X-Git-Url: https://git.saurik.com/wxWidgets.git/blobdiff_plain/f4c90fdfaf2fd2ea1652f958c66e16fc9faff602..03773e400b6ff2692f2a9ba045d652c3b7b7109d:/include/wx/string.h?ds=sidebyside diff --git a/include/wx/string.h b/include/wx/string.h index 735e75ef77..517abb1185 100644 --- a/include/wx/string.h +++ b/include/wx/string.h @@ -24,6 +24,7 @@ #include "wx/defs.h" // everybody should include this +#ifndef __WXPALMOS5__ #if defined(__WXMAC__) || defined(__VISAGECPP__) #include #endif @@ -46,19 +47,60 @@ #ifdef HAVE_STRCASECMP_IN_STRINGS_H #include // for strcasecmp() #endif // HAVE_STRCASECMP_IN_STRINGS_H +#endif // ! __WXPALMOS5__ -#ifdef __WXPALMOS__ - #include -#endif - -#include "wx/wxchar.h" // for wxChar, wxStrlen() etc. +#include "wx/wxcrtbase.h" // for wxChar, wxStrlen() etc. #include "wx/strvararg.h" #include "wx/buffer.h" // for wxCharBuffer #include "wx/strconv.h" // for wxConvertXXX() macros and wxMBConv classes #include "wx/stringimpl.h" +#include "wx/stringops.h" #include "wx/unichar.h" -class WXDLLIMPEXP_BASE wxString; +// by default we cache the mapping of the positions in UTF-8 string to the byte +// offset as this results in noticeable performance improvements for loops over +// strings using indices; comment out this line to disable this +// +// notice that this optimization is well worth using even in debug builds as it +// changes asymptotic complexity of algorithms using indices to iterate over +// wxString back to expected linear from quadratic +// +// also notice that wxTLS_TYPE() (__declspec(thread) in this case) is unsafe to +// use in DLL build under pre-Vista Windows so we disable this code for now, if +// anybody really needs to use UTF-8 build under Windows with this optimization +// it would have to be re-tested and probably corrected +// CS: under OSX release builds the string destructor/cache cleanup sometimes +// crashes, disable until we find the true reason or a better workaround +#if wxUSE_UNICODE_UTF8 && !defined(__WXMSW__) && !defined(__WXOSX__) + #define wxUSE_STRING_POS_CACHE 1 +#else + #define wxUSE_STRING_POS_CACHE 0 +#endif + +#if wxUSE_STRING_POS_CACHE + #include "wx/tls.h" + + // change this 0 to 1 to enable additional (very expensive) asserts + // verifying that string caching logic works as expected + #if 0 + #define wxSTRING_CACHE_ASSERT(cond) wxASSERT(cond) + #else + #define wxSTRING_CACHE_ASSERT(cond) + #endif +#endif // wxUSE_STRING_POS_CACHE + +class WXDLLIMPEXP_FWD_BASE wxString; + +// unless this symbol is predefined to disable the compatibility functions, do +// use them +#ifndef WXWIN_COMPATIBILITY_STRING_PTR_AS_ITER + #define WXWIN_COMPATIBILITY_STRING_PTR_AS_ITER 1 +#endif + +namespace wxPrivate +{ + template struct wxStringAsBufHelper; +} // --------------------------------------------------------------------------- // macros @@ -118,8 +160,6 @@ inline int Stricmp(const char *psz1, const char *psz2) return _stricmp(psz1, psz2); #elif defined(__SC__) return _stricmp(psz1, psz2); -#elif defined(__SALFORDC__) - return stricmp(psz1, psz2); #elif defined(__BORLANDC__) return stricmp(psz1, psz2); #elif defined(__WATCOMC__) @@ -179,40 +219,62 @@ private: public: // Ctor constructs the object from char literal; they are needed to make // operator?: compile and they intentionally take char*, not const char* - wxCStrData(char *buf); - wxCStrData(wchar_t *buf); + inline wxCStrData(char *buf); + inline wxCStrData(wchar_t *buf); + inline wxCStrData(const wxCStrData& data); - ~wxCStrData(); + inline ~wxCStrData(); - // FIXME: we'll need convertors for both char* and wchar_t* and NONE - // for wxChar*, but that's after completing the transition to - // "smart" wxUniChar class. For now, just have conversion to - // char* in ANSI build and wchar_t in Unicode build. -#if wxUSE_UNICODE + // methods defined inline below must be declared inline or mingw32 3.4.5 + // warns about " defined locally after being referenced with + // dllimport linkage" +#if wxUSE_UNICODE_WCHAR + inline +#endif const wchar_t* AsWChar() const; operator const wchar_t*() const { return AsWChar(); } -#else + +#if !wxUSE_UNICODE || wxUSE_UTF8_LOCALE_ONLY + inline +#endif const char* AsChar() const; const unsigned char* AsUnsignedChar() const { return (const unsigned char *) AsChar(); } - operator const void*() const { return AsChar(); } operator const char*() const { return AsChar(); } operator const unsigned char*() const { return AsUnsignedChar(); } -#endif - wxString AsString() const; + operator const void*() const { return AsChar(); } + + // returns buffers that are valid as long as the associated wxString exists + const wxScopedCharBuffer AsCharBuf() const + { + return wxScopedCharBuffer::CreateNonOwned(AsChar()); + } + + const wxScopedWCharBuffer AsWCharBuf() const + { + return wxScopedWCharBuffer::CreateNonOwned(AsWChar()); + } + + inline wxString AsString() const; + + // returns the value as C string in internal representation (equivalent + // to AsString().wx_str(), but more efficient) + const wxStringCharType *AsInternal() const; // allow expressions like "c_str()[0]": + inline wxUniChar operator[](size_t n) const; wxUniChar operator[](int n) const { return operator[](size_t(n)); } - wxUniChar operator[](size_t n) const; wxUniChar operator[](long n) const { return operator[](size_t(n)); } #ifndef wxSIZE_T_IS_UINT wxUniChar operator[](unsigned int n) const { return operator[](size_t(n)); } #endif // size_t != unsigned int - // these operators are needed to emulate the pointer semantics of c_str(): + // These operators are needed to emulate the pointer semantics of c_str(): // expressions like "wxChar *p = str.c_str() + 1;" should continue to work - // (we need both versions to resolve ambiguities): + // (we need both versions to resolve ambiguities). Note that this means + // the 'n' value is interpreted as addition to char*/wchar_t* pointer, it + // is *not* number of Unicode characters in wxString. wxCStrData operator+(int n) const { return wxCStrData(m_str, m_offset + n, m_owned); } wxCStrData operator+(long n) const @@ -220,36 +282,34 @@ public: wxCStrData operator+(size_t n) const { return wxCStrData(m_str, m_offset + n, m_owned); } - // and these for "str.c_str() + n - 2": - wxCStrData operator-(int n) const - { - wxASSERT_MSG( n <= (int)m_offset, - _T("attempt to construct address before the beginning of the string") ); - return wxCStrData(m_str, m_offset - n, m_owned); - } - wxCStrData operator-(long n) const - { - wxASSERT_MSG( n <= (int)m_offset, - _T("attempt to construct address before the beginning of the string") ); - return wxCStrData(m_str, m_offset - n, m_owned); - } - wxCStrData operator-(size_t n) const + // and these for "str.c_str() + (p2 - p1)" (it also works for any integer + // expression but it must be ptrdiff_t and not e.g. int to work in this + // example): + wxCStrData operator-(ptrdiff_t n) const { - wxASSERT_MSG( n <= (int)m_offset, - _T("attempt to construct address before the beginning of the string") ); + wxASSERT_MSG( n <= (ptrdiff_t)m_offset, + wxT("attempt to construct address before the beginning of the string") ); return wxCStrData(m_str, m_offset - n, m_owned); } // this operator is needed to make expressions like "*c_str()" or // "*(c_str() + 2)" work - wxUniChar operator*() const; + inline wxUniChar operator*() const; private: + // the wxString this object was returned for const wxString *m_str; + // Offset into c_str() return value. Note that this is *not* offset in + // m_str in Unicode characters. Instead, it is index into the + // char*/wchar_t* buffer returned by c_str(). It's interpretation depends + // on how is the wxCStrData instance used: if it is eventually cast to + // const char*, m_offset will be in bytes form string's start; if it is + // cast to const wchar_t*, it will be in wchar_t values. size_t m_offset; + // should m_str be deleted, i.e. is it owned by us? bool m_owned; - friend class WXDLLIMPEXP_BASE wxString; + friend class WXDLLIMPEXP_FWD_BASE wxString; }; // ---------------------------------------------------------------------------- @@ -278,8 +338,14 @@ class WXDLLIMPEXP_BASE wxStringPrintfMixinBase protected: wxStringPrintfMixinBase() {} - int DoPrintf(const wxChar *format, ...) ATTRIBUTE_PRINTF_2; - static wxString DoFormat(const wxChar *format, ...) ATTRIBUTE_PRINTF_1; +#if !wxUSE_UTF8_LOCALE_ONLY + int DoPrintfWchar(const wxChar *format, ...); + static wxString DoFormatWchar(const wxChar *format, ...); +#endif +#if wxUSE_UNICODE_UTF8 + int DoPrintfUtf8(const char *format, ...); + static wxString DoFormatUtf8(const char *format, ...); +#endif }; // this class contains template wrappers for wxString's vararg methods, it's @@ -302,13 +368,31 @@ public: // these are duplicated wxString methods, they're also declared below // if !wxNEEDS_WXSTRING_PRINTF_MIXIN: - // int Printf(const wxChar *pszFormat, ...); - WX_DEFINE_VARARG_FUNC(int, Printf, DoPrintf) - // static wxString Format(const wxChar *pszFormat, ...) ATTRIBUTE_PRINTF_1; - WX_DEFINE_VARARG_FUNC(static typename StringReturnType::type, - Format, DoFormat) - // int sprintf(const wxChar *pszFormat, ...) ATTRIBUTE_PRINTF_2; - WX_DEFINE_VARARG_FUNC(int, sprintf, DoPrintf) + // static wxString Format(const wString& format, ...) WX_ATTRIBUTE_PRINTF_1; + WX_DEFINE_VARARG_FUNC_SANS_N0(static typename StringReturnType::type, + Format, 1, (const wxFormatString&), + DoFormatWchar, DoFormatUtf8) + // We have to implement the version without template arguments manually + // because of the StringReturnType<> hack, although WX_DEFINE_VARARG_FUNC + // normally does it itself. It has to be a template so that we can use + // the hack, even though there's no real template parameter. We can't move + // it to wxStrig, because it would shadow these versions of Format() then. + template + inline static typename StringReturnType::type + Format(const T& fmt) + { + // NB: this doesn't compile if T is not (some form of) a string; + // this makes Format's prototype equivalent to + // Format(const wxFormatString& fmt) + return DoFormatWchar(wxFormatString(fmt)); + } + + // int Printf(const wxString& format, ...); + WX_DEFINE_VARARG_FUNC(int, Printf, 1, (const wxFormatString&), + DoPrintfWchar, DoPrintfUtf8) + // int sprintf(const wxString& format, ...) WX_ATTRIBUTE_PRINTF_2; + WX_DEFINE_VARARG_FUNC(int, sprintf, 1, (const wxFormatString&), + DoPrintfWchar, DoPrintfUtf8) protected: wxStringPrintfMixin() : wxStringPrintfMixinBase() {} @@ -324,9 +408,47 @@ protected: #ifdef wxNEEDS_WXSTRING_PRINTF_MIXIN // "non dll-interface class 'wxStringPrintfMixin' used as base interface // for dll-interface class 'wxString'" -- this is OK in our case + #pragma warning (push) #pragma warning (disable:4275) #endif +#if wxUSE_UNICODE_UTF8 +// see the comment near wxString::iterator for why we need this +class WXDLLIMPEXP_BASE wxStringIteratorNode +{ +public: + wxStringIteratorNode() + : m_str(NULL), m_citer(NULL), m_iter(NULL), m_prev(NULL), m_next(NULL) {} + wxStringIteratorNode(const wxString *str, + wxStringImpl::const_iterator *citer) + { DoSet(str, citer, NULL); } + wxStringIteratorNode(const wxString *str, wxStringImpl::iterator *iter) + { DoSet(str, NULL, iter); } + ~wxStringIteratorNode() + { clear(); } + + inline void set(const wxString *str, wxStringImpl::const_iterator *citer) + { clear(); DoSet(str, citer, NULL); } + inline void set(const wxString *str, wxStringImpl::iterator *iter) + { clear(); DoSet(str, NULL, iter); } + + const wxString *m_str; + wxStringImpl::const_iterator *m_citer; + wxStringImpl::iterator *m_iter; + wxStringIteratorNode *m_prev, *m_next; + +private: + inline void clear(); + inline void DoSet(const wxString *str, + wxStringImpl::const_iterator *citer, + wxStringImpl::iterator *iter); + + // the node belongs to a particular iterator instance, it's not copied + // when a copy of the iterator is made + wxDECLARE_NO_COPY_CLASS(wxStringIteratorNode); +}; +#endif // wxUSE_UNICODE_UTF8 + class WXDLLIMPEXP_BASE wxString #ifdef wxNEEDS_WXSTRING_PRINTF_MIXIN : public wxStringPrintfMixin @@ -358,26 +480,29 @@ private: // buffer for holding temporary substring when using any of the methods // that take (char*,size_t) or (wchar_t*,size_t) arguments: - // FIXME-UTF8: This will need changes when UTF8 build is introduced template struct SubstrBufFromType { T data; size_t len; - SubstrBufFromType() {} SubstrBufFromType(const T& data_, size_t len_) - : data(data_), len(len_) {} + : data(data_), len(len_) + { + wxASSERT_MSG( len != npos, "must have real length" ); + } }; #if wxUSE_UNICODE_UTF8 - // FIXME-UTF8: this will have to use slightly different type + // even char* -> char* needs conversion, from locale charset to UTF-8 + typedef SubstrBufFromType SubstrBufFromWC; + typedef SubstrBufFromType SubstrBufFromMB; #elif wxUSE_UNICODE_WCHAR - typedef SubstrBufFromType SubstrBufFromWC; - typedef SubstrBufFromType SubstrBufFromMB; + typedef SubstrBufFromType SubstrBufFromWC; + typedef SubstrBufFromType SubstrBufFromMB; #else - typedef SubstrBufFromType SubstrBufFromMB; - typedef SubstrBufFromType SubstrBufFromWC; + typedef SubstrBufFromType SubstrBufFromMB; + typedef SubstrBufFromType SubstrBufFromWC; #endif @@ -386,8 +511,12 @@ private: // between UTF-8 and wchar_t* representations of the string are mostly // contained here. -#if wxUSE_UNICODE - // FIXME-UTF8: This will need changes when UTF8 build is introduced +#if wxUSE_UNICODE_UTF8 + static SubstrBufFromMB ConvertStr(const char *psz, size_t nLength, + const wxMBConv& conv); + static SubstrBufFromWC ConvertStr(const wchar_t *pwz, size_t nLength, + const wxMBConv& conv); +#elif wxUSE_UNICODE_WCHAR static SubstrBufFromMB ConvertStr(const char *psz, size_t nLength, const wxMBConv& conv); #else @@ -399,175 +528,345 @@ private: // returns C string encoded as the implementation expects: #if wxUSE_UNICODE static const wchar_t* ImplStr(const wchar_t* str) - { return str; } + { return str ? str : wxT(""); } static const SubstrBufFromWC ImplStr(const wchar_t* str, size_t n) - { return SubstrBufFromWC(str, n == npos ? wxWcslen(str) : n); } - static wxWCharBuffer ImplStr(const char* str) - { return ConvertStr(str, npos, wxConvLibc).data; } - static SubstrBufFromMB ImplStr(const char* str, size_t n) - { return ConvertStr(str, n, wxConvLibc); } + { return SubstrBufFromWC(str, (str && n == npos) ? wxWcslen(str) : n); } + static wxScopedWCharBuffer ImplStr(const char* str, + const wxMBConv& conv = wxConvLibc) + { return ConvertStr(str, npos, conv).data; } + static SubstrBufFromMB ImplStr(const char* str, size_t n, + const wxMBConv& conv = wxConvLibc) + { return ConvertStr(str, n, conv); } #else - static const char* ImplStr(const char* str) - { return str; } - static const SubstrBufFromMB ImplStr(const char* str, size_t n) - { return SubstrBufFromMB(str, n == npos ? wxStrlen(str) : n); } - static wxCharBuffer ImplStr(const wchar_t* str) + static const char* ImplStr(const char* str, + const wxMBConv& WXUNUSED(conv) = wxConvLibc) + { return str ? str : ""; } + static const SubstrBufFromMB ImplStr(const char* str, size_t n, + const wxMBConv& WXUNUSED(conv) = wxConvLibc) + { return SubstrBufFromMB(str, (str && n == npos) ? wxStrlen(str) : n); } + static wxScopedCharBuffer ImplStr(const wchar_t* str) { return ConvertStr(str, npos, wxConvLibc).data; } static SubstrBufFromWC ImplStr(const wchar_t* str, size_t n) { return ConvertStr(str, n, wxConvLibc); } #endif - // moves the iterator to the next Unicode character - static void IncIter(wxStringImpl::iterator& i) { ++i; } - static void IncIter(wxStringImpl::const_iterator& i) { ++i; } - // moves the iterator to the previous Unicode character - static void DecIter(wxStringImpl::iterator& i) { --i; } - static void DecIter(wxStringImpl::const_iterator& i) { --i; } - // moves the iterator by n Unicode characters - static wxStringImpl::iterator AddToIter(wxStringImpl::iterator i, int n) - { return i + n; } - static wxStringImpl::const_iterator AddToIter(wxStringImpl::const_iterator i, int n) - { return i + n; } - // returns distance of the two iterators in Unicode characters - static int DiffIters(wxStringImpl::iterator i1, wxStringImpl::iterator i2) - { return i1 - i2; } - static int DiffIters(wxStringImpl::const_iterator i1, wxStringImpl::const_iterator i2) - { return i1 - i2; } - - // encodes the character to a form used to represent it in internal - // representation (returns a string in UTF8 version) - static wxChar EncodeChar(wxUniChar ch) { return (wxChar)ch; } - // translates position index in wxString to/from index in underlying // wxStringImpl: static size_t PosToImpl(size_t pos) { return pos; } static void PosLenToImpl(size_t pos, size_t len, size_t *implPos, size_t *implLen) { *implPos = pos; *implLen = len; } + static size_t LenToImpl(size_t len) { return len; } static size_t PosFromImpl(size_t pos) { return pos; } + // we don't want to define these as empty inline functions as it could + // result in noticeable (and quite unnecessary in non-UTF-8 build) slowdown + // in debug build where the inline functions are not effectively inlined + #define wxSTRING_INVALIDATE_CACHE() + #define wxSTRING_INVALIDATE_CACHED_LENGTH() + #define wxSTRING_UPDATE_CACHED_LENGTH(n) + #define wxSTRING_SET_CACHED_LENGTH(n) + #else // wxUSE_UNICODE_UTF8 - typedef char Utf8CharBuffer[5]; - static Utf8CharBuffer EncodeChar(wxUniChar ch); - // returns n copies of ch encoded in UTF-8 string - static wxCharBuffer EncodeNChars(size_t n, wxUniChar ch); + static wxScopedCharBuffer ImplStr(const char* str, + const wxMBConv& conv = wxConvLibc) + { return ConvertStr(str, npos, conv).data; } + static SubstrBufFromMB ImplStr(const char* str, size_t n, + const wxMBConv& conv = wxConvLibc) + { return ConvertStr(str, n, conv); } - size_t PosToImpl(size_t pos) const + static wxScopedCharBuffer ImplStr(const wchar_t* str) + { return ConvertStr(str, npos, wxMBConvUTF8()).data; } + static SubstrBufFromWC ImplStr(const wchar_t* str, size_t n) + { return ConvertStr(str, n, wxMBConvUTF8()); } + +#if wxUSE_STRING_POS_CACHE + // this is an extremely simple cache used by PosToImpl(): each cache element + // contains the string it applies to and the index corresponding to the last + // used position in this wxString in its m_impl string + // + // NB: notice that this struct (and nested Element one) must be a POD or we + // wouldn't be able to use a thread-local variable of this type, in + // particular it should have no ctor -- we rely on statics being + // initialized to 0 instead + struct Cache { - if ( pos == 0 || pos == npos ) - return pos; - else - return wxStringImpl::const_iterator(begin() + pos) - m_impl.begin(); + enum { SIZE = 8 }; + + struct Element + { + const wxString *str; // the string to which this element applies + size_t pos, // the cached index in this string + impl, // the corresponding position in its m_impl + len; // cached length or npos if unknown + + // reset cached index to 0 + void ResetPos() { pos = impl = 0; } + + // reset position and length + void Reset() { ResetPos(); len = npos; } + }; + + // cache the indices mapping for the last few string used + Element cached[SIZE]; + + // the last used index + unsigned lastUsed; + }; + +#ifndef wxHAS_COMPILER_TLS + // we must use an accessor function and not a static variable when the TLS + // variables support is implemented in the library (and not by the compiler) + // because the global s_cache variable could be not yet initialized when a + // ctor of another global object is executed and if that ctor uses any + // wxString methods, bad things happen + // + // however notice that this approach does not work when compiler TLS is used, + // at least not with g++ 4.1.2 under amd64 as it apparently compiles code + // using this accessor incorrectly when optimizations are enabled (-O2 is + // enough) -- luckily we don't need it then neither as static __thread + // variables are initialized by 0 anyhow then and so we can use the variable + // directly + WXEXPORT static Cache& GetCache() + { + static wxTLS_TYPE(Cache) s_cache; + + return wxTLS_VALUE(s_cache); } - size_t PosFromImpl(size_t pos) const + // this helper struct is used to ensure that GetCache() is called during + // static initialization time, i.e. before any threads creation, as otherwise + // the static s_cache construction inside GetCache() wouldn't be MT-safe + friend struct wxStrCacheInitializer; +#else // wxHAS_COMPILER_TLS + static wxTLS_TYPE(Cache) ms_cache; + static Cache& GetCache() { return wxTLS_VALUE(ms_cache); } +#endif // !wxHAS_COMPILER_TLS/wxHAS_COMPILER_TLS + + static Cache::Element *GetCacheBegin() { return GetCache().cached; } + static Cache::Element *GetCacheEnd() { return GetCacheBegin() + Cache::SIZE; } + static unsigned& LastUsedCacheElement() { return GetCache().lastUsed; } + + // this is used in debug builds only to provide a convenient function, + // callable from a debugger, to show the cache contents + friend struct wxStrCacheDumper; + + // uncomment this to have access to some profiling statistics on program + // termination + //#define wxPROFILE_STRING_CACHE + +#ifdef wxPROFILE_STRING_CACHE + static struct PosToImplCacheStats { - if ( pos == 0 || pos == npos ) - return pos; - else - return const_iterator(m_impl.begin() + pos) - begin(); + unsigned postot, // total non-trivial calls to PosToImpl + poshits, // cache hits from PosToImpl() + mishits, // cached position beyond the needed one + sumpos, // sum of all positions, used to compute the + // average position after dividing by postot + sumofs, // sum of all offsets after using the cache, used to + // compute the average after dividing by hits + lentot, // number of total calls to length() + lenhits; // number of cache hits in length() + } ms_cacheStats; + + friend struct wxStrCacheStatsDumper; + + #define wxCACHE_PROFILE_FIELD_INC(field) ms_cacheStats.field++ + #define wxCACHE_PROFILE_FIELD_ADD(field, val) ms_cacheStats.field += (val) +#else // !wxPROFILE_STRING_CACHE + #define wxCACHE_PROFILE_FIELD_INC(field) + #define wxCACHE_PROFILE_FIELD_ADD(field, val) +#endif // wxPROFILE_STRING_CACHE/!wxPROFILE_STRING_CACHE + + // note: it could seem that the functions below shouldn't be inline because + // they are big, contain loops and so the compiler shouldn't be able to + // inline them anyhow, however moving them into string.cpp does decrease the + // code performance by ~5%, at least when using g++ 4.1 so do keep them here + // unless tests show that it's not advantageous any more + + // return the pointer to the cache element for this string or NULL if not + // cached + Cache::Element *FindCacheElement() const + { + // profiling seems to show a small but consistent gain if we use this + // simple loop instead of starting from the last used element (there are + // a lot of misses in this function...) + Cache::Element * const cacheBegin = GetCacheBegin(); +#ifndef wxHAS_COMPILER_TLS + // during destruction tls calls may return NULL, in this case return NULL + // immediately without accessing anything else + if ( cacheBegin == NULL ) + return NULL; +#endif + Cache::Element * const cacheEnd = GetCacheEnd(); + for ( Cache::Element *c = cacheBegin; c != cacheEnd; c++ ) + { + if ( c->str == this ) + return c; + } + + return NULL; } - // FIXME: return as-is without copying under UTF8 locale, return - // converted string under other locales - needs wxCharBuffer - // changes - static wxCharBuffer ImplStr(const char* str); + // unlike FindCacheElement(), this one always returns a valid pointer to the + // cache element for this string, it may have valid last cached position and + // its corresponding index in the byte string or not + Cache::Element *GetCacheElement() const + { + Cache::Element * const cacheBegin = GetCacheBegin(); + Cache::Element * const cacheEnd = GetCacheEnd(); + Cache::Element * const cacheStart = cacheBegin + LastUsedCacheElement(); + + // check the last used first, this does no (measurable) harm for a miss + // but does help for simple loops addressing the same string all the time + if ( cacheStart->str == this ) + return cacheStart; + + // notice that we're going to check cacheStart again inside this call but + // profiling shows that it's still faster to use a simple loop like + // inside FindCacheElement() than manually looping with wrapping starting + // from the cache entry after the start one + Cache::Element *c = FindCacheElement(); + if ( !c ) + { + // claim the next cache entry for this string + c = cacheStart; + if ( ++c == cacheEnd ) + c = cacheBegin; - static wxCharBuffer ImplStr(const wchar_t* str) - { return wxConvUTF8.cWC2MB(str); } -#endif // !wxUSE_UNICODE_UTF8/wxUSE_UNICODE_UTF8 + c->str = this; + c->Reset(); + + // and remember the last used element + LastUsedCacheElement() = c - cacheBegin; + } + return c; + } -public: - // constructors and destructor - // ctor for an empty string - wxString() {} - // copy ctor - wxString(const wxStringImpl& stringSrc) : m_impl(stringSrc) { } - wxString(const wxString& stringSrc) : m_impl(stringSrc) { } - // string containing nRepeat copies of ch - wxString(wxUniChar ch, size_t nRepeat = 1) - : m_impl(nRepeat, ch) { } - wxString(size_t nRepeat, wxUniChar ch) - : m_impl(nRepeat, ch) { } - wxString(wxUniCharRef ch, size_t nRepeat = 1) - : m_impl(nRepeat, ch) { } - wxString(size_t nRepeat, wxUniCharRef ch) - : m_impl(nRepeat, ch) { } - wxString(char ch, size_t nRepeat = 1) - : m_impl(nRepeat, ch) { } - wxString(size_t nRepeat, char ch) - : m_impl(nRepeat, ch) { } - wxString(wchar_t ch, size_t nRepeat = 1) - : m_impl(nRepeat, ch) { } - wxString(size_t nRepeat, wchar_t ch) - : m_impl(nRepeat, ch) { } - // ctor takes first nLength characters from C string - // (default value of npos means take all the string) - wxString(const wxChar *psz) - : m_impl(psz ? psz : wxT("")) { } - wxString(const wxChar *psz, size_t nLength) - : m_impl(psz, nLength) { } - wxString(const wxChar *psz, - const wxMBConv& WXUNUSED(conv), - size_t nLength = npos) - : m_impl(psz, nLength == npos ? wxStrlen(psz) : nLength) { } + size_t DoPosToImpl(size_t pos) const + { + wxCACHE_PROFILE_FIELD_INC(postot); - // even if we're not built with wxUSE_STL == 1 it is very convenient to allow - // implicit conversions from std::string to wxString as this allows to use - // the same strings in non-GUI and GUI code, however we don't want to - // unconditionally add this ctor as it would make wx lib dependent on - // libstdc++ on some Linux versions which is bad, so instead we ask the - // client code to define this wxUSE_STD_STRING symbol if they need it -#if wxUSE_STD_STRING && !wxUSE_STL_BASED_WXSTRING - wxString(const wxStdString& s) - : m_impl(s.c_str()) { } // FIXME-UTF8: this is broken for embedded 0s -#endif // wxUSE_STD_STRING && !wxUSE_STL_BASED_WXSTRING + // NB: although the case of pos == 1 (and offset from cached position + // equal to 1) are common, nothing is gained by writing special code + // for handling them, the compiler (at least g++ 4.1 used) seems to + // optimize the code well enough on its own -#if wxUSE_UNICODE - // from multibyte string - wxString(const char *psz, - const wxMBConv& conv = wxConvLibc, - size_t nLength = npos); - // from multibyte string for ANSI compatibility, with wxConvLibc - wxString(const char *psz, size_t nLength); - // from wxWCharBuffer (i.e. return from wxGetString) - wxString(const wxWCharBuffer& psz) : m_impl(psz.data()) { } -#else // ANSI - // from C string (for compilers using unsigned char) - wxString(const unsigned char* psz) - : m_impl((const char*)psz) { } - // from part of C string (for compilers using unsigned char) - wxString(const unsigned char* psz, size_t nLength) - : m_impl((const char*)psz, nLength) { } - -#if wxUSE_WCHAR_T - // from wide (Unicode) string - wxString(const wchar_t *pwz, - const wxMBConv& conv = wxConvLibc, - size_t nLength = npos); - // from wide string for Unicode compatibility, with wxConvLibc - wxString(const wchar_t *pwz, size_t nLength); -#endif // !wxUSE_WCHAR_T - - // from wxCharBuffer - wxString(const wxCharBuffer& psz) - : m_impl(psz) { } -#endif // Unicode/ANSI + wxCACHE_PROFILE_FIELD_ADD(sumpos, pos); - wxString(const wxCStrData& cstr) - : m_impl(cstr.AsString().m_impl) { } + Cache::Element * const cache = GetCacheElement(); - // as we provide both ctors with this signature for both char and unsigned - // char string, we need to provide one for wxCStrData to resolve ambiguity - wxString(const wxCStrData& cstr, size_t nLength) - { assign(cstr.AsString(), nLength); } + // cached position can't be 0 so if it is, it means that this entry was + // used for length caching only so far, i.e. it doesn't count as a hit + // from our point of view + if ( cache->pos ) + { + wxCACHE_PROFILE_FIELD_INC(poshits); + } - // and because wxString is convertible to wxCStrData and const wxChar * - // we also need to provide this one - wxString(const wxString& str, size_t nLength) - { assign(str, nLength); } + if ( pos == cache->pos ) + return cache->impl; + + // this seems to happen only rarely so just reset the cache in this case + // instead of complicating code even further by seeking backwards in this + // case + if ( cache->pos > pos ) + { + wxCACHE_PROFILE_FIELD_INC(mishits); + + cache->ResetPos(); + } + + wxCACHE_PROFILE_FIELD_ADD(sumofs, pos - cache->pos); + + + wxStringImpl::const_iterator i(m_impl.begin() + cache->impl); + for ( size_t n = cache->pos; n < pos; n++ ) + wxStringOperations::IncIter(i); + + cache->pos = pos; + cache->impl = i - m_impl.begin(); + + wxSTRING_CACHE_ASSERT( + (int)cache->impl == (begin() + pos).impl() - m_impl.begin() ); + + return cache->impl; + } + + void InvalidateCache() + { + Cache::Element * const cache = FindCacheElement(); + if ( cache ) + cache->Reset(); + } + + void InvalidateCachedLength() + { + Cache::Element * const cache = FindCacheElement(); + if ( cache ) + cache->len = npos; + } + + void SetCachedLength(size_t len) + { + // we optimistically cache the length here even if the string wasn't + // present in the cache before, this seems to do no harm and the + // potential for avoiding length recomputation for long strings looks + // interesting + GetCacheElement()->len = len; + } + + void UpdateCachedLength(ptrdiff_t delta) + { + Cache::Element * const cache = FindCacheElement(); + if ( cache && cache->len != npos ) + { + wxSTRING_CACHE_ASSERT( (ptrdiff_t)cache->len + delta >= 0 ); + + cache->len += delta; + } + } + + #define wxSTRING_INVALIDATE_CACHE() InvalidateCache() + #define wxSTRING_INVALIDATE_CACHED_LENGTH() InvalidateCachedLength() + #define wxSTRING_UPDATE_CACHED_LENGTH(n) UpdateCachedLength(n) + #define wxSTRING_SET_CACHED_LENGTH(n) SetCachedLength(n) +#else // !wxUSE_STRING_POS_CACHE + size_t DoPosToImpl(size_t pos) const + { + return (begin() + pos).impl() - m_impl.begin(); + } + + #define wxSTRING_INVALIDATE_CACHE() + #define wxSTRING_INVALIDATE_CACHED_LENGTH() + #define wxSTRING_UPDATE_CACHED_LENGTH(n) + #define wxSTRING_SET_CACHED_LENGTH(n) +#endif // wxUSE_STRING_POS_CACHE/!wxUSE_STRING_POS_CACHE + + size_t PosToImpl(size_t pos) const + { + return pos == 0 || pos == npos ? pos : DoPosToImpl(pos); + } + + void PosLenToImpl(size_t pos, size_t len, size_t *implPos, size_t *implLen) const; + + size_t LenToImpl(size_t len) const + { + size_t pos, len2; + PosLenToImpl(0, len, &pos, &len2); + return len2; + } + + size_t PosFromImpl(size_t pos) const + { + if ( pos == 0 || pos == npos ) + return pos; + else + return const_iterator(this, m_impl.begin() + pos) - begin(); + } +#endif // !wxUSE_UNICODE_UTF8/wxUSE_UNICODE_UTF8 public: // standard types @@ -580,66 +879,65 @@ public: typedef size_t size_type; typedef wxUniChar const_reference; -#if wxUSE_STL - #define WX_STR_ITERATOR_TAG std::random_access_iterator_tag +#if wxUSE_STD_STRING + #if wxUSE_UNICODE_UTF8 + // random access is not O(1), as required by Random Access Iterator + #define WX_STR_ITERATOR_TAG std::bidirectional_iterator_tag + #else + #define WX_STR_ITERATOR_TAG std::random_access_iterator_tag + #endif + #define WX_DEFINE_ITERATOR_CATEGORY(cat) typedef cat iterator_category; #else - #define WX_STR_ITERATOR_TAG void /* dummy type */ + // not defining iterator_category at all in this case is better than defining + // it as some dummy type -- at least it results in more intelligible error + // messages + #define WX_DEFINE_ITERATOR_CATEGORY(cat) #endif - #define WX_STR_ITERATOR_IMPL(iterator_name, pointer_type, \ - reference_type, reference_ctor) \ + #define WX_STR_ITERATOR_IMPL(iterator_name, pointer_type, reference_type) \ private: \ typedef wxStringImpl::iterator_name underlying_iterator; \ public: \ - typedef WX_STR_ITERATOR_TAG iterator_category; \ + WX_DEFINE_ITERATOR_CATEGORY(WX_STR_ITERATOR_TAG) \ typedef wxUniChar value_type; \ typedef int difference_type; \ typedef reference_type reference; \ typedef pointer_type pointer; \ \ - iterator_name(const iterator_name& i) : m_cur(i.m_cur) {} \ - \ - reference operator*() const { return reference_ctor; } \ reference operator[](size_t n) const { return *(*this + n); } \ \ iterator_name& operator++() \ - { wxString::IncIter(m_cur); return *this; } \ + { wxStringOperations::IncIter(m_cur); return *this; } \ iterator_name& operator--() \ - { wxString::DecIter(m_cur); return *this; } \ + { wxStringOperations::DecIter(m_cur); return *this; } \ iterator_name operator++(int) \ { \ iterator_name tmp = *this; \ - wxString::IncIter(m_cur); \ + wxStringOperations::IncIter(m_cur); \ return tmp; \ } \ iterator_name operator--(int) \ { \ iterator_name tmp = *this; \ - wxString::DecIter(m_cur); \ + wxStringOperations::DecIter(m_cur); \ return tmp; \ } \ \ - iterator_name operator+(int n) const \ - { return iterator_name(wxString::AddToIter(m_cur, n)); } \ - iterator_name operator+(size_t n) const \ - { return iterator_name(wxString::AddToIter(m_cur, (int)n)); } \ - iterator_name operator-(int n) const \ - { return iterator_name(wxString::AddToIter(m_cur, -n)); } \ - iterator_name operator-(size_t n) const \ - { return iterator_name(wxString::AddToIter(m_cur, -(int)n)); } \ - iterator_name operator+=(int n) \ - { m_cur = wxString::AddToIter(m_cur, n); return *this; } \ - iterator_name operator+=(size_t n) \ - { m_cur = wxString::AddToIter(m_cur, (int)n); return *this; } \ - iterator_name operator-=(int n) \ - { m_cur = wxString::AddToIter(m_cur, -n); return *this; } \ - iterator_name operator-=(size_t n) \ - { m_cur = wxString::AddToIter(m_cur, -(int)n); return *this; } \ + iterator_name& operator+=(ptrdiff_t n) \ + { \ + m_cur = wxStringOperations::AddToIter(m_cur, n); \ + return *this; \ + } \ + iterator_name& operator-=(ptrdiff_t n) \ + { \ + m_cur = wxStringOperations::AddToIter(m_cur, -n); \ + return *this; \ + } \ \ - unsigned operator-(const iterator_name& i) const \ - { return wxString::DiffIters(m_cur, i.m_cur); } \ + difference_type operator-(const iterator_name& i) const \ + { return wxStringOperations::DiffIters(m_cur, i.m_cur); } \ \ - bool operator==(const iterator_name&i) const \ + bool operator==(const iterator_name& i) const \ { return m_cur == i.m_cur; } \ bool operator!=(const iterator_name& i) const \ { return m_cur != i.m_cur; } \ @@ -655,36 +953,174 @@ public: \ private: \ /* for internal wxString use only: */ \ - iterator_name(underlying_iterator ptr) : m_cur(ptr) {} \ - operator underlying_iterator() const { return m_cur; } \ + underlying_iterator impl() const { return m_cur; } \ \ - friend class WXDLLIMPEXP_BASE wxString; \ - friend class WXDLLIMPEXP_BASE wxCStrData; \ + friend class wxString; \ + friend class wxCStrData; \ \ private: \ - underlying_iterator m_cur; + underlying_iterator m_cur + + class WXDLLIMPEXP_FWD_BASE const_iterator; + +#if wxUSE_UNICODE_UTF8 + // NB: In UTF-8 build, (non-const) iterator needs to keep reference + // to the underlying wxStringImpl, because UTF-8 is variable-length + // encoding and changing the value pointer to by an iterator (using + // its operator*) requires calling wxStringImpl::replace() if the old + // and new values differ in their encoding's length. + // + // Furthermore, the replace() call may invalid all iterators for the + // string, so we have to keep track of outstanding iterators and update + // them if replace() happens. + // + // This is implemented by maintaining linked list of iterators for every + // string and traversing it in wxUniCharRef::operator=(). Head of the + // list is stored in wxString. (FIXME-UTF8) + + class WXDLLIMPEXP_BASE iterator + { + WX_STR_ITERATOR_IMPL(iterator, wxChar*, wxUniCharRef); + + public: + iterator() {} + iterator(const iterator& i) + : m_cur(i.m_cur), m_node(i.str(), &m_cur) {} + iterator& operator=(const iterator& i) + { + if (&i != this) + { + m_cur = i.m_cur; + m_node.set(i.str(), &m_cur); + } + return *this; + } + + reference operator*() + { return wxUniCharRef::CreateForString(*str(), m_cur); } + + iterator operator+(ptrdiff_t n) const + { return iterator(str(), wxStringOperations::AddToIter(m_cur, n)); } + iterator operator-(ptrdiff_t n) const + { return iterator(str(), wxStringOperations::AddToIter(m_cur, -n)); } + + private: + iterator(wxString *wxstr, underlying_iterator ptr) + : m_cur(ptr), m_node(wxstr, &m_cur) {} + + wxString* str() const { return const_cast(m_node.m_str); } + + wxStringIteratorNode m_node; + + friend class const_iterator; + }; + + class WXDLLIMPEXP_BASE const_iterator + { + // NB: reference_type is intentionally value, not reference, the character + // may be encoded differently in wxString data: + WX_STR_ITERATOR_IMPL(const_iterator, const wxChar*, wxUniChar); + + public: + const_iterator() {} + const_iterator(const const_iterator& i) + : m_cur(i.m_cur), m_node(i.str(), &m_cur) {} + const_iterator(const iterator& i) + : m_cur(i.m_cur), m_node(i.str(), &m_cur) {} - class const_iterator; + const_iterator& operator=(const const_iterator& i) + { + if (&i != this) + { + m_cur = i.m_cur; + m_node.set(i.str(), &m_cur); + } + return *this; + } + const_iterator& operator=(const iterator& i) + { m_cur = i.m_cur; m_node.set(i.str(), &m_cur); return *this; } + + reference operator*() const + { return wxStringOperations::DecodeChar(m_cur); } + + const_iterator operator+(ptrdiff_t n) const + { return const_iterator(str(), wxStringOperations::AddToIter(m_cur, n)); } + const_iterator operator-(ptrdiff_t n) const + { return const_iterator(str(), wxStringOperations::AddToIter(m_cur, -n)); } + + private: + // for internal wxString use only: + const_iterator(const wxString *wxstr, underlying_iterator ptr) + : m_cur(ptr), m_node(wxstr, &m_cur) {} + + const wxString* str() const { return m_node.m_str; } + + wxStringIteratorNode m_node; + }; + + size_t IterToImplPos(wxString::iterator i) const + { return wxStringImpl::const_iterator(i.impl()) - m_impl.begin(); } - class iterator + iterator GetIterForNthChar(size_t n) + { return iterator(this, m_impl.begin() + PosToImpl(n)); } + const_iterator GetIterForNthChar(size_t n) const + { return const_iterator(this, m_impl.begin() + PosToImpl(n)); } +#else // !wxUSE_UNICODE_UTF8 + + class WXDLLIMPEXP_BASE iterator { - WX_STR_ITERATOR_IMPL(iterator, wxChar*, wxUniCharRef, - wxUniCharRef::CreateForString(m_cur)) + WX_STR_ITERATOR_IMPL(iterator, wxChar*, wxUniCharRef); + + public: + iterator() {} + iterator(const iterator& i) : m_cur(i.m_cur) {} + + reference operator*() + { return wxUniCharRef::CreateForString(m_cur); } + + iterator operator+(ptrdiff_t n) const + { return iterator(wxStringOperations::AddToIter(m_cur, n)); } + iterator operator-(ptrdiff_t n) const + { return iterator(wxStringOperations::AddToIter(m_cur, -n)); } + + private: + // for internal wxString use only: + iterator(underlying_iterator ptr) : m_cur(ptr) {} + iterator(wxString *WXUNUSED(str), underlying_iterator ptr) : m_cur(ptr) {} friend class const_iterator; }; - class const_iterator + class WXDLLIMPEXP_BASE const_iterator { // NB: reference_type is intentionally value, not reference, the character // may be encoded differently in wxString data: - WX_STR_ITERATOR_IMPL(const_iterator, const wxChar*, wxUniChar, - wxUniChar(*m_cur)) + WX_STR_ITERATOR_IMPL(const_iterator, const wxChar*, wxUniChar); public: + const_iterator() {} + const_iterator(const const_iterator& i) : m_cur(i.m_cur) {} const_iterator(const iterator& i) : m_cur(i.m_cur) {} + + reference operator*() const + { return wxStringOperations::DecodeChar(m_cur); } + + const_iterator operator+(ptrdiff_t n) const + { return const_iterator(wxStringOperations::AddToIter(m_cur, n)); } + const_iterator operator-(ptrdiff_t n) const + { return const_iterator(wxStringOperations::AddToIter(m_cur, -n)); } + + private: + // for internal wxString use only: + const_iterator(underlying_iterator ptr) : m_cur(ptr) {} + const_iterator(const wxString *WXUNUSED(str), underlying_iterator ptr) + : m_cur(ptr) {} }; + iterator GetIterForNthChar(size_t n) { return begin() + n; } + const_iterator GetIterForNthChar(size_t n) const { return begin() + n; } +#endif // wxUSE_UNICODE_UTF8/!wxUSE_UNICODE_UTF8 + #undef WX_STR_ITERATOR_TAG #undef WX_STR_ITERATOR_IMPL @@ -697,12 +1133,13 @@ public: public: typedef T iterator_type; - typedef typename T::iterator_category iterator_category; + WX_DEFINE_ITERATOR_CATEGORY(typename T::iterator_category) typedef typename T::value_type value_type; typedef typename T::difference_type difference_type; typedef typename T::reference reference; typedef typename T::pointer *pointer; + reverse_iterator_impl() {} reverse_iterator_impl(iterator_type i) : m_cur(i) {} reverse_iterator_impl(const reverse_iterator_impl& ri) : m_cur(ri.m_cur) {} @@ -721,21 +1158,14 @@ public: reverse_iterator_impl operator--(int) { reverse_iterator_impl tmp = *this; ++m_cur; return tmp; } - reverse_iterator_impl operator+(int n) const - { return reverse_iterator_impl(m_cur - n); } - reverse_iterator_impl operator+(size_t n) const - { return reverse_iterator_impl(m_cur - n); } - reverse_iterator_impl operator-(int n) const - { return reverse_iterator_impl(m_cur + n); } - reverse_iterator_impl operator-(size_t n) const - { return reverse_iterator_impl(m_cur + n); } - reverse_iterator_impl operator+=(int n) + // NB: explicit in the functions below is to keep BCC 5.5 happy + reverse_iterator_impl operator+(ptrdiff_t n) const + { return reverse_iterator_impl(m_cur - n); } + reverse_iterator_impl operator-(ptrdiff_t n) const + { return reverse_iterator_impl(m_cur + n); } + reverse_iterator_impl operator+=(ptrdiff_t n) { m_cur -= n; return *this; } - reverse_iterator_impl operator+=(size_t n) - { m_cur -= n; return *this; } - reverse_iterator_impl operator-=(int n) - { m_cur += n; return *this; } - reverse_iterator_impl operator-=(size_t n) + reverse_iterator_impl operator-=(ptrdiff_t n) { m_cur += n; return *this; } unsigned operator-(const reverse_iterator_impl& i) const @@ -762,62 +1192,285 @@ public: typedef reverse_iterator_impl reverse_iterator; typedef reverse_iterator_impl const_reverse_iterator; - // first valid index position - const_iterator begin() const { return const_iterator(m_impl.begin()); } - iterator begin() { return iterator(m_impl.begin()); } - // position one after the last valid one - const_iterator end() const { return const_iterator(m_impl.end()); } - iterator end() { return iterator(m_impl.end()); } +private: + // used to transform an expression built using c_str() (and hence of type + // wxCStrData) to an iterator into the string + static const_iterator CreateConstIterator(const wxCStrData& data) + { + return const_iterator(data.m_str, + (data.m_str->begin() + data.m_offset).impl()); + } - // first element of the reversed string - const_reverse_iterator rbegin() const - { return const_reverse_iterator(end()); } - reverse_iterator rbegin() - { return reverse_iterator(end()); } - // one beyond the end of the reversed string - const_reverse_iterator rend() const - { return const_reverse_iterator(begin()); } - reverse_iterator rend() - { return reverse_iterator(begin()); } + // in UTF-8 STL build, creation from std::string requires conversion under + // non-UTF8 locales, so we can't have and use wxString(wxStringImpl) ctor; + // instead we define dummy type that lets us have wxString ctor for creation + // from wxStringImpl that couldn't be used by user code (in all other builds, + // "standard" ctors can be used): +#if wxUSE_UNICODE_UTF8 && wxUSE_STL_BASED_WXSTRING + struct CtorFromStringImplTag {}; - // std::string methods: -#if wxUSE_UNICODE_UTF8 - size_t length() const { return end() - begin(); } // FIXME-UTF8: optimize! + wxString(CtorFromStringImplTag* WXUNUSED(dummy), const wxStringImpl& src) + : m_impl(src) {} + + static wxString FromImpl(const wxStringImpl& src) + { return wxString((CtorFromStringImplTag*)NULL, src); } #else - size_t length() const { return m_impl.length(); } + #if !wxUSE_STL_BASED_WXSTRING + wxString(const wxStringImpl& src) : m_impl(src) { } + // else: already defined as wxString(wxStdString) below + #endif + static wxString FromImpl(const wxStringImpl& src) { return wxString(src); } #endif - size_type size() const { return length(); } - size_type max_size() const { return npos; } - - bool empty() const { return m_impl.empty(); } +public: + // constructors and destructor + // ctor for an empty string + wxString() {} + + // copy ctor + wxString(const wxString& stringSrc) : m_impl(stringSrc.m_impl) { } + + // string containing nRepeat copies of ch + wxString(wxUniChar ch, size_t nRepeat = 1 ) + { assign(nRepeat, ch); } + wxString(size_t nRepeat, wxUniChar ch) + { assign(nRepeat, ch); } + wxString(wxUniCharRef ch, size_t nRepeat = 1) + { assign(nRepeat, ch); } + wxString(size_t nRepeat, wxUniCharRef ch) + { assign(nRepeat, ch); } + wxString(char ch, size_t nRepeat = 1) + { assign(nRepeat, ch); } + wxString(size_t nRepeat, char ch) + { assign(nRepeat, ch); } + wxString(wchar_t ch, size_t nRepeat = 1) + { assign(nRepeat, ch); } + wxString(size_t nRepeat, wchar_t ch) + { assign(nRepeat, ch); } + + // ctors from char* strings: + wxString(const char *psz) + : m_impl(ImplStr(psz)) {} + wxString(const char *psz, const wxMBConv& conv) + : m_impl(ImplStr(psz, conv)) {} + wxString(const char *psz, size_t nLength) + { assign(psz, nLength); } + wxString(const char *psz, const wxMBConv& conv, size_t nLength) + { + SubstrBufFromMB str(ImplStr(psz, nLength, conv)); + m_impl.assign(str.data, str.len); + } + + // and unsigned char*: + wxString(const unsigned char *psz) + : m_impl(ImplStr((const char*)psz)) {} + wxString(const unsigned char *psz, const wxMBConv& conv) + : m_impl(ImplStr((const char*)psz, conv)) {} + wxString(const unsigned char *psz, size_t nLength) + { assign((const char*)psz, nLength); } + wxString(const unsigned char *psz, const wxMBConv& conv, size_t nLength) + { + SubstrBufFromMB str(ImplStr((const char*)psz, nLength, conv)); + m_impl.assign(str.data, str.len); + } + + // ctors from wchar_t* strings: + wxString(const wchar_t *pwz) + : m_impl(ImplStr(pwz)) {} + wxString(const wchar_t *pwz, const wxMBConv& WXUNUSED(conv)) + : m_impl(ImplStr(pwz)) {} + wxString(const wchar_t *pwz, size_t nLength) + { assign(pwz, nLength); } + wxString(const wchar_t *pwz, const wxMBConv& WXUNUSED(conv), size_t nLength) + { assign(pwz, nLength); } + + wxString(const wxScopedCharBuffer& buf) + { assign(buf.data(), buf.length()); } + wxString(const wxScopedWCharBuffer& buf) + { assign(buf.data(), buf.length()); } + + // NB: this version uses m_impl.c_str() to force making a copy of the + // string, so that "wxString(str.c_str())" idiom for passing strings + // between threads works + wxString(const wxCStrData& cstr) + : m_impl(cstr.AsString().m_impl.c_str()) { } + + // as we provide both ctors with this signature for both char and unsigned + // char string, we need to provide one for wxCStrData to resolve ambiguity + wxString(const wxCStrData& cstr, size_t nLength) + : m_impl(cstr.AsString().Mid(0, nLength).m_impl) {} + + // and because wxString is convertible to wxCStrData and const wxChar * + // we also need to provide this one + wxString(const wxString& str, size_t nLength) + { assign(str, nLength); } + + +#if wxUSE_STRING_POS_CACHE + ~wxString() + { + // we need to invalidate our cache entry as another string could be + // recreated at the same address (unlikely, but still possible, with the + // heap-allocated strings but perfectly common with stack-allocated ones) + InvalidateCache(); + } +#endif // wxUSE_STRING_POS_CACHE + + // even if we're not built with wxUSE_STL == 1 it is very convenient to allow + // implicit conversions from std::string to wxString and vice verse as this + // allows to use the same strings in non-GUI and GUI code, however we don't + // want to unconditionally add this ctor as it would make wx lib dependent on + // libstdc++ on some Linux versions which is bad, so instead we ask the + // client code to define this wxUSE_STD_STRING symbol if they need it +#if wxUSE_STD_STRING + #if wxUSE_UNICODE_WCHAR + wxString(const wxStdWideString& str) : m_impl(str) {} + #else // UTF-8 or ANSI + wxString(const wxStdWideString& str) + { assign(str.c_str(), str.length()); } + #endif + + #if !wxUSE_UNICODE // ANSI build + // FIXME-UTF8: do this in UTF8 build #if wxUSE_UTF8_LOCALE_ONLY, too + wxString(const std::string& str) : m_impl(str) {} + #else // Unicode + wxString(const std::string& str) + { assign(str.c_str(), str.length()); } + #endif +#endif // wxUSE_STD_STRING + + // Unlike ctor from std::string, we provide conversion to std::string only + // if wxUSE_STL and not merely wxUSE_STD_STRING (which is on by default), + // because it conflicts with operator const char/wchar_t*: +#if wxUSE_STL + #if wxUSE_UNICODE_WCHAR && wxUSE_STL_BASED_WXSTRING + // wxStringImpl is std::string in the encoding we want + operator const wxStdWideString&() const { return m_impl; } + #else + // wxStringImpl is either not std::string or needs conversion + operator wxStdWideString() const + { + wxScopedWCharBuffer buf(wc_str()); + return wxStdWideString(buf.data(), buf.length()); + } + #endif + + #if (!wxUSE_UNICODE || wxUSE_UTF8_LOCALE_ONLY) && wxUSE_STL_BASED_WXSTRING + // wxStringImpl is std::string in the encoding we want + operator const std::string&() const { return m_impl; } + #else + // wxStringImpl is either not std::string or needs conversion + operator std::string() const + { + wxScopedCharBuffer buf(mb_str()); + return std::string(buf.data(), buf.length()); + } + #endif +#endif // wxUSE_STL + + wxString Clone() const + { + // make a deep copy of the string, i.e. the returned string will have + // ref count = 1 with refcounted implementation + return wxString::FromImpl(wxStringImpl(m_impl.c_str(), m_impl.length())); + } + + // first valid index position + const_iterator begin() const { return const_iterator(this, m_impl.begin()); } + iterator begin() { return iterator(this, m_impl.begin()); } + // position one after the last valid one + const_iterator end() const { return const_iterator(this, m_impl.end()); } + iterator end() { return iterator(this, m_impl.end()); } + + // first element of the reversed string + const_reverse_iterator rbegin() const + { return const_reverse_iterator(end()); } + reverse_iterator rbegin() + { return reverse_iterator(end()); } + // one beyond the end of the reversed string + const_reverse_iterator rend() const + { return const_reverse_iterator(begin()); } + reverse_iterator rend() + { return reverse_iterator(begin()); } + + // std::string methods: +#if wxUSE_UNICODE_UTF8 + size_t length() const + { +#if wxUSE_STRING_POS_CACHE + wxCACHE_PROFILE_FIELD_INC(lentot); + + Cache::Element * const cache = GetCacheElement(); + + if ( cache->len == npos ) + { + // it's probably not worth trying to be clever and using cache->pos + // here as it's probably 0 anyhow -- you usually call length() before + // starting to index the string + cache->len = end() - begin(); + } + else + { + wxCACHE_PROFILE_FIELD_INC(lenhits); + + wxSTRING_CACHE_ASSERT( (int)cache->len == end() - begin() ); + } + + return cache->len; +#else // !wxUSE_STRING_POS_CACHE + return end() - begin(); +#endif // wxUSE_STRING_POS_CACHE/!wxUSE_STRING_POS_CACHE + } +#else + size_t length() const { return m_impl.length(); } +#endif + + size_type size() const { return length(); } + size_type max_size() const { return npos; } - size_type capacity() const { return m_impl.capacity(); } // FIXME-UTF8 - void reserve(size_t sz) { m_impl.reserve(sz); } // FIXME-UTF8 + bool empty() const { return m_impl.empty(); } + + // NB: these methods don't have a well-defined meaning in UTF-8 case + size_type capacity() const { return m_impl.capacity(); } + void reserve(size_t sz) { m_impl.reserve(sz); } void resize(size_t nSize, wxUniChar ch = wxT('\0')) { + const size_t len = length(); + if ( nSize == len) + return; + #if wxUSE_UNICODE_UTF8 + if ( nSize < len ) + { + wxSTRING_INVALIDATE_CACHE(); + + // we can't use wxStringImpl::resize() for truncating the string as it + // counts in bytes, not characters + erase(nSize); + return; + } + + // we also can't use (presumably more efficient) resize() if we have to + // append characters taking more than one byte if ( !ch.IsAscii() ) { - size_t len = length(); - if ( nSize == len) - return; - else if ( nSize < len ) - erase(nSize); - else - append(nSize - len, ch); + append(nSize - len, ch); } - else -#endif + else // can use (presumably faster) resize() version +#endif // wxUSE_UNICODE_UTF8 + { + wxSTRING_INVALIDATE_CACHED_LENGTH(); + m_impl.resize(nSize, (wxStringCharType)ch); + } } wxString substr(size_t nStart = 0, size_t nLen = npos) const { size_t pos, len; PosLenToImpl(nStart, nLen, &pos, &len); - return m_impl.substr(pos, len); + return FromImpl(m_impl.substr(pos, len)); } // generic attributes & operations @@ -830,18 +1483,9 @@ public: // truncate the string to given length wxString& Truncate(size_t uiLen); // empty string contents - void Empty() - { - Truncate(0); - - wxASSERT_MSG( empty(), _T("string not empty after call to Empty()?") ); - } + void Empty() { clear(); } // empty the string and free memory - void Clear() - { - wxString tmp(wxEmptyString); - swap(tmp); - } + void Clear() { clear(); } // contents test // Is an ascii value @@ -854,32 +1498,31 @@ public: // data access (all indexes are 0 based) // read access wxUniChar at(size_t n) const - { return *(begin() + n); } // FIXME-UTF8: optimize? + { return wxStringOperations::DecodeChar(m_impl.begin() + PosToImpl(n)); } wxUniChar GetChar(size_t n) const { return at(n); } // read/write access wxUniCharRef at(size_t n) - { return *(begin() + n); } // FIXME-UTF8: optimize? + { return *GetIterForNthChar(n); } wxUniCharRef GetWritableChar(size_t n) { return at(n); } // write access - void SetChar(size_t n, wxUniChar ch) + void SetChar(size_t n, wxUniChar ch) { at(n) = ch; } // get last character wxUniChar Last() const - { - wxASSERT_MSG( !empty(), _T("wxString: index out of bounds") ); - - return at(length() - 1); - } + { + wxASSERT_MSG( !empty(), wxT("wxString: index out of bounds") ); + return *rbegin(); + } // get writable last character wxUniCharRef Last() - { - wxASSERT_MSG( !empty(), _T("wxString: index out of bounds") ); - return at(length() - 1); - } + { + wxASSERT_MSG( !empty(), wxT("wxString: index out of bounds") ); + return *rbegin(); + } /* Note that we we must define all of the overloads below to avoid @@ -908,20 +1551,107 @@ public: { return at(n); } #endif // size_t != unsigned int - // explicit conversion to C string (use this with printf()!) + + /* + Overview of wxString conversions, implicit and explicit: + + - wxString has a std::[w]string-like c_str() method, however it does + not return a C-style string directly but instead returns wxCStrData + helper object which is convertible to either "char *" narrow string + or "wchar_t *" wide string. Usually the correct conversion will be + applied by the compiler automatically but if this doesn't happen you + need to explicitly choose one using wxCStrData::AsChar() or AsWChar() + methods or another wxString conversion function. + + - One of the places where the conversion does *NOT* happen correctly is + when c_str() is passed to a vararg function such as printf() so you + must *NOT* use c_str() with them. Either use wxPrintf() (all wx + functions do handle c_str() correctly, even if they appear to be + vararg (but they're not, really)) or add an explicit AsChar() or, if + compatibility with previous wxWidgets versions is important, add a + cast to "const char *". + + - In non-STL mode only, wxString is also implicitly convertible to + wxCStrData. The same warning as above applies. + + - c_str() is polymorphic as it can be converted to either narrow or + wide string. If you explicitly need one or the other, choose to use + mb_str() (for narrow) or wc_str() (for wide) instead. Notice that + these functions can return either the pointer to string directly (if + this is what the string uses internally) or a temporary buffer + containing the string and convertible to it. Again, conversion will + usually be done automatically by the compiler but beware of the + vararg functions: you need an explicit cast when using them. + + - There are also non-const versions of mb_str() and wc_str() called + char_str() and wchar_str(). They are only meant to be used with + non-const-correct functions and they always return buffers. + + - Finally wx_str() returns whatever string representation is used by + wxString internally. It may be either a narrow or wide string + depending on wxWidgets build mode but it will always be a raw pointer + (and not a buffer). + */ + + // explicit conversion to wxCStrData wxCStrData c_str() const { return wxCStrData(this); } wxCStrData data() const { return c_str(); } - // implicit conversion to C string + // implicit conversion to wxCStrData operator wxCStrData() const { return c_str(); } - operator const wxChar*() const { return c_str(); } + + // the first two operators conflict with operators for conversion to + // std::string and they must be disabled in STL build; the next one only + // makes sense if conversions to char* are also defined and not defining it + // in STL build also helps us to get more clear error messages for the code + // which relies on implicit conversion to char* in STL build +#if !wxUSE_STL + operator const char*() const { return c_str(); } + operator const wchar_t*() const { return c_str(); } + + // implicit conversion to untyped pointer for compatibility with previous + // wxWidgets versions: this is the same as conversion to const char * so it + // may fail! + operator const void*() const { return c_str(); } +#endif // wxUSE_STL // identical to c_str(), for MFC compatibility const wxCStrData GetData() const { return c_str(); } // explicit conversion to C string in internal representation (char*, // wchar_t*, UTF-8-encoded char*, depending on the build): - const_pointer wx_str() const { return m_impl.c_str(); } + const wxStringCharType *wx_str() const { return m_impl.c_str(); } + + // conversion to *non-const* multibyte or widestring buffer; modifying + // returned buffer won't affect the string, these methods are only useful + // for passing values to const-incorrect functions + wxWritableCharBuffer char_str(const wxMBConv& conv = wxConvLibc) const + { return mb_str(conv); } + wxWritableWCharBuffer wchar_str() const { return wc_str(); } + + // conversion to the buffer of the given type T (= char or wchar_t) and + // also optionally return the buffer length + // + // this is mostly/only useful for the template functions + // + // FIXME-VC6: the second argument only exists for VC6 which doesn't support + // explicit template function selection, do not use it unless + // you must support VC6! + template + wxCharTypeBuffer tchar_str(size_t *len = NULL, + T * WXUNUSED(dummy) = NULL) const + { +#if wxUSE_UNICODE + // we need a helper dispatcher depending on type + return wxPrivate::wxStringAsBufHelper::Get(*this, len); +#else // ANSI + // T can only be char in ANSI build + if ( len ) + *len = length(); + + return wxCharTypeBuffer::CreateNonOwned(wx_str(), length()); +#endif // Unicode build kind + } // conversion to/from plain (i.e. 7 bit) ASCII: this is useful for // converting numbers or strings which are certain not to contain special @@ -930,15 +1660,131 @@ public: // the behaviour of these functions with the strings containing anything // else than 7 bit ASCII characters is undefined, use at your own risk. #if wxUSE_UNICODE - static wxString FromAscii(const char *ascii); // string - static wxString FromAscii(const char ascii); // char - const wxCharBuffer ToAscii() const; + static wxString FromAscii(const char *ascii, size_t len); + static wxString FromAscii(const char *ascii); + static wxString FromAscii(char ascii); + const wxScopedCharBuffer ToAscii() const; #else // ANSI static wxString FromAscii(const char *ascii) { return wxString( ascii ); } - static wxString FromAscii(const char ascii) { return wxString( ascii ); } + static wxString FromAscii(const char *ascii, size_t len) + { return wxString( ascii, len ); } + static wxString FromAscii(char ascii) { return wxString( ascii ); } const char *ToAscii() const { return c_str(); } #endif // Unicode/!Unicode + // also provide unsigned char overloads as signed/unsigned doesn't matter + // for 7 bit ASCII characters + static wxString FromAscii(const unsigned char *ascii) + { return FromAscii((const char *)ascii); } + static wxString FromAscii(const unsigned char *ascii, size_t len) + { return FromAscii((const char *)ascii, len); } + + // conversion to/from UTF-8: +#if wxUSE_UNICODE_UTF8 + static wxString FromUTF8Unchecked(const char *utf8) + { + if ( !utf8 ) + return wxEmptyString; + + wxASSERT( wxStringOperations::IsValidUtf8String(utf8) ); + return FromImpl(wxStringImpl(utf8)); + } + static wxString FromUTF8Unchecked(const char *utf8, size_t len) + { + if ( !utf8 ) + return wxEmptyString; + if ( len == npos ) + return FromUTF8Unchecked(utf8); + + wxASSERT( wxStringOperations::IsValidUtf8String(utf8, len) ); + return FromImpl(wxStringImpl(utf8, len)); + } + + static wxString FromUTF8(const char *utf8) + { + if ( !utf8 || !wxStringOperations::IsValidUtf8String(utf8) ) + return ""; + + return FromImpl(wxStringImpl(utf8)); + } + static wxString FromUTF8(const char *utf8, size_t len) + { + if ( len == npos ) + return FromUTF8(utf8); + + if ( !utf8 || !wxStringOperations::IsValidUtf8String(utf8, len) ) + return ""; + + return FromImpl(wxStringImpl(utf8, len)); + } + + const wxScopedCharBuffer utf8_str() const + { return wxCharBuffer::CreateNonOwned(m_impl.c_str(), m_impl.length()); } + + // this function exists in UTF-8 build only and returns the length of the + // internal UTF-8 representation + size_t utf8_length() const { return m_impl.length(); } +#elif wxUSE_UNICODE_WCHAR + static wxString FromUTF8(const char *utf8, size_t len = npos) + { return wxString(utf8, wxMBConvUTF8(), len); } + static wxString FromUTF8Unchecked(const char *utf8, size_t len = npos) + { + const wxString s(utf8, wxMBConvUTF8(), len); + wxASSERT_MSG( !utf8 || !*utf8 || !s.empty(), + "string must be valid UTF-8" ); + return s; + } + const wxScopedCharBuffer utf8_str() const { return mb_str(wxMBConvUTF8()); } +#else // ANSI + static wxString FromUTF8(const char *utf8) + { return wxString(wxMBConvUTF8().cMB2WC(utf8)); } + static wxString FromUTF8(const char *utf8, size_t len) + { + size_t wlen; + wxScopedWCharBuffer buf(wxMBConvUTF8().cMB2WC(utf8, len == npos ? wxNO_LEN : len, &wlen)); + return wxString(buf.data(), wlen); + } + static wxString FromUTF8Unchecked(const char *utf8, size_t len = npos) + { + size_t wlen; + wxScopedWCharBuffer buf + ( + wxMBConvUTF8().cMB2WC + ( + utf8, + len == npos ? wxNO_LEN : len, + &wlen + ) + ); + wxASSERT_MSG( !utf8 || !*utf8 || wlen, + "string must be valid UTF-8" ); + + return wxString(buf.data(), wlen); + } + const wxScopedCharBuffer utf8_str() const + { return wxMBConvUTF8().cWC2MB(wc_str()); } +#endif + + const wxScopedCharBuffer ToUTF8() const { return utf8_str(); } + + // functions for storing binary data in wxString: +#if wxUSE_UNICODE + static wxString From8BitData(const char *data, size_t len) + { return wxString(data, wxConvISO8859_1, len); } + // version for NUL-terminated data: + static wxString From8BitData(const char *data) + { return wxString(data, wxConvISO8859_1); } + const wxScopedCharBuffer To8BitData() const + { return mb_str(wxConvISO8859_1); } +#else // ANSI + static wxString From8BitData(const char *data, size_t len) + { return wxString(data, len); } + // version for NUL-terminated data: + static wxString From8BitData(const char *data) + { return wxString(data); } + const char *To8BitData() const { return c_str(); } +#endif // Unicode/ANSI + // conversions with (possible) format conversions: have to return a // buffer with temporary data // @@ -948,82 +1794,162 @@ public: // accepting the file names. The return value is always the same, but the // type differs because a function may either return pointer to the buffer // directly or have to use intermediate buffer for translation. + #if wxUSE_UNICODE - const wxCharBuffer mb_str(const wxMBConv& conv = wxConvLibc) const; - const wxWX2MBbuf mbc_str() const { return mb_str(*wxConvCurrent); } + // this is an optimization: even though using mb_str(wxConvLibc) does the + // same thing (i.e. returns pointer to internal representation as locale is + // always an UTF-8 one) in wxUSE_UTF8_LOCALE_ONLY case, we can avoid the + // extra checks and the temporary buffer construction by providing a + // separate mb_str() overload +#if wxUSE_UTF8_LOCALE_ONLY + const char* mb_str() const { return wx_str(); } + const wxScopedCharBuffer mb_str(const wxMBConv& conv) const + { + return AsCharBuf(conv); + } +#else // !wxUSE_UTF8_LOCALE_ONLY + const wxScopedCharBuffer mb_str(const wxMBConv& conv = wxConvLibc) const + { + return AsCharBuf(conv); + } +#endif // wxUSE_UTF8_LOCALE_ONLY/!wxUSE_UTF8_LOCALE_ONLY - const wxChar* wc_str() const { return c_str(); } + const wxWX2MBbuf mbc_str() const { return mb_str(*wxConvCurrent); } +#if wxUSE_UNICODE_WCHAR + const wchar_t* wc_str() const { return wx_str(); } +#elif wxUSE_UNICODE_UTF8 + const wxScopedWCharBuffer wc_str() const + { return AsWCharBuf(wxMBConvStrictUTF8()); } +#endif // for compatibility with !wxUSE_UNICODE version - const wxChar* wc_str(const wxMBConv& WXUNUSED(conv)) const { return c_str(); } + const wxWX2WCbuf wc_str(const wxMBConv& WXUNUSED(conv)) const + { return wc_str(); } #if wxMBFILES - const wxCharBuffer fn_str() const { return mb_str(wxConvFile); } + const wxScopedCharBuffer fn_str() const { return mb_str(wxConvFile); } #else // !wxMBFILES - const wxChar* fn_str() const { return c_str(); } + const wxWX2WCbuf fn_str() const { return wc_str(); } #endif // wxMBFILES/!wxMBFILES + #else // ANSI - const wxChar* mb_str() const { return c_str(); } + const char* mb_str() const { return wx_str(); } // for compatibility with wxUSE_UNICODE version - const wxChar* mb_str(const wxMBConv& WXUNUSED(conv)) const { return c_str(); } + const char* mb_str(const wxMBConv& WXUNUSED(conv)) const { return wx_str(); } const wxWX2MBbuf mbc_str() const { return mb_str(); } -#if wxUSE_WCHAR_T - const wxWCharBuffer wc_str(const wxMBConv& conv) const; -#endif // wxUSE_WCHAR_T -#ifdef __WXOSX__ - const wxCharBuffer fn_str() const { return wxConvFile.cWC2WX( wc_str( wxConvLocal ) ); } + const wxScopedWCharBuffer wc_str(const wxMBConv& conv = wxConvLibc) const + { return AsWCharBuf(conv); } + + const wxScopedCharBuffer fn_str() const + { return wxConvFile.cWC2WX( wc_str( wxConvLibc ) ); } +#endif // Unicode/ANSI + +#if wxUSE_UNICODE_UTF8 + const wxScopedWCharBuffer t_str() const { return wc_str(); } +#elif wxUSE_UNICODE_WCHAR + const wchar_t* t_str() const { return wx_str(); } #else - const wxChar* fn_str() const { return c_str(); } + const char* t_str() const { return wx_str(); } #endif -#endif // Unicode/ANSI + // overloaded assignment // from another wxString - wxString& operator=(const wxStringImpl& stringSrc) - { m_impl = stringSrc; return *this; } + wxString& operator=(const wxString& stringSrc) + { + if ( this != &stringSrc ) + { + wxSTRING_INVALIDATE_CACHE(); + + m_impl = stringSrc.m_impl; + } + + return *this; + } + wxString& operator=(const wxCStrData& cstr) { return *this = cstr.AsString(); } // from a character wxString& operator=(wxUniChar ch) - { m_impl = EncodeChar(ch); return *this; } + { + wxSTRING_INVALIDATE_CACHE(); + +#if wxUSE_UNICODE_UTF8 + if ( !ch.IsAscii() ) + m_impl = wxStringOperations::EncodeChar(ch); + else +#endif // wxUSE_UNICODE_UTF8 + m_impl = (wxStringCharType)ch; + return *this; + } + wxString& operator=(wxUniCharRef ch) { return operator=((wxUniChar)ch); } wxString& operator=(char ch) { return operator=(wxUniChar(ch)); } + wxString& operator=(unsigned char ch) + { return operator=(wxUniChar(ch)); } wxString& operator=(wchar_t ch) { return operator=(wxUniChar(ch)); } // from a C string - STL probably will crash on NULL, // so we need to compensate in that case #if wxUSE_STL_BASED_WXSTRING - wxString& operator=(const wxChar *psz) - { if(psz) m_impl = psz; else Clear(); return *this; } -#else - wxString& operator=(const wxChar *psz) - { m_impl = psz; return *this; } -#endif + wxString& operator=(const char *psz) + { + wxSTRING_INVALIDATE_CACHE(); -#if wxUSE_UNICODE - // from wxWCharBuffer - wxString& operator=(const wxWCharBuffer& s) - { (void) operator=((const wchar_t *)s); return *this; } - // from C string - wxString& operator=(const char* psz) - { return operator=(wxString(psz)); } -#else // ANSI - // from another kind of C string - wxString& operator=(const unsigned char* psz); -#if wxUSE_WCHAR_T - // from a wide string - wxString& operator=(const wchar_t *pwz); -#endif - // from wxCharBuffer - wxString& operator=(const wxCharBuffer& psz) - { (void) operator=((const char *)psz); return *this; } -#endif // Unicode/ANSI + if ( psz ) + m_impl = ImplStr(psz); + else + clear(); + + return *this; + } + + wxString& operator=(const wchar_t *pwz) + { + wxSTRING_INVALIDATE_CACHE(); + + if ( pwz ) + m_impl = ImplStr(pwz); + else + clear(); + + return *this; + } +#else // !wxUSE_STL_BASED_WXSTRING + wxString& operator=(const char *psz) + { + wxSTRING_INVALIDATE_CACHE(); + + m_impl = ImplStr(psz); + + return *this; + } + + wxString& operator=(const wchar_t *pwz) + { + wxSTRING_INVALIDATE_CACHE(); + + m_impl = ImplStr(pwz); + + return *this; + } +#endif // wxUSE_STL_BASED_WXSTRING/!wxUSE_STL_BASED_WXSTRING + + wxString& operator=(const unsigned char *psz) + { return operator=((const char*)psz); } + + // from wxScopedWCharBuffer + wxString& operator=(const wxScopedWCharBuffer& s) + { return assign(s); } + // from wxScopedCharBuffer + wxString& operator=(const wxScopedCharBuffer& s) + { return assign(s); } // string concatenation // in place concatenation @@ -1037,7 +1963,7 @@ public: { #if WXWIN_COMPATIBILITY_2_8 && !wxUSE_STL_BASED_WXSTRING && !wxUSE_UNICODE_UTF8 wxASSERT_MSG( s.IsValid(), - _T("did you forget to call UngetWriteBuf()?") ); + wxT("did you forget to call UngetWriteBuf()?") ); #endif append(s); @@ -1054,18 +1980,14 @@ public: wxString& operator<<(wxUniChar ch) { append(1, ch); return *this; } wxString& operator<<(wxUniCharRef ch) { append(1, ch); return *this; } wxString& operator<<(char ch) { append(1, ch); return *this; } + wxString& operator<<(unsigned char ch) { append(1, ch); return *this; } wxString& operator<<(wchar_t ch) { append(1, ch); return *this; } // string += buffer (i.e. from wxGetString) - wxString& operator<<(const wxWCharBuffer& s) - { return operator<<((const wchar_t *)s); } - wxString& operator+=(const wxWCharBuffer& s) - { return operator<<((const wchar_t *)s); } - - wxString& operator<<(const wxCharBuffer& s) - { return operator<<((const char *)s); } - wxString& operator+=(const wxCharBuffer& s) - { return operator<<((const char *)s); } + wxString& operator<<(const wxScopedWCharBuffer& s) + { return append(s); } + wxString& operator<<(const wxScopedCharBuffer& s) + { return append(s); } // string += C string wxString& Append(const wxString& s) @@ -1077,12 +1999,26 @@ public: append(s); return *this; } - wxString& Append(const wxCStrData& psz) - { append(psz); return *this; } wxString& Append(const char* psz) { append(psz); return *this; } wxString& Append(const wchar_t* pwz) { append(pwz); return *this; } + wxString& Append(const wxCStrData& psz) + { append(psz); return *this; } + wxString& Append(const wxScopedCharBuffer& psz) + { append(psz); return *this; } + wxString& Append(const wxScopedWCharBuffer& psz) + { append(psz); return *this; } + wxString& Append(const char* psz, size_t nLen) + { append(psz, nLen); return *this; } + wxString& Append(const wchar_t* pwz, size_t nLen) + { append(pwz, nLen); return *this; } + wxString& Append(const wxCStrData& psz, size_t nLen) + { append(psz, nLen); return *this; } + wxString& Append(const wxScopedCharBuffer& psz, size_t nLen) + { append(psz, nLen); return *this; } + wxString& Append(const wxScopedWCharBuffer& psz, size_t nLen) + { append(psz, nLen); return *this; } // append count copies of given character wxString& Append(wxUniChar ch, size_t count = 1u) { append(count, ch); return *this; } @@ -1090,12 +2026,10 @@ public: { append(count, ch); return *this; } wxString& Append(char ch, size_t count = 1u) { append(count, ch); return *this; } + wxString& Append(unsigned char ch, size_t count = 1u) + { append(count, ch); return *this; } wxString& Append(wchar_t ch, size_t count = 1u) { append(count, ch); return *this; } - wxString& Append(const char* psz, size_t nLen) - { append(psz, nLen); return *this; } - wxString& Append(const wchar_t* pwz, size_t nLen) - { append(pwz, nLen); return *this; } // prepend a string, return the string itself wxString& Prepend(const wxString& str) @@ -1123,36 +2057,34 @@ public: // stream-like functions // insert an int into string wxString& operator<<(int i) - { return (*this) << Format(_T("%d"), i); } + { return (*this) << Format(wxT("%d"), i); } // insert an unsigned int into string wxString& operator<<(unsigned int ui) - { return (*this) << Format(_T("%u"), ui); } + { return (*this) << Format(wxT("%u"), ui); } // insert a long into string wxString& operator<<(long l) - { return (*this) << Format(_T("%ld"), l); } + { return (*this) << Format(wxT("%ld"), l); } // insert an unsigned long into string wxString& operator<<(unsigned long ul) - { return (*this) << Format(_T("%lu"), ul); } + { return (*this) << Format(wxT("%lu"), ul); } #if defined wxLongLong_t && !defined wxLongLongIsLong // insert a long long if they exist and aren't longs wxString& operator<<(wxLongLong_t ll) { - const wxChar *fmt = _T("%") wxLongLongFmtSpec _T("d"); - return (*this) << Format(fmt, ll); + return (*this) << Format("%" wxLongLongFmtSpec "d", ll); } // insert an unsigned long long wxString& operator<<(wxULongLong_t ull) { - const wxChar *fmt = _T("%") wxLongLongFmtSpec _T("u"); - return (*this) << Format(fmt , ull); + return (*this) << Format("%" wxLongLongFmtSpec "u" , ull); } -#endif +#endif // wxLongLong_t && !wxLongLongIsLong // insert a float into string wxString& operator<<(float f) - { return (*this) << Format(_T("%f"), f); } + { return (*this) << Format(wxT("%f"), f); } // insert a double into string wxString& operator<<(double d) - { return (*this) << Format(_T("%g"), d); } + { return (*this) << Format(wxT("%g"), d); } // string comparison // case-sensitive comparison (returns a value < 0, = 0 or > 0) @@ -1162,24 +2094,50 @@ public: { return compare(pwz); } int Cmp(const wxString& s) const { return compare(s); } + int Cmp(const wxCStrData& s) const + { return compare(s); } + int Cmp(const wxScopedCharBuffer& s) const + { return compare(s); } + int Cmp(const wxScopedWCharBuffer& s) const + { return compare(s); } // same as Cmp() but not case-sensitive int CmpNoCase(const wxString& s) const; - int CmpNoCase(const char *psz) const - { return CmpNoCase(wxString(psz)); } - int CmpNoCase(const wchar_t *pwz) const - { return CmpNoCase(wxString(pwz)); } + // test for the string equality, either considering case or not // (if compareWithCase then the case matters) - bool IsSameAs(const char *psz, bool compareWithCase = true) const - { return (compareWithCase ? Cmp(psz) : CmpNoCase(psz)) == 0; } - bool IsSameAs(const wchar_t *pwz, bool compareWithCase = true) const - { return (compareWithCase ? Cmp(pwz) : CmpNoCase(pwz)) == 0; } + bool IsSameAs(const wxString& str, bool compareWithCase = true) const + { +#if !wxUSE_UNICODE_UTF8 + // in UTF-8 build, length() is O(n) and doing this would be _slower_ + if ( length() != str.length() ) + return false; +#endif + return (compareWithCase ? Cmp(str) : CmpNoCase(str)) == 0; + } + bool IsSameAs(const char *str, bool compareWithCase = true) const + { return (compareWithCase ? Cmp(str) : CmpNoCase(str)) == 0; } + bool IsSameAs(const wchar_t *str, bool compareWithCase = true) const + { return (compareWithCase ? Cmp(str) : CmpNoCase(str)) == 0; } + + bool IsSameAs(const wxCStrData& str, bool compareWithCase = true) const + { return IsSameAs(str.AsString(), compareWithCase); } + bool IsSameAs(const wxScopedCharBuffer& str, bool compareWithCase = true) const + { return IsSameAs(str.data(), compareWithCase); } + bool IsSameAs(const wxScopedWCharBuffer& str, bool compareWithCase = true) const + { return IsSameAs(str.data(), compareWithCase); } // comparison with a single character: returns true if equal - bool IsSameAs(wxUniChar c, bool compareWithCase = true) const - { - return (length() == 1) && (compareWithCase ? GetChar(0u) == c - : wxToupper(GetChar(0u)) == wxToupper(c)); - } + bool IsSameAs(wxUniChar c, bool compareWithCase = true) const; + // FIXME-UTF8: remove these overloads + bool IsSameAs(wxUniCharRef c, bool compareWithCase = true) const + { return IsSameAs(wxUniChar(c), compareWithCase); } + bool IsSameAs(char c, bool compareWithCase = true) const + { return IsSameAs(wxUniChar(c), compareWithCase); } + bool IsSameAs(unsigned char c, bool compareWithCase = true) const + { return IsSameAs(wxUniChar(c), compareWithCase); } + bool IsSameAs(wchar_t c, bool compareWithCase = true) const + { return IsSameAs(wxUniChar(c), compareWithCase); } + bool IsSameAs(int c, bool compareWithCase = true) const + { return IsSameAs(wxUniChar(c), compareWithCase); } // simple sub-string extraction // return substring starting at nFirst of length nCount (or till the end @@ -1193,26 +2151,26 @@ public: // check if the string starts with the given prefix and return the rest // of the string in the provided pointer if it is not NULL; otherwise // return false - bool StartsWith(const wxChar *prefix, wxString *rest = NULL) const; + bool StartsWith(const wxString& prefix, wxString *rest = NULL) const; // check if the string ends with the given suffix and return the // beginning of the string before the suffix in the provided pointer if // it is not NULL; otherwise return false - bool EndsWith(const wxChar *suffix, wxString *rest = NULL) const; + bool EndsWith(const wxString& suffix, wxString *rest = NULL) const; // get first nCount characters wxString Left(size_t nCount) const; // get last nCount characters wxString Right(size_t nCount) const; - // get all characters before the first occurance of ch + // get all characters before the first occurrence of ch // (returns the whole string if ch not found) wxString BeforeFirst(wxUniChar ch) const; - // get all characters before the last occurence of ch + // get all characters before the last occurrence of ch // (returns empty string if ch not found) wxString BeforeLast(wxUniChar ch) const; - // get all characters after the first occurence of ch + // get all characters after the first occurrence of ch // (returns empty string if ch not found) wxString AfterFirst(wxUniChar ch) const; - // get all characters after the last occurence of ch + // get all characters after the last occurrence of ch // (returns the whole string if ch not found) wxString AfterLast(wxUniChar ch) const; @@ -1224,12 +2182,17 @@ public: // convert to upper case in place, return the string itself wxString& MakeUpper(); // convert to upper case, return the copy of the string - // Here's something to remember: BC++ doesn't like returns in inlines. - wxString Upper() const ; + wxString Upper() const { return wxString(*this).MakeUpper(); } // convert to lower case in place, return the string itself wxString& MakeLower(); // convert to lower case, return the copy of the string - wxString Lower() const ; + wxString Lower() const { return wxString(*this).MakeLower(); } + // convert the first character to the upper case and the rest to the + // lower one, return the modified string itself + wxString& MakeCapitalized(); + // convert the first character to the upper case and the rest to the + // lower one, return the copy of the string + wxString Capitalize() const { return wxString(*this).MakeCapitalized(); } // trimming/padding whitespace (either side) and truncating // remove spaces from left or from right (default) side @@ -1240,50 +2203,113 @@ public: // searching and replacing // searching (return starting index, or -1 if not found) int Find(wxUniChar ch, bool bFromEnd = false) const; // like strchr/strrchr + int Find(wxUniCharRef ch, bool bFromEnd = false) const + { return Find(wxUniChar(ch), bFromEnd); } + int Find(char ch, bool bFromEnd = false) const + { return Find(wxUniChar(ch), bFromEnd); } + int Find(unsigned char ch, bool bFromEnd = false) const + { return Find(wxUniChar(ch), bFromEnd); } + int Find(wchar_t ch, bool bFromEnd = false) const + { return Find(wxUniChar(ch), bFromEnd); } // searching (return starting index, or -1 if not found) - int Find(const wxChar *pszSub) const; // like strstr - // replace first (or all of bReplaceAll) occurences of substring with + int Find(const wxString& sub) const // like strstr + { + size_type idx = find(sub); + return (idx == npos) ? wxNOT_FOUND : (int)idx; + } + int Find(const char *sub) const // like strstr + { + size_type idx = find(sub); + return (idx == npos) ? wxNOT_FOUND : (int)idx; + } + int Find(const wchar_t *sub) const // like strstr + { + size_type idx = find(sub); + return (idx == npos) ? wxNOT_FOUND : (int)idx; + } + + int Find(const wxCStrData& sub) const + { return Find(sub.AsString()); } + int Find(const wxScopedCharBuffer& sub) const + { return Find(sub.data()); } + int Find(const wxScopedWCharBuffer& sub) const + { return Find(sub.data()); } + + // replace first (or all of bReplaceAll) occurrences of substring with // another string, returns the number of replacements made - size_t Replace(const wxChar *szOld, - const wxChar *szNew, + size_t Replace(const wxString& strOld, + const wxString& strNew, bool bReplaceAll = true); // check if the string contents matches a mask containing '*' and '?' - bool Matches(const wxChar *szMask) const; - - // conversion to numbers: all functions return true only if the whole - // string is a number and put the value of this number into the pointer - // provided, the base is the numeric base in which the conversion should be - // done and must be comprised between 2 and 36 or be 0 in which case the - // standard C rules apply (leading '0' => octal, "0x" => hex) - // convert to a signed integer - bool ToLong(long *val, int base = 10) const; - // convert to an unsigned integer - bool ToULong(unsigned long *val, int base = 10) const; - // convert to wxLongLong + bool Matches(const wxString& mask) const; + + // conversion to numbers: all functions return true only if the whole + // string is a number and put the value of this number into the pointer + // provided, the base is the numeric base in which the conversion should be + // done and must be comprised between 2 and 36 or be 0 in which case the + // standard C rules apply (leading '0' => octal, "0x" => hex) + // convert to a signed integer + bool ToLong(long *val, int base = 10) const; + // convert to an unsigned integer + bool ToULong(unsigned long *val, int base = 10) const; + // convert to wxLongLong #if defined(wxLongLong_t) - bool ToLongLong(wxLongLong_t *val, int base = 10) const; - // convert to wxULongLong - bool ToULongLong(wxULongLong_t *val, int base = 10) const; + bool ToLongLong(wxLongLong_t *val, int base = 10) const; + // convert to wxULongLong + bool ToULongLong(wxULongLong_t *val, int base = 10) const; #endif // wxLongLong_t - // convert to a double - bool ToDouble(double *val) const; - + // convert to a double + bool ToDouble(double *val) const; + +#if wxUSE_XLOCALE + // conversions to numbers using C locale + // convert to a signed integer + bool ToCLong(long *val, int base = 10) const; + // convert to an unsigned integer + bool ToCULong(unsigned long *val, int base = 10) const; + // convert to a double + bool ToCDouble(double *val) const; +#endif #ifndef wxNEEDS_WXSTRING_PRINTF_MIXIN // formatted input/output // as sprintf(), returns the number of characters written or < 0 on error // (take 'this' into account in attribute parameter count) - // int Printf(const wxChar *pszFormat, ...); - WX_DEFINE_VARARG_FUNC(int, Printf, DoPrintf) + // int Printf(const wxString& format, ...); + WX_DEFINE_VARARG_FUNC(int, Printf, 1, (const wxFormatString&), + DoPrintfWchar, DoPrintfUtf8) +#ifdef __WATCOMC__ + // workaround for http://bugzilla.openwatcom.org/show_bug.cgi?id=351 + WX_VARARG_WATCOM_WORKAROUND(int, Printf, 1, (const wxString&), + (wxFormatString(f1))); + WX_VARARG_WATCOM_WORKAROUND(int, Printf, 1, (const wxCStrData&), + (wxFormatString(f1))); + WX_VARARG_WATCOM_WORKAROUND(int, Printf, 1, (const char*), + (wxFormatString(f1))); + WX_VARARG_WATCOM_WORKAROUND(int, Printf, 1, (const wchar_t*), + (wxFormatString(f1))); +#endif #endif // !wxNEEDS_WXSTRING_PRINTF_MIXIN // as vprintf(), returns the number of characters written or < 0 on error int PrintfV(const wxString& format, va_list argptr); #ifndef wxNEEDS_WXSTRING_PRINTF_MIXIN // returns the string containing the result of Printf() to it - // static wxString Format(const wxChar *pszFormat, ...) ATTRIBUTE_PRINTF_1; - WX_DEFINE_VARARG_FUNC(static wxString, Format, DoFormat) + // static wxString Format(const wxString& format, ...) WX_ATTRIBUTE_PRINTF_1; + WX_DEFINE_VARARG_FUNC(static wxString, Format, 1, (const wxFormatString&), + DoFormatWchar, DoFormatUtf8) +#ifdef __WATCOMC__ + // workaround for http://bugzilla.openwatcom.org/show_bug.cgi?id=351 + WX_VARARG_WATCOM_WORKAROUND(static wxString, Format, 1, (const wxString&), + (wxFormatString(f1))); + WX_VARARG_WATCOM_WORKAROUND(static wxString, Format, 1, (const wxCStrData&), + (wxFormatString(f1))); + WX_VARARG_WATCOM_WORKAROUND(static wxString, Format, 1, (const char*), + (wxFormatString(f1))); + WX_VARARG_WATCOM_WORKAROUND(static wxString, Format, 1, (const wchar_t*), + (wxFormatString(f1))); +#endif #endif // the same as above, but takes a va_list static wxString FormatV(const wxString& format, va_list argptr); @@ -1291,7 +2317,7 @@ public: // raw access to string memory // ensure that string has space for at least nLen characters // only works if the data of this string is not shared - bool Alloc(size_t nLen) { reserve(nLen); /*return capacity() >= nLen;*/ return true; } + bool Alloc(size_t nLen) { reserve(nLen); return capacity() >= nLen; } // minimize the string's memory // only works if the data of this string is not shared bool Shrink(); @@ -1300,7 +2326,7 @@ public: // // get writable buffer of at least nLen bytes. Unget() *must* be called // a.s.a.p. to put string back in a reasonable state! - wxDEPRECATED( wxChar *GetWriteBuf(size_t nLen) ); + wxDEPRECATED( wxStringCharType *GetWriteBuf(size_t nLen) ); // call this immediately after GetWriteBuf() has been used wxDEPRECATED( void UngetWriteBuf() ); wxDEPRECATED( void UngetWriteBuf(size_t nLen) ); @@ -1319,12 +2345,24 @@ public: #ifndef wxNEEDS_WXSTRING_PRINTF_MIXIN // use Printf() // (take 'this' into account in attribute parameter count) - // int sprintf(const wxChar *pszFormat, ...) ATTRIBUTE_PRINTF_2; - WX_DEFINE_VARARG_FUNC(int, sprintf, DoPrintf) + // int sprintf(const wxString& format, ...) WX_ATTRIBUTE_PRINTF_2; + WX_DEFINE_VARARG_FUNC(int, sprintf, 1, (const wxFormatString&), + DoPrintfWchar, DoPrintfUtf8) +#ifdef __WATCOMC__ + // workaround for http://bugzilla.openwatcom.org/show_bug.cgi?id=351 + WX_VARARG_WATCOM_WORKAROUND(int, sprintf, 1, (const wxString&), + (wxFormatString(f1))); + WX_VARARG_WATCOM_WORKAROUND(int, sprintf, 1, (const wxCStrData&), + (wxFormatString(f1))); + WX_VARARG_WATCOM_WORKAROUND(int, sprintf, 1, (const char*), + (wxFormatString(f1))); + WX_VARARG_WATCOM_WORKAROUND(int, sprintf, 1, (const wchar_t*), + (wxFormatString(f1))); +#endif #endif // wxNEEDS_WXSTRING_PRINTF_MIXIN // use Cmp() - inline int CompareTo(const wxChar* psz, caseCompare cmp = exact) const + int CompareTo(const wxChar* psz, caseCompare cmp = exact) const { return cmp == exact ? Cmp(psz) : CmpNoCase(psz); } // use length() @@ -1350,10 +2388,11 @@ public: // use Find() int First( wxUniChar ch ) const { return Find(ch); } + int First( wxUniCharRef ch ) const { return Find(ch); } int First( char ch ) const { return Find(ch); } + int First( unsigned char ch ) const { return Find(ch); } int First( wchar_t ch ) const { return Find(ch); } - int First( const wxChar* psz ) const { return Find(psz); } - int First( const wxString &str ) const { return Find(str); } + int First( const wxString& str ) const { return Find(str); } int Last( wxUniChar ch ) const { return Find(ch, true); } bool Contains(const wxString& str) const { return Find(str) != wxNOT_FOUND; } @@ -1364,114 +2403,289 @@ public: // take nLen chars starting at nPos wxString(const wxString& str, size_t nPos, size_t nLen) - : m_impl(str.m_impl, nPos, nLen) { } - // take all characters from pStart to pEnd - wxString(const void *pStart, const void *pEnd) - : m_impl((const wxChar*)pStart, (const wxChar*)pEnd) { } + { assign(str, nPos, nLen); } + // take all characters from first to last wxString(const_iterator first, const_iterator last) - : m_impl(first, last) { } - wxString(iterator first, iterator last) - : m_impl(first, last) { } + : m_impl(first.impl(), last.impl()) { } +#if WXWIN_COMPATIBILITY_STRING_PTR_AS_ITER + // the 2 overloads below are for compatibility with the existing code using + // pointers instead of iterators + wxString(const char *first, const char *last) + { + SubstrBufFromMB str(ImplStr(first, last - first)); + m_impl.assign(str.data, str.len); + } + wxString(const wchar_t *first, const wchar_t *last) + { + SubstrBufFromWC str(ImplStr(first, last - first)); + m_impl.assign(str.data, str.len); + } + // and this one is needed to compile code adding offsets to c_str() result + wxString(const wxCStrData& first, const wxCStrData& last) + : m_impl(CreateConstIterator(first).impl(), + CreateConstIterator(last).impl()) + { + wxASSERT_MSG( first.m_str == last.m_str, + wxT("pointers must be into the same string") ); + } +#endif // WXWIN_COMPATIBILITY_STRING_PTR_AS_ITER // lib.string.modifiers // append elements str[pos], ..., str[pos+n] wxString& append(const wxString& str, size_t pos, size_t n) { - size_t from, len; - str.PosLenToImpl(pos, n, &from, &len); - m_impl.append(str.m_impl, from, len); - return *this; + wxSTRING_UPDATE_CACHED_LENGTH(n); + + size_t from, len; + str.PosLenToImpl(pos, n, &from, &len); + m_impl.append(str.m_impl, from, len); + return *this; } // append a string wxString& append(const wxString& str) - { m_impl.append(str.m_impl); return *this; } - wxString& append(const wxCStrData& str) - { m_impl.append(str.AsString().m_impl); return *this; } + { + wxSTRING_UPDATE_CACHED_LENGTH(str.length()); + + m_impl.append(str.m_impl); + return *this; + } + // append first n (or all if n == npos) characters of sz wxString& append(const char *sz) - { m_impl.append(ImplStr(sz)); return *this; } + { + wxSTRING_INVALIDATE_CACHED_LENGTH(); + + m_impl.append(ImplStr(sz)); + return *this; + } + wxString& append(const wchar_t *sz) - { m_impl.append(ImplStr(sz)); return *this; } + { + wxSTRING_INVALIDATE_CACHED_LENGTH(); + + m_impl.append(ImplStr(sz)); + return *this; + } + wxString& append(const char *sz, size_t n) { - SubstrBufFromMB str(ImplStr(sz, n)); - m_impl.append(str.data, str.len); - return *this; + wxSTRING_INVALIDATE_CACHED_LENGTH(); + + SubstrBufFromMB str(ImplStr(sz, n)); + m_impl.append(str.data, str.len); + return *this; } wxString& append(const wchar_t *sz, size_t n) { - SubstrBufFromWC str(ImplStr(sz, n)); - m_impl.append(str.data, str.len); - return *this; + wxSTRING_UPDATE_CACHED_LENGTH(n); + + SubstrBufFromWC str(ImplStr(sz, n)); + m_impl.append(str.data, str.len); + return *this; } + + wxString& append(const wxCStrData& str) + { return append(str.AsString()); } + wxString& append(const wxScopedCharBuffer& str) + { return append(str.data(), str.length()); } + wxString& append(const wxScopedWCharBuffer& str) + { return append(str.data(), str.length()); } + wxString& append(const wxCStrData& str, size_t n) + { return append(str.AsString(), 0, n); } + wxString& append(const wxScopedCharBuffer& str, size_t n) + { return append(str.data(), n); } + wxString& append(const wxScopedWCharBuffer& str, size_t n) + { return append(str.data(), n); } + // append n copies of ch wxString& append(size_t n, wxUniChar ch) { #if wxUSE_UNICODE_UTF8 - if ( !ch.IsAscii() ) - m_impl.append(EncodeNChars(n, ch)); - else + if ( !ch.IsAscii() ) + { + wxSTRING_INVALIDATE_CACHED_LENGTH(); + + m_impl.append(wxStringOperations::EncodeNChars(n, ch)); + } + else // ASCII #endif - m_impl.append(n, (wxStringCharType)ch); - return *this; + { + wxSTRING_UPDATE_CACHED_LENGTH(n); + + m_impl.append(n, (wxStringCharType)ch); + } + + return *this; } + + wxString& append(size_t n, wxUniCharRef ch) + { return append(n, wxUniChar(ch)); } + wxString& append(size_t n, char ch) + { return append(n, wxUniChar(ch)); } + wxString& append(size_t n, unsigned char ch) + { return append(n, wxUniChar(ch)); } + wxString& append(size_t n, wchar_t ch) + { return append(n, wxUniChar(ch)); } + // append from first to last wxString& append(const_iterator first, const_iterator last) - { m_impl.append(first, last); return *this; } + { + wxSTRING_INVALIDATE_CACHED_LENGTH(); + + m_impl.append(first.impl(), last.impl()); + return *this; + } +#if WXWIN_COMPATIBILITY_STRING_PTR_AS_ITER + wxString& append(const char *first, const char *last) + { return append(first, last - first); } + wxString& append(const wchar_t *first, const wchar_t *last) + { return append(first, last - first); } + wxString& append(const wxCStrData& first, const wxCStrData& last) + { return append(CreateConstIterator(first), CreateConstIterator(last)); } +#endif // WXWIN_COMPATIBILITY_STRING_PTR_AS_ITER // same as `this_string = str' wxString& assign(const wxString& str) - { m_impl = str.m_impl; return *this; } + { + wxSTRING_SET_CACHED_LENGTH(str.length()); + + m_impl = str.m_impl; + + return *this; + } + + wxString& assign(const wxString& str, size_t len) + { + wxSTRING_SET_CACHED_LENGTH(len); + + m_impl.assign(str.m_impl, 0, str.LenToImpl(len)); + + return *this; + } + // same as ` = str[pos..pos + n] wxString& assign(const wxString& str, size_t pos, size_t n) { - size_t from, len; - str.PosLenToImpl(pos, n, &from, &len); - m_impl.assign(str.m_impl, from, len); - return *this; + size_t from, len; + str.PosLenToImpl(pos, n, &from, &len); + m_impl.assign(str.m_impl, from, len); + + // it's important to call this after PosLenToImpl() above in case str is + // the same string as this one + wxSTRING_SET_CACHED_LENGTH(n); + + return *this; } + // same as `= first n (or all if n == npos) characters of sz' wxString& assign(const char *sz) - { m_impl.assign(ImplStr(sz)); return *this; } + { + wxSTRING_INVALIDATE_CACHE(); + + m_impl.assign(ImplStr(sz)); + + return *this; + } + wxString& assign(const wchar_t *sz) - { m_impl.assign(ImplStr(sz)); return *this; } + { + wxSTRING_INVALIDATE_CACHE(); + + m_impl.assign(ImplStr(sz)); + + return *this; + } + wxString& assign(const char *sz, size_t n) { - SubstrBufFromMB str(ImplStr(sz, n)); - m_impl.assign(str.data, str.len); - return *this; + wxSTRING_SET_CACHED_LENGTH(n); + + SubstrBufFromMB str(ImplStr(sz, n)); + m_impl.assign(str.data, str.len); + + return *this; } + wxString& assign(const wchar_t *sz, size_t n) { - SubstrBufFromWC str(ImplStr(sz, n)); - m_impl.assign(str.data, str.len); - return *this; + wxSTRING_SET_CACHED_LENGTH(n); + + SubstrBufFromWC str(ImplStr(sz, n)); + m_impl.assign(str.data, str.len); + + return *this; } + + wxString& assign(const wxCStrData& str) + { return assign(str.AsString()); } + wxString& assign(const wxScopedCharBuffer& str) + { return assign(str.data(), str.length()); } + wxString& assign(const wxScopedWCharBuffer& str) + { return assign(str.data(), str.length()); } + wxString& assign(const wxCStrData& str, size_t len) + { return assign(str.AsString(), len); } + wxString& assign(const wxScopedCharBuffer& str, size_t len) + { return assign(str.data(), len); } + wxString& assign(const wxScopedWCharBuffer& str, size_t len) + { return assign(str.data(), len); } + // same as `= n copies of ch' wxString& assign(size_t n, wxUniChar ch) { + wxSTRING_SET_CACHED_LENGTH(n); + #if wxUSE_UNICODE_UTF8 - if ( !ch.IsAscii() ) - m_impl.assign(EncodeNChars(n, ch)); - else + if ( !ch.IsAscii() ) + m_impl.assign(wxStringOperations::EncodeNChars(n, ch)); + else #endif - m_impl.assign(n, (wxStringCharType)ch); - return *this; + m_impl.assign(n, (wxStringCharType)ch); + + return *this; } + + wxString& assign(size_t n, wxUniCharRef ch) + { return assign(n, wxUniChar(ch)); } + wxString& assign(size_t n, char ch) + { return assign(n, wxUniChar(ch)); } + wxString& assign(size_t n, unsigned char ch) + { return assign(n, wxUniChar(ch)); } + wxString& assign(size_t n, wchar_t ch) + { return assign(n, wxUniChar(ch)); } + // assign from first to last wxString& assign(const_iterator first, const_iterator last) - { m_impl.assign(first, last); return *this; } + { + wxSTRING_INVALIDATE_CACHE(); + + m_impl.assign(first.impl(), last.impl()); + + return *this; + } +#if WXWIN_COMPATIBILITY_STRING_PTR_AS_ITER + wxString& assign(const char *first, const char *last) + { return assign(first, last - first); } + wxString& assign(const wchar_t *first, const wchar_t *last) + { return assign(first, last - first); } + wxString& assign(const wxCStrData& first, const wxCStrData& last) + { return assign(CreateConstIterator(first), CreateConstIterator(last)); } +#endif // WXWIN_COMPATIBILITY_STRING_PTR_AS_ITER // string comparison int compare(const wxString& str) const; + int compare(const char* sz) const; + int compare(const wchar_t* sz) const; + int compare(const wxCStrData& str) const + { return compare(str.AsString()); } + int compare(const wxScopedCharBuffer& str) const + { return compare(str.data()); } + int compare(const wxScopedWCharBuffer& str) const + { return compare(str.data()); } // comparison with a substring int compare(size_t nStart, size_t nLen, const wxString& str) const; // comparison of 2 substrings int compare(size_t nStart, size_t nLen, const wxString& str, size_t nStart2, size_t nLen2) const; - // just like strcmp() - int compare(const char* sz) const; - int compare(const wchar_t* sz) const; // substring comparison with first nCount characters of sz int compare(size_t nStart, size_t nLen, const char* sz, size_t nCount = npos) const; @@ -1480,183 +2694,356 @@ public: // insert another string wxString& insert(size_t nPos, const wxString& str) - { insert(begin() + nPos, str.begin(), str.end()); return *this; } + { insert(GetIterForNthChar(nPos), str.begin(), str.end()); return *this; } // insert n chars of str starting at nStart (in str) wxString& insert(size_t nPos, const wxString& str, size_t nStart, size_t n) { - size_t from, len; - str.PosLenToImpl(nStart, n, &from, &len); - m_impl.insert(PosToImpl(nPos), str.m_impl, from, len); - return *this; + wxSTRING_UPDATE_CACHED_LENGTH(n); + + size_t from, len; + str.PosLenToImpl(nStart, n, &from, &len); + m_impl.insert(PosToImpl(nPos), str.m_impl, from, len); + + return *this; } + // insert first n (or all if n == npos) characters of sz wxString& insert(size_t nPos, const char *sz) - { m_impl.insert(PosToImpl(nPos), ImplStr(sz)); return *this; } + { + wxSTRING_INVALIDATE_CACHE(); + + m_impl.insert(PosToImpl(nPos), ImplStr(sz)); + + return *this; + } + wxString& insert(size_t nPos, const wchar_t *sz) - { m_impl.insert(PosToImpl(nPos), ImplStr(sz)); return *this; } + { + wxSTRING_INVALIDATE_CACHE(); + + m_impl.insert(PosToImpl(nPos), ImplStr(sz)); return *this; + } + wxString& insert(size_t nPos, const char *sz, size_t n) { - SubstrBufFromMB str(ImplStr(sz, n)); - m_impl.insert(PosToImpl(nPos), str.data, str.len); - return *this; + wxSTRING_UPDATE_CACHED_LENGTH(n); + + SubstrBufFromMB str(ImplStr(sz, n)); + m_impl.insert(PosToImpl(nPos), str.data, str.len); + + return *this; } + wxString& insert(size_t nPos, const wchar_t *sz, size_t n) { - SubstrBufFromWC str(ImplStr(sz, n)); - m_impl.insert(PosToImpl(nPos), str.data, str.len); - return *this; + wxSTRING_UPDATE_CACHED_LENGTH(n); + + SubstrBufFromWC str(ImplStr(sz, n)); + m_impl.insert(PosToImpl(nPos), str.data, str.len); + + return *this; } + // insert n copies of ch wxString& insert(size_t nPos, size_t n, wxUniChar ch) { + wxSTRING_UPDATE_CACHED_LENGTH(n); + #if wxUSE_UNICODE_UTF8 - if ( !ch.IsAscii() ) - m_impl.insert(begin() + nPos, EncodeNChars(n, ch)); - else + if ( !ch.IsAscii() ) + m_impl.insert(PosToImpl(nPos), wxStringOperations::EncodeNChars(n, ch)); + else #endif - m_impl.insert(begin() + nPos, n, (wxStringCharType)ch); - return *this; + m_impl.insert(PosToImpl(nPos), n, (wxStringCharType)ch); + return *this; } + iterator insert(iterator it, wxUniChar ch) - { return iterator(m_impl.insert(it, EncodeChar(ch))); } + { + wxSTRING_UPDATE_CACHED_LENGTH(1); + +#if wxUSE_UNICODE_UTF8 + if ( !ch.IsAscii() ) + { + size_t pos = IterToImplPos(it); + m_impl.insert(pos, wxStringOperations::EncodeChar(ch)); + return iterator(this, m_impl.begin() + pos); + } + else +#endif + return iterator(this, m_impl.insert(it.impl(), (wxStringCharType)ch)); + } + void insert(iterator it, const_iterator first, const_iterator last) - { m_impl.insert(it, first, last); } + { + wxSTRING_INVALIDATE_CACHE(); + + m_impl.insert(it.impl(), first.impl(), last.impl()); + } + +#if WXWIN_COMPATIBILITY_STRING_PTR_AS_ITER + void insert(iterator it, const char *first, const char *last) + { insert(it - begin(), first, last - first); } + void insert(iterator it, const wchar_t *first, const wchar_t *last) + { insert(it - begin(), first, last - first); } + void insert(iterator it, const wxCStrData& first, const wxCStrData& last) + { insert(it, CreateConstIterator(first), CreateConstIterator(last)); } +#endif // WXWIN_COMPATIBILITY_STRING_PTR_AS_ITER + void insert(iterator it, size_type n, wxUniChar ch) { + wxSTRING_UPDATE_CACHED_LENGTH(n); + #if wxUSE_UNICODE_UTF8 - if ( !ch.IsAscii() ) - m_impl.insert(it, EncodeNChars(n, ch)); - else + if ( !ch.IsAscii() ) + m_impl.insert(IterToImplPos(it), wxStringOperations::EncodeNChars(n, ch)); + else #endif - m_impl.insert(it, n, (wxStringCharType)ch); + m_impl.insert(it.impl(), n, (wxStringCharType)ch); } // delete characters from nStart to nStart + nLen wxString& erase(size_type pos = 0, size_type n = npos) { - size_t from, len; - PosLenToImpl(pos, n, &from, &len); - m_impl.erase(from, len); - return *this; + wxSTRING_INVALIDATE_CACHE(); + + size_t from, len; + PosLenToImpl(pos, n, &from, &len); + m_impl.erase(from, len); + + return *this; } + + // delete characters from first up to last iterator erase(iterator first, iterator last) - { return iterator(m_impl.erase(first, last)); } + { + wxSTRING_INVALIDATE_CACHE(); + + return iterator(this, m_impl.erase(first.impl(), last.impl())); + } + iterator erase(iterator first) - { return iterator(m_impl.erase(first)); } + { + wxSTRING_UPDATE_CACHED_LENGTH(-1); + + return iterator(this, m_impl.erase(first.impl())); + } #ifdef wxSTRING_BASE_HASNT_CLEAR void clear() { erase(); } #else - void clear() { m_impl.clear(); } + void clear() + { + wxSTRING_SET_CACHED_LENGTH(0); + + m_impl.clear(); + } #endif // replaces the substring of length nLen starting at nStart wxString& replace(size_t nStart, size_t nLen, const char* sz) { - size_t from, len; - PosLenToImpl(nStart, nLen, &from, &len); - m_impl.replace(from, len, ImplStr(sz)); - return *this; + wxSTRING_INVALIDATE_CACHE(); + + size_t from, len; + PosLenToImpl(nStart, nLen, &from, &len); + m_impl.replace(from, len, ImplStr(sz)); + + return *this; } + wxString& replace(size_t nStart, size_t nLen, const wchar_t* sz) { - size_t from, len; - PosLenToImpl(nStart, nLen, &from, &len); - m_impl.replace(from, len, ImplStr(sz)); - return *this; + wxSTRING_INVALIDATE_CACHE(); + + size_t from, len; + PosLenToImpl(nStart, nLen, &from, &len); + m_impl.replace(from, len, ImplStr(sz)); + + return *this; } + // replaces the substring of length nLen starting at nStart wxString& replace(size_t nStart, size_t nLen, const wxString& str) { - size_t from, len; - PosLenToImpl(nStart, nLen, &from, &len); - m_impl.replace(from, len, str.m_impl); - return *this; + wxSTRING_INVALIDATE_CACHE(); + + size_t from, len; + PosLenToImpl(nStart, nLen, &from, &len); + m_impl.replace(from, len, str.m_impl); + + return *this; } + // replaces the substring with nCount copies of ch wxString& replace(size_t nStart, size_t nLen, size_t nCount, wxUniChar ch) { - size_t from, len; - PosLenToImpl(nStart, nLen, &from, &len); + wxSTRING_INVALIDATE_CACHE(); + + size_t from, len; + PosLenToImpl(nStart, nLen, &from, &len); #if wxUSE_UNICODE_UTF8 - if ( !ch.IsAscii() ) - m_impl.replace(from, len, EncodeNChars(nCount, ch)); - else + if ( !ch.IsAscii() ) + m_impl.replace(from, len, wxStringOperations::EncodeNChars(nCount, ch)); + else #endif - m_impl.replace(from, len, nCount, (wxStringCharType)ch); - return *this; + m_impl.replace(from, len, nCount, (wxStringCharType)ch); + + return *this; } + // replaces a substring with another substring wxString& replace(size_t nStart, size_t nLen, const wxString& str, size_t nStart2, size_t nLen2) { - size_t from, len; - PosLenToImpl(nStart, nLen, &from, &len); + wxSTRING_INVALIDATE_CACHE(); - size_t from2, len2; - str.PosLenToImpl(nStart2, nLen2, &from2, &len2); + size_t from, len; + PosLenToImpl(nStart, nLen, &from, &len); - m_impl.replace(from, len, str.m_impl, from2, len2); - return *this; + size_t from2, len2; + str.PosLenToImpl(nStart2, nLen2, &from2, &len2); + + m_impl.replace(from, len, str.m_impl, from2, len2); + + return *this; } + // replaces the substring with first nCount chars of sz wxString& replace(size_t nStart, size_t nLen, const char* sz, size_t nCount) { - size_t from, len; - PosLenToImpl(nStart, nLen, &from, &len); + wxSTRING_INVALIDATE_CACHE(); - SubstrBufFromMB str(ImplStr(sz, nCount)); + size_t from, len; + PosLenToImpl(nStart, nLen, &from, &len); - m_impl.replace(from, len, str.data, str.len); - return *this; + SubstrBufFromMB str(ImplStr(sz, nCount)); + + m_impl.replace(from, len, str.data, str.len); + + return *this; } + wxString& replace(size_t nStart, size_t nLen, const wchar_t* sz, size_t nCount) { - size_t from, len; - PosLenToImpl(nStart, nLen, &from, &len); + wxSTRING_INVALIDATE_CACHE(); - SubstrBufFromWC str(ImplStr(sz, nCount)); + size_t from, len; + PosLenToImpl(nStart, nLen, &from, &len); - m_impl.replace(from, len, str.data, str.len); - return *this; + SubstrBufFromWC str(ImplStr(sz, nCount)); + + m_impl.replace(from, len, str.data, str.len); + + return *this; + } + + wxString& replace(size_t nStart, size_t nLen, + const wxString& s, size_t nCount) + { + wxSTRING_INVALIDATE_CACHE(); + + size_t from, len; + PosLenToImpl(nStart, nLen, &from, &len); + m_impl.replace(from, len, s.m_impl.c_str(), s.LenToImpl(nCount)); + + return *this; } + wxString& replace(iterator first, iterator last, const char* s) - { m_impl.replace(first, last, ImplStr(s)); return *this; } + { + wxSTRING_INVALIDATE_CACHE(); + + m_impl.replace(first.impl(), last.impl(), ImplStr(s)); + + return *this; + } + wxString& replace(iterator first, iterator last, const wchar_t* s) - { m_impl.replace(first, last, ImplStr(s)); return *this; } + { + wxSTRING_INVALIDATE_CACHE(); + + m_impl.replace(first.impl(), last.impl(), ImplStr(s)); + + return *this; + } + wxString& replace(iterator first, iterator last, const char* s, size_type n) { - SubstrBufFromMB str(ImplStr(s, n)); - m_impl.replace(first, last, str.data, str.len); - return *this; + wxSTRING_INVALIDATE_CACHE(); + + SubstrBufFromMB str(ImplStr(s, n)); + m_impl.replace(first.impl(), last.impl(), str.data, str.len); + + return *this; } + wxString& replace(iterator first, iterator last, const wchar_t* s, size_type n) { - SubstrBufFromWC str(ImplStr(s, n)); - m_impl.replace(first, last, str.data, str.len); - return *this; + wxSTRING_INVALIDATE_CACHE(); + + SubstrBufFromWC str(ImplStr(s, n)); + m_impl.replace(first.impl(), last.impl(), str.data, str.len); + + return *this; } + wxString& replace(iterator first, iterator last, const wxString& s) - { m_impl.replace(first, last, s.m_impl); return *this; } + { + wxSTRING_INVALIDATE_CACHE(); + + m_impl.replace(first.impl(), last.impl(), s.m_impl); + + return *this; + } + wxString& replace(iterator first, iterator last, size_type n, wxUniChar ch) { + wxSTRING_INVALIDATE_CACHE(); + #if wxUSE_UNICODE_UTF8 - if ( !ch.IsAscii() ) - m_impl.replace(first, last, EncodeNChars(n, ch)); - else + if ( !ch.IsAscii() ) + m_impl.replace(first.impl(), last.impl(), + wxStringOperations::EncodeNChars(n, ch)); + else #endif - m_impl.replace(first, last, n, (wxStringCharType)ch); - return *this; + m_impl.replace(first.impl(), last.impl(), n, (wxStringCharType)ch); + + return *this; } + wxString& replace(iterator first, iterator last, const_iterator first1, const_iterator last1) - { m_impl.replace(first, last, first1, last1); return *this; } + { + wxSTRING_INVALIDATE_CACHE(); + + m_impl.replace(first.impl(), last.impl(), first1.impl(), last1.impl()); + + return *this; + } + + wxString& replace(iterator first, iterator last, + const char *first1, const char *last1) + { replace(first, last, first1, last1 - first1); return *this; } + wxString& replace(iterator first, iterator last, + const wchar_t *first1, const wchar_t *last1) + { replace(first, last, first1, last1 - first1); return *this; } // swap two strings void swap(wxString& str) - { m_impl.swap(str.m_impl); } + { +#if wxUSE_STRING_POS_CACHE + // we modify not only this string but also the other one directly so we + // need to invalidate cache for both of them (we could also try to + // exchange their cache entries but it seems unlikely to be worth it) + InvalidateCache(); + str.InvalidateCache(); +#endif // wxUSE_STRING_POS_CACHE + + m_impl.swap(str.m_impl); + } // find a substring size_t find(const wxString& str, size_t nStart = 0) const @@ -1673,16 +3060,34 @@ public: SubstrBufFromWC str(ImplStr(sz, n)); return PosFromImpl(m_impl.find(str.data, PosToImpl(nStart), str.len)); } - - // find the first occurence of character ch after nStart + size_t find(const wxScopedCharBuffer& s, size_t nStart = 0, size_t n = npos) const + { return find(s.data(), nStart, n); } + size_t find(const wxScopedWCharBuffer& s, size_t nStart = 0, size_t n = npos) const + { return find(s.data(), nStart, n); } + size_t find(const wxCStrData& s, size_t nStart = 0, size_t n = npos) const + { return find(s.AsWChar(), nStart, n); } + + // find the first occurrence of character ch after nStart size_t find(wxUniChar ch, size_t nStart = 0) const - { return PosFromImpl(m_impl.find(EncodeChar(ch), PosToImpl(nStart))); } - size_t find(wxUniCharRef ch, size_t nStart = 0) const - { return find(wxUniChar(ch), nStart); } - size_t find(char ch, size_t nStart = 0) const - { return find(wxUniChar(ch), nStart); } - size_t find(wchar_t ch, size_t nStart = 0) const - { return find(wxUniChar(ch), nStart); } + { +#if wxUSE_UNICODE_UTF8 + if ( !ch.IsAscii() ) + return PosFromImpl(m_impl.find(wxStringOperations::EncodeChar(ch), + PosToImpl(nStart))); + else +#endif + return PosFromImpl(m_impl.find((wxStringCharType)ch, + PosToImpl(nStart))); + + } + size_t find(wxUniCharRef ch, size_t nStart = 0) const + { return find(wxUniChar(ch), nStart); } + size_t find(char ch, size_t nStart = 0) const + { return find(wxUniChar(ch), nStart); } + size_t find(unsigned char ch, size_t nStart = 0) const + { return find(wxUniChar(ch), nStart); } + size_t find(wchar_t ch, size_t nStart = 0) const + { return find(wxUniChar(ch), nStart); } // rfind() family is exactly like find() but works right to left @@ -1701,17 +3106,34 @@ public: SubstrBufFromWC str(ImplStr(sz, n)); return PosFromImpl(m_impl.rfind(str.data, PosToImpl(nStart), str.len)); } + size_t rfind(const wxScopedCharBuffer& s, size_t nStart = npos, size_t n = npos) const + { return rfind(s.data(), nStart, n); } + size_t rfind(const wxScopedWCharBuffer& s, size_t nStart = npos, size_t n = npos) const + { return rfind(s.data(), nStart, n); } + size_t rfind(const wxCStrData& s, size_t nStart = npos, size_t n = npos) const + { return rfind(s.AsWChar(), nStart, n); } // as find, but from the end size_t rfind(wxUniChar ch, size_t nStart = npos) const - { return PosFromImpl(m_impl.rfind(EncodeChar(ch), PosToImpl(nStart))); } + { +#if wxUSE_UNICODE_UTF8 + if ( !ch.IsAscii() ) + return PosFromImpl(m_impl.rfind(wxStringOperations::EncodeChar(ch), + PosToImpl(nStart))); + else +#endif + return PosFromImpl(m_impl.rfind((wxStringCharType)ch, + PosToImpl(nStart))); + } size_t rfind(wxUniCharRef ch, size_t nStart = npos) const { return rfind(wxUniChar(ch), nStart); } size_t rfind(char ch, size_t nStart = npos) const { return rfind(wxUniChar(ch), nStart); } + size_t rfind(unsigned char ch, size_t nStart = npos) const + { return rfind(wxUniChar(ch), nStart); } size_t rfind(wchar_t ch, size_t nStart = npos) const { return rfind(wxUniChar(ch), nStart); } - // find first/last occurence of any character (not) in the set: + // find first/last occurrence of any character (not) in the set: #if wxUSE_STL_BASED_WXSTRING && !wxUSE_UNICODE_UTF8 // FIXME-UTF8: this is not entirely correct, because it doesn't work if // sizeof(wchar_t)==2 and surrogates are present in the string; @@ -1773,7 +3195,11 @@ public: // as strpbrk() but starts at nStart, returns npos if not found size_t find_first_of(const wxString& str, size_t nStart = 0) const - { return find_first_of((const wxChar*)str.c_str(), nStart); } +#if wxUSE_UNICODE // FIXME-UTF8: temporary + { return find_first_of(str.wc_str(), nStart); } +#else + { return find_first_of(str.mb_str(), nStart); } +#endif // same as above size_t find_first_of(const char* sz, size_t nStart = 0) const; size_t find_first_of(const wchar_t* sz, size_t nStart = 0) const; @@ -1784,7 +3210,11 @@ public: { return find(c, nStart); } // find the last (starting from nStart) char from str in this string size_t find_last_of (const wxString& str, size_t nStart = npos) const - { return find_last_of((const wxChar*)str.c_str(), nStart); } +#if wxUSE_UNICODE // FIXME-UTF8: temporary + { return find_last_of(str.wc_str(), nStart); } +#else + { return find_last_of(str.mb_str(), nStart); } +#endif // same as above size_t find_last_of (const char* sz, size_t nStart = npos) const; size_t find_last_of (const wchar_t* sz, size_t nStart = npos) const; @@ -1794,11 +3224,15 @@ public: size_t find_last_of(wxUniChar c, size_t nStart = npos) const { return rfind(c, nStart); } - // find first/last occurence of any character not in the set + // find first/last occurrence of any character not in the set // as strspn() (starting from nStart), returns npos on failure size_t find_first_not_of(const wxString& str, size_t nStart = 0) const - { return find_first_not_of((const wxChar*)str.c_str(), nStart); } +#if wxUSE_UNICODE // FIXME-UTF8: temporary + { return find_first_not_of(str.wc_str(), nStart); } +#else + { return find_first_not_of(str.mb_str(), nStart); } +#endif // same as above size_t find_first_not_of(const char* sz, size_t nStart = 0) const; size_t find_first_not_of(const wchar_t* sz, size_t nStart = 0) const; @@ -1808,7 +3242,11 @@ public: size_t find_first_not_of(wxUniChar ch, size_t nStart = 0) const; // as strcspn() size_t find_last_not_of(const wxString& str, size_t nStart = npos) const - { return find_last_not_of((const wxChar*)str.c_str(), nStart); } +#if wxUSE_UNICODE // FIXME-UTF8: temporary + { return find_last_not_of(str.wc_str(), nStart); } +#else + { return find_last_not_of(str.mb_str(), nStart); } +#endif // same as above size_t find_last_not_of(const char* sz, size_t nStart = npos) const; size_t find_last_not_of(const wchar_t* sz, size_t nStart = npos) const; @@ -1824,40 +3262,135 @@ public: { return find_first_of(wxUniChar(ch), nStart); } size_t find_first_of(char ch, size_t nStart = 0) const { return find_first_of(wxUniChar(ch), nStart); } + size_t find_first_of(unsigned char ch, size_t nStart = 0) const + { return find_first_of(wxUniChar(ch), nStart); } size_t find_first_of(wchar_t ch, size_t nStart = 0) const { return find_first_of(wxUniChar(ch), nStart); } size_t find_last_of(wxUniCharRef ch, size_t nStart = npos) const { return find_last_of(wxUniChar(ch), nStart); } size_t find_last_of(char ch, size_t nStart = npos) const { return find_last_of(wxUniChar(ch), nStart); } + size_t find_last_of(unsigned char ch, size_t nStart = npos) const + { return find_last_of(wxUniChar(ch), nStart); } size_t find_last_of(wchar_t ch, size_t nStart = npos) const { return find_last_of(wxUniChar(ch), nStart); } size_t find_first_not_of(wxUniCharRef ch, size_t nStart = 0) const { return find_first_not_of(wxUniChar(ch), nStart); } size_t find_first_not_of(char ch, size_t nStart = 0) const { return find_first_not_of(wxUniChar(ch), nStart); } + size_t find_first_not_of(unsigned char ch, size_t nStart = 0) const + { return find_first_not_of(wxUniChar(ch), nStart); } size_t find_first_not_of(wchar_t ch, size_t nStart = 0) const { return find_first_not_of(wxUniChar(ch), nStart); } size_t find_last_not_of(wxUniCharRef ch, size_t nStart = npos) const { return find_last_not_of(wxUniChar(ch), nStart); } size_t find_last_not_of(char ch, size_t nStart = npos) const { return find_last_not_of(wxUniChar(ch), nStart); } + size_t find_last_not_of(unsigned char ch, size_t nStart = npos) const + { return find_last_not_of(wxUniChar(ch), nStart); } size_t find_last_not_of(wchar_t ch, size_t nStart = npos) const { return find_last_not_of(wxUniChar(ch), nStart); } + // and additional overloads for the versions taking strings: + size_t find_first_of(const wxCStrData& sz, size_t nStart = 0) const + { return find_first_of(sz.AsString(), nStart); } + size_t find_first_of(const wxScopedCharBuffer& sz, size_t nStart = 0) const + { return find_first_of(sz.data(), nStart); } + size_t find_first_of(const wxScopedWCharBuffer& sz, size_t nStart = 0) const + { return find_first_of(sz.data(), nStart); } + size_t find_first_of(const wxCStrData& sz, size_t nStart, size_t n) const + { return find_first_of(sz.AsWChar(), nStart, n); } + size_t find_first_of(const wxScopedCharBuffer& sz, size_t nStart, size_t n) const + { return find_first_of(sz.data(), nStart, n); } + size_t find_first_of(const wxScopedWCharBuffer& sz, size_t nStart, size_t n) const + { return find_first_of(sz.data(), nStart, n); } + + size_t find_last_of(const wxCStrData& sz, size_t nStart = 0) const + { return find_last_of(sz.AsString(), nStart); } + size_t find_last_of(const wxScopedCharBuffer& sz, size_t nStart = 0) const + { return find_last_of(sz.data(), nStart); } + size_t find_last_of(const wxScopedWCharBuffer& sz, size_t nStart = 0) const + { return find_last_of(sz.data(), nStart); } + size_t find_last_of(const wxCStrData& sz, size_t nStart, size_t n) const + { return find_last_of(sz.AsWChar(), nStart, n); } + size_t find_last_of(const wxScopedCharBuffer& sz, size_t nStart, size_t n) const + { return find_last_of(sz.data(), nStart, n); } + size_t find_last_of(const wxScopedWCharBuffer& sz, size_t nStart, size_t n) const + { return find_last_of(sz.data(), nStart, n); } + + size_t find_first_not_of(const wxCStrData& sz, size_t nStart = 0) const + { return find_first_not_of(sz.AsString(), nStart); } + size_t find_first_not_of(const wxScopedCharBuffer& sz, size_t nStart = 0) const + { return find_first_not_of(sz.data(), nStart); } + size_t find_first_not_of(const wxScopedWCharBuffer& sz, size_t nStart = 0) const + { return find_first_not_of(sz.data(), nStart); } + size_t find_first_not_of(const wxCStrData& sz, size_t nStart, size_t n) const + { return find_first_not_of(sz.AsWChar(), nStart, n); } + size_t find_first_not_of(const wxScopedCharBuffer& sz, size_t nStart, size_t n) const + { return find_first_not_of(sz.data(), nStart, n); } + size_t find_first_not_of(const wxScopedWCharBuffer& sz, size_t nStart, size_t n) const + { return find_first_not_of(sz.data(), nStart, n); } + + size_t find_last_not_of(const wxCStrData& sz, size_t nStart = 0) const + { return find_last_not_of(sz.AsString(), nStart); } + size_t find_last_not_of(const wxScopedCharBuffer& sz, size_t nStart = 0) const + { return find_last_not_of(sz.data(), nStart); } + size_t find_last_not_of(const wxScopedWCharBuffer& sz, size_t nStart = 0) const + { return find_last_not_of(sz.data(), nStart); } + size_t find_last_not_of(const wxCStrData& sz, size_t nStart, size_t n) const + { return find_last_not_of(sz.AsWChar(), nStart, n); } + size_t find_last_not_of(const wxScopedCharBuffer& sz, size_t nStart, size_t n) const + { return find_last_not_of(sz.data(), nStart, n); } + size_t find_last_not_of(const wxScopedWCharBuffer& sz, size_t nStart, size_t n) const + { return find_last_not_of(sz.data(), nStart, n); } + // string += string wxString& operator+=(const wxString& s) - { m_impl += s.m_impl; return *this; } + { + wxSTRING_INVALIDATE_CACHED_LENGTH(); + + m_impl += s.m_impl; + return *this; + } // string += C string wxString& operator+=(const char *psz) - { m_impl += ImplStr(psz); return *this; } + { + wxSTRING_INVALIDATE_CACHED_LENGTH(); + + m_impl += ImplStr(psz); + return *this; + } wxString& operator+=(const wchar_t *pwz) - { m_impl += ImplStr(pwz); return *this; } + { + wxSTRING_INVALIDATE_CACHED_LENGTH(); + + m_impl += ImplStr(pwz); + return *this; + } wxString& operator+=(const wxCStrData& s) - { m_impl += s.AsString().m_impl; return *this; } + { + wxSTRING_INVALIDATE_CACHED_LENGTH(); + + m_impl += s.AsString().m_impl; + return *this; + } + wxString& operator+=(const wxScopedCharBuffer& s) + { return append(s); } + wxString& operator+=(const wxScopedWCharBuffer& s) + { return append(s); } // string += char wxString& operator+=(wxUniChar ch) - { m_impl += EncodeChar(ch); return *this; } + { + wxSTRING_UPDATE_CACHED_LENGTH(1); + +#if wxUSE_UNICODE_UTF8 + if ( !ch.IsAscii() ) + m_impl += wxStringOperations::EncodeChar(ch); + else +#endif + m_impl += (wxStringCharType)ch; + return *this; + } wxString& operator+=(wxUniCharRef ch) { return *this += wxUniChar(ch); } wxString& operator+=(int ch) { return *this += wxUniChar(ch); } wxString& operator+=(char ch) { return *this += wxUniChar(ch); } @@ -1865,22 +3398,37 @@ public: wxString& operator+=(wchar_t ch) { return *this += wxUniChar(ch); } private: -#if !wxUSE_STL_BASED_WXSTRING && !wxUSE_UNICODE_UTF8 +#if !wxUSE_STL_BASED_WXSTRING // helpers for wxStringBuffer and wxStringBufferLength wxStringCharType *DoGetWriteBuf(size_t nLen) - { return m_impl.DoGetWriteBuf(nLen); } + { + return m_impl.DoGetWriteBuf(nLen); + } + void DoUngetWriteBuf() - { m_impl.DoUngetWriteBuf(); } + { + wxSTRING_INVALIDATE_CACHE(); + + m_impl.DoUngetWriteBuf(); + } + void DoUngetWriteBuf(size_t nLen) - { m_impl.DoUngetWriteBuf(nLen); } + { + wxSTRING_SET_CACHED_LENGTH(nLen); - friend class WXDLLIMPEXP_BASE wxStringBuffer; - friend class WXDLLIMPEXP_BASE wxStringBufferLength; -#endif // !wxUSE_STL_BASED_WXSTRING && !wxUSE_UNICODE_UTF8 + m_impl.DoUngetWriteBuf(nLen); + } +#endif // !wxUSE_STL_BASED_WXSTRING #ifndef wxNEEDS_WXSTRING_PRINTF_MIXIN - int DoPrintf(const wxChar *format, ...) ATTRIBUTE_PRINTF_2; - static wxString DoFormat(const wxChar *format, ...) ATTRIBUTE_PRINTF_1; + #if !wxUSE_UTF8_LOCALE_ONLY + int DoPrintfWchar(const wxChar *format, ...); + static wxString DoFormatWchar(const wxChar *format, ...); + #endif + #if wxUSE_UNICODE_UTF8 + int DoPrintfUtf8(const char *format, ...); + static wxString DoFormatUtf8(const char *format, ...); + #endif #endif #if !wxUSE_STL_BASED_WXSTRING @@ -1890,29 +3438,155 @@ private: private: wxStringImpl m_impl; + + // buffers for compatibility conversion from (char*)c_str() and + // (wchar_t*)c_str(): the pointers returned by these functions should remain + // valid until the string itself is modified for compatibility with the + // existing code and consistency with std::string::c_str() so returning a + // temporary buffer won't do and we need to cache the conversion results + + // TODO-UTF8: benchmark various approaches to keeping compatibility buffers + template + struct ConvertedBuffer + { + // notice that there is no need to initialize m_len here as it's unused + // as long as m_str is NULL + ConvertedBuffer() : m_str(NULL) {} + ~ConvertedBuffer() + { free(m_str); } + + bool Extend(size_t len) + { + // add extra 1 for the trailing NUL + void * const str = realloc(m_str, sizeof(T)*(len + 1)); + if ( !str ) + return false; + + m_str = static_cast(str); + m_len = len; + + return true; + } + + const wxScopedCharTypeBuffer AsScopedBuffer() const + { + return wxScopedCharTypeBuffer::CreateNonOwned(m_str, m_len); + } + + T *m_str; // pointer to the string data + size_t m_len; // length, not size, i.e. in chars and without last NUL + }; + + +#if wxUSE_UNICODE + // common mb_str() and wxCStrData::AsChar() helper: performs the conversion + // and returns either m_convertedToChar.m_str (in which case its m_len is + // also updated) or NULL if it failed + // + // there is an important exception: in wxUSE_UNICODE_UTF8 build if conv is a + // UTF-8 one, we return m_impl.c_str() directly, without doing any conversion + // as optimization and so the caller needs to check for this before using + // m_convertedToChar + // + // NB: AsChar() returns char* in any build, unlike mb_str() + const char *AsChar(const wxMBConv& conv) const; + + // mb_str() implementation helper + wxScopedCharBuffer AsCharBuf(const wxMBConv& conv) const + { +#if wxUSE_UNICODE_UTF8 + // avoid conversion if we can + if ( conv.IsUTF8() ) + { + return wxScopedCharBuffer::CreateNonOwned(m_impl.c_str(), + m_impl.length()); + } +#endif // wxUSE_UNICODE_UTF8 + + // call this solely in order to fill in m_convertedToChar as AsChar() + // updates it as a side effect: this is a bit ugly but it's a completely + // internal function so the users of this class shouldn't care or know + // about it and doing it like this, i.e. having a separate AsChar(), + // allows us to avoid the creation and destruction of a temporary buffer + // when using wxCStrData without duplicating any code + if ( !AsChar(conv) ) + { + // although it would be probably more correct to return NULL buffer + // from here if the conversion fails, a lot of existing code doesn't + // expect mb_str() (or wc_str()) to ever return NULL so return an + // empty string otherwise to avoid crashes in it + // + // also, some existing code does check for the conversion success and + // so asserting here would be bad too -- even if it does mean that + // silently losing data is possible for badly written code + return wxScopedCharBuffer::CreateNonOwned("", 0); + } + + return m_convertedToChar.AsScopedBuffer(); + } + + ConvertedBuffer m_convertedToChar; +#endif // !wxUSE_UNICODE + +#if !wxUSE_UNICODE_WCHAR + // common wc_str() and wxCStrData::AsWChar() helper for both UTF-8 and ANSI + // builds: converts the string contents into m_convertedToWChar and returns + // NULL if the conversion failed (this can only happen in ANSI build) + // + // NB: AsWChar() returns wchar_t* in any build, unlike wc_str() + const wchar_t *AsWChar(const wxMBConv& conv) const; + + // wc_str() implementation helper + wxScopedWCharBuffer AsWCharBuf(const wxMBConv& conv) const + { + if ( !AsWChar(conv) ) + return wxScopedWCharBuffer::CreateNonOwned(L"", 0); + + return m_convertedToWChar.AsScopedBuffer(); + } + + ConvertedBuffer m_convertedToWChar; +#endif // !wxUSE_UNICODE_WCHAR + +#if wxUSE_UNICODE_UTF8 + // FIXME-UTF8: (try to) move this elsewhere (TLS) or solve differently + // assigning to character pointer to by wxString::iterator may + // change the underlying wxStringImpl iterator, so we have to + // keep track of all iterators and update them as necessary: + struct wxStringIteratorNodeHead + { + wxStringIteratorNodeHead() : ptr(NULL) {} + wxStringIteratorNode *ptr; + + // copying is disallowed as it would result in more than one pointer into + // the same linked list + wxDECLARE_NO_COPY_CLASS(wxStringIteratorNodeHead); + }; + + wxStringIteratorNodeHead m_iterators; + + friend class WXDLLIMPEXP_FWD_BASE wxStringIteratorNode; + friend class WXDLLIMPEXP_FWD_BASE wxUniCharRef; +#endif // wxUSE_UNICODE_UTF8 + + friend class WXDLLIMPEXP_FWD_BASE wxCStrData; + friend class wxStringInternalBuffer; + friend class wxStringInternalBufferLength; }; #ifdef wxNEEDS_WXSTRING_PRINTF_MIXIN - #pragma warning (default:4275) + #pragma warning (pop) #endif // string iterator operators that satisfy STL Random Access Iterator // requirements: -inline wxString::iterator operator+(int n, wxString::iterator i) - { return i + n; } -inline wxString::iterator operator+(size_t n, wxString::iterator i) +inline wxString::iterator operator+(ptrdiff_t n, wxString::iterator i) { return i + n; } -inline wxString::const_iterator operator+(int n, wxString::const_iterator i) +inline wxString::const_iterator operator+(ptrdiff_t n, wxString::const_iterator i) { return i + n; } -inline wxString::const_iterator operator+(size_t n, wxString::const_iterator i) +inline wxString::reverse_iterator operator+(ptrdiff_t n, wxString::reverse_iterator i) { return i + n; } -inline wxString::reverse_iterator operator+(int n, wxString::reverse_iterator i) - { return i + n; } -inline wxString::reverse_iterator operator+(size_t n, wxString::reverse_iterator i) - { return i + n; } -inline wxString::const_reverse_iterator operator+(int n, wxString::const_reverse_iterator i) - { return i + n; } -inline wxString::const_reverse_iterator operator+(size_t n, wxString::const_reverse_iterator i) +inline wxString::const_reverse_iterator operator+(ptrdiff_t n, wxString::const_reverse_iterator i) { return i + n; } // notice that even though for many compilers the friend declarations above are @@ -1942,140 +3616,346 @@ inline wxString operator+(wchar_t ch, const wxString& string) { return wxUniChar(ch) + string; } -#if wxUSE_STL_BASED_WXSTRING - // return an empty wxString (not very useful with wxUSE_STL == 1) - inline const wxString wxGetEmptyString() { return wxString(); } -#else // !wxUSE_STL_BASED_WXSTRING - // return an empty wxString (more efficient than wxString() here) - inline const wxString& wxGetEmptyString() +#define wxGetEmptyString() wxString() + +// ---------------------------------------------------------------------------- +// helper functions which couldn't be defined inline +// ---------------------------------------------------------------------------- + +namespace wxPrivate +{ + +#if wxUSE_UNICODE_WCHAR + +template <> +struct wxStringAsBufHelper +{ + static wxScopedCharBuffer Get(const wxString& s, size_t *len) { - return *(wxString *)&wxEmptyString; + wxScopedCharBuffer buf(s.mb_str()); + if ( len ) + *len = buf ? strlen(buf) : 0; + return buf; } -#endif // wxUSE_STL_BASED_WXSTRING/!wxUSE_STL_BASED_WXSTRING +}; + +template <> +struct wxStringAsBufHelper +{ + static wxScopedWCharBuffer Get(const wxString& s, size_t *len) + { + const size_t length = s.length(); + if ( len ) + *len = length; + return wxScopedWCharBuffer::CreateNonOwned(s.wx_str(), length); + } +}; + +#elif wxUSE_UNICODE_UTF8 + +template <> +struct wxStringAsBufHelper +{ + static wxScopedCharBuffer Get(const wxString& s, size_t *len) + { + const size_t length = s.utf8_length(); + if ( len ) + *len = length; + return wxScopedCharBuffer::CreateNonOwned(s.wx_str(), length); + } +}; + +template <> +struct wxStringAsBufHelper +{ + static wxScopedWCharBuffer Get(const wxString& s, size_t *len) + { + wxScopedWCharBuffer wbuf(s.wc_str()); + if ( len ) + *len = wxWcslen(wbuf); + return wbuf; + } +}; + +#endif // Unicode build kind + +} // namespace wxPrivate // ---------------------------------------------------------------------------- // wxStringBuffer: a tiny class allowing to get a writable pointer into string // ---------------------------------------------------------------------------- -#if wxUSE_STL_BASED_WXSTRING || wxUSE_UNICODE_UTF8 - -class WXDLLIMPEXP_BASE wxStringBuffer +#if !wxUSE_STL_BASED_WXSTRING +// string buffer for direct access to string data in their native +// representation: +class wxStringInternalBuffer { public: - wxStringBuffer(wxString& str, size_t lenWanted = 1024) - : m_str(str), m_buf(lenWanted) - { } + typedef wxStringCharType CharType; - ~wxStringBuffer() { m_str.assign(m_buf.data(), wxStrlen(m_buf.data())); } + wxStringInternalBuffer(wxString& str, size_t lenWanted = 1024) + : m_str(str), m_buf(NULL) + { m_buf = m_str.DoGetWriteBuf(lenWanted); } - operator wxChar*() { return m_buf.data(); } + ~wxStringInternalBuffer() { m_str.DoUngetWriteBuf(); } + + operator wxStringCharType*() const { return m_buf; } private: - wxString& m_str; -#if wxUSE_UNICODE - wxWCharBuffer m_buf; -#else - wxCharBuffer m_buf; -#endif + wxString& m_str; + wxStringCharType *m_buf; - DECLARE_NO_COPY_CLASS(wxStringBuffer) + wxDECLARE_NO_COPY_CLASS(wxStringInternalBuffer); }; -class WXDLLIMPEXP_BASE wxStringBufferLength +class wxStringInternalBufferLength { public: - wxStringBufferLength(wxString& str, size_t lenWanted = 1024) - : m_str(str), m_buf(lenWanted), m_len(0), m_lenSet(false) - { } + typedef wxStringCharType CharType; - ~wxStringBufferLength() + wxStringInternalBufferLength(wxString& str, size_t lenWanted = 1024) + : m_str(str), m_buf(NULL), m_len(0), m_lenSet(false) + { + m_buf = m_str.DoGetWriteBuf(lenWanted); + wxASSERT(m_buf != NULL); + } + + ~wxStringInternalBufferLength() { wxASSERT(m_lenSet); - m_str.assign(m_buf.data(), m_len); + m_str.DoUngetWriteBuf(m_len); } - operator wxChar*() { return m_buf.data(); } + operator wxStringCharType*() const { return m_buf; } void SetLength(size_t length) { m_len = length; m_lenSet = true; } private: - wxString& m_str; -#if wxUSE_UNICODE - wxWCharBuffer m_buf; -#else - wxCharBuffer m_buf; -#endif - size_t m_len; - bool m_lenSet; + wxString& m_str; + wxStringCharType *m_buf; + size_t m_len; + bool m_lenSet; - DECLARE_NO_COPY_CLASS(wxStringBufferLength) + wxDECLARE_NO_COPY_CLASS(wxStringInternalBufferLength); }; -#else // if !wxUSE_STL_BASED_WXSTRING && !wxUSE_UNICODE_UTF8 +#endif // !wxUSE_STL_BASED_WXSTRING -class WXDLLIMPEXP_BASE wxStringBuffer +template +class wxStringTypeBufferBase { public: - wxStringBuffer(wxString& str, size_t lenWanted = 1024) - : m_str(str), m_buf(NULL) - { m_buf = m_str.DoGetWriteBuf(lenWanted); } + typedef T CharType; - ~wxStringBuffer() { m_str.DoUngetWriteBuf(); } + wxStringTypeBufferBase(wxString& str, size_t lenWanted = 1024) + : m_str(str), m_buf(lenWanted) + { + // for compatibility with old wxStringBuffer which provided direct + // access to wxString internal buffer, initialize ourselves with the + // string initial contents + + // FIXME-VC6: remove the ugly (CharType *)NULL and use normal + // tchar_str + size_t len; + const wxCharTypeBuffer buf(str.tchar_str(&len, (CharType *)NULL)); + if ( buf ) + { + if ( len > lenWanted ) + { + // in this case there is not enough space for terminating NUL, + // ensure that we still put it there + m_buf.data()[lenWanted] = 0; + len = lenWanted - 1; + } + + memcpy(m_buf.data(), buf, (len + 1)*sizeof(CharType)); + } + //else: conversion failed, this can happen when trying to get Unicode + // string contents into a char string + } - operator wxChar*() const { return m_buf; } + operator CharType*() { return m_buf.data(); } -private: +protected: wxString& m_str; - wxChar *m_buf; + wxCharTypeBuffer m_buf; +}; + +template +class wxStringTypeBufferLengthBase : public wxStringTypeBufferBase +{ +public: + wxStringTypeBufferLengthBase(wxString& str, size_t lenWanted = 1024) + : wxStringTypeBufferBase(str, lenWanted), + m_len(0), + m_lenSet(false) + { } + + ~wxStringTypeBufferLengthBase() + { + wxASSERT_MSG( this->m_lenSet, "forgot to call SetLength()" ); + } + + void SetLength(size_t length) { m_len = length; m_lenSet = true; } - DECLARE_NO_COPY_CLASS(wxStringBuffer) +protected: + size_t m_len; + bool m_lenSet; }; -class WXDLLIMPEXP_BASE wxStringBufferLength +template +class wxStringTypeBuffer : public wxStringTypeBufferBase { public: - wxStringBufferLength(wxString& str, size_t lenWanted = 1024) - : m_str(str), m_buf(NULL), m_len(0), m_lenSet(false) + wxStringTypeBuffer(wxString& str, size_t lenWanted = 1024) + : wxStringTypeBufferBase(str, lenWanted) + { } + + ~wxStringTypeBuffer() { - m_buf = m_str.DoGetWriteBuf(lenWanted); - wxASSERT(m_buf != NULL); + this->m_str.assign(this->m_buf.data()); } - ~wxStringBufferLength() + wxDECLARE_NO_COPY_CLASS(wxStringTypeBuffer); +}; + +template +class wxStringTypeBufferLength : public wxStringTypeBufferLengthBase +{ +public: + wxStringTypeBufferLength(wxString& str, size_t lenWanted = 1024) + : wxStringTypeBufferLengthBase(str, lenWanted) + { } + + ~wxStringTypeBufferLength() { - wxASSERT(m_lenSet); - m_str.DoUngetWriteBuf(m_len); + this->m_str.assign(this->m_buf.data(), this->m_len); } - operator wxChar*() const { return m_buf; } - void SetLength(size_t length) { m_len = length; m_lenSet = true; } + wxDECLARE_NO_COPY_CLASS(wxStringTypeBufferLength); +}; -private: - wxString& m_str; - wxChar *m_buf; - size_t m_len; - bool m_lenSet; +#if wxUSE_STL_BASED_WXSTRING + +WXDLLIMPEXP_TEMPLATE_INSTANCE_BASE( wxStringTypeBufferBase ) + +class wxStringInternalBuffer : public wxStringTypeBufferBase +{ +public: + wxStringInternalBuffer(wxString& str, size_t lenWanted = 1024) + : wxStringTypeBufferBase(str, lenWanted) {} + ~wxStringInternalBuffer() + { m_str.m_impl.assign(m_buf.data()); } + + wxDECLARE_NO_COPY_CLASS(wxStringInternalBuffer); +}; + +WXDLLIMPEXP_TEMPLATE_INSTANCE_BASE( + wxStringTypeBufferLengthBase ) + +class wxStringInternalBufferLength + : public wxStringTypeBufferLengthBase +{ +public: + wxStringInternalBufferLength(wxString& str, size_t lenWanted = 1024) + : wxStringTypeBufferLengthBase(str, lenWanted) {} - DECLARE_NO_COPY_CLASS(wxStringBufferLength) + ~wxStringInternalBufferLength() + { + m_str.m_impl.assign(m_buf.data(), m_len); + } + + wxDECLARE_NO_COPY_CLASS(wxStringInternalBufferLength); }; +#endif // wxUSE_STL_BASED_WXSTRING + + +#if wxUSE_STL_BASED_WXSTRING || wxUSE_UNICODE_UTF8 +typedef wxStringTypeBuffer wxStringBuffer; +typedef wxStringTypeBufferLength wxStringBufferLength; +#else // if !wxUSE_STL_BASED_WXSTRING && !wxUSE_UNICODE_UTF8 +typedef wxStringInternalBuffer wxStringBuffer; +typedef wxStringInternalBufferLength wxStringBufferLength; #endif // !wxUSE_STL_BASED_WXSTRING && !wxUSE_UNICODE_UTF8 +#if wxUSE_UNICODE_UTF8 +typedef wxStringInternalBuffer wxUTF8StringBuffer; +typedef wxStringInternalBufferLength wxUTF8StringBufferLength; +#elif wxUSE_UNICODE_WCHAR + +WXDLLIMPEXP_TEMPLATE_INSTANCE_BASE( wxStringTypeBufferBase ) + +// Note about inlined dtors in the classes below: this is done not for +// performance reasons but just to avoid linking errors in the MSVC DLL build +// under Windows: if a class has non-inline methods it must be declared as +// being DLL-exported but, due to an extremely interesting feature of MSVC 7 +// and later, any template class which is used as a base of a DLL-exported +// class is implicitly made DLL-exported too, as explained at the bottom of +// http://msdn.microsoft.com/en-us/library/twa2aw10.aspx (just to confirm: yes, +// _inheriting_ from a class can change whether it is being exported from DLL) +// +// But this results in link errors because the base template class is not DLL- +// exported, whether it is declared with WXDLLIMPEXP_BASE or not, because it +// does have only inline functions. So the simplest fix is to just make all the +// functions of these classes inline too. + +class wxUTF8StringBuffer : public wxStringTypeBufferBase +{ +public: + wxUTF8StringBuffer(wxString& str, size_t lenWanted = 1024) + : wxStringTypeBufferBase(str, lenWanted) {} + ~wxUTF8StringBuffer() + { + wxMBConvStrictUTF8 conv; + size_t wlen = conv.ToWChar(NULL, 0, m_buf); + wxCHECK_RET( wlen != wxCONV_FAILED, "invalid UTF-8 data in string buffer?" ); + + wxStringInternalBuffer wbuf(m_str, wlen); + conv.ToWChar(wbuf, wlen, m_buf); + } + + wxDECLARE_NO_COPY_CLASS(wxUTF8StringBuffer); +}; + +WXDLLIMPEXP_TEMPLATE_INSTANCE_BASE( wxStringTypeBufferLengthBase ) + +class wxUTF8StringBufferLength : public wxStringTypeBufferLengthBase +{ +public: + wxUTF8StringBufferLength(wxString& str, size_t lenWanted = 1024) + : wxStringTypeBufferLengthBase(str, lenWanted) {} + ~wxUTF8StringBufferLength() + { + wxCHECK_RET(m_lenSet, "length not set"); + + wxMBConvStrictUTF8 conv; + size_t wlen = conv.ToWChar(NULL, 0, m_buf, m_len); + wxCHECK_RET( wlen != wxCONV_FAILED, "invalid UTF-8 data in string buffer?" ); + + wxStringInternalBufferLength wbuf(m_str, wlen); + conv.ToWChar(wbuf, wlen, m_buf, m_len); + wbuf.SetLength(wlen); + } + + wxDECLARE_NO_COPY_CLASS(wxUTF8StringBufferLength); +}; +#endif // wxUSE_UNICODE_UTF8/wxUSE_UNICODE_WCHAR + + // --------------------------------------------------------------------------- // wxString comparison functions: operator versions are always case sensitive // --------------------------------------------------------------------------- -#define wxCMP_WXCHAR_STRING(p, s, op) s.Cmp(p) op 0 +#define wxCMP_WXCHAR_STRING(p, s, op) 0 op s.Cmp(p) wxDEFINE_ALL_COMPARISONS(const wxChar *, const wxString&, wxCMP_WXCHAR_STRING) #undef wxCMP_WXCHAR_STRING -// note that there is an optimization in operator==() and !=(): we (quickly) -// checks the strings length first, before comparing their data inline bool operator==(const wxString& s1, const wxString& s2) - { return (s1.Len() == s2.Len()) && (s1.Cmp(s2) == 0); } + { return s1.IsSameAs(s2); } inline bool operator!=(const wxString& s1, const wxString& s2) - { return (s1.Len() != s2.Len()) || (s1.Cmp(s2) != 0); } + { return !s1.IsSameAs(s2); } inline bool operator< (const wxString& s1, const wxString& s2) { return s1.Cmp(s2) < 0; } inline bool operator> (const wxString& s1, const wxString& s2) @@ -2085,37 +3965,42 @@ inline bool operator<=(const wxString& s1, const wxString& s2) inline bool operator>=(const wxString& s1, const wxString& s2) { return s1.Cmp(s2) >= 0; } -#if wxUSE_UNICODE -inline bool operator==(const wxString& s1, const wxWCharBuffer& s2) +inline bool operator==(const wxString& s1, const wxCStrData& s2) + { return s1 == s2.AsString(); } +inline bool operator==(const wxCStrData& s1, const wxString& s2) + { return s1.AsString() == s2; } +inline bool operator!=(const wxString& s1, const wxCStrData& s2) + { return s1 != s2.AsString(); } +inline bool operator!=(const wxCStrData& s1, const wxString& s2) + { return s1.AsString() != s2; } + +inline bool operator==(const wxString& s1, const wxScopedWCharBuffer& s2) { return (s1.Cmp((const wchar_t *)s2) == 0); } -inline bool operator==(const wxWCharBuffer& s1, const wxString& s2) +inline bool operator==(const wxScopedWCharBuffer& s1, const wxString& s2) { return (s2.Cmp((const wchar_t *)s1) == 0); } -inline bool operator!=(const wxString& s1, const wxWCharBuffer& s2) +inline bool operator!=(const wxString& s1, const wxScopedWCharBuffer& s2) { return (s1.Cmp((const wchar_t *)s2) != 0); } -inline bool operator!=(const wxWCharBuffer& s1, const wxString& s2) +inline bool operator!=(const wxScopedWCharBuffer& s1, const wxString& s2) { return (s2.Cmp((const wchar_t *)s1) != 0); } -#else // !wxUSE_UNICODE -inline bool operator==(const wxString& s1, const wxCharBuffer& s2) + +inline bool operator==(const wxString& s1, const wxScopedCharBuffer& s2) { return (s1.Cmp((const char *)s2) == 0); } -inline bool operator==(const wxCharBuffer& s1, const wxString& s2) +inline bool operator==(const wxScopedCharBuffer& s1, const wxString& s2) { return (s2.Cmp((const char *)s1) == 0); } -inline bool operator!=(const wxString& s1, const wxCharBuffer& s2) +inline bool operator!=(const wxString& s1, const wxScopedCharBuffer& s2) { return (s1.Cmp((const char *)s2) != 0); } -inline bool operator!=(const wxCharBuffer& s1, const wxString& s2) +inline bool operator!=(const wxScopedCharBuffer& s1, const wxString& s2) { return (s2.Cmp((const char *)s1) != 0); } -#endif // wxUSE_UNICODE/!wxUSE_UNICODE -#if wxUSE_UNICODE -inline wxString operator+(const wxString& string, const wxWCharBuffer& buf) +inline wxString operator+(const wxString& string, const wxScopedWCharBuffer& buf) { return string + (const wchar_t *)buf; } -inline wxString operator+(const wxWCharBuffer& buf, const wxString& string) +inline wxString operator+(const wxScopedWCharBuffer& buf, const wxString& string) { return (const wchar_t *)buf + string; } -#else // !wxUSE_UNICODE -inline wxString operator+(const wxString& string, const wxCharBuffer& buf) + +inline wxString operator+(const wxString& string, const wxScopedCharBuffer& buf) { return string + (const char *)buf; } -inline wxString operator+(const wxCharBuffer& buf, const wxString& string) +inline wxString operator+(const wxScopedCharBuffer& buf, const wxString& string) { return (const char *)buf + string; } -#endif // wxUSE_UNICODE/!wxUSE_UNICODE // comparison with char inline bool operator==(const wxUniChar& c, const wxString& s) { return s.IsSameAs(c); } @@ -2156,12 +4041,8 @@ wxDEFINE_ALL_COMPARISONS(const char *, const wxString&, wxCMP_CHAR_STRING) #define wxCMP_CHAR_CSTRDATA(p, s, op) p op s.AsChar() #define wxCMP_WCHAR_CSTRDATA(p, s, op) p op s.AsWChar() -// FIXME: these ifdefs must be removed when wxCStrData has both conversions -#if wxUSE_UNICODE - wxDEFINE_ALL_COMPARISONS(const wchar_t *, const wxCStrData&, wxCMP_WCHAR_CSTRDATA) -#else - wxDEFINE_ALL_COMPARISONS(const char *, const wxCStrData&, wxCMP_CHAR_CSTRDATA) -#endif +wxDEFINE_ALL_COMPARISONS(const wchar_t *, const wxCStrData&, wxCMP_WCHAR_CSTRDATA) +wxDEFINE_ALL_COMPARISONS(const char *, const wxCStrData&, wxCMP_CHAR_CSTRDATA) #undef wxCMP_CHAR_CSTRDATA #undef wxCMP_WCHAR_CSTRDATA @@ -2176,8 +4057,20 @@ wxDEFINE_ALL_COMPARISONS(const char *, const wxString&, wxCMP_CHAR_STRING) WXDLLIMPEXP_BASE wxSTD ostream& operator<<(wxSTD ostream&, const wxString&); WXDLLIMPEXP_BASE wxSTD ostream& operator<<(wxSTD ostream&, const wxCStrData&); +WXDLLIMPEXP_BASE wxSTD ostream& operator<<(wxSTD ostream&, const wxScopedCharBuffer&); +#ifndef __BORLANDC__ +WXDLLIMPEXP_BASE wxSTD ostream& operator<<(wxSTD ostream&, const wxScopedWCharBuffer&); +#endif + +#if wxUSE_UNICODE && defined(HAVE_WOSTREAM) + +WXDLLIMPEXP_BASE wxSTD wostream& operator<<(wxSTD wostream&, const wxString&); +WXDLLIMPEXP_BASE wxSTD wostream& operator<<(wxSTD wostream&, const wxCStrData&); +WXDLLIMPEXP_BASE wxSTD wostream& operator<<(wxSTD wostream&, const wxScopedWCharBuffer&); + +#endif // wxUSE_UNICODE && defined(HAVE_WOSTREAM) -#endif // wxSTD_STRING_COMPATIBILITY +#endif // wxUSE_STD_IOSTREAM // --------------------------------------------------------------------------- // wxCStrData implementation @@ -2188,20 +4081,66 @@ inline wxCStrData::wxCStrData(char *buf) inline wxCStrData::wxCStrData(wchar_t *buf) : m_str(new wxString(buf)), m_offset(0), m_owned(true) {} +inline wxCStrData::wxCStrData(const wxCStrData& data) + : m_str(data.m_owned ? new wxString(*data.m_str) : data.m_str), + m_offset(data.m_offset), + m_owned(data.m_owned) +{ +} + inline wxCStrData::~wxCStrData() { if ( m_owned ) - delete m_str; + delete const_cast(m_str); // cast to silence warnings } -#if wxUSE_UNICODE +// AsChar() and AsWChar() implementations simply forward to wxString methods + inline const wchar_t* wxCStrData::AsWChar() const +{ + const wchar_t * const p = +#if wxUSE_UNICODE_WCHAR + m_str->wc_str(); +#elif wxUSE_UNICODE_UTF8 + m_str->AsWChar(wxMBConvStrictUTF8()); #else -inline const char* wxCStrData::AsChar() const + m_str->AsWChar(wxConvLibc); #endif + + // in Unicode build the string always has a valid Unicode representation + // and even if a conversion is needed (as in UTF8 case) it can't fail + // + // but in ANSI build the string contents might be not convertible to + // Unicode using the current locale encoding so we do need to check for + // errors +#if !wxUSE_UNICODE + if ( !p ) + { + // if conversion fails, return empty string and not NULL to avoid + // crashes in code written with either wxWidgets 2 wxString or + // std::string behaviour in mind: neither of them ever returns NULL + // from its c_str() and so we shouldn't neither + // + // notice that the same is done in AsChar() below and + // wxString::wc_str() and mb_str() for the same reasons + return L""; + } +#endif // !wxUSE_UNICODE + + return p + m_offset; +} + +inline const char* wxCStrData::AsChar() const { - // FIXME-UTF8: incorrect position, incorrect charset - return m_str->wx_str() + m_offset; +#if wxUSE_UNICODE && !wxUSE_UTF8_LOCALE_ONLY + const char * const p = m_str->AsChar(wxConvLibc); + if ( !p ) + return ""; +#else // !wxUSE_UNICODE || wxUSE_UTF8_LOCALE_ONLY + const char * const p = m_str->mb_str(); +#endif // wxUSE_UNICODE && !wxUSE_UTF8_LOCALE_ONLY + + return p + m_offset; } inline wxString wxCStrData::AsString() const @@ -2212,27 +4151,124 @@ inline wxString wxCStrData::AsString() const return m_str->Mid(m_offset); } +inline const wxStringCharType *wxCStrData::AsInternal() const +{ +#if wxUSE_UNICODE_UTF8 + return wxStringOperations::AddToIter(m_str->wx_str(), m_offset); +#else + return m_str->wx_str() + m_offset; +#endif +} + inline wxUniChar wxCStrData::operator*() const { if ( m_str->empty() ) - return wxUniChar(_T('\0')); + return wxUniChar(wxT('\0')); else return (*m_str)[m_offset]; } inline wxUniChar wxCStrData::operator[](size_t n) const { - return m_str->at(m_offset + n); + // NB: we intentionally use operator[] and not at() here because the former + // works for the terminating NUL while the latter does not + return (*m_str)[m_offset + n]; +} + +// ---------------------------------------------------------------------------- +// more wxCStrData operators +// ---------------------------------------------------------------------------- + +// we need to define those to allow "size_t pos = p - s.c_str()" where p is +// some pointer into the string +inline size_t operator-(const char *p, const wxCStrData& cs) +{ + return p - cs.AsChar(); +} + +inline size_t operator-(const wchar_t *p, const wxCStrData& cs) +{ + return p - cs.AsWChar(); } // ---------------------------------------------------------------------------- // implementation of wx[W]CharBuffer inline methods using wxCStrData // ---------------------------------------------------------------------------- -// FIXME-UTF8: move this to buffer.h; provide versions for both variants -inline wxWxCharBuffer::wxWxCharBuffer(const wxCStrData& cstr) - : wxCharTypeBufferBase((const wxChar *)cstr) +// FIXME-UTF8: move this to buffer.h +inline wxCharBuffer::wxCharBuffer(const wxCStrData& cstr) + : wxCharTypeBufferBase(cstr.AsCharBuf()) +{ +} + +inline wxWCharBuffer::wxWCharBuffer(const wxCStrData& cstr) + : wxCharTypeBufferBase(cstr.AsWCharBuf()) { } -#endif // _WX_WXSTRING_H__ +#if wxUSE_UNICODE_UTF8 +// ---------------------------------------------------------------------------- +// implementation of wxStringIteratorNode inline methods +// ---------------------------------------------------------------------------- + +void wxStringIteratorNode::DoSet(const wxString *str, + wxStringImpl::const_iterator *citer, + wxStringImpl::iterator *iter) +{ + m_prev = NULL; + m_iter = iter; + m_citer = citer; + m_str = str; + if ( str ) + { + m_next = str->m_iterators.ptr; + const_cast(m_str)->m_iterators.ptr = this; + if ( m_next ) + m_next->m_prev = this; + } + else + { + m_next = NULL; + } +} + +void wxStringIteratorNode::clear() +{ + if ( m_next ) + m_next->m_prev = m_prev; + if ( m_prev ) + m_prev->m_next = m_next; + else if ( m_str ) // first in the list + const_cast(m_str)->m_iterators.ptr = m_next; + + m_next = m_prev = NULL; + m_citer = NULL; + m_iter = NULL; + m_str = NULL; +} +#endif // wxUSE_UNICODE_UTF8 + +#if WXWIN_COMPATIBILITY_2_8 + // lot of code out there doesn't explicitly include wx/crt.h, but uses + // CRT wrappers that are now declared in wx/wxcrt.h and wx/wxcrtvararg.h, + // so let's include this header now that wxString is defined and it's safe + // to do it: + #include "wx/crt.h" +#endif + +// ---------------------------------------------------------------------------- +// Checks on wxString characters +// ---------------------------------------------------------------------------- + +template + inline bool wxStringCheck(const wxString& val) + { + for ( wxString::const_iterator i = val.begin(); + i != val.end(); + ++i ) + if (T(*i) == 0) + return false; + return true; + } + +#endif // _WX_WXSTRING_H_