find_last_not_of(STRCONV(sz, n, NULL), nStart, n); }
1107
1108 #undef wxOtherCharType
1109 #undef STRCONV
1110
1111 #endif // !wxUSE_STL_BASED_WXSTRING || wxUSE_UNICODE_UTF8
1112
1113 // ===========================================================================
1114 // other common string functions
1115 // ===========================================================================
1116
1117 int wxString::CmpNoCase(const wxString& s) const
1118 {
1119 #if wxUSE_UNICODE_UTF8
1120 // FIXME-UTF8: use wxUniChar::ToLower/ToUpper once added
1121
1122 const_iterator i1 = begin();
1123 const_iterator end1 = end();
1124 const_iterator i2 = s.begin();
1125 const_iterator end2 = s.end();
1126
1127 for ( ; i1 != end1 && i2 != end2; ++i1, ++i2 )
1128 {
1129 wxUniChar lower1 = (wxChar)wxTolower(*i1);
1130 wxUniChar lower2 = (wxChar)wxTolower(*i2);
1131 if ( lower1 != lower2 )
1132 return lower1 < lower2 ? -1 : 1;
1133 }
1134
1135 size_t len1 = length();
1136 size_t len2 = s.length();
1137
1138 if ( len1 < len2 )
1139 return -1;
1140 else if ( len1 > len2 )
1141 return 1;
1142 return 0;
1143 #else // wxUSE_UNICODE_WCHAR or ANSI
1144 return wxStricmp(m_impl.c_str(), s.m_impl.c_str());
1145 #endif
1146 }
1147
1148
1149 #if wxUSE_UNICODE
1150
1151 #ifdef __MWERKS__
1152 #ifndef __SCHAR_MAX__
1153 #define __SCHAR_MAX__ 127
1154 #endif
1155 #endif
1156
1157 wxString wxString::FromAscii(const char *ascii, size_t len)
1158 {
1159 if (!ascii || len == 0)
1160 return wxEmptyString;
1161
1162 wxString res;
1163
1164 {
1165 wxStringInternalBuffer buf(res, len);
1166 wxStringCharType *dest = buf;
1167
1168 for ( ; len > 0; --len )
1169 {
1170 unsigned char c = (unsigned char)*ascii++;
1171 wxASSERT_MSG( c < 0x80,
1172 _T("Non-ASCII value passed to FromAscii().") );
1173
1174 *dest++ = (wchar_t)c;
1175 }
1176 }
1177
1178 return res;
1179 }
1180
1181 wxString wxString::FromAscii(const char *ascii)
1182 {
1183 return FromAscii(ascii, wxStrlen(ascii));
1184 }
1185
1186 wxString wxString::FromAscii(char ascii)
1187 {
1188 // What do we do with '\0' ?
1189
1190 unsigned char c = (unsigned char)ascii;
1191
1192 wxASSERT_MSG( c < 0x80, _T("Non-ASCII value passed to FromAscii().") );
1193
1194 // NB: the cast to wchar_t causes interpretation of 'ascii' as Latin1 value
1195 return wxString(wxUniChar((wchar_t)c));
1196 }
1197
1198 const wxCharBuffer wxString::ToAscii() const
1199 {
1200 // this will allocate enough space for the terminating NUL too
1201 wxCharBuffer buffer(length());
1202 char *dest = buffer.data();
1203
1204 for ( const_iterator i = begin(); i != end(); ++i )
1205 {
1206 wxUniChar c(*i);
1207 // FIXME-UTF8: unify substituted char ('_') with wxUniChar ('?')
1208 *dest++ = c.IsAscii() ? (char)c : '_';
1209
1210 // the output string can't have embedded NULs anyhow, so we can safely
1211 // stop at first of them even if we do have any
1212 if ( !c )
1213 break;
1214 }
1215
1216 return buffer;
1217 }
1218
1219 #endif // wxUSE_UNICODE
1220
1221 // extract string of length nCount starting at nFirst
1222 wxString wxString::Mid(size_t nFirst, size_t nCount) const
1223 {
1224 size_t nLen = length();
1225
1226 // default value of nCount is npos and means "till the end"
1227 if ( nCount == npos )
1228 {
1229 nCount = nLen - nFirst;
1230 }
1231
1232 // out-of-bounds requests return sensible things
1233 if ( nFirst + nCount > nLen )
1234 {
1235 nCount = nLen - nFirst;
1236 }
1237
1238 if ( nFirst > nLen )
1239 {
1240 // AllocCopy() will return empty string
1241 return wxEmptyString;
1242 }
1243
1244 wxString dest(*this, nFirst, nCount);
1245 if ( dest.length() != nCount )
1246 {
1247 wxFAIL_MSG( _T("out of memory in wxString::Mid") );
1248 }
1249
1250 return dest;
1251 }
1252
1253 // check that the string starts with prefix and return the rest of the string
1254 // in the provided pointer if it is not NULL, otherwise return false
1255 bool wxString::StartsWith(const wxString& prefix, wxString *rest) const
1256 {
1257 if ( compare(0, prefix.length(), prefix) != 0 )
1258 return false;
1259
1260 if ( rest )
1261 {
1262 // put the rest of the string into provided pointer
1263 rest->assign(*this, prefix.length(), npos);
1264 }
1265
1266 return true;
1267 }
1268
1269
1270 // check that the string ends with suffix and return the rest of it in the
1271 // provided pointer if it is not NULL, otherwise return false
1272 bool wxString::EndsWith(const wxString& suffix, wxString *rest) const
1273 {
1274 int start = length() - suffix.length();
1275
1276 if ( start < 0 || compare(start, npos, suffix) != 0 )
1277 return false;
1278
1279 if ( rest )
1280 {
1281 // put the rest of the string into provided pointer
1282 rest->assign(*this, 0, start);
1283 }
1284
1285 return true;
1286 }
1287
1288
1289 // extract nCount last (rightmost) characters
1290 wxString wxString::Right(size_t nCount) const
1291 {
1292 if ( nCount > length() )
1293 nCount = length();
1294
1295 wxString dest(*this, length() - nCount, nCount);
1296 if ( dest.length() != nCount ) {
1297 wxFAIL_MSG( _T("out of memory in wxString::Right") );
1298 }
1299 return dest;
1300 }
1301
1302 // get all characters after the last occurrence of ch
1303 // (returns the whole string if ch not found)
1304 wxString wxString::AfterLast(wxUniChar ch) const
1305 {
1306 wxString str;
1307 int iPos = Find(ch, true);
1308 if ( iPos == wxNOT_FOUND )
1309 str = *this;
1310 else
1311 str.assign(*this, iPos + 1, npos);
1312
1313 return str;
1314 }
1315
1316 // extract nCount first (leftmost) characters
1317 wxString wxString::Left(size_t nCount) const
1318 {
1319 if ( nCount > length() )
1320 nCount = length();
1321
1322 wxString dest(*this, 0, nCount);
1323 if ( dest.length() != nCount ) {
1324 wxFAIL_MSG( _T("out of memory in wxString::Left") );
1325 }
1326 return dest;
1327 }
1328
1329 // get all characters before the first occurrence of ch
1330 // (returns the whole string if ch not found)
1331 wxString wxString::BeforeFirst(wxUniChar ch) const
1332 {
1333 int iPos = Find(ch);
1334 if ( iPos == wxNOT_FOUND )
1335 iPos = length();
1336 return wxString(*this, 0, iPos);
1337 }
1338
1339 /// get all characters before the last occurrence of ch
1340 /// (returns empty string if ch not found)
1341 wxString wxString::BeforeLast(wxUniChar ch) const
1342 {
1343 wxString str;
1344 int iPos = Find(ch, true);
1345 if ( iPos != wxNOT_FOUND && iPos != 0 )
1346 str = wxString(c_str(), iPos);
1347
1348 return str;
1349 }
1350
1351 /// get all characters after the first occurrence of ch
1352 /// (returns empty string if ch not found)
1353 wxString wxString::AfterFirst(wxUniChar ch) const
1354 {
1355 wxString str;
1356 int iPos = Find(ch);
1357 if ( iPos != wxNOT_FOUND )
1358 str.assign(*this, iPos + 1, npos);
1359
1360 return str;
1361 }
1362
1363 // replace first (or all) occurrences of some substring with another one
1364 size_t wxString::Replace(const wxString& strOld,
1365 const wxString& strNew, bool bReplaceAll)
1366 {
1367 // if we tried to replace an empty string we'd enter an infinite loop below
1368 wxCHECK_MSG( !strOld.empty(), 0,
1369 _T("wxString::Replace(): invalid parameter") );
1370
1371 wxSTRING_INVALIDATE_CACHE();
1372
1373 size_t uiCount = 0; // count of replacements made
1374
1375 // optimize the special common case: replacement of one character by
1376 // another one (in UTF-8 case we can only do this for ASCII characters)
1377 //
1378 // benchmarks show that this special version is around 3 times faster
1379 // (depending on the proportion of matching characters and UTF-8/wchar_t
1380 // build)
1381 if ( strOld.m_impl.length() == 1 && strNew.m_impl.length() == 1 )
1382 {
1383 const wxStringCharType chOld = strOld.m_impl[0],
1384 chNew = strNew.m_impl[0];
1385
1386 // this loop is the simplified version of the one below
1387 for ( size_t pos = 0; ; )
1388 {
1389 pos = m_impl.find(chOld, pos);
1390 if ( pos == npos )
1391 break;
1392
1393 m_impl[pos++] = chNew;
1394
1395 uiCount++;
1396
1397 if ( !bReplaceAll )
1398 break;
1399 }
1400 }
1401 else if ( !bReplaceAll)
1402 {
1403 size_t pos = m_impl.find(strOld, 0);
1404 if ( pos != npos )
1405 {
1406 m_impl.replace(pos, strOld.m_impl.length(), strNew.m_impl);
1407 uiCount = 1;
1408 }
1409 }
1410 else // replace all occurrences
1411 {
1412 const size_t uiOldLen = strOld.m_impl.length();
1413 const size_t uiNewLen = strNew.m_impl.length();
1414
1415 // first scan the string to find all positions at which the replacement
1416 // should be made
1417 wxVector<size_t> replacePositions;
1418
1419 size_t pos;
1420 for ( pos = m_impl.find(strOld.m_impl, 0);
1421 pos != npos;
1422 pos = m_impl.find(strOld.m_impl, pos + uiOldLen))
1423 {
1424 replacePositions.push_back(pos);
1425 ++uiCount;
1426 }
1427
1428 if ( !uiCount )
1429 return 0;
1430
1431 // allocate enough memory for the whole new string
1432 wxString tmp;
1433 tmp.m_impl.reserve(m_impl.length() + uiCount*(uiNewLen - uiOldLen));
1434
1435 // copy this string to tmp doing replacements on the fly
1436 size_t replNum = 0;
1437 for ( pos = 0; replNum < uiCount; replNum++ )
1438 {
1439 const size_t nextReplPos = replacePositions[replNum];
1440
1441 if ( pos != nextReplPos )
1442 {
1443 tmp.m_impl.append(m_impl, pos, nextReplPos - pos);
1444 }
1445
1446 tmp.m_impl.append(strNew.m_impl);
1447 pos = nextReplPos + uiOldLen;
1448 }
1449
1450 if ( pos != m_impl.length() )
1451 {
1452 // append the rest of the string unchanged
1453 tmp.m_impl.append(m_impl, pos, m_impl.length() - pos);
1454 }
1455
1456 swap(tmp);
1457 }
1458
1459 return uiCount;
1460 }
1461
1462 bool wxString::IsAscii() const
1463 {
1464 for ( const_iterator i = begin(); i != end(); ++i )
1465 {
1466 if ( !(*i).IsAscii() )
1467 return false;
1468 }
1469
1470 return true;
1471 }
1472
1473 bool wxString::IsWord() const
1474 {
1475 for ( const_iterator i = begin(); i != end(); ++i )
1476 {
1477 if ( !wxIsalpha(*i) )
1478 return false;
1479 }
1480
1481 return true;
1482 }
1483
1484 bool wxString::IsNumber() const
1485 {
1486 if ( empty() )
1487 return true;
1488
1489 const_iterator i = begin();
1490
1491 if ( *i == _T('-') || *i == _T('+') )
1492 ++i;
1493
1494 for ( ; i != end(); ++i )
1495 {
1496 if ( !wxIsdigit(*i) )
1497 return false;
1498 }
1499
1500 return true;
1501 }
1502
1503 wxString wxString::Strip(stripType w) const
1504 {
1505 wxString s = *this;
1506 if ( w & leading ) s.Trim(false);
1507 if ( w & trailing ) s.Trim(true);
1508 return s;
1509 }
1510
1511 // ---------------------------------------------------------------------------
1512 // case conversion
1513 // ---------------------------------------------------------------------------
1514
1515 wxString& wxString::MakeUpper()
1516 {
1517 for ( iterator it = begin(), en = end(); it != en; ++it )
1518 *it = (wxChar)wxToupper(*it);
1519
1520 return *this;
1521 }
1522
1523 wxString& wxString::MakeLower()
1524 {
1525 for ( iterator it = begin(), en = end(); it != en; ++it )
1526 *it = (wxChar)wxTolower(*it);
1527
1528 return *this;
1529 }
1530
1531 wxString& wxString::MakeCapitalized()
1532 {
1533 const iterator en = end();
1534 iterator it = begin();
1535 if ( it != en )
1536 {
1537 *it = (wxChar)wxToupper(*it);
1538 for ( ++it; it != en; ++it )
1539 *it = (wxChar)wxTolower(*it);
1540 }
1541
1542 return *this;
1543 }
1544
1545 // ---------------------------------------------------------------------------
1546 // trimming and padding
1547 // ---------------------------------------------------------------------------
1548
1549 // some compilers (VC++ 6.0 not to name them) return true for a call to
1550 // isspace('\xEA') in the C locale which seems to be broken to me, but we have
1551 // to live with this by checking that the character is a 7 bit one - even if
1552 // this may fail to detect some spaces (I don't know if Unicode doesn't have
1553 // space-like symbols somewhere except in the first 128 chars), it is arguably
1554 // still better than trimming away accented letters
1555 inline int wxSafeIsspace(wxChar ch) { return (ch < 127) && wxIsspace(ch); }
1556
1557 // trims spaces (in the sense of isspace) from left or right side
1558 wxString& wxString::Trim(bool bFromRight)
1559 {
1560 // first check if we're going to modify the string at all
1561 if ( !empty() &&
1562 (
1563 (bFromRight && wxSafeIsspace(GetChar(length() - 1))) ||
1564 (!bFromRight && wxSafeIsspace(GetChar(0u)))
1565 )
1566 )
1567 {
1568 if ( bFromRight )
1569 {
1570 // find last non-space character
1571 reverse_iterator psz = rbegin();
1572 while ( (psz != rend()) && wxSafeIsspace(*psz) )
1573 ++psz;
1574
1575 // truncate at trailing space start
1576 erase(psz.base(), end());
1577 }
1578 else
1579 {
1580 // find first non-space character
1581 iterator psz = begin();
1582 while ( (psz != end()) && wxSafeIsspace(*psz) )
1583 ++psz;
1584
1585 // fix up data and length
1586 erase(begin(), psz);
1587 }
1588 }
1589
1590 return *this;
1591 }
1592
1593 // adds nCount characters chPad to the string from either side
1594 wxString& wxString::Pad(size_t nCount, wxUniChar chPad, bool bFromRight)
1595 {
1596 wxString s(chPad, nCount);
1597
1598 if ( bFromRight )
1599 *this += s;
1600 else
1601 {
1602 s += *this;
1603 swap(s);
1604 }
1605
1606 return *this;
1607 }
1608
1609 // truncate the string
1610 wxString& wxString::Truncate(size_t uiLen)
1611 {
1612 if ( uiLen < length() )
1613 {
1614 erase(begin() + uiLen, end());
1615 }
1616 //else: nothing to do, string is already short enough
1617
1618 return *this;
1619 }
1620
1621 // ---------------------------------------------------------------------------
1622 // finding (return wxNOT_FOUND if not found and index otherwise)
1623 // ---------------------------------------------------------------------------
1624
1625 // find a character
1626 int wxString::Find(wxUniChar ch, bool bFromEnd) const
1627 {
1628 size_type idx = bFromEnd ? find_last_of(ch) : find_first_of(ch);
1629
1630 return (idx == npos) ? wxNOT_FOUND : (int)idx;
1631 }
1632
1633 // ----------------------------------------------------------------------------
1634 // conversion to numbers
1635 // ----------------------------------------------------------------------------
1636
1637 // The implementation of all the functions below is exactly the same so factor
1638 // it out. Note that number extraction works correctly on UTF-8 strings, so
1639 // we can use wxStringCharType and wx_str() for maximum efficiency.
1640
1641 #ifndef __WXWINCE__
1642 #define DO_IF_NOT_WINCE(x) x
1643 #else
1644 #define DO_IF_NOT_WINCE(x)
1645 #endif
1646
1647 #define WX_STRING_TO_INT_TYPE(out, base, func, T) \
1648 wxCHECK_MSG( out, false, _T("NULL output pointer") ); \
1649 wxASSERT_MSG( !base || (base > 1 && base <= 36), _T("invalid base") ); \
1650 \
1651 DO_IF_NOT_WINCE( errno = 0; ) \
1652 \
1653 const wxStringCharType *start = wx_str(); \
1654 wxStringCharType *end; \
1655 T val = func(start, &end, base); \
1656 \
1657 /* return true only if scan was stopped by the terminating NUL and */ \
1658 /* if the string was not empty to start with and no under/overflow */ \
1659 /* occurred: */ \
1660 if ( *end || end == start DO_IF_NOT_WINCE(|| errno == ERANGE) ) \
1661 return false; \
1662 *out = val; \
1663 return true
1664
1665 bool wxString::ToLong(long *pVal, int base) const
1666 {
1667 WX_STRING_TO_INT_TYPE(pVal, base, wxStrtol, long);
1668 }
1669
1670 bool wxString::ToULong(unsigned long *pVal, int base) const
1671 {
1672 WX_STRING_TO_INT_TYPE(pVal, base, wxStrtoul, unsigned long);
1673 }
1674
1675 bool wxString::ToLongLong(wxLongLong_t *pVal, int base) const
1676 {
1677 WX_STRING_TO_INT_TYPE(pVal, base, wxStrtoll, wxLongLong_t);
1678 }
1679
1680 bool wxString::ToULongLong(wxULongLong_t *pVal, int base) const
1681 {
1682 WX_STRING_TO_INT_TYPE(pVal, base, wxStrtoull, wxULongLong_t);
1683 }
1684
1685 bool wxString::ToDouble(double *pVal) const
1686 {
1687 wxCHECK_MSG( pVal, false, _T("NULL output pointer") );
1688
1689 DO_IF_NOT_WINCE( errno = 0; )
1690
1691 const wxChar *start = c_str();
1692 wxChar *end;
1693 double val = wxStrtod(start, &end);
1694
1695 // return true only if scan was stopped by the terminating NUL and if the
1696 // string was not empty to start with and no under/overflow occurred
1697 if ( *end || end == start DO_IF_NOT_WINCE(|| errno == ERANGE) )
1698 return false;
1699
1700 *pVal = val;
1701
1702 return true;
1703 }
1704
1705 // ---------------------------------------------------------------------------
1706 // formatted output
1707 // ---------------------------------------------------------------------------
1708
1709 #if !wxUSE_UTF8_LOCALE_ONLY
1710 /* static */
1711 #ifdef wxNEEDS_WXSTRING_PRINTF_MIXIN
1712 wxString wxStringPrintfMixinBase::DoFormatWchar(const wxChar *format, ...)
1713 #else
1714 wxString wxString::DoFormatWchar(const wxChar *format, ...)
1715 #endif
1716 {
1717 va_list argptr;
1718 va_start(argptr, format);
1719
1720 wxString s;
1721 s.PrintfV(format, argptr);
1722
1723 va_end(argptr);
1724
1725 return s;
1726 }
1727 #endif // !wxUSE_UTF8_LOCALE_ONLY
1728
1729 #if wxUSE_UNICODE_UTF8
1730 /* static */
1731 wxString wxString::DoFormatUtf8(const char *format, ...)
1732 {
1733 va_list argptr;
1734 va_start(argptr, format);
1735
1736 wxString s;
1737 s.PrintfV(format, argptr);
1738
1739 va_end(argptr);
1740
1741 return s;
1742 }
1743 #endif // wxUSE_UNICODE_UTF8
1744
1745 /* static */
1746 wxString wxString::FormatV(const wxString& format, va_list argptr)
1747 {
1748 wxString s;
1749 s.PrintfV(format, argptr);
1750 return s;
1751 }
1752
1753 #if !wxUSE_UTF8_LOCALE_ONLY
1754 #ifdef wxNEEDS_WXSTRING_PRINTF_MIXIN
1755 int wxStringPrintfMixinBase::DoPrintfWchar(const wxChar *format, ...)
1756 #else
1757 int wxString::DoPrintfWchar(const wxChar *format, ...)
1758 #endif
1759 {
1760 va_list argptr;
1761 va_start(argptr, format);
1762
1763 #ifdef wxNEEDS_WXSTRING_PRINTF_MIXIN
1764 // get a pointer to the wxString instance; we have to use dynamic_cast<>
1765 // because it's the only cast that works safely for downcasting when
1766 // multiple inheritance is used:
1767 wxString *str = static_cast<wxString*>(this);
1768 #else
1769 wxString *str = this;
1770 #endif
1771
1772 int iLen = str->PrintfV(format, argptr);
1773
1774 va_end(argptr);
1775
1776 return iLen;
1777 }
1778 #endif // !wxUSE_UTF8_LOCALE_ONLY
1779
1780 #if wxUSE_UNICODE_UTF8
1781 int wxString::DoPrintfUtf8(const char *format, ...)
1782 {
1783 va_list argptr;
1784 va_start(argptr, format);
1785
1786 int iLen = PrintfV(format, argptr);
1787
1788 va_end(argptr);
1789
1790 return iLen;
1791 }
1792 #endif // wxUSE_UNICODE_UTF8
1793
1794 /*
1795 Uses wxVsnprintf and places the result into the this string.
1796
1797 In ANSI build, wxVsnprintf is effectively vsnprintf but in Unicode build
1798 it is vswprintf. Due to a discrepancy between vsnprintf and vswprintf in
1799 the ISO C99 (and thus SUSv3) standard the return value for the case of
1800 an undersized buffer is inconsistent. For conforming vsnprintf
1801 implementations the function must return the number of characters that
1802 would have been printed had the buffer been large enough. For conforming
1803 vswprintf implementations the function must return a negative number
1804 and set errno.
1805
1806 What vswprintf sets errno to is undefined but Darwin seems to set it to
1807 EOVERFLOW. The only expected errno are EILSEQ and EINVAL. Both of
1808 those are defined in the standard and backed up by several conformance
1809 statements. Note that ENOMEM mentioned in the manual page does not
1810 apply to swprintf, only wprintf and fwprintf.
1811
1812 Official manual page:
1813 http://www.opengroup.org/onlinepubs/009695399/functions/swprintf.html
1814
1815 Some conformance statements (AIX, Solaris):
1816 http://www.opengroup.org/csq/view.mhtml?RID=ibm%2FSD1%2F3
1817 http://www.theopengroup.org/csq/view.mhtml?norationale=1&noreferences=1&RID=Fujitsu%2FSE2%2F10
1818
1819 Since EILSEQ and EINVAL are rather common but EOVERFLOW is not and since
1820 EILSEQ and EINVAL are specifically defined to mean the error is other than
1821 an undersized buffer and no other errno are defined we treat those two
1822 as meaning hard errors and everything else gets the old behavior which
1823 is to keep looping and increasing buffer size until the function succeeds.
1824
1825 In practice it's impossible to determine before compilation which behavior
1826 may be used. The vswprintf function may have vsnprintf-like behavior or
1827 vice-versa. Behavior detected on one release can theoretically change
1828 with an updated release. Not to mention that configure testing for it
1829 would require the test to be run on the host system, not the build system
1830 which makes cross compilation difficult. Therefore, we make no assumptions
1831 about behavior and try our best to handle every known case, including the
1832 case where wxVsnprintf returns a negative number and fails to set errno.
1833
1834 There is yet one more non-standard implementation and that is our own.
1835 Fortunately, that can be detected at compile-time.
1836
1837 On top of all that, ISO C99 explicitly defines snprintf to write a null
1838 character to the last position of the specified buffer. That would be at
1839 at the given buffer size minus 1. It is supposed to do this even if it
1840 turns out that the buffer is sized too small.
1841
1842 Darwin (tested on 10.5) follows the C99 behavior exactly.
1843
1844 Glibc 2.6 almost follows the C99 behavior except vswprintf never sets
1845 errno even when it fails. However, it only seems to ever fail due
1846 to an undersized buffer.
1847 */
1848 #if wxUSE_UNICODE_UTF8
1849 template<typename BufferType>
1850 #else
1851 // we only need one version in non-UTF8 builds and at least two Windows
1852 // compilers have problems with this function template, so use just one
1853 // normal function here
1854 #endif
1855 static int DoStringPrintfV(wxString& str,
1856 const wxString& format, va_list argptr)
1857 {
1858 int size = 1024;
1859
1860 for ( ;; )
1861 {
1862 #if wxUSE_UNICODE_UTF8
1863 BufferType tmp(str, size + 1);
1864 typename BufferType::CharType *buf = tmp;
1865 #else
1866 wxStringBuffer tmp(str, size + 1);
1867 wxChar *buf = tmp;
1868 #endif
1869
1870 if ( !buf )
1871 {
1872 // out of memory
1873
1874 // in UTF-8 build, leaving uninitialized junk in the buffer
1875 // could result in invalid non-empty UTF-8 string, so just
1876 // reset the string to empty on failure:
1877 buf[0] = '\0';
1878 return -1;
1879 }
1880
1881 // wxVsnprintf() may modify the original arg pointer, so pass it
1882 // only a copy
1883 va_list argptrcopy;
1884 wxVaCopy(argptrcopy, argptr);
1885
1886 #ifndef __WXWINCE__
1887 // Set errno to 0 to make it determinate if wxVsnprintf fails to set it.
1888 errno = 0;
1889 #endif
1890 int len = wxVsnprintf(buf, size, format, argptrcopy);
1891 va_end(argptrcopy);
1892
1893 // some implementations of vsnprintf() don't NUL terminate
1894 // the string if there is not enough space for it so
1895 // always do it manually
1896 // FIXME: This really seems to be the wrong and would be an off-by-one
1897 // bug except the code above allocates an extra character.
1898 buf[size] = _T('\0');
1899
1900 // vsnprintf() may return either -1 (traditional Unix behaviour) or the
1901 // total number of characters which would have been written if the
1902 // buffer were large enough (newer standards such as Unix98)
1903 if ( len < 0 )
1904 {
1905 // NB: wxVsnprintf() may call either wxCRT_VsnprintfW or
1906 // wxCRT_VsnprintfA in UTF-8 build; wxUSE_WXVSNPRINTF
1907 // is true if *both* of them use our own implementation,
1908 // otherwise we can't be sure
1909 #if wxUSE_WXVSNPRINTF
1910 // we know that our own implementation of wxVsnprintf() returns -1
1911 // only for a format error - thus there's something wrong with
1912 // the user's format string
1913 buf[0] = '\0';
1914 return -1;
1915 #else // possibly using system version
1916 // assume it only returns error if there is not enough space, but
1917 // as we don't know how much we need, double the current size of
1918 // the buffer
1919 #ifndef __WXWINCE__
1920 if( (errno == EILSEQ) || (errno == EINVAL) )
1921 // If errno was set to one of the two well-known hard errors
1922 // then fail immediately to avoid an infinite loop.
1923 return -1;
1924 else
1925 #endif // __WXWINCE__
1926 // still not enough, as we don't know how much we need, double the
1927 // current size of the buffer
1928 size *= 2;
1929 #endif // wxUSE_WXVSNPRINTF/!wxUSE_WXVSNPRINTF
1930 }
1931 else if ( len >= size )
1932 {
1933 #if wxUSE_WXVSNPRINTF
1934 // we know that our own implementation of wxVsnprintf() returns
1935 // size+1 when there's not enough space but that's not the size
1936 // of the required buffer!
1937 size *= 2; // so we just double the current size of the buffer
1938 #else
1939 // some vsnprintf() implementations NUL-terminate the buffer and
1940 // some don't in len == size case, to be safe always add 1
1941 // FIXME: I don't quite understand this comment. The vsnprintf
1942 // function is specifically defined to return the number of
1943 // characters printed not including the null terminator.
1944 // So OF COURSE you need to add 1 to get the right buffer size.
1945 // The following line is definitely correct, no question.
1946 size = len + 1;
1947 #endif
1948 }
1949 else // ok, there was enough space
1950 {
1951 break;
1952 }
1953 }
1954
1955 // we could have overshot
1956 str.Shrink();
1957
1958 return str.length();
1959 }
1960
1961 int wxString::PrintfV(const wxString& format, va_list argptr)
1962 {
1963 #if wxUSE_UNICODE_UTF8
1964 #if wxUSE_STL_BASED_WXSTRING
1965 typedef wxStringTypeBuffer<char> Utf8Buffer;
1966 #else
1967 typedef wxStringInternalBuffer Utf8Buffer;
1968 #endif
1969 #endif
1970
1971 #if wxUSE_UTF8_LOCALE_ONLY
1972 return DoStringPrintfV<Utf8Buffer>(*this, format, argptr);
1973 #else
1974 #if wxUSE_UNICODE_UTF8
1975 if ( wxLocaleIsUtf8 )
1976 return DoStringPrintfV<Utf8Buffer>(*this, format, argptr);
1977 else
1978 // wxChar* version
1979 return DoStringPrintfV<wxStringBuffer>(*this, format, argptr);
1980 #else
1981 return DoStringPrintfV(*this, format, argptr);
1982 #endif // UTF8/WCHAR
1983 #endif
1984 }
1985
1986 // ----------------------------------------------------------------------------
1987 // misc other operations
1988 // ----------------------------------------------------------------------------
1989
1990 // returns true if the string matches the pattern which may contain '*' and
1991 // '?' metacharacters (as usual, '?' matches any character and '*' any number
1992 // of them)
1993 bool wxString::Matches(const wxString& mask) const
1994 {
1995 // I disable this code as it doesn't seem to be faster (in fact, it seems
1996 // to be much slower) than the old, hand-written code below and using it
1997 // here requires always linking with libregex even if the user code doesn't
1998 // use it
1999 #if 0 // wxUSE_REGEX
2000 // first translate the shell-like mask into a regex
2001 wxString pattern;
2002 pattern.reserve(wxStrlen(pszMask));
2003
2004 pattern += _T('^');
2005 while ( *pszMask )
2006 {
2007 switch ( *pszMask )
2008 {
2009 case _T('?'):
2010 pattern += _T('.');
2011 break;
2012
2013 case _T('*'):
2014 pattern += _T(".*");
2015 break;
2016
2017 case _T('^'):
2018 case _T('.'):
2019 case _T('$'):
2020 case _T('('):
2021 case _T(')'):
2022 case _T('|'):
2023 case _T('+'):
2024 case _T('\\'):
2025 // these characters are special in a RE, quote them
2026 // (however note that we don't quote '[' and ']' to allow
2027 // using them for Unix shell like matching)
2028 pattern += _T('\\');
2029 // fall through
2030
2031 default:
2032 pattern += *pszMask;
2033 }
2034
2035 pszMask++;
2036 }
2037 pattern += _T('$');
2038
2039 // and now use it
2040 return wxRegEx(pattern, wxRE_NOSUB | wxRE_EXTENDED).Matches(c_str());
2041 #else // !wxUSE_REGEX
2042 // TODO: this is, of course, awfully inefficient...
2043
2044 // FIXME-UTF8: implement using iterators, remove #if
2045 #if wxUSE_UNICODE_UTF8
2046 wxWCharBuffer maskBuf = mask.wc_str();
2047 wxWCharBuffer txtBuf = wc_str();
2048 const wxChar *pszMask = maskBuf.data();
2049 const wxChar *pszTxt = txtBuf.data();
2050 #else
2051 const wxChar *pszMask = mask.wx_str();
2052 // the char currently being checked
2053 const wxChar *pszTxt = wx_str();
2054 #endif
2055
2056 // the last location where '*' matched
2057 const wxChar *pszLastStarInText = NULL;
2058 const wxChar *pszLastStarInMask = NULL;
2059
2060 match:
2061 for ( ; *pszMask != wxT('\0'); pszMask++, pszTxt++ ) {
2062 switch ( *pszMask ) {
2063 case wxT('?'):
2064 if ( *pszTxt == wxT('\0') )
2065 return false;
2066
2067 // pszTxt and pszMask will be incremented in the loop statement
2068
2069 break;
2070
2071 case wxT('*'):
2072 {
2073 // remember where we started to be able to backtrack later
2074 pszLastStarInText = pszTxt;
2075 pszLastStarInMask = pszMask;
2076
2077 // ignore special chars immediately following this one
2078 // (should this be an error?)
2079 while ( *pszMask == wxT('*') || *pszMask == wxT('?') )
2080 pszMask++;
2081
2082 // if there is nothing more, match
2083 if ( *pszMask == wxT('\0') )
2084 return true;
2085
2086 // are there any other metacharacters in the mask?
2087 size_t uiLenMask;
2088 const wxChar *pEndMask = wxStrpbrk(pszMask, wxT("*?"));
2089
2090 if ( pEndMask != NULL ) {
2091 // we have to match the string between two metachars
2092 uiLenMask = pEndMask - pszMask;
2093 }
2094 else {
2095 // we have to match the remainder of the string
2096 uiLenMask = wxStrlen(pszMask);
2097 }
2098
2099 wxString strToMatch(pszMask, uiLenMask);
2100 const wxChar* pMatch = wxStrstr(pszTxt, strToMatch);
2101 if ( pMatch == NULL )
2102 return false;
2103
2104 // -1 to compensate "++" in the loop
2105 pszTxt = pMatch + uiLenMask - 1;
2106 pszMask += uiLenMask - 1;
2107 }
2108 break;
2109
2110 default:
2111 if ( *pszMask != *pszTxt )
2112 return false;
2113 break;
2114 }
2115 }
2116
2117 // match only if nothing left
2118 if ( *pszTxt == wxT('\0') )
2119 return true;
2120
2121 // if we failed to match, backtrack if we can
2122 if ( pszLastStarInText ) {
2123 pszTxt = pszLastStarInText + 1;
2124 pszMask = pszLastStarInMask;
2125
2126 pszLastStarInText = NULL;
2127
2128 // don't bother resetting pszLastStarInMask, it's unnecessary
2129
2130 goto match;
2131 }
2132
2133 return false;
2134 #endif // wxUSE_REGEX/!wxUSE_REGEX
2135 }
2136
2137 // Count the number of chars
2138 int wxString::Freq(wxUniChar ch) const
2139 {
2140 int count = 0;
2141 for ( const_iterator i = begin(); i != end(); ++i )
2142 {
2143 if ( *i == ch )
2144 count ++;
2145 }
2146 return count;
2147 }
2148