]> git.saurik.com Git - wxWidgets.git/blob - src/common/string.cpp
fix After{First,Last}() to work for strings with non-ASCII characters in UTF-8 build...
[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
40 // string handling functions used by wxString:
41 #if wxUSE_UNICODE_UTF8
42 #define wxStringMemcpy memcpy
43 #define wxStringMemcmp memcmp
44 #define wxStringMemchr memchr
45 #define wxStringStrlen strlen
46 #else
47 #define wxStringMemcpy wxTmemcpy
48 #define wxStringMemcmp wxTmemcmp
49 #define wxStringMemchr wxTmemchr
50 #define wxStringStrlen wxStrlen
51 #endif
52
53
54 // ---------------------------------------------------------------------------
55 // static class variables definition
56 // ---------------------------------------------------------------------------
57
58 //According to STL _must_ be a -1 size_t
59 const size_t wxString::npos = (size_t) -1;
60
61 #if wxUSE_STRING_POS_CACHE
62
63 #ifdef wxHAS_COMPILER_TLS
64
65 wxTLS_TYPE(wxString::Cache) wxString::ms_cache;
66
67 #else // !wxHAS_COMPILER_TLS
68
69 struct wxStrCacheInitializer
70 {
71 wxStrCacheInitializer()
72 {
73 // calling this function triggers s_cache initialization in it, and
74 // from now on it becomes safe to call from multiple threads
75 wxString::GetCache();
76 }
77 };
78
79 /*
80 wxString::Cache& wxString::GetCache()
81 {
82 static wxTLS_TYPE(Cache) s_cache;
83
84 return wxTLS_VALUE(s_cache);
85 }
86 */
87
88 static wxStrCacheInitializer gs_stringCacheInit;
89
90 #endif // wxHAS_COMPILER_TLS/!wxHAS_COMPILER_TLS
91
92 // gdb seems to be unable to display thread-local variables correctly, at least
93 // not my 6.4.98 version under amd64, so provide this debugging helper to do it
94 #ifdef __WXDEBUG__
95
96 struct wxStrCacheDumper
97 {
98 static void ShowAll()
99 {
100 puts("*** wxString cache dump:");
101 for ( unsigned n = 0; n < wxString::Cache::SIZE; n++ )
102 {
103 const wxString::Cache::Element&
104 c = wxString::GetCacheBegin()[n];
105
106 printf("\t%u%s\t%p: pos=(%lu, %lu), len=%ld\n",
107 n,
108 n == wxString::LastUsedCacheElement() ? " [*]" : "",
109 c.str,
110 (unsigned long)c.pos,
111 (unsigned long)c.impl,
112 (long)c.len);
113 }
114 }
115 };
116
117 void wxDumpStrCache() { wxStrCacheDumper::ShowAll(); }
118
119 #endif // __WXDEBUG__
120
121 #ifdef wxPROFILE_STRING_CACHE
122
123 wxString::CacheStats wxString::ms_cacheStats;
124
125 struct wxStrCacheStatsDumper
126 {
127 ~wxStrCacheStatsDumper()
128 {
129 const wxString::CacheStats& stats = wxString::ms_cacheStats;
130
131 if ( stats.postot )
132 {
133 puts("*** wxString cache statistics:");
134 printf("\tTotal non-trivial calls to PosToImpl(): %u\n",
135 stats.postot);
136 printf("\tHits %u (of which %u not used) or %.2f%%\n",
137 stats.poshits,
138 stats.mishits,
139 100.*float(stats.poshits - stats.mishits)/stats.postot);
140 printf("\tAverage position requested: %.2f\n",
141 float(stats.sumpos) / stats.postot);
142 printf("\tAverage offset after cached hint: %.2f\n",
143 float(stats.sumofs) / stats.postot);
144 }
145
146 if ( stats.lentot )
147 {
148 printf("\tNumber of calls to length(): %u, hits=%.2f%%\n",
149 stats.lentot, 100.*float(stats.lenhits)/stats.lentot);
150 }
151 }
152 };
153
154 static wxStrCacheStatsDumper s_showCacheStats;
155
156 #endif // wxPROFILE_STRING_CACHE
157
158 #endif // wxUSE_STRING_POS_CACHE
159
160 // ----------------------------------------------------------------------------
161 // global functions
162 // ----------------------------------------------------------------------------
163
164 #if wxUSE_STD_IOSTREAM
165
166 #include <iostream>
167
168 wxSTD ostream& operator<<(wxSTD ostream& os, const wxCStrData& str)
169 {
170 #if wxUSE_UNICODE && !wxUSE_UNICODE_UTF8
171 return os << (const char *)str.AsCharBuf();
172 #else
173 return os << str.AsInternal();
174 #endif
175 }
176
177 wxSTD ostream& operator<<(wxSTD ostream& os, const wxString& str)
178 {
179 return os << str.c_str();
180 }
181
182 wxSTD ostream& operator<<(wxSTD ostream& os, const wxCharBuffer& str)
183 {
184 return os << str.data();
185 }
186
187 #ifndef __BORLANDC__
188 wxSTD ostream& operator<<(wxSTD ostream& os, const wxWCharBuffer& str)
189 {
190 return os << str.data();
191 }
192 #endif
193
194 #if wxUSE_UNICODE && defined(HAVE_WOSTREAM)
195
196 wxSTD wostream& operator<<(wxSTD wostream& wos, const wxString& str)
197 {
198 return wos << str.wc_str();
199 }
200
201 wxSTD wostream& operator<<(wxSTD wostream& wos, const wxCStrData& str)
202 {
203 return wos << str.AsWChar();
204 }
205
206 wxSTD wostream& operator<<(wxSTD wostream& wos, const wxWCharBuffer& str)
207 {
208 return wos << str.data();
209 }
210
211 #endif // wxUSE_UNICODE && defined(HAVE_WOSTREAM)
212
213 #endif // wxUSE_STD_IOSTREAM
214
215 // ===========================================================================
216 // wxString class core
217 // ===========================================================================
218
219 #if wxUSE_UNICODE_UTF8
220
221 void wxString::PosLenToImpl(size_t pos, size_t len,
222 size_t *implPos, size_t *implLen) const
223 {
224 if ( pos == npos )
225 {
226 *implPos = npos;
227 }
228 else // have valid start position
229 {
230 const const_iterator b = GetIterForNthChar(pos);
231 *implPos = wxStringImpl::const_iterator(b.impl()) - m_impl.begin();
232 if ( len == npos )
233 {
234 *implLen = npos;
235 }
236 else // have valid length too
237 {
238 // we need to handle the case of length specifying a substring
239 // going beyond the end of the string, just as std::string does
240 const const_iterator e(end());
241 const_iterator i(b);
242 while ( len && i <= e )
243 {
244 ++i;
245 --len;
246 }
247
248 *implLen = i.impl() - b.impl();
249 }
250 }
251 }
252
253 #endif // wxUSE_UNICODE_UTF8
254
255 // ----------------------------------------------------------------------------
256 // wxCStrData converted strings caching
257 // ----------------------------------------------------------------------------
258
259 // FIXME-UTF8: temporarily disabled because it doesn't work with global
260 // string objects; re-enable after fixing this bug and benchmarking
261 // performance to see if using a hash is a good idea at all
262 #if 0
263
264 // For backward compatibility reasons, it must be possible to assign the value
265 // returned by wxString::c_str() to a char* or wchar_t* variable and work with
266 // it. Returning wxCharBuffer from (const char*)c_str() wouldn't do the trick,
267 // because the memory would be freed immediately, but it has to be valid as long
268 // as the string is not modified, so that code like this still works:
269 //
270 // const wxChar *s = str.c_str();
271 // while ( s ) { ... }
272
273 // FIXME-UTF8: not thread safe!
274 // FIXME-UTF8: we currently clear the cached conversion only when the string is
275 // destroyed, but we should do it when the string is modified, to
276 // keep memory usage down
277 // FIXME-UTF8: we do the conversion every time As[W]Char() is called, but if we
278 // invalidated the cache on every change, we could keep the previous
279 // conversion
280 // FIXME-UTF8: add tracing of usage of these two methods - new code is supposed
281 // to use mb_str() or wc_str() instead of (const [w]char*)c_str()
282
283 template<typename T>
284 static inline void DeleteStringFromConversionCache(T& hash, const wxString *s)
285 {
286 typename T::iterator i = hash.find(wxConstCast(s, wxString));
287 if ( i != hash.end() )
288 {
289 free(i->second);
290 hash.erase(i);
291 }
292 }
293
294 #if wxUSE_UNICODE
295 // NB: non-STL implementation doesn't compile with "const wxString*" key type,
296 // so we have to use wxString* here and const-cast when used
297 WX_DECLARE_HASH_MAP(wxString*, char*, wxPointerHash, wxPointerEqual,
298 wxStringCharConversionCache);
299 static wxStringCharConversionCache gs_stringsCharCache;
300
301 const char* wxCStrData::AsChar() const
302 {
303 // remove previously cache value, if any (see FIXMEs above):
304 DeleteStringFromConversionCache(gs_stringsCharCache, m_str);
305
306 // convert the string and keep it:
307 const char *s = gs_stringsCharCache[wxConstCast(m_str, wxString)] =
308 m_str->mb_str().release();
309
310 return s + m_offset;
311 }
312 #endif // wxUSE_UNICODE
313
314 #if !wxUSE_UNICODE_WCHAR
315 WX_DECLARE_HASH_MAP(wxString*, wchar_t*, wxPointerHash, wxPointerEqual,
316 wxStringWCharConversionCache);
317 static wxStringWCharConversionCache gs_stringsWCharCache;
318
319 const wchar_t* wxCStrData::AsWChar() const
320 {
321 // remove previously cache value, if any (see FIXMEs above):
322 DeleteStringFromConversionCache(gs_stringsWCharCache, m_str);
323
324 // convert the string and keep it:
325 const wchar_t *s = gs_stringsWCharCache[wxConstCast(m_str, wxString)] =
326 m_str->wc_str().release();
327
328 return s + m_offset;
329 }
330 #endif // !wxUSE_UNICODE_WCHAR
331
332 wxString::~wxString()
333 {
334 #if wxUSE_UNICODE
335 // FIXME-UTF8: do this only if locale is not UTF8 if wxUSE_UNICODE_UTF8
336 DeleteStringFromConversionCache(gs_stringsCharCache, this);
337 #endif
338 #if !wxUSE_UNICODE_WCHAR
339 DeleteStringFromConversionCache(gs_stringsWCharCache, this);
340 #endif
341 }
342 #endif
343
344 #if wxUSE_UNICODE && !wxUSE_UTF8_LOCALE_ONLY
345 const char* wxCStrData::AsChar() const
346 {
347 #if wxUSE_UNICODE_UTF8
348 if ( wxLocaleIsUtf8 )
349 return AsInternal();
350 #endif
351 // under non-UTF8 locales, we have to convert the internal UTF-8
352 // representation using wxConvLibc and cache the result
353
354 wxString *str = wxConstCast(m_str, wxString);
355
356 // convert the string:
357 //
358 // FIXME-UTF8: we'd like to do the conversion in the existing buffer (if we
359 // have it) but it's unfortunately not obvious to implement
360 // because we don't know how big buffer do we need for the
361 // given string length (in case of multibyte encodings, e.g.
362 // ISO-2022-JP or UTF-8 when internal representation is wchar_t)
363 //
364 // One idea would be to store more than just m_convertedToChar
365 // in wxString: then we could record the length of the string
366 // which was converted the last time and try to reuse the same
367 // buffer if the current length is not greater than it (this
368 // could still fail because string could have been modified in
369 // place but it would work most of the time, so we'd do it and
370 // only allocate the new buffer if in-place conversion returned
371 // an error). We could also store a bit saying if the string
372 // was modified since the last conversion (and update it in all
373 // operation modifying the string, of course) to avoid unneeded
374 // consequential conversions. But both of these ideas require
375 // adding more fields to wxString and require profiling results
376 // to be sure that we really gain enough from them to justify
377 // doing it.
378 wxCharBuffer buf(str->mb_str());
379
380 // if it failed, return empty string and not NULL to avoid crashes in code
381 // written with either wxWidgets 2 wxString or std::string behaviour in
382 // mind: neither of them ever returns NULL and so we shouldn't neither
383 if ( !buf )
384 return "";
385
386 if ( str->m_convertedToChar &&
387 strlen(buf) == strlen(str->m_convertedToChar) )
388 {
389 // keep the same buffer for as long as possible, so that several calls
390 // to c_str() in a row still work:
391 strcpy(str->m_convertedToChar, buf);
392 }
393 else
394 {
395 str->m_convertedToChar = buf.release();
396 }
397
398 // and keep it:
399 return str->m_convertedToChar + m_offset;
400 }
401 #endif // wxUSE_UNICODE && !wxUSE_UTF8_LOCALE_ONLY
402
403 #if !wxUSE_UNICODE_WCHAR
404 const wchar_t* wxCStrData::AsWChar() const
405 {
406 wxString *str = wxConstCast(m_str, wxString);
407
408 // convert the string:
409 wxWCharBuffer buf(str->wc_str());
410
411 // notice that here, unlike above in AsChar(), conversion can't fail as our
412 // internal UTF-8 is always well-formed -- or the string was corrupted and
413 // all bets are off anyhow
414
415 // FIXME-UTF8: do the conversion in-place in the existing buffer
416 if ( str->m_convertedToWChar &&
417 wxWcslen(buf) == wxWcslen(str->m_convertedToWChar) )
418 {
419 // keep the same buffer for as long as possible, so that several calls
420 // to c_str() in a row still work:
421 memcpy(str->m_convertedToWChar, buf, sizeof(wchar_t) * wxWcslen(buf));
422 }
423 else
424 {
425 str->m_convertedToWChar = buf.release();
426 }
427
428 // and keep it:
429 return str->m_convertedToWChar + m_offset;
430 }
431 #endif // !wxUSE_UNICODE_WCHAR
432
433 // ===========================================================================
434 // wxString class core
435 // ===========================================================================
436
437 // ---------------------------------------------------------------------------
438 // construction and conversion
439 // ---------------------------------------------------------------------------
440
441 #if wxUSE_UNICODE_WCHAR
442 /* static */
443 wxString::SubstrBufFromMB wxString::ConvertStr(const char *psz, size_t nLength,
444 const wxMBConv& conv)
445 {
446 // anything to do?
447 if ( !psz || nLength == 0 )
448 return SubstrBufFromMB(L"", 0);
449
450 if ( nLength == npos )
451 nLength = wxNO_LEN;
452
453 size_t wcLen;
454 wxWCharBuffer wcBuf(conv.cMB2WC(psz, nLength, &wcLen));
455 if ( !wcLen )
456 return SubstrBufFromMB(_T(""), 0);
457 else
458 return SubstrBufFromMB(wcBuf, wcLen);
459 }
460 #endif // wxUSE_UNICODE_WCHAR
461
462 #if wxUSE_UNICODE_UTF8
463 /* static */
464 wxString::SubstrBufFromMB wxString::ConvertStr(const char *psz, size_t nLength,
465 const wxMBConv& conv)
466 {
467 // anything to do?
468 if ( !psz || nLength == 0 )
469 return SubstrBufFromMB("", 0);
470
471 // if psz is already in UTF-8, we don't have to do the roundtrip to
472 // wchar_t* and back:
473 if ( conv.IsUTF8() )
474 {
475 // we need to validate the input because UTF8 iterators assume valid
476 // UTF-8 sequence and psz may be invalid:
477 if ( wxStringOperations::IsValidUtf8String(psz, nLength) )
478 {
479 // we must pass the real string length to SubstrBufFromMB ctor
480 if ( nLength == npos )
481 nLength = psz ? strlen(psz) : 0;
482 return SubstrBufFromMB(wxCharBuffer::CreateNonOwned(psz), nLength);
483 }
484 // else: do the roundtrip through wchar_t*
485 }
486
487 if ( nLength == npos )
488 nLength = wxNO_LEN;
489
490 // first convert to wide string:
491 size_t wcLen;
492 wxWCharBuffer wcBuf(conv.cMB2WC(psz, nLength, &wcLen));
493 if ( !wcLen )
494 return SubstrBufFromMB("", 0);
495
496 // and then to UTF-8:
497 SubstrBufFromMB buf(ConvertStr(wcBuf, wcLen, wxMBConvStrictUTF8()));
498 // widechar -> UTF-8 conversion isn't supposed to ever fail:
499 wxASSERT_MSG( buf.data, _T("conversion to UTF-8 failed") );
500
501 return buf;
502 }
503 #endif // wxUSE_UNICODE_UTF8
504
505 #if wxUSE_UNICODE_UTF8 || !wxUSE_UNICODE
506 /* static */
507 wxString::SubstrBufFromWC wxString::ConvertStr(const wchar_t *pwz, size_t nLength,
508 const wxMBConv& conv)
509 {
510 // anything to do?
511 if ( !pwz || nLength == 0 )
512 return SubstrBufFromWC("", 0);
513
514 if ( nLength == npos )
515 nLength = wxNO_LEN;
516
517 size_t mbLen;
518 wxCharBuffer mbBuf(conv.cWC2MB(pwz, nLength, &mbLen));
519 if ( !mbLen )
520 return SubstrBufFromWC("", 0);
521 else
522 return SubstrBufFromWC(mbBuf, mbLen);
523 }
524 #endif // wxUSE_UNICODE_UTF8 || !wxUSE_UNICODE
525
526
527 #if wxUSE_UNICODE_WCHAR
528
529 //Convert wxString in Unicode mode to a multi-byte string
530 const wxCharBuffer wxString::mb_str(const wxMBConv& conv) const
531 {
532 return conv.cWC2MB(wx_str(), length() + 1 /* size, not length */, NULL);
533 }
534
535 #elif wxUSE_UNICODE_UTF8
536
537 const wxWCharBuffer wxString::wc_str() const
538 {
539 return wxMBConvStrictUTF8().cMB2WC
540 (
541 m_impl.c_str(),
542 m_impl.length() + 1, // size, not length
543 NULL
544 );
545 }
546
547 const wxCharBuffer wxString::mb_str(const wxMBConv& conv) const
548 {
549 if ( conv.IsUTF8() )
550 return wxCharBuffer::CreateNonOwned(m_impl.c_str());
551
552 // FIXME-UTF8: use wc_str() here once we have buffers with length
553
554 size_t wcLen;
555 wxWCharBuffer wcBuf(wxMBConvStrictUTF8().cMB2WC
556 (
557 m_impl.c_str(),
558 m_impl.length() + 1, // size
559 &wcLen
560 ));
561 if ( !wcLen )
562 return wxCharBuffer("");
563
564 return conv.cWC2MB(wcBuf, wcLen+1, NULL);
565 }
566
567 #else // ANSI
568
569 //Converts this string to a wide character string if unicode
570 //mode is not enabled and wxUSE_WCHAR_T is enabled
571 const wxWCharBuffer wxString::wc_str(const wxMBConv& conv) const
572 {
573 return conv.cMB2WC(wx_str(), length() + 1 /* size, not length */, NULL);
574 }
575
576 #endif // Unicode/ANSI
577
578 // shrink to minimal size (releasing extra memory)
579 bool wxString::Shrink()
580 {
581 wxString tmp(begin(), end());
582 swap(tmp);
583 return tmp.length() == length();
584 }
585
586 // deprecated compatibility code:
587 #if WXWIN_COMPATIBILITY_2_8 && !wxUSE_STL_BASED_WXSTRING && !wxUSE_UNICODE_UTF8
588 wxStringCharType *wxString::GetWriteBuf(size_t nLen)
589 {
590 return DoGetWriteBuf(nLen);
591 }
592
593 void wxString::UngetWriteBuf()
594 {
595 DoUngetWriteBuf();
596 }
597
598 void wxString::UngetWriteBuf(size_t nLen)
599 {
600 DoUngetWriteBuf(nLen);
601 }
602 #endif // WXWIN_COMPATIBILITY_2_8 && !wxUSE_STL_BASED_WXSTRING && !wxUSE_UNICODE_UTF8
603
604
605 // ---------------------------------------------------------------------------
606 // data access
607 // ---------------------------------------------------------------------------
608
609 // all functions are inline in string.h
610
611 // ---------------------------------------------------------------------------
612 // concatenation operators
613 // ---------------------------------------------------------------------------
614
615 /*
616 * concatenation functions come in 5 flavours:
617 * string + string
618 * char + string and string + char
619 * C str + string and string + C str
620 */
621
622 wxString operator+(const wxString& str1, const wxString& str2)
623 {
624 #if !wxUSE_STL_BASED_WXSTRING
625 wxASSERT( str1.IsValid() );
626 wxASSERT( str2.IsValid() );
627 #endif
628
629 wxString s = str1;
630 s += str2;
631
632 return s;
633 }
634
635 wxString operator+(const wxString& str, wxUniChar ch)
636 {
637 #if !wxUSE_STL_BASED_WXSTRING
638 wxASSERT( str.IsValid() );
639 #endif
640
641 wxString s = str;
642 s += ch;
643
644 return s;
645 }
646
647 wxString operator+(wxUniChar ch, const wxString& str)
648 {
649 #if !wxUSE_STL_BASED_WXSTRING
650 wxASSERT( str.IsValid() );
651 #endif
652
653 wxString s = ch;
654 s += str;
655
656 return s;
657 }
658
659 wxString operator+(const wxString& str, const char *psz)
660 {
661 #if !wxUSE_STL_BASED_WXSTRING
662 wxASSERT( str.IsValid() );
663 #endif
664
665 wxString s;
666 if ( !s.Alloc(strlen(psz) + str.length()) ) {
667 wxFAIL_MSG( _T("out of memory in wxString::operator+") );
668 }
669 s += str;
670 s += psz;
671
672 return s;
673 }
674
675 wxString operator+(const wxString& str, const wchar_t *pwz)
676 {
677 #if !wxUSE_STL_BASED_WXSTRING
678 wxASSERT( str.IsValid() );
679 #endif
680
681 wxString s;
682 if ( !s.Alloc(wxWcslen(pwz) + str.length()) ) {
683 wxFAIL_MSG( _T("out of memory in wxString::operator+") );
684 }
685 s += str;
686 s += pwz;
687
688 return s;
689 }
690
691 wxString operator+(const char *psz, const wxString& str)
692 {
693 #if !wxUSE_STL_BASED_WXSTRING
694 wxASSERT( str.IsValid() );
695 #endif
696
697 wxString s;
698 if ( !s.Alloc(strlen(psz) + str.length()) ) {
699 wxFAIL_MSG( _T("out of memory in wxString::operator+") );
700 }
701 s = psz;
702 s += str;
703
704 return s;
705 }
706
707 wxString operator+(const wchar_t *pwz, const wxString& str)
708 {
709 #if !wxUSE_STL_BASED_WXSTRING
710 wxASSERT( str.IsValid() );
711 #endif
712
713 wxString s;
714 if ( !s.Alloc(wxWcslen(pwz) + str.length()) ) {
715 wxFAIL_MSG( _T("out of memory in wxString::operator+") );
716 }
717 s = pwz;
718 s += str;
719
720 return s;
721 }
722
723 // ---------------------------------------------------------------------------
724 // string comparison
725 // ---------------------------------------------------------------------------
726
727 bool wxString::IsSameAs(wxUniChar c, bool compareWithCase) const
728 {
729 return (length() == 1) && (compareWithCase ? GetChar(0u) == c
730 : wxToupper(GetChar(0u)) == wxToupper(c));
731 }
732
733 #ifdef HAVE_STD_STRING_COMPARE
734
735 // NB: Comparison code (both if HAVE_STD_STRING_COMPARE and if not) works with
736 // UTF-8 encoded strings too, thanks to UTF-8's design which allows us to
737 // sort strings in characters code point order by sorting the byte sequence
738 // in byte values order (i.e. what strcmp() and memcmp() do).
739
740 int wxString::compare(const wxString& str) const
741 {
742 return m_impl.compare(str.m_impl);
743 }
744
745 int wxString::compare(size_t nStart, size_t nLen,
746 const wxString& str) const
747 {
748 size_t pos, len;
749 PosLenToImpl(nStart, nLen, &pos, &len);
750 return m_impl.compare(pos, len, str.m_impl);
751 }
752
753 int wxString::compare(size_t nStart, size_t nLen,
754 const wxString& str,
755 size_t nStart2, size_t nLen2) const
756 {
757 size_t pos, len;
758 PosLenToImpl(nStart, nLen, &pos, &len);
759
760 size_t pos2, len2;
761 str.PosLenToImpl(nStart2, nLen2, &pos2, &len2);
762
763 return m_impl.compare(pos, len, str.m_impl, pos2, len2);
764 }
765
766 int wxString::compare(const char* sz) const
767 {
768 return m_impl.compare(ImplStr(sz));
769 }
770
771 int wxString::compare(const wchar_t* sz) const
772 {
773 return m_impl.compare(ImplStr(sz));
774 }
775
776 int wxString::compare(size_t nStart, size_t nLen,
777 const char* sz, size_t nCount) const
778 {
779 size_t pos, len;
780 PosLenToImpl(nStart, nLen, &pos, &len);
781
782 SubstrBufFromMB str(ImplStr(sz, nCount));
783
784 return m_impl.compare(pos, len, str.data, str.len);
785 }
786
787 int wxString::compare(size_t nStart, size_t nLen,
788 const wchar_t* sz, size_t nCount) const
789 {
790 size_t pos, len;
791 PosLenToImpl(nStart, nLen, &pos, &len);
792
793 SubstrBufFromWC str(ImplStr(sz, nCount));
794
795 return m_impl.compare(pos, len, str.data, str.len);
796 }
797
798 #else // !HAVE_STD_STRING_COMPARE
799
800 static inline int wxDoCmp(const wxStringCharType* s1, size_t l1,
801 const wxStringCharType* s2, size_t l2)
802 {
803 if( l1 == l2 )
804 return wxStringMemcmp(s1, s2, l1);
805 else if( l1 < l2 )
806 {
807 int ret = wxStringMemcmp(s1, s2, l1);
808 return ret == 0 ? -1 : ret;
809 }
810 else
811 {
812 int ret = wxStringMemcmp(s1, s2, l2);
813 return ret == 0 ? +1 : ret;
814 }
815 }
816
817 int wxString::compare(const wxString& str) const
818 {
819 return ::wxDoCmp(m_impl.data(), m_impl.length(),
820 str.m_impl.data(), str.m_impl.length());
821 }
822
823 int wxString::compare(size_t nStart, size_t nLen,
824 const wxString& str) const
825 {
826 wxASSERT(nStart <= length());
827 size_type strLen = length() - nStart;
828 nLen = strLen < nLen ? strLen : nLen;
829
830 size_t pos, len;
831 PosLenToImpl(nStart, nLen, &pos, &len);
832
833 return ::wxDoCmp(m_impl.data() + pos, len,
834 str.m_impl.data(), str.m_impl.length());
835 }
836
837 int wxString::compare(size_t nStart, size_t nLen,
838 const wxString& str,
839 size_t nStart2, size_t nLen2) const
840 {
841 wxASSERT(nStart <= length());
842 wxASSERT(nStart2 <= str.length());
843 size_type strLen = length() - nStart,
844 strLen2 = str.length() - nStart2;
845 nLen = strLen < nLen ? strLen : nLen;
846 nLen2 = strLen2 < nLen2 ? strLen2 : nLen2;
847
848 size_t pos, len;
849 PosLenToImpl(nStart, nLen, &pos, &len);
850 size_t pos2, len2;
851 str.PosLenToImpl(nStart2, nLen2, &pos2, &len2);
852
853 return ::wxDoCmp(m_impl.data() + pos, len,
854 str.m_impl.data() + pos2, len2);
855 }
856
857 int wxString::compare(const char* sz) const
858 {
859 SubstrBufFromMB str(ImplStr(sz, npos));
860 if ( str.len == npos )
861 str.len = wxStringStrlen(str.data);
862 return ::wxDoCmp(m_impl.data(), m_impl.length(), str.data, str.len);
863 }
864
865 int wxString::compare(const wchar_t* sz) const
866 {
867 SubstrBufFromWC str(ImplStr(sz, npos));
868 if ( str.len == npos )
869 str.len = wxStringStrlen(str.data);
870 return ::wxDoCmp(m_impl.data(), m_impl.length(), str.data, str.len);
871 }
872
873 int wxString::compare(size_t nStart, size_t nLen,
874 const char* sz, size_t nCount) const
875 {
876 wxASSERT(nStart <= length());
877 size_type strLen = length() - nStart;
878 nLen = strLen < nLen ? strLen : nLen;
879
880 size_t pos, len;
881 PosLenToImpl(nStart, nLen, &pos, &len);
882
883 SubstrBufFromMB str(ImplStr(sz, nCount));
884 if ( str.len == npos )
885 str.len = wxStringStrlen(str.data);
886
887 return ::wxDoCmp(m_impl.data() + pos, len, str.data, str.len);
888 }
889
890 int wxString::compare(size_t nStart, size_t nLen,
891 const wchar_t* sz, size_t nCount) const
892 {
893 wxASSERT(nStart <= length());
894 size_type strLen = length() - nStart;
895 nLen = strLen < nLen ? strLen : nLen;
896
897 size_t pos, len;
898 PosLenToImpl(nStart, nLen, &pos, &len);
899
900 SubstrBufFromWC str(ImplStr(sz, nCount));
901 if ( str.len == npos )
902 str.len = wxStringStrlen(str.data);
903
904 return ::wxDoCmp(m_impl.data() + pos, len, str.data, str.len);
905 }
906
907 #endif // HAVE_STD_STRING_COMPARE/!HAVE_STD_STRING_COMPARE
908
909
910 // ---------------------------------------------------------------------------
911 // find_{first,last}_[not]_of functions
912 // ---------------------------------------------------------------------------
913
914 #if !wxUSE_STL_BASED_WXSTRING || wxUSE_UNICODE_UTF8
915
916 // NB: All these functions are implemented with the argument being wxChar*,
917 // i.e. widechar string in any Unicode build, even though native string
918 // representation is char* in the UTF-8 build. This is because we couldn't
919 // use memchr() to determine if a character is in a set encoded as UTF-8.
920
921 size_t wxString::find_first_of(const wxChar* sz, size_t nStart) const
922 {
923 return find_first_of(sz, nStart, wxStrlen(sz));
924 }
925
926 size_t wxString::find_first_not_of(const wxChar* sz, size_t nStart) const
927 {
928 return find_first_not_of(sz, nStart, wxStrlen(sz));
929 }
930
931 size_t wxString::find_first_of(const wxChar* sz, size_t nStart, size_t n) const
932 {
933 wxASSERT_MSG( nStart <= length(), _T("invalid index") );
934
935 size_t idx = nStart;
936 for ( const_iterator i = begin() + nStart; i != end(); ++idx, ++i )
937 {
938 if ( wxTmemchr(sz, *i, n) )
939 return idx;
940 }
941
942 return npos;
943 }
944
945 size_t wxString::find_first_not_of(const wxChar* sz, size_t nStart, size_t n) const
946 {
947 wxASSERT_MSG( nStart <= length(), _T("invalid index") );
948
949 size_t idx = nStart;
950 for ( const_iterator i = begin() + nStart; i != end(); ++idx, ++i )
951 {
952 if ( !wxTmemchr(sz, *i, n) )
953 return idx;
954 }
955
956 return npos;
957 }
958
959
960 size_t wxString::find_last_of(const wxChar* sz, size_t nStart) const
961 {
962 return find_last_of(sz, nStart, wxStrlen(sz));
963 }
964
965 size_t wxString::find_last_not_of(const wxChar* sz, size_t nStart) const
966 {
967 return find_last_not_of(sz, nStart, wxStrlen(sz));
968 }
969
970 size_t wxString::find_last_of(const wxChar* sz, size_t nStart, size_t n) const
971 {
972 size_t len = length();
973
974 if ( nStart == npos )
975 {
976 nStart = len - 1;
977 }
978 else
979 {
980 wxASSERT_MSG( nStart <= len, _T("invalid index") );
981 }
982
983 size_t idx = nStart;
984 for ( const_reverse_iterator i = rbegin() + (len - nStart - 1);
985 i != rend(); --idx, ++i )
986 {
987 if ( wxTmemchr(sz, *i, n) )
988 return idx;
989 }
990
991 return npos;
992 }
993
994 size_t wxString::find_last_not_of(const wxChar* sz, size_t nStart, size_t n) const
995 {
996 size_t len = length();
997
998 if ( nStart == npos )
999 {
1000 nStart = len - 1;
1001 }
1002 else
1003 {
1004 wxASSERT_MSG( nStart <= len, _T("invalid index") );
1005 }
1006
1007 size_t idx = nStart;
1008 for ( const_reverse_iterator i = rbegin() + (len - nStart - 1);
1009 i != rend(); --idx, ++i )
1010 {
1011 if ( !wxTmemchr(sz, *i, n) )
1012 return idx;
1013 }
1014
1015 return npos;
1016 }
1017
1018 size_t wxString::find_first_not_of(wxUniChar ch, size_t nStart) const
1019 {
1020 wxASSERT_MSG( nStart <= length(), _T("invalid index") );
1021
1022 size_t idx = nStart;
1023 for ( const_iterator i = begin() + nStart; i != end(); ++idx, ++i )
1024 {
1025 if ( *i != ch )
1026 return idx;
1027 }
1028
1029 return npos;
1030 }
1031
1032 size_t wxString::find_last_not_of(wxUniChar ch, size_t nStart) const
1033 {
1034 size_t len = length();
1035
1036 if ( nStart == npos )
1037 {
1038 nStart = len - 1;
1039 }
1040 else
1041 {
1042 wxASSERT_MSG( nStart <= len, _T("invalid index") );
1043 }
1044
1045 size_t idx = nStart;
1046 for ( const_reverse_iterator i = rbegin() + (len - nStart - 1);
1047 i != rend(); --idx, ++i )
1048 {
1049 if ( *i != ch )
1050 return idx;
1051 }
1052
1053 return npos;
1054 }
1055
1056 // the functions above were implemented for wchar_t* arguments in Unicode
1057 // build and char* in ANSI build; below are implementations for the other
1058 // version:
1059 #if wxUSE_UNICODE
1060 #define wxOtherCharType char
1061 #define STRCONV (const wxChar*)wxConvLibc.cMB2WC
1062 #else
1063 #define wxOtherCharType wchar_t
1064 #define STRCONV (const wxChar*)wxConvLibc.cWC2MB
1065 #endif
1066
1067 size_t wxString::find_first_of(const wxOtherCharType* sz, size_t nStart) const
1068 { return find_first_of(STRCONV(sz), nStart); }
1069
1070 size_t wxString::find_first_of(const wxOtherCharType* sz, size_t nStart,
1071 size_t n) const
1072 { return find_first_of(STRCONV(sz, n, NULL), nStart, n); }
1073 size_t wxString::find_last_of(const wxOtherCharType* sz, size_t nStart) const
1074 { return find_last_of(STRCONV(sz), nStart); }
1075 size_t wxString::find_last_of(const wxOtherCharType* sz, size_t nStart,
1076 size_t n) const
1077 { return find_last_of(STRCONV(sz, n, NULL), nStart, n); }
1078 size_t wxString::find_first_not_of(const wxOtherCharType* sz, size_t nStart) const
1079 { return find_first_not_of(STRCONV(sz), nStart); }
1080 size_t wxString::find_first_not_of(const wxOtherCharType* sz, size_t nStart,
1081 size_t n) const
1082 { return find_first_not_of(STRCONV(sz, n, NULL), nStart, n); }
1083 size_t wxString::find_last_not_of(const wxOtherCharType* sz, size_t nStart) const
1084 { return find_last_not_of(STRCONV(sz), nStart); }
1085 size_t wxString::find_last_not_of(const wxOtherCharType* sz, size_t nStart,
1086 size_t n) const
1087 { return find_last_not_of(STRCONV(sz, n, NULL), nStart, n); }
1088
1089 #undef wxOtherCharType
1090 #undef STRCONV
1091
1092 #endif // !wxUSE_STL_BASED_WXSTRING || wxUSE_UNICODE_UTF8
1093
1094 // ===========================================================================
1095 // other common string functions
1096 // ===========================================================================
1097
1098 int wxString::CmpNoCase(const wxString& s) const
1099 {
1100 // FIXME-UTF8: use wxUniChar::ToLower/ToUpper once added
1101
1102 const_iterator i1 = begin();
1103 const_iterator end1 = end();
1104 const_iterator i2 = s.begin();
1105 const_iterator end2 = s.end();
1106
1107 for ( ; i1 != end1 && i2 != end2; ++i1, ++i2 )
1108 {
1109 wxUniChar lower1 = (wxChar)wxTolower(*i1);
1110 wxUniChar lower2 = (wxChar)wxTolower(*i2);
1111 if ( lower1 != lower2 )
1112 return lower1 < lower2 ? -1 : 1;
1113 }
1114
1115 size_t len1 = length();
1116 size_t len2 = s.length();
1117
1118 if ( len1 < len2 )
1119 return -1;
1120 else if ( len1 > len2 )
1121 return 1;
1122 return 0;
1123 }
1124
1125
1126 #if wxUSE_UNICODE
1127
1128 #ifdef __MWERKS__
1129 #ifndef __SCHAR_MAX__
1130 #define __SCHAR_MAX__ 127
1131 #endif
1132 #endif
1133
1134 wxString wxString::FromAscii(const char *ascii, size_t len)
1135 {
1136 if (!ascii || len == 0)
1137 return wxEmptyString;
1138
1139 wxString res;
1140
1141 {
1142 wxStringInternalBuffer buf(res, len);
1143 wxStringCharType *dest = buf;
1144
1145 for ( ; len > 0; --len )
1146 {
1147 unsigned char c = (unsigned char)*ascii++;
1148 wxASSERT_MSG( c < 0x80,
1149 _T("Non-ASCII value passed to FromAscii().") );
1150
1151 *dest++ = (wchar_t)c;
1152 }
1153 }
1154
1155 return res;
1156 }
1157
1158 wxString wxString::FromAscii(const char *ascii)
1159 {
1160 return FromAscii(ascii, wxStrlen(ascii));
1161 }
1162
1163 wxString wxString::FromAscii(char ascii)
1164 {
1165 // What do we do with '\0' ?
1166
1167 unsigned char c = (unsigned char)ascii;
1168
1169 wxASSERT_MSG( c < 0x80, _T("Non-ASCII value passed to FromAscii().") );
1170
1171 // NB: the cast to wchar_t causes interpretation of 'ascii' as Latin1 value
1172 return wxString(wxUniChar((wchar_t)c));
1173 }
1174
1175 const wxCharBuffer wxString::ToAscii() const
1176 {
1177 // this will allocate enough space for the terminating NUL too
1178 wxCharBuffer buffer(length());
1179 char *dest = buffer.data();
1180
1181 for ( const_iterator i = begin(); i != end(); ++i )
1182 {
1183 wxUniChar c(*i);
1184 // FIXME-UTF8: unify substituted char ('_') with wxUniChar ('?')
1185 *dest++ = c.IsAscii() ? (char)c : '_';
1186
1187 // the output string can't have embedded NULs anyhow, so we can safely
1188 // stop at first of them even if we do have any
1189 if ( !c )
1190 break;
1191 }
1192
1193 return buffer;
1194 }
1195
1196 #endif // wxUSE_UNICODE
1197
1198 // extract string of length nCount starting at nFirst
1199 wxString wxString::Mid(size_t nFirst, size_t nCount) const
1200 {
1201 size_t nLen = length();
1202
1203 // default value of nCount is npos and means "till the end"
1204 if ( nCount == npos )
1205 {
1206 nCount = nLen - nFirst;
1207 }
1208
1209 // out-of-bounds requests return sensible things
1210 if ( nFirst + nCount > nLen )
1211 {
1212 nCount = nLen - nFirst;
1213 }
1214
1215 if ( nFirst > nLen )
1216 {
1217 // AllocCopy() will return empty string
1218 return wxEmptyString;
1219 }
1220
1221 wxString dest(*this, nFirst, nCount);
1222 if ( dest.length() != nCount )
1223 {
1224 wxFAIL_MSG( _T("out of memory in wxString::Mid") );
1225 }
1226
1227 return dest;
1228 }
1229
1230 // check that the string starts with prefix and return the rest of the string
1231 // in the provided pointer if it is not NULL, otherwise return false
1232 bool wxString::StartsWith(const wxString& prefix, wxString *rest) const
1233 {
1234 if ( compare(0, prefix.length(), prefix) != 0 )
1235 return false;
1236
1237 if ( rest )
1238 {
1239 // put the rest of the string into provided pointer
1240 rest->assign(*this, prefix.length(), npos);
1241 }
1242
1243 return true;
1244 }
1245
1246
1247 // check that the string ends with suffix and return the rest of it in the
1248 // provided pointer if it is not NULL, otherwise return false
1249 bool wxString::EndsWith(const wxString& suffix, wxString *rest) const
1250 {
1251 int start = length() - suffix.length();
1252
1253 if ( start < 0 || compare(start, npos, suffix) != 0 )
1254 return false;
1255
1256 if ( rest )
1257 {
1258 // put the rest of the string into provided pointer
1259 rest->assign(*this, 0, start);
1260 }
1261
1262 return true;
1263 }
1264
1265
1266 // extract nCount last (rightmost) characters
1267 wxString wxString::Right(size_t nCount) const
1268 {
1269 if ( nCount > length() )
1270 nCount = length();
1271
1272 wxString dest(*this, length() - nCount, nCount);
1273 if ( dest.length() != nCount ) {
1274 wxFAIL_MSG( _T("out of memory in wxString::Right") );
1275 }
1276 return dest;
1277 }
1278
1279 // get all characters after the last occurence of ch
1280 // (returns the whole string if ch not found)
1281 wxString wxString::AfterLast(wxUniChar ch) const
1282 {
1283 wxString str;
1284 int iPos = Find(ch, true);
1285 if ( iPos == wxNOT_FOUND )
1286 str = *this;
1287 else
1288 str.assign(*this, iPos + 1, npos);
1289
1290 return str;
1291 }
1292
1293 // extract nCount first (leftmost) characters
1294 wxString wxString::Left(size_t nCount) const
1295 {
1296 if ( nCount > length() )
1297 nCount = length();
1298
1299 wxString dest(*this, 0, nCount);
1300 if ( dest.length() != nCount ) {
1301 wxFAIL_MSG( _T("out of memory in wxString::Left") );
1302 }
1303 return dest;
1304 }
1305
1306 // get all characters before the first occurence of ch
1307 // (returns the whole string if ch not found)
1308 wxString wxString::BeforeFirst(wxUniChar ch) const
1309 {
1310 int iPos = Find(ch);
1311 if ( iPos == wxNOT_FOUND )
1312 iPos = length();
1313 return wxString(*this, 0, iPos);
1314 }
1315
1316 /// get all characters before the last occurence of ch
1317 /// (returns empty string if ch not found)
1318 wxString wxString::BeforeLast(wxUniChar ch) const
1319 {
1320 wxString str;
1321 int iPos = Find(ch, true);
1322 if ( iPos != wxNOT_FOUND && iPos != 0 )
1323 str = wxString(c_str(), iPos);
1324
1325 return str;
1326 }
1327
1328 /// get all characters after the first occurence of ch
1329 /// (returns empty string if ch not found)
1330 wxString wxString::AfterFirst(wxUniChar ch) const
1331 {
1332 wxString str;
1333 int iPos = Find(ch);
1334 if ( iPos != wxNOT_FOUND )
1335 str.assign(*this, iPos + 1, npos);
1336
1337 return str;
1338 }
1339
1340 // replace first (or all) occurences of some substring with another one
1341 size_t wxString::Replace(const wxString& strOld,
1342 const wxString& strNew, bool bReplaceAll)
1343 {
1344 // if we tried to replace an empty string we'd enter an infinite loop below
1345 wxCHECK_MSG( !strOld.empty(), 0,
1346 _T("wxString::Replace(): invalid parameter") );
1347
1348 wxSTRING_INVALIDATE_CACHE();
1349
1350 size_t uiCount = 0; // count of replacements made
1351
1352 // optimize the special common case: replacement of one character by
1353 // another one (in UTF-8 case we can only do this for ASCII characters)
1354 //
1355 // benchmarks show that this special version is around 3 times faster
1356 // (depending on the proportion of matching characters and UTF-8/wchar_t
1357 // build)
1358 if ( strOld.m_impl.length() == 1 && strNew.m_impl.length() == 1 )
1359 {
1360 const wxStringCharType chOld = strOld.m_impl[0],
1361 chNew = strNew.m_impl[0];
1362
1363 // this loop is the simplified version of the one below
1364 for ( size_t pos = 0; ; )
1365 {
1366 pos = m_impl.find(chOld, pos);
1367 if ( pos == npos )
1368 break;
1369
1370 m_impl[pos++] = chNew;
1371
1372 uiCount++;
1373
1374 if ( !bReplaceAll )
1375 break;
1376 }
1377 }
1378 else // general case
1379 {
1380 const size_t uiOldLen = strOld.m_impl.length();
1381 const size_t uiNewLen = strNew.m_impl.length();
1382
1383 for ( size_t pos = 0; ; )
1384 {
1385 pos = m_impl.find(strOld.m_impl, pos);
1386 if ( pos == npos )
1387 break;
1388
1389 // replace this occurrence of the old string with the new one
1390 m_impl.replace(pos, uiOldLen, strNew.m_impl);
1391
1392 // move up pos past the string that was replaced
1393 pos += uiNewLen;
1394
1395 // increase replace count
1396 uiCount++;
1397
1398 // stop after the first one?
1399 if ( !bReplaceAll )
1400 break;
1401 }
1402 }
1403
1404 return uiCount;
1405 }
1406
1407 bool wxString::IsAscii() const
1408 {
1409 for ( const_iterator i = begin(); i != end(); ++i )
1410 {
1411 if ( !(*i).IsAscii() )
1412 return false;
1413 }
1414
1415 return true;
1416 }
1417
1418 bool wxString::IsWord() const
1419 {
1420 for ( const_iterator i = begin(); i != end(); ++i )
1421 {
1422 if ( !wxIsalpha(*i) )
1423 return false;
1424 }
1425
1426 return true;
1427 }
1428
1429 bool wxString::IsNumber() const
1430 {
1431 if ( empty() )
1432 return true;
1433
1434 const_iterator i = begin();
1435
1436 if ( *i == _T('-') || *i == _T('+') )
1437 ++i;
1438
1439 for ( ; i != end(); ++i )
1440 {
1441 if ( !wxIsdigit(*i) )
1442 return false;
1443 }
1444
1445 return true;
1446 }
1447
1448 wxString wxString::Strip(stripType w) const
1449 {
1450 wxString s = *this;
1451 if ( w & leading ) s.Trim(false);
1452 if ( w & trailing ) s.Trim(true);
1453 return s;
1454 }
1455
1456 // ---------------------------------------------------------------------------
1457 // case conversion
1458 // ---------------------------------------------------------------------------
1459
1460 wxString& wxString::MakeUpper()
1461 {
1462 for ( iterator it = begin(), en = end(); it != en; ++it )
1463 *it = (wxChar)wxToupper(*it);
1464
1465 return *this;
1466 }
1467
1468 wxString& wxString::MakeLower()
1469 {
1470 for ( iterator it = begin(), en = end(); it != en; ++it )
1471 *it = (wxChar)wxTolower(*it);
1472
1473 return *this;
1474 }
1475
1476 wxString& wxString::MakeCapitalized()
1477 {
1478 const iterator en = end();
1479 iterator it = begin();
1480 if ( it != en )
1481 {
1482 *it = (wxChar)wxToupper(*it);
1483 for ( ++it; it != en; ++it )
1484 *it = (wxChar)wxTolower(*it);
1485 }
1486
1487 return *this;
1488 }
1489
1490 // ---------------------------------------------------------------------------
1491 // trimming and padding
1492 // ---------------------------------------------------------------------------
1493
1494 // some compilers (VC++ 6.0 not to name them) return true for a call to
1495 // isspace('\xEA') in the C locale which seems to be broken to me, but we have
1496 // to live with this by checking that the character is a 7 bit one - even if
1497 // this may fail to detect some spaces (I don't know if Unicode doesn't have
1498 // space-like symbols somewhere except in the first 128 chars), it is arguably
1499 // still better than trimming away accented letters
1500 inline int wxSafeIsspace(wxChar ch) { return (ch < 127) && wxIsspace(ch); }
1501
1502 // trims spaces (in the sense of isspace) from left or right side
1503 wxString& wxString::Trim(bool bFromRight)
1504 {
1505 // first check if we're going to modify the string at all
1506 if ( !empty() &&
1507 (
1508 (bFromRight && wxSafeIsspace(GetChar(length() - 1))) ||
1509 (!bFromRight && wxSafeIsspace(GetChar(0u)))
1510 )
1511 )
1512 {
1513 if ( bFromRight )
1514 {
1515 // find last non-space character
1516 reverse_iterator psz = rbegin();
1517 while ( (psz != rend()) && wxSafeIsspace(*psz) )
1518 ++psz;
1519
1520 // truncate at trailing space start
1521 erase(psz.base(), end());
1522 }
1523 else
1524 {
1525 // find first non-space character
1526 iterator psz = begin();
1527 while ( (psz != end()) && wxSafeIsspace(*psz) )
1528 ++psz;
1529
1530 // fix up data and length
1531 erase(begin(), psz);
1532 }
1533 }
1534
1535 return *this;
1536 }
1537
1538 // adds nCount characters chPad to the string from either side
1539 wxString& wxString::Pad(size_t nCount, wxUniChar chPad, bool bFromRight)
1540 {
1541 wxString s(chPad, nCount);
1542
1543 if ( bFromRight )
1544 *this += s;
1545 else
1546 {
1547 s += *this;
1548 swap(s);
1549 }
1550
1551 return *this;
1552 }
1553
1554 // truncate the string
1555 wxString& wxString::Truncate(size_t uiLen)
1556 {
1557 if ( uiLen < length() )
1558 {
1559 erase(begin() + uiLen, end());
1560 }
1561 //else: nothing to do, string is already short enough
1562
1563 return *this;
1564 }
1565
1566 // ---------------------------------------------------------------------------
1567 // finding (return wxNOT_FOUND if not found and index otherwise)
1568 // ---------------------------------------------------------------------------
1569
1570 // find a character
1571 int wxString::Find(wxUniChar ch, bool bFromEnd) const
1572 {
1573 size_type idx = bFromEnd ? find_last_of(ch) : find_first_of(ch);
1574
1575 return (idx == npos) ? wxNOT_FOUND : (int)idx;
1576 }
1577
1578 // ----------------------------------------------------------------------------
1579 // conversion to numbers
1580 // ----------------------------------------------------------------------------
1581
1582 // The implementation of all the functions below is exactly the same so factor
1583 // it out. Note that number extraction works correctly on UTF-8 strings, so
1584 // we can use wxStringCharType and wx_str() for maximum efficiency.
1585
1586 #ifndef __WXWINCE__
1587 #define DO_IF_NOT_WINCE(x) x
1588 #else
1589 #define DO_IF_NOT_WINCE(x)
1590 #endif
1591
1592 #define WX_STRING_TO_INT_TYPE(out, base, func, T) \
1593 wxCHECK_MSG( out, false, _T("NULL output pointer") ); \
1594 wxASSERT_MSG( !base || (base > 1 && base <= 36), _T("invalid base") ); \
1595 \
1596 DO_IF_NOT_WINCE( errno = 0; ) \
1597 \
1598 const wxStringCharType *start = wx_str(); \
1599 wxStringCharType *end; \
1600 T val = func(start, &end, base); \
1601 \
1602 /* return true only if scan was stopped by the terminating NUL and */ \
1603 /* if the string was not empty to start with and no under/overflow */ \
1604 /* occurred: */ \
1605 if ( *end || end == start DO_IF_NOT_WINCE(|| errno == ERANGE) ) \
1606 return false; \
1607 *out = val; \
1608 return true
1609
1610 bool wxString::ToLong(long *pVal, int base) const
1611 {
1612 WX_STRING_TO_INT_TYPE(pVal, base, wxStrtol, long);
1613 }
1614
1615 bool wxString::ToULong(unsigned long *pVal, int base) const
1616 {
1617 WX_STRING_TO_INT_TYPE(pVal, base, wxStrtoul, unsigned long);
1618 }
1619
1620 bool wxString::ToLongLong(wxLongLong_t *pVal, int base) const
1621 {
1622 WX_STRING_TO_INT_TYPE(pVal, base, wxStrtoll, wxLongLong_t);
1623 }
1624
1625 bool wxString::ToULongLong(wxULongLong_t *pVal, int base) const
1626 {
1627 WX_STRING_TO_INT_TYPE(pVal, base, wxStrtoull, wxULongLong_t);
1628 }
1629
1630 bool wxString::ToDouble(double *pVal) const
1631 {
1632 wxCHECK_MSG( pVal, false, _T("NULL output pointer") );
1633
1634 DO_IF_NOT_WINCE( errno = 0; )
1635
1636 const wxChar *start = c_str();
1637 wxChar *end;
1638 double val = wxStrtod(start, &end);
1639
1640 // return true only if scan was stopped by the terminating NUL and if the
1641 // string was not empty to start with and no under/overflow occurred
1642 if ( *end || end == start DO_IF_NOT_WINCE(|| errno == ERANGE) )
1643 return false;
1644
1645 *pVal = val;
1646
1647 return true;
1648 }
1649
1650 // ---------------------------------------------------------------------------
1651 // formatted output
1652 // ---------------------------------------------------------------------------
1653
1654 #if !wxUSE_UTF8_LOCALE_ONLY
1655 /* static */
1656 #ifdef wxNEEDS_WXSTRING_PRINTF_MIXIN
1657 wxString wxStringPrintfMixinBase::DoFormatWchar(const wxChar *format, ...)
1658 #else
1659 wxString wxString::DoFormatWchar(const wxChar *format, ...)
1660 #endif
1661 {
1662 va_list argptr;
1663 va_start(argptr, format);
1664
1665 wxString s;
1666 s.PrintfV(format, argptr);
1667
1668 va_end(argptr);
1669
1670 return s;
1671 }
1672 #endif // !wxUSE_UTF8_LOCALE_ONLY
1673
1674 #if wxUSE_UNICODE_UTF8
1675 /* static */
1676 wxString wxString::DoFormatUtf8(const char *format, ...)
1677 {
1678 va_list argptr;
1679 va_start(argptr, format);
1680
1681 wxString s;
1682 s.PrintfV(format, argptr);
1683
1684 va_end(argptr);
1685
1686 return s;
1687 }
1688 #endif // wxUSE_UNICODE_UTF8
1689
1690 /* static */
1691 wxString wxString::FormatV(const wxString& format, va_list argptr)
1692 {
1693 wxString s;
1694 s.PrintfV(format, argptr);
1695 return s;
1696 }
1697
1698 #if !wxUSE_UTF8_LOCALE_ONLY
1699 #ifdef wxNEEDS_WXSTRING_PRINTF_MIXIN
1700 int wxStringPrintfMixinBase::DoPrintfWchar(const wxChar *format, ...)
1701 #else
1702 int wxString::DoPrintfWchar(const wxChar *format, ...)
1703 #endif
1704 {
1705 va_list argptr;
1706 va_start(argptr, format);
1707
1708 #ifdef wxNEEDS_WXSTRING_PRINTF_MIXIN
1709 // get a pointer to the wxString instance; we have to use dynamic_cast<>
1710 // because it's the only cast that works safely for downcasting when
1711 // multiple inheritance is used:
1712 wxString *str = static_cast<wxString*>(this);
1713 #else
1714 wxString *str = this;
1715 #endif
1716
1717 int iLen = str->PrintfV(format, argptr);
1718
1719 va_end(argptr);
1720
1721 return iLen;
1722 }
1723 #endif // !wxUSE_UTF8_LOCALE_ONLY
1724
1725 #if wxUSE_UNICODE_UTF8
1726 int wxString::DoPrintfUtf8(const char *format, ...)
1727 {
1728 va_list argptr;
1729 va_start(argptr, format);
1730
1731 int iLen = PrintfV(format, argptr);
1732
1733 va_end(argptr);
1734
1735 return iLen;
1736 }
1737 #endif // wxUSE_UNICODE_UTF8
1738
1739 /*
1740 Uses wxVsnprintf and places the result into the this string.
1741
1742 In ANSI build, wxVsnprintf is effectively vsnprintf but in Unicode build
1743 it is vswprintf. Due to a discrepancy between vsnprintf and vswprintf in
1744 the ISO C99 (and thus SUSv3) standard the return value for the case of
1745 an undersized buffer is inconsistent. For conforming vsnprintf
1746 implementations the function must return the number of characters that
1747 would have been printed had the buffer been large enough. For conforming
1748 vswprintf implementations the function must return a negative number
1749 and set errno.
1750
1751 What vswprintf sets errno to is undefined but Darwin seems to set it to
1752 EOVERFLOW. The only expected errno are EILSEQ and EINVAL. Both of
1753 those are defined in the standard and backed up by several conformance
1754 statements. Note that ENOMEM mentioned in the manual page does not
1755 apply to swprintf, only wprintf and fwprintf.
1756
1757 Official manual page:
1758 http://www.opengroup.org/onlinepubs/009695399/functions/swprintf.html
1759
1760 Some conformance statements (AIX, Solaris):
1761 http://www.opengroup.org/csq/view.mhtml?RID=ibm%2FSD1%2F3
1762 http://www.theopengroup.org/csq/view.mhtml?norationale=1&noreferences=1&RID=Fujitsu%2FSE2%2F10
1763
1764 Since EILSEQ and EINVAL are rather common but EOVERFLOW is not and since
1765 EILSEQ and EINVAL are specifically defined to mean the error is other than
1766 an undersized buffer and no other errno are defined we treat those two
1767 as meaning hard errors and everything else gets the old behavior which
1768 is to keep looping and increasing buffer size until the function succeeds.
1769
1770 In practice it's impossible to determine before compilation which behavior
1771 may be used. The vswprintf function may have vsnprintf-like behavior or
1772 vice-versa. Behavior detected on one release can theoretically change
1773 with an updated release. Not to mention that configure testing for it
1774 would require the test to be run on the host system, not the build system
1775 which makes cross compilation difficult. Therefore, we make no assumptions
1776 about behavior and try our best to handle every known case, including the
1777 case where wxVsnprintf returns a negative number and fails to set errno.
1778
1779 There is yet one more non-standard implementation and that is our own.
1780 Fortunately, that can be detected at compile-time.
1781
1782 On top of all that, ISO C99 explicitly defines snprintf to write a null
1783 character to the last position of the specified buffer. That would be at
1784 at the given buffer size minus 1. It is supposed to do this even if it
1785 turns out that the buffer is sized too small.
1786
1787 Darwin (tested on 10.5) follows the C99 behavior exactly.
1788
1789 Glibc 2.6 almost follows the C99 behavior except vswprintf never sets
1790 errno even when it fails. However, it only seems to ever fail due
1791 to an undersized buffer.
1792 */
1793 #if wxUSE_UNICODE_UTF8
1794 template<typename BufferType>
1795 #else
1796 // we only need one version in non-UTF8 builds and at least two Windows
1797 // compilers have problems with this function template, so use just one
1798 // normal function here
1799 #endif
1800 static int DoStringPrintfV(wxString& str,
1801 const wxString& format, va_list argptr)
1802 {
1803 int size = 1024;
1804
1805 for ( ;; )
1806 {
1807 #if wxUSE_UNICODE_UTF8
1808 BufferType tmp(str, size + 1);
1809 typename BufferType::CharType *buf = tmp;
1810 #else
1811 wxStringBuffer tmp(str, size + 1);
1812 wxChar *buf = tmp;
1813 #endif
1814
1815 if ( !buf )
1816 {
1817 // out of memory
1818
1819 // in UTF-8 build, leaving uninitialized junk in the buffer
1820 // could result in invalid non-empty UTF-8 string, so just
1821 // reset the string to empty on failure:
1822 buf[0] = '\0';
1823 return -1;
1824 }
1825
1826 // wxVsnprintf() may modify the original arg pointer, so pass it
1827 // only a copy
1828 va_list argptrcopy;
1829 wxVaCopy(argptrcopy, argptr);
1830
1831 #ifndef __WXWINCE__
1832 // Set errno to 0 to make it determinate if wxVsnprintf fails to set it.
1833 errno = 0;
1834 #endif
1835 int len = wxVsnprintf(buf, size, format, argptrcopy);
1836 va_end(argptrcopy);
1837
1838 // some implementations of vsnprintf() don't NUL terminate
1839 // the string if there is not enough space for it so
1840 // always do it manually
1841 // FIXME: This really seems to be the wrong and would be an off-by-one
1842 // bug except the code above allocates an extra character.
1843 buf[size] = _T('\0');
1844
1845 // vsnprintf() may return either -1 (traditional Unix behaviour) or the
1846 // total number of characters which would have been written if the
1847 // buffer were large enough (newer standards such as Unix98)
1848 if ( len < 0 )
1849 {
1850 // NB: wxVsnprintf() may call either wxCRT_VsnprintfW or
1851 // wxCRT_VsnprintfA in UTF-8 build; wxUSE_WXVSNPRINTF
1852 // is true if *both* of them use our own implementation,
1853 // otherwise we can't be sure
1854 #if wxUSE_WXVSNPRINTF
1855 // we know that our own implementation of wxVsnprintf() returns -1
1856 // only for a format error - thus there's something wrong with
1857 // the user's format string
1858 buf[0] = '\0';
1859 return -1;
1860 #else // possibly using system version
1861 // assume it only returns error if there is not enough space, but
1862 // as we don't know how much we need, double the current size of
1863 // the buffer
1864 #ifndef __WXWINCE__
1865 if( (errno == EILSEQ) || (errno == EINVAL) )
1866 // If errno was set to one of the two well-known hard errors
1867 // then fail immediately to avoid an infinite loop.
1868 return -1;
1869 else
1870 #endif // __WXWINCE__
1871 // still not enough, as we don't know how much we need, double the
1872 // current size of the buffer
1873 size *= 2;
1874 #endif // wxUSE_WXVSNPRINTF/!wxUSE_WXVSNPRINTF
1875 }
1876 else if ( len >= size )
1877 {
1878 #if wxUSE_WXVSNPRINTF
1879 // we know that our own implementation of wxVsnprintf() returns
1880 // size+1 when there's not enough space but that's not the size
1881 // of the required buffer!
1882 size *= 2; // so we just double the current size of the buffer
1883 #else
1884 // some vsnprintf() implementations NUL-terminate the buffer and
1885 // some don't in len == size case, to be safe always add 1
1886 // FIXME: I don't quite understand this comment. The vsnprintf
1887 // function is specifically defined to return the number of
1888 // characters printed not including the null terminator.
1889 // So OF COURSE you need to add 1 to get the right buffer size.
1890 // The following line is definitely correct, no question.
1891 size = len + 1;
1892 #endif
1893 }
1894 else // ok, there was enough space
1895 {
1896 break;
1897 }
1898 }
1899
1900 // we could have overshot
1901 str.Shrink();
1902
1903 return str.length();
1904 }
1905
1906 int wxString::PrintfV(const wxString& format, va_list argptr)
1907 {
1908 #if wxUSE_UNICODE_UTF8
1909 #if wxUSE_STL_BASED_WXSTRING
1910 typedef wxStringTypeBuffer<char> Utf8Buffer;
1911 #else
1912 typedef wxStringInternalBuffer Utf8Buffer;
1913 #endif
1914 #endif
1915
1916 #if wxUSE_UTF8_LOCALE_ONLY
1917 return DoStringPrintfV<Utf8Buffer>(*this, format, argptr);
1918 #else
1919 #if wxUSE_UNICODE_UTF8
1920 if ( wxLocaleIsUtf8 )
1921 return DoStringPrintfV<Utf8Buffer>(*this, format, argptr);
1922 else
1923 // wxChar* version
1924 return DoStringPrintfV<wxStringBuffer>(*this, format, argptr);
1925 #else
1926 return DoStringPrintfV(*this, format, argptr);
1927 #endif // UTF8/WCHAR
1928 #endif
1929 }
1930
1931 // ----------------------------------------------------------------------------
1932 // misc other operations
1933 // ----------------------------------------------------------------------------
1934
1935 // returns true if the string matches the pattern which may contain '*' and
1936 // '?' metacharacters (as usual, '?' matches any character and '*' any number
1937 // of them)
1938 bool wxString::Matches(const wxString& mask) const
1939 {
1940 // I disable this code as it doesn't seem to be faster (in fact, it seems
1941 // to be much slower) than the old, hand-written code below and using it
1942 // here requires always linking with libregex even if the user code doesn't
1943 // use it
1944 #if 0 // wxUSE_REGEX
1945 // first translate the shell-like mask into a regex
1946 wxString pattern;
1947 pattern.reserve(wxStrlen(pszMask));
1948
1949 pattern += _T('^');
1950 while ( *pszMask )
1951 {
1952 switch ( *pszMask )
1953 {
1954 case _T('?'):
1955 pattern += _T('.');
1956 break;
1957
1958 case _T('*'):
1959 pattern += _T(".*");
1960 break;
1961
1962 case _T('^'):
1963 case _T('.'):
1964 case _T('$'):
1965 case _T('('):
1966 case _T(')'):
1967 case _T('|'):
1968 case _T('+'):
1969 case _T('\\'):
1970 // these characters are special in a RE, quote them
1971 // (however note that we don't quote '[' and ']' to allow
1972 // using them for Unix shell like matching)
1973 pattern += _T('\\');
1974 // fall through
1975
1976 default:
1977 pattern += *pszMask;
1978 }
1979
1980 pszMask++;
1981 }
1982 pattern += _T('$');
1983
1984 // and now use it
1985 return wxRegEx(pattern, wxRE_NOSUB | wxRE_EXTENDED).Matches(c_str());
1986 #else // !wxUSE_REGEX
1987 // TODO: this is, of course, awfully inefficient...
1988
1989 // FIXME-UTF8: implement using iterators, remove #if
1990 #if wxUSE_UNICODE_UTF8
1991 wxWCharBuffer maskBuf = mask.wc_str();
1992 wxWCharBuffer txtBuf = wc_str();
1993 const wxChar *pszMask = maskBuf.data();
1994 const wxChar *pszTxt = txtBuf.data();
1995 #else
1996 const wxChar *pszMask = mask.wx_str();
1997 // the char currently being checked
1998 const wxChar *pszTxt = wx_str();
1999 #endif
2000
2001 // the last location where '*' matched
2002 const wxChar *pszLastStarInText = NULL;
2003 const wxChar *pszLastStarInMask = NULL;
2004
2005 match:
2006 for ( ; *pszMask != wxT('\0'); pszMask++, pszTxt++ ) {
2007 switch ( *pszMask ) {
2008 case wxT('?'):
2009 if ( *pszTxt == wxT('\0') )
2010 return false;
2011
2012 // pszTxt and pszMask will be incremented in the loop statement
2013
2014 break;
2015
2016 case wxT('*'):
2017 {
2018 // remember where we started to be able to backtrack later
2019 pszLastStarInText = pszTxt;
2020 pszLastStarInMask = pszMask;
2021
2022 // ignore special chars immediately following this one
2023 // (should this be an error?)
2024 while ( *pszMask == wxT('*') || *pszMask == wxT('?') )
2025 pszMask++;
2026
2027 // if there is nothing more, match
2028 if ( *pszMask == wxT('\0') )
2029 return true;
2030
2031 // are there any other metacharacters in the mask?
2032 size_t uiLenMask;
2033 const wxChar *pEndMask = wxStrpbrk(pszMask, wxT("*?"));
2034
2035 if ( pEndMask != NULL ) {
2036 // we have to match the string between two metachars
2037 uiLenMask = pEndMask - pszMask;
2038 }
2039 else {
2040 // we have to match the remainder of the string
2041 uiLenMask = wxStrlen(pszMask);
2042 }
2043
2044 wxString strToMatch(pszMask, uiLenMask);
2045 const wxChar* pMatch = wxStrstr(pszTxt, strToMatch);
2046 if ( pMatch == NULL )
2047 return false;
2048
2049 // -1 to compensate "++" in the loop
2050 pszTxt = pMatch + uiLenMask - 1;
2051 pszMask += uiLenMask - 1;
2052 }
2053 break;
2054
2055 default:
2056 if ( *pszMask != *pszTxt )
2057 return false;
2058 break;
2059 }
2060 }
2061
2062 // match only if nothing left
2063 if ( *pszTxt == wxT('\0') )
2064 return true;
2065
2066 // if we failed to match, backtrack if we can
2067 if ( pszLastStarInText ) {
2068 pszTxt = pszLastStarInText + 1;
2069 pszMask = pszLastStarInMask;
2070
2071 pszLastStarInText = NULL;
2072
2073 // don't bother resetting pszLastStarInMask, it's unnecessary
2074
2075 goto match;
2076 }
2077
2078 return false;
2079 #endif // wxUSE_REGEX/!wxUSE_REGEX
2080 }
2081
2082 // Count the number of chars
2083 int wxString::Freq(wxUniChar ch) const
2084 {
2085 int count = 0;
2086 for ( const_iterator i = begin(); i != end(); ++i )
2087 {
2088 if ( *i == ch )
2089 count ++;
2090 }
2091 return count;
2092 }
2093
2094 // ----------------------------------------------------------------------------
2095 // wxUTF8StringBuffer
2096 // ----------------------------------------------------------------------------
2097
2098 #if wxUSE_UNICODE_WCHAR
2099 wxUTF8StringBuffer::~wxUTF8StringBuffer()
2100 {
2101 wxMBConvStrictUTF8 conv;
2102 size_t wlen = conv.ToWChar(NULL, 0, m_buf);
2103 wxCHECK_RET( wlen != wxCONV_FAILED, "invalid UTF-8 data in string buffer?" );
2104
2105 wxStringInternalBuffer wbuf(m_str, wlen);
2106 conv.ToWChar(wbuf, wlen, m_buf);
2107 }
2108
2109 wxUTF8StringBufferLength::~wxUTF8StringBufferLength()
2110 {
2111 wxCHECK_RET(m_lenSet, "length not set");
2112
2113 wxMBConvStrictUTF8 conv;
2114 size_t wlen = conv.ToWChar(NULL, 0, m_buf, m_len);
2115 wxCHECK_RET( wlen != wxCONV_FAILED, "invalid UTF-8 data in string buffer?" );
2116
2117 wxStringInternalBufferLength wbuf(m_str, wlen);
2118 conv.ToWChar(wbuf, wlen, m_buf, m_len);
2119 wbuf.SetLength(wlen);
2120 }
2121 #endif // wxUSE_UNICODE_WCHAR
2122
2123 // ----------------------------------------------------------------------------
2124 // wxCharBufferType<T>
2125 // ----------------------------------------------------------------------------
2126
2127 #ifndef __VMS_BROKEN_TEMPLATES
2128 template<>
2129 #endif
2130 wxCharTypeBuffer<char>::Data
2131 wxCharTypeBuffer<char>::NullData(NULL);
2132
2133 #ifndef __VMS_BROKEN_TEMPLATES
2134 template<>
2135 #endif
2136 wxCharTypeBuffer<wchar_t>::Data
2137 wxCharTypeBuffer<wchar_t>::NullData(NULL);