1 // Copyright 2017 The PDFium Authors 2 // Use of this source code is governed by a BSD-style license that can be 3 // found in the LICENSE file. 4 5 // Original code copyright 2014 Foxit Software Inc. http://www.foxitsoftware.com 6 7 #ifndef CORE_FXCRT_WIDESTRING_H_ 8 #define CORE_FXCRT_WIDESTRING_H_ 9 10 #include <stdarg.h> 11 #include <stddef.h> 12 #include <stdint.h> 13 #include <wchar.h> 14 15 #include <functional> 16 #include <iosfwd> 17 #include <iterator> 18 #include <utility> 19 20 #include "core/fxcrt/retain_ptr.h" 21 #include "core/fxcrt/string_data_template.h" 22 #include "core/fxcrt/string_view_template.h" 23 #include "third_party/abseil-cpp/absl/types/optional.h" 24 #include "third_party/base/check.h" 25 #include "third_party/base/span.h" 26 27 namespace fxcrt { 28 29 class ByteString; 30 31 // A mutable string with shared buffers using copy-on-write semantics that 32 // avoids the cost of std::string's iterator stability guarantees. 33 class WideString { 34 public: 35 using CharType = wchar_t; 36 using const_iterator = const CharType*; 37 using const_reverse_iterator = std::reverse_iterator<const_iterator>; 38 39 [[nodiscard]] static WideString FormatInteger(int i); 40 [[nodiscard]] static WideString Format(const wchar_t* pFormat, ...); 41 [[nodiscard]] static WideString FormatV(const wchar_t* lpszFormat, 42 va_list argList); 43 44 WideString(); 45 WideString(const WideString& other); 46 47 // Move-construct a WideString. After construction, |other| is empty. 48 WideString(WideString&& other) noexcept; 49 50 // Make a one-character string from one wide char. 51 explicit WideString(wchar_t ch); 52 53 // Deliberately implicit to avoid calling on every string literal. 54 // NOLINTNEXTLINE(runtime/explicit) 55 WideString(const wchar_t* ptr); 56 57 // No implicit conversions from byte strings. 58 // NOLINTNEXTLINE(runtime/explicit) 59 WideString(char) = delete; 60 61 WideString(const wchar_t* pStr, size_t len); 62 63 explicit WideString(WideStringView str); 64 WideString(WideStringView str1, WideStringView str2); 65 WideString(const std::initializer_list<WideStringView>& list); 66 67 ~WideString(); 68 69 [[nodiscard]] static WideString FromASCII(ByteStringView str); 70 [[nodiscard]] static WideString FromLatin1(ByteStringView str); 71 [[nodiscard]] static WideString FromDefANSI(ByteStringView str); 72 [[nodiscard]] static WideString FromUTF8(ByteStringView str); 73 [[nodiscard]] static WideString FromUTF16LE(const unsigned short* str, 74 size_t len); 75 [[nodiscard]] static WideString FromUTF16BE(const unsigned short* wstr, 76 size_t wlen); 77 78 [[nodiscard]] static size_t WStringLength(const unsigned short* str); 79 80 // Explicit conversion to C-style wide string. 81 // Note: Any subsequent modification of |this| will invalidate the result. c_str()82 const wchar_t* c_str() const { return m_pData ? m_pData->m_String : L""; } 83 84 // Explicit conversion to WideStringView. 85 // Note: Any subsequent modification of |this| will invalidate the result. AsStringView()86 WideStringView AsStringView() const { 87 return WideStringView(c_str(), GetLength()); 88 } 89 90 // Explicit conversion to span. 91 // Note: Any subsequent modification of |this| will invalidate the result. span()92 pdfium::span<const wchar_t> span() const { 93 return pdfium::make_span(m_pData ? m_pData->m_String : nullptr, 94 GetLength()); 95 } 96 97 // Note: Any subsequent modification of |this| will invalidate iterators. begin()98 const_iterator begin() const { return m_pData ? m_pData->m_String : nullptr; } end()99 const_iterator end() const { 100 return m_pData ? m_pData->m_String + m_pData->m_nDataLength : nullptr; 101 } 102 103 // Note: Any subsequent modification of |this| will invalidate iterators. rbegin()104 const_reverse_iterator rbegin() const { 105 return const_reverse_iterator(end()); 106 } rend()107 const_reverse_iterator rend() const { 108 return const_reverse_iterator(begin()); 109 } 110 111 // Holds on to buffer if possible for later re-use. Assign WideString() 112 // to force immediate release if desired. 113 void clear(); 114 GetLength()115 size_t GetLength() const { return m_pData ? m_pData->m_nDataLength : 0; } GetStringLength()116 size_t GetStringLength() const { 117 return m_pData ? wcslen(m_pData->m_String) : 0; 118 } IsEmpty()119 bool IsEmpty() const { return !GetLength(); } IsValidIndex(size_t index)120 bool IsValidIndex(size_t index) const { return index < GetLength(); } IsValidLength(size_t length)121 bool IsValidLength(size_t length) const { return length <= GetLength(); } 122 123 WideString& operator=(const wchar_t* str); 124 WideString& operator=(WideStringView str); 125 WideString& operator=(const WideString& that); 126 127 // Move-assign a WideString. After assignment, |that| is empty. 128 WideString& operator=(WideString&& that) noexcept; 129 130 WideString& operator+=(const wchar_t* str); 131 WideString& operator+=(wchar_t ch); 132 WideString& operator+=(const WideString& str); 133 WideString& operator+=(WideStringView str); 134 135 bool operator==(const wchar_t* ptr) const; 136 bool operator==(WideStringView str) const; 137 bool operator==(const WideString& other) const; 138 139 bool operator!=(const wchar_t* ptr) const { return !(*this == ptr); } 140 bool operator!=(WideStringView str) const { return !(*this == str); } 141 bool operator!=(const WideString& other) const { return !(*this == other); } 142 143 bool operator<(const wchar_t* ptr) const; 144 bool operator<(WideStringView str) const; 145 bool operator<(const WideString& other) const; 146 147 CharType operator[](const size_t index) const { 148 CHECK(IsValidIndex(index)); 149 return m_pData->m_String[index]; 150 } 151 Front()152 CharType Front() const { return GetLength() ? (*this)[0] : 0; } Back()153 CharType Back() const { return GetLength() ? (*this)[GetLength() - 1] : 0; } 154 155 void SetAt(size_t index, wchar_t c); 156 157 int Compare(const wchar_t* str) const; 158 int Compare(const WideString& str) const; 159 int CompareNoCase(const wchar_t* str) const; 160 161 WideString Substr(size_t offset) const; 162 WideString Substr(size_t first, size_t count) const; 163 WideString First(size_t count) const; 164 WideString Last(size_t count) const; 165 166 size_t Insert(size_t index, wchar_t ch); InsertAtFront(wchar_t ch)167 size_t InsertAtFront(wchar_t ch) { return Insert(0, ch); } InsertAtBack(wchar_t ch)168 size_t InsertAtBack(wchar_t ch) { return Insert(GetLength(), ch); } 169 size_t Delete(size_t index, size_t count = 1); 170 171 void MakeLower(); 172 void MakeUpper(); 173 174 void Trim(); 175 void Trim(wchar_t target); 176 void Trim(WideStringView targets); 177 178 void TrimLeft(); 179 void TrimLeft(wchar_t target); 180 void TrimLeft(WideStringView targets); 181 182 void TrimRight(); 183 void TrimRight(wchar_t target); 184 void TrimRight(WideStringView targets); 185 186 void Reserve(size_t len); 187 188 // Note: any modification of the string (including ReleaseBuffer()) may 189 // invalidate the span, which must not outlive its buffer. 190 pdfium::span<wchar_t> GetBuffer(size_t nMinBufLength); 191 void ReleaseBuffer(size_t nNewLength); 192 193 int GetInteger() const; 194 195 absl::optional<size_t> Find(WideStringView subStr, size_t start = 0) const; 196 absl::optional<size_t> Find(wchar_t ch, size_t start = 0) const; 197 absl::optional<size_t> ReverseFind(wchar_t ch) const; 198 199 bool Contains(WideStringView lpszSub, size_t start = 0) const { 200 return Find(lpszSub, start).has_value(); 201 } 202 203 bool Contains(char ch, size_t start = 0) const { 204 return Find(ch, start).has_value(); 205 } 206 207 size_t Replace(WideStringView pOld, WideStringView pNew); 208 size_t Remove(wchar_t ch); 209 IsASCII()210 bool IsASCII() const { return AsStringView().IsASCII(); } EqualsASCII(ByteStringView that)211 bool EqualsASCII(ByteStringView that) const { 212 return AsStringView().EqualsASCII(that); 213 } EqualsASCIINoCase(ByteStringView that)214 bool EqualsASCIINoCase(ByteStringView that) const { 215 return AsStringView().EqualsASCIINoCase(that); 216 } 217 218 ByteString ToASCII() const; 219 ByteString ToLatin1() const; 220 ByteString ToDefANSI() const; 221 ByteString ToUTF8() const; 222 223 // This method will add \0\0 to the end of the string to represent the 224 // wide string terminator. These values are in the string, not just the data, 225 // so GetLength() will include them. 226 ByteString ToUTF16LE() const; 227 228 // Replace the characters &<>'" with HTML entities. 229 WideString EncodeEntities() const; 230 231 protected: 232 using StringData = StringDataTemplate<wchar_t>; 233 234 void ReallocBeforeWrite(size_t nNewLength); 235 void AllocBeforeWrite(size_t nNewLength); 236 void AllocCopy(WideString& dest, size_t nCopyLen, size_t nCopyIndex) const; 237 void AssignCopy(const wchar_t* pSrcData, size_t nSrcLen); 238 void Concat(const wchar_t* pSrcData, size_t nSrcLen); 239 intptr_t ReferenceCountForTesting() const; 240 241 RetainPtr<StringData> m_pData; 242 243 friend class WideString_Assign_Test; 244 friend class WideString_ConcatInPlace_Test; 245 friend class WideString_Construct_Test; 246 friend class StringPool_WideString_Test; 247 }; 248 249 inline WideString operator+(WideStringView str1, WideStringView str2) { 250 return WideString(str1, str2); 251 } 252 inline WideString operator+(WideStringView str1, const wchar_t* str2) { 253 return WideString(str1, str2); 254 } 255 inline WideString operator+(const wchar_t* str1, WideStringView str2) { 256 return WideString(str1, str2); 257 } 258 inline WideString operator+(WideStringView str1, wchar_t ch) { 259 return WideString(str1, WideStringView(ch)); 260 } 261 inline WideString operator+(wchar_t ch, WideStringView str2) { 262 return WideString(WideStringView(ch), str2); 263 } 264 inline WideString operator+(const WideString& str1, const WideString& str2) { 265 return WideString(str1.AsStringView(), str2.AsStringView()); 266 } 267 inline WideString operator+(const WideString& str1, wchar_t ch) { 268 return WideString(str1.AsStringView(), WideStringView(ch)); 269 } 270 inline WideString operator+(wchar_t ch, const WideString& str2) { 271 return WideString(WideStringView(ch), str2.AsStringView()); 272 } 273 inline WideString operator+(const WideString& str1, const wchar_t* str2) { 274 return WideString(str1.AsStringView(), str2); 275 } 276 inline WideString operator+(const wchar_t* str1, const WideString& str2) { 277 return WideString(str1, str2.AsStringView()); 278 } 279 inline WideString operator+(const WideString& str1, WideStringView str2) { 280 return WideString(str1.AsStringView(), str2); 281 } 282 inline WideString operator+(WideStringView str1, const WideString& str2) { 283 return WideString(str1, str2.AsStringView()); 284 } 285 inline bool operator==(const wchar_t* lhs, const WideString& rhs) { 286 return rhs == lhs; 287 } 288 inline bool operator==(WideStringView lhs, const WideString& rhs) { 289 return rhs == lhs; 290 } 291 inline bool operator!=(const wchar_t* lhs, const WideString& rhs) { 292 return rhs != lhs; 293 } 294 inline bool operator!=(WideStringView lhs, const WideString& rhs) { 295 return rhs != lhs; 296 } 297 inline bool operator<(const wchar_t* lhs, const WideString& rhs) { 298 return rhs.Compare(lhs) > 0; 299 } 300 301 std::wostream& operator<<(std::wostream& os, const WideString& str); 302 std::ostream& operator<<(std::ostream& os, const WideString& str); 303 std::wostream& operator<<(std::wostream& os, WideStringView str); 304 std::ostream& operator<<(std::ostream& os, WideStringView str); 305 306 } // namespace fxcrt 307 308 using WideString = fxcrt::WideString; 309 310 uint32_t FX_HashCode_GetW(WideStringView str); 311 uint32_t FX_HashCode_GetLoweredW(WideStringView str); 312 313 namespace std { 314 315 template <> 316 struct hash<WideString> { 317 size_t operator()(const WideString& str) const { 318 return FX_HashCode_GetW(str.AsStringView()); 319 } 320 }; 321 322 } // namespace std 323 324 extern template struct std::hash<WideString>; 325 326 #endif // CORE_FXCRT_WIDESTRING_H_ 327