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