base/strings/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 // You can use StringPiece as a function or method parameter.  A StringPiece
   9 // parameter can receive a double-quoted string literal argument, a "const
  10 // char*" argument, a string argument, or a StringPiece argument with no data
  11 // copying.  Systematic use of StringPiece for arguments reduces data
  12 // copies and strlen() calls.
  13 //
  14 // Prefer passing StringPieces by value:
  15 //   void MyFunction(StringPiece arg);
  16 // If circumstances require, you may also pass by const reference:
  17 //   void MyFunction(const StringPiece& arg);  // not preferred
  18 // Both of these have the same lifetime semantics.  Passing by value
  19 // generates slightly smaller code.  For more discussion, Googlers can see
  20 // the thread go/stringpiecebyvalue on c-users.
  21
  22 #ifndef BASE_STRINGS_STRING_PIECE_H_
  23 #define BASE_STRINGS_STRING_PIECE_H_
  24
  25 #include <stddef.h>
  26
  27 #include <iosfwd>
  28 #include <string>
  29
  30 #include "base/base_export.h"
  31 #include "base/basictypes.h"
  32 #include "base/containers/hash_tables.h"
  33 #include "base/strings/string16.h"
  34
  35 namespace base {
  36
  37 template <typename STRING_TYPE> class BasicStringPiece;
  38 typedef BasicStringPiece<std::string> StringPiece;
  39 typedef BasicStringPiece<string16> StringPiece16;
  40
  41 // internal --------------------------------------------------------------------
  42
  43 // Many of the StringPiece functions use different implementations for the
  44 // 8-bit and 16-bit versions, and we don't want lots of template expansions in
  45 // this (very common) header that will slow down compilation.
  46 //
  47 // So here we define overloaded functions called by the StringPiece template.
  48 // For those that share an implementation, the two versions will expand to a
  49 // template internal to the .cc file.
  50 namespace internal {
  51
  52 BASE_EXPORT void CopyToString(const StringPiece& self, std::string* target);
  53 BASE_EXPORT void CopyToString(const StringPiece16& self, string16* target);
  54
  55 BASE_EXPORT void AppendToString(const StringPiece& self, std::string* target);
  56 BASE_EXPORT void AppendToString(const StringPiece16& self, string16* target);
  57
  58 BASE_EXPORT size_t copy(const StringPiece& self,
  59                         char* buf,
  60                         size_t n,
  61                         size_t pos);
  62 BASE_EXPORT size_t copy(const StringPiece16& self,
  63                         char16* buf,
  64                         size_t n,
  65                         size_t pos);
  66
  67 BASE_EXPORT size_t find(const StringPiece& self,
  68                         const StringPiece& s,
  69                         size_t pos);
  70 BASE_EXPORT size_t find(const StringPiece16& self,
  71                         const StringPiece16& s,
  72                         size_t pos);
  73 BASE_EXPORT size_t find(const StringPiece& self,
  74                         char c,
  75                         size_t pos);
  76 BASE_EXPORT size_t find(const StringPiece16& self,
  77                         char16 c,
  78                         size_t pos);
  79
  80 BASE_EXPORT size_t rfind(const StringPiece& self,
  81                          const StringPiece& s,
  82                          size_t pos);
  83 BASE_EXPORT size_t rfind(const StringPiece16& self,
  84                          const StringPiece16& s,
  85                          size_t pos);
  86 BASE_EXPORT size_t rfind(const StringPiece& self,
  87                          char c,
  88                          size_t pos);
  89 BASE_EXPORT size_t rfind(const StringPiece16& self,
  90                          char16 c,
  91                          size_t pos);
  92
  93 BASE_EXPORT size_t find_first_of(const StringPiece& self,
  94                                  const StringPiece& s,
  95                                  size_t pos);
  96 BASE_EXPORT size_t find_first_of(const StringPiece16& self,
  97                                  const StringPiece16& s,
  98                                  size_t pos);
  99
 100 BASE_EXPORT size_t find_first_not_of(const StringPiece& self,
 101                                      const StringPiece& s,
 102                                      size_t pos);
 103 BASE_EXPORT size_t find_first_not_of(const StringPiece16& self,
 104                                      const StringPiece16& s,
 105                                      size_t pos);
 106 BASE_EXPORT size_t find_first_not_of(const StringPiece& self,
 107                                      char c,
 108                                      size_t pos);
 109 BASE_EXPORT size_t find_first_not_of(const StringPiece16& self,
 110                                      char16 c,
 111                                      size_t pos);
 112
 113 BASE_EXPORT size_t find_last_of(const StringPiece& self,
 114                                 const StringPiece& s,
 115                                 size_t pos);
 116 BASE_EXPORT size_t find_last_of(const StringPiece16& self,
 117                                 const StringPiece16& s,
 118                                 size_t pos);
 119 BASE_EXPORT size_t find_last_of(const StringPiece& self,
 120                                 char c,
 121                                 size_t pos);
 122 BASE_EXPORT size_t find_last_of(const StringPiece16& self,
 123                                 char16 c,
 124                                 size_t pos);
 125
 126 BASE_EXPORT size_t find_last_not_of(const StringPiece& self,
 127                                     const StringPiece& s,
 128                                     size_t pos);
 129 BASE_EXPORT size_t find_last_not_of(const StringPiece16& self,
 130                                     const StringPiece16& s,
 131                                     size_t pos);
 132 BASE_EXPORT size_t find_last_not_of(const StringPiece16& self,
 133                                     char16 c,
 134                                     size_t pos);
 135 BASE_EXPORT size_t find_last_not_of(const StringPiece& self,
 136                                     char c,
 137                                     size_t pos);
 138
 139 BASE_EXPORT StringPiece substr(const StringPiece& self,
 140                                size_t pos,
 141                                size_t n);
 142 BASE_EXPORT StringPiece16 substr(const StringPiece16& self,
 143                                  size_t pos,
 144                                  size_t n);
 145
 146 #if !defined(NDEBUG) || defined(DCHECK_ALWAYS_ON)
 147 // Asserts that begin <= end to catch some errors with iterator usage.
 148 BASE_EXPORT void AssertIteratorsInOrder(std::string::const_iterator begin,
 149                                         std::string::const_iterator end);
 150 BASE_EXPORT void AssertIteratorsInOrder(string16::const_iterator begin,
 151                                         string16::const_iterator end);
 152 #endif
 153
 154 }  // namespace internal
 155
 156 // BasicStringPiece ------------------------------------------------------------
 157
 158 // Defines the types, methods, operators, and data members common to both
 159 // StringPiece and StringPiece16. Do not refer to this class directly, but
 160 // rather to BasicStringPiece, StringPiece, or StringPiece16.
 161 //
 162 // This is templatized by string class type rather than character type, so
 163 // BasicStringPiece<std::string> or BasicStringPiece<base::string16>.
 164 template <typename STRING_TYPE> class BasicStringPiece {
 165  public:
 166   // Standard STL container boilerplate.
 167   typedef size_t size_type;
 168   typedef typename STRING_TYPE::value_type value_type;
 169   typedef const value_type* pointer;
 170   typedef const value_type& reference;
 171   typedef const value_type& const_reference;
 172   typedef ptrdiff_t difference_type;
 173   typedef const value_type* const_iterator;
 174   typedef std::reverse_iterator<const_iterator> const_reverse_iterator;
 175
 176   static const size_type npos;
 177
 178  public:
 179   // We provide non-explicit singleton constructors so users can pass
 180   // in a "const char*" or a "string" wherever a "StringPiece" is
 181   // expected (likewise for char16, string16, StringPiece16).
 182   BasicStringPiece() : ptr_(NULL), length_(0) {}
 183   BasicStringPiece(const value_type* str)
 184       : ptr_(str),
 185         length_((str == NULL) ? 0 : STRING_TYPE::traits_type::length(str)) {}
 186   BasicStringPiece(const STRING_TYPE& str)
 187       : ptr_(str.data()), length_(str.size()) {}
 188   BasicStringPiece(const value_type* offset, size_type len)
 189       : ptr_(offset), length_(len) {}
 190   BasicStringPiece(const typename STRING_TYPE::const_iterator& begin,
 191                    const typename STRING_TYPE::const_iterator& end) {
 192 #if !defined(NDEBUG) || defined(DCHECK_ALWAYS_ON)
 193     // This assertion is done out-of-line to avoid bringing in logging.h and
 194     // instantiating logging macros for every instantiation.
 195     internal::AssertIteratorsInOrder(begin, end);
 196 #endif
 197     length_ = static_cast<size_t>(std::distance(begin, end));
 198
 199     // The length test before assignment is to avoid dereferencing an iterator
 200     // that may point to the end() of a string.
 201     ptr_ = length_ > 0 ? &*begin : nullptr;
 202   }
 203
 204   // data() may return a pointer to a buffer with embedded NULs, and the
 205   // returned buffer may or may not be null terminated.  Therefore it is
 206   // typically a mistake to pass data() to a routine that expects a NUL
 207   // terminated string.
 208   const value_type* data() const { return ptr_; }
 209   size_type size() const { return length_; }
 210   size_type length() const { return length_; }
 211   bool empty() const { return length_ == 0; }
 212
 213   void clear() {
 214     ptr_ = NULL;
 215     length_ = 0;
 216   }
 217   void set(const value_type* data, size_type len) {
 218     ptr_ = data;
 219     length_ = len;
 220   }
 221   void set(const value_type* str) {
 222     ptr_ = str;
 223     length_ = str ? STRING_TYPE::traits_type::length(str) : 0;
 224   }
 225
 226   value_type operator[](size_type i) const { return ptr_[i]; }
 227
 228   void remove_prefix(size_type n) {
 229     ptr_ += n;
 230     length_ -= n;
 231   }
 232
 233   void remove_suffix(size_type n) {
 234     length_ -= n;
 235   }
 236
 237   int compare(const BasicStringPiece<STRING_TYPE>& x) const {
 238     int r = wordmemcmp(
 239         ptr_, x.ptr_, (length_ < x.length_ ? length_ : x.length_));
 240     if (r == 0) {
 241       if (length_ < x.length_) r = -1;
 242       else if (length_ > x.length_) r = +1;
 243     }
 244     return r;
 245   }
 246
 247   STRING_TYPE as_string() const {
 248     // std::string doesn't like to take a NULL pointer even with a 0 size.
 249     return empty() ? STRING_TYPE() : STRING_TYPE(data(), size());
 250   }
 251
 252   const_iterator begin() const { return ptr_; }
 253   const_iterator end() const { return ptr_ + length_; }
 254   const_reverse_iterator rbegin() const {
 255     return const_reverse_iterator(ptr_ + length_);
 256   }
 257   const_reverse_iterator rend() const {
 258     return const_reverse_iterator(ptr_);
 259   }
 260
 261   size_type max_size() const { return length_; }
 262   size_type capacity() const { return length_; }
 263
 264   static int wordmemcmp(const value_type* p,
 265                         const value_type* p2,
 266                         size_type N) {
 267     return STRING_TYPE::traits_type::compare(p, p2, N);
 268   }
 269
 270   // Sets the value of the given string target type to be the current string.
 271   // This saves a temporary over doing |a = b.as_string()|
 272   void CopyToString(STRING_TYPE* target) const {
 273     internal::CopyToString(*this, target);
 274   }
 275
 276   void AppendToString(STRING_TYPE* target) const {
 277     internal::AppendToString(*this, target);
 278   }
 279
 280   size_type copy(value_type* buf, size_type n, size_type pos = 0) const {
 281     return internal::copy(*this, buf, n, pos);
 282   }
 283
 284   // Does "this" start with "x"
 285   bool starts_with(const BasicStringPiece& x) const {
 286     return ((this->length_ >= x.length_) &&
 287             (wordmemcmp(this->ptr_, x.ptr_, x.length_) == 0));
 288   }
 289
 290   // Does "this" end with "x"
 291   bool ends_with(const BasicStringPiece& x) const {
 292     return ((this->length_ >= x.length_) &&
 293             (wordmemcmp(this->ptr_ + (this->length_-x.length_),
 294                         x.ptr_, x.length_) == 0));
 295   }
 296
 297   // find: Search for a character or substring at a given offset.
 298   size_type find(const BasicStringPiece<STRING_TYPE>& s,
 299                  size_type pos = 0) const {
 300     return internal::find(*this, s, pos);
 301   }
 302   size_type find(value_type c, size_type pos = 0) const {
 303     return internal::find(*this, c, pos);
 304   }
 305
 306   // rfind: Reverse find.
 307   size_type rfind(const BasicStringPiece& s,
 308                   size_type pos = BasicStringPiece::npos) const {
 309     return internal::rfind(*this, s, pos);
 310   }
 311   size_type rfind(value_type c, size_type pos = BasicStringPiece::npos) const {
 312     return internal::rfind(*this, c, pos);
 313   }
 314
 315   // find_first_of: Find the first occurence of one of a set of characters.
 316   size_type find_first_of(const BasicStringPiece& s,
 317                           size_type pos = 0) const {
 318     return internal::find_first_of(*this, s, pos);
 319   }
 320   size_type find_first_of(value_type c, size_type pos = 0) const {
 321     return find(c, pos);
 322   }
 323
 324   // find_first_not_of: Find the first occurence not of a set of characters.
 325   size_type find_first_not_of(const BasicStringPiece& s,
 326                               size_type pos = 0) const {
 327     return internal::find_first_not_of(*this, s, pos);
 328   }
 329   size_type find_first_not_of(value_type c, size_type pos = 0) const {
 330     return internal::find_first_not_of(*this, c, pos);
 331   }
 332
 333   // find_last_of: Find the last occurence of one of a set of characters.
 334   size_type find_last_of(const BasicStringPiece& s,
 335                          size_type pos = BasicStringPiece::npos) const {
 336     return internal::find_last_of(*this, s, pos);
 337   }
 338   size_type find_last_of(value_type c,
 339                          size_type pos = BasicStringPiece::npos) const {
 340     return rfind(c, pos);
 341   }
 342
 343   // find_last_not_of: Find the last occurence not of a set of characters.
 344   size_type find_last_not_of(const BasicStringPiece& s,
 345                              size_type pos = BasicStringPiece::npos) const {
 346     return internal::find_last_not_of(*this, s, pos);
 347   }
 348   size_type find_last_not_of(value_type c,
 349                              size_type pos = BasicStringPiece::npos) const {
 350     return internal::find_last_not_of(*this, c, pos);
 351   }
 352
 353   // substr.
 354   BasicStringPiece substr(size_type pos,
 355                           size_type n = BasicStringPiece::npos) const {
 356     return internal::substr(*this, pos, n);
 357   }
 358
 359  protected:
 360   const value_type* ptr_;
 361   size_type     length_;
 362 };
 363
 364 template <typename STRING_TYPE>
 365 const typename BasicStringPiece<STRING_TYPE>::size_type
 366 BasicStringPiece<STRING_TYPE>::npos =
 367     typename BasicStringPiece<STRING_TYPE>::size_type(-1);
 368
 369 // MSVC doesn't like complex extern templates and DLLs.
 370 #if !defined(COMPILER_MSVC)
 371 extern template class BASE_EXPORT BasicStringPiece<std::string>;
 372 extern template class BASE_EXPORT BasicStringPiece<string16>;
 373 #endif
 374
 375 // StingPiece operators --------------------------------------------------------
 376
 377 BASE_EXPORT bool operator==(const StringPiece& x, const StringPiece& y);
 378
 379 inline bool operator!=(const StringPiece& x, const StringPiece& y) {
 380   return !(x == y);
 381 }
 382
 383 inline bool operator<(const StringPiece& x, const StringPiece& y) {
 384   const int r = StringPiece::wordmemcmp(
 385       x.data(), y.data(), (x.size() < y.size() ? x.size() : y.size()));
 386   return ((r < 0) || ((r == 0) && (x.size() < y.size())));
 387 }
 388
 389 inline bool operator>(const StringPiece& x, const StringPiece& y) {
 390   return y < x;
 391 }
 392
 393 inline bool operator<=(const StringPiece& x, const StringPiece& y) {
 394   return !(x > y);
 395 }
 396
 397 inline bool operator>=(const StringPiece& x, const StringPiece& y) {
 398   return !(x < y);
 399 }
 400
 401 // StringPiece16 operators -----------------------------------------------------
 402
 403 inline bool operator==(const StringPiece16& x, const StringPiece16& y) {
 404   if (x.size() != y.size())
 405     return false;
 406
 407   return StringPiece16::wordmemcmp(x.data(), y.data(), x.size()) == 0;
 408 }
 409
 410 inline bool operator!=(const StringPiece16& x, const StringPiece16& y) {
 411   return !(x == y);
 412 }
 413
 414 inline bool operator<(const StringPiece16& x, const StringPiece16& y) {
 415   const int r = StringPiece16::wordmemcmp(
 416       x.data(), y.data(), (x.size() < y.size() ? x.size() : y.size()));
 417   return ((r < 0) || ((r == 0) && (x.size() < y.size())));
 418 }
 419
 420 inline bool operator>(const StringPiece16& x, const StringPiece16& y) {
 421   return y < x;
 422 }
 423
 424 inline bool operator<=(const StringPiece16& x, const StringPiece16& y) {
 425   return !(x > y);
 426 }
 427
 428 inline bool operator>=(const StringPiece16& x, const StringPiece16& y) {
 429   return !(x < y);
 430 }
 431
 432 BASE_EXPORT std::ostream& operator<<(std::ostream& o,
 433                                      const StringPiece& piece);
 434
 435 }  // namespace base
 436
 437 // Hashing ---------------------------------------------------------------------
 438
 439 // We provide appropriate hash functions so StringPiece and StringPiece16 can
 440 // be used as keys in hash sets and maps.
 441
 442 // This hash function is copied from base/containers/hash_tables.h. We don't
 443 // use the ones already defined for string and string16 directly because it
 444 // would require the string constructors to be called, which we don't want.
 445 #define HASH_STRING_PIECE(StringPieceType, string_piece)                \
 446   std::size_t result = 0;                                               \
 447   for (StringPieceType::const_iterator i = string_piece.begin();        \
 448        i != string_piece.end(); ++i)                                    \
 449     result = (result * 131) + *i;                                       \
 450   return result;                                                        \
 451
 452 namespace BASE_HASH_NAMESPACE {
 453
 454 template<>
 455 struct hash<base::StringPiece> {
 456   std::size_t operator()(const base::StringPiece& sp) const {
 457     HASH_STRING_PIECE(base::StringPiece, sp);
 458   }
 459 };
 460 template<>
 461 struct hash<base::StringPiece16> {
 462   std::size_t operator()(const base::StringPiece16& sp16) const {
 463     HASH_STRING_PIECE(base::StringPiece16, sp16);
 464   }
 465 };
 466
 467 }  // namespace BASE_HASH_NAMESPACE
 468
 469 #endif  // BASE_STRINGS_STRING_PIECE_H_