base/string_piece.h

   1 // Copyright (c) 2012 The Chromium Authors. All rights reserved.
   2 // Use of this source code is governed by a BSD-style license that can be
   3 // found in the LICENSE file.
   4 // Copied from strings/stringpiece.h with modifications
   5 //
   6 // A string-like object that points to a sized piece of memory.
   7 //
   8 // Functions or methods may use const StringPiece& parameters to accept either
   9 // a "const char*" or a "string" value that will be implicitly converted to
  10 // a StringPiece.  The implicit conversion means that it is often appropriate
  11 // to include this .h file in other files rather than forward-declaring
  12 // StringPiece as would be appropriate for most other Google classes.
  13 //
  14 // Systematic usage of StringPiece is encouraged as it will reduce unnecessary
  15 // conversions from "const char*" to "string" and back again.
  16 //
  17 // StringPiece16 is similar to StringPiece but for base::string16 instead of
  18 // std::string. We do not define as large of a subset of the STL functions
  19 // from basic_string as in StringPiece, but this can be changed if these
  20 // functions (find, find_first_of, etc.) are found to be useful in this context.
  21 //
  22
  23 #ifndef BASE_STRING_PIECE_H_
  24 #define BASE_STRING_PIECE_H_
  25
  26 #include <stddef.h>
  27
  28 #include <iosfwd>
  29 #include <string>
  30
  31 #include "base/base_export.h"
  32 #include "base/basictypes.h"
  33 #include "base/hash_tables.h"
  34 #include "base/string16.h"
  35
  36 namespace base {
  37
  38 template <typename STRING_TYPE> class BasicStringPiece;
  39 typedef BasicStringPiece<std::string> StringPiece;
  40 typedef BasicStringPiece<string16> StringPiece16;
  41
  42 namespace internal {
  43
  44 // Defines the types, methods, operators, and data members common to both
  45 // StringPiece and StringPiece16. Do not refer to this class directly, but
  46 // rather to BasicStringPiece, StringPiece, or StringPiece16.
  47 template <typename STRING_TYPE> class StringPieceDetail {
  48  public:
  49   // standard STL container boilerplate
  50   typedef size_t size_type;
  51   typedef typename STRING_TYPE::value_type value_type;
  52   typedef const value_type* pointer;
  53   typedef const value_type& reference;
  54   typedef const value_type& const_reference;
  55   typedef ptrdiff_t difference_type;
  56   typedef const value_type* const_iterator;
  57   typedef std::reverse_iterator<const_iterator> const_reverse_iterator;
  58
  59   static const size_type npos;
  60
  61  public:
  62   // We provide non-explicit singleton constructors so users can pass
  63   // in a "const char*" or a "string" wherever a "StringPiece" is
  64   // expected (likewise for char16, string16, StringPiece16).
  65   StringPieceDetail() : ptr_(NULL), length_(0) {}
  66   StringPieceDetail(const value_type* str)
  67       : ptr_(str),
  68         length_((str == NULL) ? 0 : STRING_TYPE::traits_type::length(str)) {}
  69   StringPieceDetail(const STRING_TYPE& str)
  70       : ptr_(str.data()), length_(str.size()) {}
  71   StringPieceDetail(const value_type* offset, size_type len)
  72       : ptr_(offset), length_(len) {}
  73   StringPieceDetail(const typename STRING_TYPE::const_iterator& begin,
  74                     const typename STRING_TYPE::const_iterator& end)
  75       : ptr_((end > begin) ? &(*begin) : NULL),
  76         length_((end > begin) ? (size_type)(end - begin) : 0) {}
  77
  78   // data() may return a pointer to a buffer with embedded NULs, and the
  79   // returned buffer may or may not be null terminated.  Therefore it is
  80   // typically a mistake to pass data() to a routine that expects a NUL
  81   // terminated string.
  82   const value_type* data() const { return ptr_; }
  83   size_type size() const { return length_; }
  84   size_type length() const { return length_; }
  85   bool empty() const { return length_ == 0; }
  86
  87   void clear() {
  88     ptr_ = NULL;
  89     length_ = 0;
  90   }
  91   void set(const value_type* data, size_type len) {
  92     ptr_ = data;
  93     length_ = len;
  94   }
  95   void set(const value_type* str) {
  96     ptr_ = str;
  97     length_ = str ? STRING_TYPE::traits_type::length(str) : 0;
  98   }
  99
 100   value_type operator[](size_type i) const { return ptr_[i]; }
 101
 102   void remove_prefix(size_type n) {
 103     ptr_ += n;
 104     length_ -= n;
 105   }
 106
 107   void remove_suffix(size_type n) {
 108     length_ -= n;
 109   }
 110
 111   int compare(const BasicStringPiece<STRING_TYPE>& x) const {
 112     int r = wordmemcmp(
 113         ptr_, x.ptr_, (length_ < x.length_ ? length_ : x.length_));
 114     if (r == 0) {
 115       if (length_ < x.length_) r = -1;
 116       else if (length_ > x.length_) r = +1;
 117     }
 118     return r;
 119   }
 120
 121   STRING_TYPE as_string() const {
 122     // std::string doesn't like to take a NULL pointer even with a 0 size.
 123     return empty() ? STRING_TYPE() : STRING_TYPE(data(), size());
 124   }
 125
 126   const_iterator begin() const { return ptr_; }
 127   const_iterator end() const { return ptr_ + length_; }
 128   const_reverse_iterator rbegin() const {
 129     return const_reverse_iterator(ptr_ + length_);
 130   }
 131   const_reverse_iterator rend() const {
 132     return const_reverse_iterator(ptr_);
 133   }
 134
 135   size_type max_size() const { return length_; }
 136   size_type capacity() const { return length_; }
 137
 138   static int wordmemcmp(const value_type* p,
 139                         const value_type* p2,
 140                         size_type N) {
 141     return STRING_TYPE::traits_type::compare(p, p2, N);
 142   }
 143
 144  protected:
 145   const value_type* ptr_;
 146   size_type     length_;
 147 };
 148
 149 template <typename STRING_TYPE>
 150 const typename StringPieceDetail<STRING_TYPE>::size_type
 151 StringPieceDetail<STRING_TYPE>::npos =
 152     typename StringPieceDetail<STRING_TYPE>::size_type(-1);
 153
 154 // MSVC doesn't like complex extern templates and DLLs.
 155 #if !defined(COMPILER_MSVC)
 156 extern template class BASE_EXPORT StringPieceDetail<std::string>;
 157 extern template class BASE_EXPORT StringPieceDetail<string16>;
 158 #endif
 159
 160 BASE_EXPORT void CopyToString(const StringPiece& self, std::string* target);
 161 BASE_EXPORT void AppendToString(const StringPiece& self, std::string* target);
 162 BASE_EXPORT StringPieceDetail<std::string>::size_type copy(
 163     const StringPiece& self,
 164     char* buf,
 165     StringPieceDetail<std::string>::size_type n,
 166     StringPieceDetail<std::string>::size_type pos);
 167 BASE_EXPORT StringPieceDetail<std::string>::size_type find(
 168     const StringPiece& self,
 169     const StringPiece& s,
 170     StringPieceDetail<std::string>::size_type pos);
 171 BASE_EXPORT StringPieceDetail<std::string>::size_type find(
 172     const StringPiece& self,
 173     char c,
 174     StringPieceDetail<std::string>::size_type pos);
 175 BASE_EXPORT StringPieceDetail<std::string>::size_type rfind(
 176     const StringPiece& self,
 177     const StringPiece& s,
 178     StringPieceDetail<std::string>::size_type pos);
 179 BASE_EXPORT StringPieceDetail<std::string>::size_type rfind(
 180     const StringPiece& self,
 181     char c,
 182     StringPieceDetail<std::string>::size_type pos);
 183 BASE_EXPORT StringPieceDetail<std::string>::size_type find_first_of(
 184     const StringPiece& self,
 185     const StringPiece& s,
 186     StringPieceDetail<std::string>::size_type pos);
 187 BASE_EXPORT StringPieceDetail<std::string>::size_type find_first_not_of(
 188     const StringPiece& self,
 189     const StringPiece& s,
 190     StringPieceDetail<std::string>::size_type pos);
 191 BASE_EXPORT StringPieceDetail<std::string>::size_type find_first_not_of(
 192     const StringPiece& self,
 193     char c,
 194     StringPieceDetail<std::string>::size_type pos);
 195 BASE_EXPORT StringPieceDetail<std::string>::size_type find_last_of(
 196     const StringPiece& self,
 197     const StringPiece& s,
 198     StringPieceDetail<std::string>::size_type pos);
 199 BASE_EXPORT StringPieceDetail<std::string>::size_type find_last_of(
 200     const StringPiece& self,
 201     char c,
 202     StringPieceDetail<std::string>::size_type pos);
 203 BASE_EXPORT StringPieceDetail<std::string>::size_type find_last_not_of(
 204     const StringPiece& self,
 205     const StringPiece& s,
 206     StringPieceDetail<std::string>::size_type pos);
 207 BASE_EXPORT StringPieceDetail<std::string>::size_type find_last_not_of(
 208     const StringPiece& self,
 209     char c,
 210     StringPieceDetail<std::string>::size_type pos);
 211 BASE_EXPORT StringPiece substr(const StringPiece& self,
 212                                StringPieceDetail<std::string>::size_type pos,
 213                                StringPieceDetail<std::string>::size_type n);
 214 }  // namespace internal
 215
 216 // Defines the template type that is instantiated as either StringPiece or
 217 // StringPiece16.
 218 template <typename STRING_TYPE> class BasicStringPiece :
 219     public internal::StringPieceDetail<STRING_TYPE> {
 220  public:
 221   typedef typename internal::StringPieceDetail<STRING_TYPE>::value_type
 222       value_type;
 223   typedef typename internal::StringPieceDetail<STRING_TYPE>::size_type
 224       size_type;
 225
 226   BasicStringPiece() {}
 227   BasicStringPiece(const value_type*str)
 228       : internal::StringPieceDetail<STRING_TYPE>(str) {}
 229   BasicStringPiece(const STRING_TYPE& str)
 230       : internal::StringPieceDetail<STRING_TYPE>(str) {}
 231   BasicStringPiece(const value_type* offset, size_type len)
 232       : internal::StringPieceDetail<STRING_TYPE>(offset, len) {}
 233   BasicStringPiece(const typename STRING_TYPE::const_iterator& begin,
 234                    const typename STRING_TYPE::const_iterator& end)
 235       : internal::StringPieceDetail<STRING_TYPE>(begin, end) {}
 236 };
 237
 238 // Specializes BasicStringPiece for std::string to add a few operations that
 239 // are not needed for string16.
 240 template <> class BasicStringPiece<std::string> :
 241     public internal::StringPieceDetail<std::string> {
 242  public:
 243   BasicStringPiece() {}
 244   BasicStringPiece(const char* str)
 245       : internal::StringPieceDetail<std::string>(str) {}
 246   BasicStringPiece(const std::string& str)
 247       : internal::StringPieceDetail<std::string>(str) {}
 248   BasicStringPiece(const char* offset, size_type len)
 249       : internal::StringPieceDetail<std::string>(offset, len) {}
 250   BasicStringPiece(const std::string::const_iterator& begin,
 251                    const std::string::const_iterator& end)
 252       : internal::StringPieceDetail<std::string>(begin, end) {}
 253
 254   // Prevent the following overload of set() from hiding the definitions in the
 255   // base class.
 256   using internal::StringPieceDetail<std::string>::set;
 257
 258   void set(const void* data, size_type len) {
 259     ptr_ = reinterpret_cast<const value_type*>(data);
 260     length_ = len;
 261   }
 262
 263   void CopyToString(std::string* target) const {
 264     internal::CopyToString(*this, target);
 265   }
 266
 267   void AppendToString(std::string* target) const {
 268     internal::AppendToString(*this, target);
 269   }
 270
 271   // Does "this" start with "x"
 272   bool starts_with(const BasicStringPiece& x) const {
 273     return ((length_ >= x.length_) &&
 274             (wordmemcmp(ptr_, x.ptr_, x.length_) == 0));
 275   }
 276
 277   // Does "this" end with "x"
 278   bool ends_with(const BasicStringPiece& x) const {
 279     return ((length_ >= x.length_) &&
 280             (wordmemcmp(ptr_ + (length_-x.length_), x.ptr_, x.length_) == 0));
 281   }
 282
 283   size_type copy(char* buf, size_type n, size_type pos = 0) const {
 284     return internal::copy(*this, buf, n, pos);
 285   }
 286
 287   size_type find(const BasicStringPiece& s, size_type pos = 0) const {
 288     return internal::find(*this, s, pos);
 289   }
 290
 291   size_type find(char c, size_type pos = 0) const {
 292     return internal::find(*this, c, pos);
 293   }
 294
 295   size_type rfind(const BasicStringPiece& s, size_type pos = npos) const {
 296     return internal::rfind(*this, s, pos);
 297   }
 298
 299   size_type rfind(char c, size_type pos = npos) const {
 300     return internal::rfind(*this, c, pos);
 301   }
 302
 303   size_type find_first_of(const BasicStringPiece& s, size_type pos = 0) const {
 304     return internal::find_first_of(*this, s, pos);
 305   }
 306
 307   size_type find_first_of(char c, size_type pos = 0) const {
 308     return find(c, pos);
 309   }
 310
 311   size_type find_first_not_of(const BasicStringPiece& s,
 312                               size_type pos = 0) const {
 313     return internal::find_first_not_of(*this, s, pos);
 314   }
 315
 316   size_type find_first_not_of(char c, size_type pos = 0) const {
 317     return internal::find_first_not_of(*this, c, pos);
 318   }
 319
 320   size_type find_last_of(const BasicStringPiece& s,
 321                          size_type pos = npos) const {
 322     return internal::find_last_of(*this, s, pos);
 323   }
 324
 325   size_type find_last_of(char c, size_type pos = npos) const {
 326     return rfind(c, pos);
 327   }
 328
 329   size_type find_last_not_of(const BasicStringPiece& s,
 330                              size_type pos = npos) const {
 331     return internal::find_last_not_of(*this, s, pos);
 332   }
 333
 334   size_type find_last_not_of(char c, size_type pos = npos) const {
 335     return internal::find_last_not_of(*this, c, pos);
 336   }
 337
 338   BasicStringPiece substr(size_type pos, size_type n = npos) const {
 339     return internal::substr(*this, pos, n);
 340   }
 341 };
 342
 343 // MSVC doesn't like complex extern templates and DLLs.
 344 #if !defined(COMPILER_MSVC)
 345 // We can't explicitly declare the std::string instantiation here because it was
 346 // already instantiated when specialized, above. Not only is it a no-op, but
 347 // currently it also crashes Clang (see http://crbug.com/107412).
 348 extern template class BASE_EXPORT BasicStringPiece<string16>;
 349 #endif
 350
 351 BASE_EXPORT bool operator==(const StringPiece& x, const StringPiece& y);
 352
 353 inline bool operator!=(const StringPiece& x, const StringPiece& y) {
 354   return !(x == y);
 355 }
 356
 357 inline bool operator<(const StringPiece& x, const StringPiece& y) {
 358   const int r = StringPiece::wordmemcmp(
 359       x.data(), y.data(), (x.size() < y.size() ? x.size() : y.size()));
 360   return ((r < 0) || ((r == 0) && (x.size() < y.size())));
 361 }
 362
 363 inline bool operator>(const StringPiece& x, const StringPiece& y) {
 364   return y < x;
 365 }
 366
 367 inline bool operator<=(const StringPiece& x, const StringPiece& y) {
 368   return !(x > y);
 369 }
 370
 371 inline bool operator>=(const StringPiece& x, const StringPiece& y) {
 372   return !(x < y);
 373 }
 374
 375 inline bool operator==(const StringPiece16& x, const StringPiece16& y) {
 376   if (x.size() != y.size())
 377     return false;
 378
 379   return StringPiece16::wordmemcmp(x.data(), y.data(), x.size()) == 0;
 380 }
 381
 382 inline bool operator!=(const StringPiece16& x, const StringPiece16& y) {
 383   return !(x == y);
 384 }
 385
 386 inline bool operator<(const StringPiece16& x, const StringPiece16& y) {
 387   const int r = StringPiece16::wordmemcmp(
 388       x.data(), y.data(), (x.size() < y.size() ? x.size() : y.size()));
 389   return ((r < 0) || ((r == 0) && (x.size() < y.size())));
 390 }
 391
 392 inline bool operator>(const StringPiece16& x, const StringPiece16& y) {
 393   return y < x;
 394 }
 395
 396 inline bool operator<=(const StringPiece16& x, const StringPiece16& y) {
 397   return !(x > y);
 398 }
 399
 400 inline bool operator>=(const StringPiece16& x, const StringPiece16& y) {
 401   return !(x < y);
 402 }
 403
 404 BASE_EXPORT std::ostream& operator<<(std::ostream& o,
 405                                      const StringPiece& piece);
 406
 407 }  // namespace base
 408
 409 // We provide appropriate hash functions so StringPiece and StringPiece16 can
 410 // be used as keys in hash sets and maps.
 411
 412 // This hash function is copied from base/hash_tables.h. We don't use the
 413 // ones already defined for string and string16 directly because it would
 414 // require the string constructors to be called, which we don't want.
 415 #define HASH_STRING_PIECE(StringPieceType, string_piece)                \
 416   std::size_t result = 0;                                               \
 417   for (StringPieceType::const_iterator i = string_piece.begin();        \
 418        i != string_piece.end(); ++i)                                    \
 419     result = (result * 131) + *i;                                       \
 420   return result;                                                        \
 421
 422 namespace BASE_HASH_NAMESPACE {
 423 #if defined(COMPILER_GCC)
 424
 425 template<>
 426 struct hash<base::StringPiece> {
 427   std::size_t operator()(const base::StringPiece& sp) const {
 428     HASH_STRING_PIECE(base::StringPiece, sp);
 429   }
 430 };
 431 template<>
 432 struct hash<base::StringPiece16> {
 433   std::size_t operator()(const base::StringPiece16& sp16) const {
 434     HASH_STRING_PIECE(base::StringPiece16, sp16);
 435   }
 436 };
 437
 438 #elif defined(COMPILER_MSVC)
 439
 440 inline size_t hash_value(const base::StringPiece& sp) {
 441   HASH_STRING_PIECE(base::StringPiece, sp);
 442 }
 443 inline size_t hash_value(const base::StringPiece16& sp16) {
 444   HASH_STRING_PIECE(base::StringPiece16, sp16);
 445 }
 446
 447 #endif  // COMPILER
 448
 449 }  // namespace BASE_HASH_NAMESPACE
 450
 451 #endif  // BASE_STRING_PIECE_H_