michael@0: // Copyright (c) 2012 The Chromium Authors. All rights reserved.
michael@0: // Use of this source code is governed by a BSD-style license that can be
michael@0: // found in the LICENSE file.
michael@0: // Copied from strings/stringpiece.h with modifications
michael@0: //
michael@0: // A string-like object that points to a sized piece of memory.
michael@0: //
michael@0: // Functions or methods may use const StringPiece& parameters to accept either
michael@0: // a "const char*" or a "string" value that will be implicitly converted to
michael@0: // a StringPiece.  The implicit conversion means that it is often appropriate
michael@0: // to include this .h file in other files rather than forward-declaring
michael@0: // StringPiece as would be appropriate for most other Google classes.
michael@0: //
michael@0: // Systematic usage of StringPiece is encouraged as it will reduce unnecessary
michael@0: // conversions from "const char*" to "string" and back again.
michael@0: //
michael@0: // StringPiece16 is similar to StringPiece but for base::string16 instead of
michael@0: // std::string. We do not define as large of a subset of the STL functions
michael@0: // from basic_string as in StringPiece, but this can be changed if these
michael@0: // functions (find, find_first_of, etc.) are found to be useful in this context.
michael@0: //
michael@0: 
michael@0: #ifndef BASE_STRINGS_STRING_PIECE_H_
michael@0: #define BASE_STRINGS_STRING_PIECE_H_
michael@0: 
michael@0: #include <stddef.h>
michael@0: 
michael@0: #include <iosfwd>
michael@0: #include <string>
michael@0: 
michael@0: #include "base/base_export.h"
michael@0: #include "base/basictypes.h"
michael@0: #include "base/containers/hash_tables.h"
michael@0: #include "base/strings/string16.h"
michael@0: 
michael@0: namespace base {
michael@0: 
michael@0: template <typename STRING_TYPE> class BasicStringPiece;
michael@0: typedef BasicStringPiece<std::string> StringPiece;
michael@0: typedef BasicStringPiece<string16> StringPiece16;
michael@0: 
michael@0: namespace internal {
michael@0: 
michael@0: // Defines the types, methods, operators, and data members common to both
michael@0: // StringPiece and StringPiece16. Do not refer to this class directly, but
michael@0: // rather to BasicStringPiece, StringPiece, or StringPiece16.
michael@0: template <typename STRING_TYPE> class StringPieceDetail {
michael@0:  public:
michael@0:   // standard STL container boilerplate
michael@0:   typedef size_t size_type;
michael@0:   typedef typename STRING_TYPE::value_type value_type;
michael@0:   typedef const value_type* pointer;
michael@0:   typedef const value_type& reference;
michael@0:   typedef const value_type& const_reference;
michael@0:   typedef ptrdiff_t difference_type;
michael@0:   typedef const value_type* const_iterator;
michael@0:   typedef std::reverse_iterator<const_iterator> const_reverse_iterator;
michael@0: 
michael@0:   static const size_type npos;
michael@0: 
michael@0:  public:
michael@0:   // We provide non-explicit singleton constructors so users can pass
michael@0:   // in a "const char*" or a "string" wherever a "StringPiece" is
michael@0:   // expected (likewise for char16, string16, StringPiece16).
michael@0:   StringPieceDetail() : ptr_(NULL), length_(0) {}
michael@0:   StringPieceDetail(const value_type* str)
michael@0:       : ptr_(str),
michael@0:         length_((str == NULL) ? 0 : STRING_TYPE::traits_type::length(str)) {}
michael@0:   StringPieceDetail(const STRING_TYPE& str)
michael@0:       : ptr_(str.data()), length_(str.size()) {}
michael@0:   StringPieceDetail(const value_type* offset, size_type len)
michael@0:       : ptr_(offset), length_(len) {}
michael@0:   StringPieceDetail(const typename STRING_TYPE::const_iterator& begin,
michael@0:                     const typename STRING_TYPE::const_iterator& end)
michael@0:       : ptr_((end > begin) ? &(*begin) : NULL),
michael@0:         length_((end > begin) ? (size_type)(end - begin) : 0) {}
michael@0: 
michael@0:   // data() may return a pointer to a buffer with embedded NULs, and the
michael@0:   // returned buffer may or may not be null terminated.  Therefore it is
michael@0:   // typically a mistake to pass data() to a routine that expects a NUL
michael@0:   // terminated string.
michael@0:   const value_type* data() const { return ptr_; }
michael@0:   size_type size() const { return length_; }
michael@0:   size_type length() const { return length_; }
michael@0:   bool empty() const { return length_ == 0; }
michael@0: 
michael@0:   void clear() {
michael@0:     ptr_ = NULL;
michael@0:     length_ = 0;
michael@0:   }
michael@0:   void set(const value_type* data, size_type len) {
michael@0:     ptr_ = data;
michael@0:     length_ = len;
michael@0:   }
michael@0:   void set(const value_type* str) {
michael@0:     ptr_ = str;
michael@0:     length_ = str ? STRING_TYPE::traits_type::length(str) : 0;
michael@0:   }
michael@0: 
michael@0:   value_type operator[](size_type i) const { return ptr_[i]; }
michael@0: 
michael@0:   void remove_prefix(size_type n) {
michael@0:     ptr_ += n;
michael@0:     length_ -= n;
michael@0:   }
michael@0: 
michael@0:   void remove_suffix(size_type n) {
michael@0:     length_ -= n;
michael@0:   }
michael@0: 
michael@0:   int compare(const BasicStringPiece<STRING_TYPE>& x) const {
michael@0:     int r = wordmemcmp(
michael@0:         ptr_, x.ptr_, (length_ < x.length_ ? length_ : x.length_));
michael@0:     if (r == 0) {
michael@0:       if (length_ < x.length_) r = -1;
michael@0:       else if (length_ > x.length_) r = +1;
michael@0:     }
michael@0:     return r;
michael@0:   }
michael@0: 
michael@0:   STRING_TYPE as_string() const {
michael@0:     // std::string doesn't like to take a NULL pointer even with a 0 size.
michael@0:     return empty() ? STRING_TYPE() : STRING_TYPE(data(), size());
michael@0:   }
michael@0: 
michael@0:   const_iterator begin() const { return ptr_; }
michael@0:   const_iterator end() const { return ptr_ + length_; }
michael@0:   const_reverse_iterator rbegin() const {
michael@0:     return const_reverse_iterator(ptr_ + length_);
michael@0:   }
michael@0:   const_reverse_iterator rend() const {
michael@0:     return const_reverse_iterator(ptr_);
michael@0:   }
michael@0: 
michael@0:   size_type max_size() const { return length_; }
michael@0:   size_type capacity() const { return length_; }
michael@0: 
michael@0:   static int wordmemcmp(const value_type* p,
michael@0:                         const value_type* p2,
michael@0:                         size_type N) {
michael@0:     return STRING_TYPE::traits_type::compare(p, p2, N);
michael@0:   }
michael@0: 
michael@0:  protected:
michael@0:   const value_type* ptr_;
michael@0:   size_type     length_;
michael@0: };
michael@0: 
michael@0: template <typename STRING_TYPE>
michael@0: const typename StringPieceDetail<STRING_TYPE>::size_type
michael@0: StringPieceDetail<STRING_TYPE>::npos =
michael@0:     typename StringPieceDetail<STRING_TYPE>::size_type(-1);
michael@0: 
michael@0: // MSVC doesn't like complex extern templates and DLLs.
michael@0: #if !defined(COMPILER_MSVC)
michael@0: extern template class BASE_EXPORT StringPieceDetail<std::string>;
michael@0: extern template class BASE_EXPORT StringPieceDetail<string16>;
michael@0: #endif
michael@0: 
michael@0: BASE_EXPORT void CopyToString(const StringPiece& self, std::string* target);
michael@0: BASE_EXPORT void AppendToString(const StringPiece& self, std::string* target);
michael@0: BASE_EXPORT StringPieceDetail<std::string>::size_type copy(
michael@0:     const StringPiece& self,
michael@0:     char* buf,
michael@0:     StringPieceDetail<std::string>::size_type n,
michael@0:     StringPieceDetail<std::string>::size_type pos);
michael@0: BASE_EXPORT StringPieceDetail<std::string>::size_type find(
michael@0:     const StringPiece& self,
michael@0:     const StringPiece& s,
michael@0:     StringPieceDetail<std::string>::size_type pos);
michael@0: BASE_EXPORT StringPieceDetail<std::string>::size_type find(
michael@0:     const StringPiece& self,
michael@0:     char c,
michael@0:     StringPieceDetail<std::string>::size_type pos);
michael@0: BASE_EXPORT StringPieceDetail<std::string>::size_type rfind(
michael@0:     const StringPiece& self,
michael@0:     const StringPiece& s,
michael@0:     StringPieceDetail<std::string>::size_type pos);
michael@0: BASE_EXPORT StringPieceDetail<std::string>::size_type rfind(
michael@0:     const StringPiece& self,
michael@0:     char c,
michael@0:     StringPieceDetail<std::string>::size_type pos);
michael@0: BASE_EXPORT StringPieceDetail<std::string>::size_type find_first_of(
michael@0:     const StringPiece& self,
michael@0:     const StringPiece& s,
michael@0:     StringPieceDetail<std::string>::size_type pos);
michael@0: BASE_EXPORT StringPieceDetail<std::string>::size_type find_first_not_of(
michael@0:     const StringPiece& self,
michael@0:     const StringPiece& s,
michael@0:     StringPieceDetail<std::string>::size_type pos);
michael@0: BASE_EXPORT StringPieceDetail<std::string>::size_type find_first_not_of(
michael@0:     const StringPiece& self,
michael@0:     char c,
michael@0:     StringPieceDetail<std::string>::size_type pos);
michael@0: BASE_EXPORT StringPieceDetail<std::string>::size_type find_last_of(
michael@0:     const StringPiece& self,
michael@0:     const StringPiece& s,
michael@0:     StringPieceDetail<std::string>::size_type pos);
michael@0: BASE_EXPORT StringPieceDetail<std::string>::size_type find_last_of(
michael@0:     const StringPiece& self,
michael@0:     char c,
michael@0:     StringPieceDetail<std::string>::size_type pos);
michael@0: BASE_EXPORT StringPieceDetail<std::string>::size_type find_last_not_of(
michael@0:     const StringPiece& self,
michael@0:     const StringPiece& s,
michael@0:     StringPieceDetail<std::string>::size_type pos);
michael@0: BASE_EXPORT StringPieceDetail<std::string>::size_type find_last_not_of(
michael@0:     const StringPiece& self,
michael@0:     char c,
michael@0:     StringPieceDetail<std::string>::size_type pos);
michael@0: BASE_EXPORT StringPiece substr(const StringPiece& self,
michael@0:                                StringPieceDetail<std::string>::size_type pos,
michael@0:                                StringPieceDetail<std::string>::size_type n);
michael@0: }  // namespace internal
michael@0: 
michael@0: // Defines the template type that is instantiated as either StringPiece or
michael@0: // StringPiece16.
michael@0: template <typename STRING_TYPE> class BasicStringPiece :
michael@0:     public internal::StringPieceDetail<STRING_TYPE> {
michael@0:  public:
michael@0:   typedef typename internal::StringPieceDetail<STRING_TYPE>::value_type
michael@0:       value_type;
michael@0:   typedef typename internal::StringPieceDetail<STRING_TYPE>::size_type
michael@0:       size_type;
michael@0: 
michael@0:   BasicStringPiece() {}
michael@0:   BasicStringPiece(const value_type*str)
michael@0:       : internal::StringPieceDetail<STRING_TYPE>(str) {}
michael@0:   BasicStringPiece(const STRING_TYPE& str)
michael@0:       : internal::StringPieceDetail<STRING_TYPE>(str) {}
michael@0:   BasicStringPiece(const value_type* offset, size_type len)
michael@0:       : internal::StringPieceDetail<STRING_TYPE>(offset, len) {}
michael@0:   BasicStringPiece(const typename STRING_TYPE::const_iterator& begin,
michael@0:                    const typename STRING_TYPE::const_iterator& end)
michael@0:       : internal::StringPieceDetail<STRING_TYPE>(begin, end) {}
michael@0: };
michael@0: 
michael@0: // Specializes BasicStringPiece for std::string to add a few operations that
michael@0: // are not needed for string16.
michael@0: template <> class BasicStringPiece<std::string> :
michael@0:     public internal::StringPieceDetail<std::string> {
michael@0:  public:
michael@0:   BasicStringPiece() {}
michael@0:   BasicStringPiece(const char* str)
michael@0:       : internal::StringPieceDetail<std::string>(str) {}
michael@0:   BasicStringPiece(const std::string& str)
michael@0:       : internal::StringPieceDetail<std::string>(str) {}
michael@0:   BasicStringPiece(const char* offset, size_type len)
michael@0:       : internal::StringPieceDetail<std::string>(offset, len) {}
michael@0:   BasicStringPiece(const std::string::const_iterator& begin,
michael@0:                    const std::string::const_iterator& end)
michael@0:       : internal::StringPieceDetail<std::string>(begin, end) {}
michael@0: 
michael@0:   // Prevent the following overload of set() from hiding the definitions in the
michael@0:   // base class.
michael@0:   using internal::StringPieceDetail<std::string>::set;
michael@0: 
michael@0:   void set(const void* data, size_type len) {
michael@0:     ptr_ = reinterpret_cast<const value_type*>(data);
michael@0:     length_ = len;
michael@0:   }
michael@0: 
michael@0:   void CopyToString(std::string* target) const {
michael@0:     internal::CopyToString(*this, target);
michael@0:   }
michael@0: 
michael@0:   void AppendToString(std::string* target) const {
michael@0:     internal::AppendToString(*this, target);
michael@0:   }
michael@0: 
michael@0:   // Does "this" start with "x"
michael@0:   bool starts_with(const BasicStringPiece& x) const {
michael@0:     return ((length_ >= x.length_) &&
michael@0:             (wordmemcmp(ptr_, x.ptr_, x.length_) == 0));
michael@0:   }
michael@0: 
michael@0:   // Does "this" end with "x"
michael@0:   bool ends_with(const BasicStringPiece& x) const {
michael@0:     return ((length_ >= x.length_) &&
michael@0:             (wordmemcmp(ptr_ + (length_-x.length_), x.ptr_, x.length_) == 0));
michael@0:   }
michael@0: 
michael@0:   size_type copy(char* buf, size_type n, size_type pos = 0) const {
michael@0:     return internal::copy(*this, buf, n, pos);
michael@0:   }
michael@0: 
michael@0:   size_type find(const BasicStringPiece& s, size_type pos = 0) const {
michael@0:     return internal::find(*this, s, pos);
michael@0:   }
michael@0: 
michael@0:   size_type find(char c, size_type pos = 0) const {
michael@0:     return internal::find(*this, c, pos);
michael@0:   }
michael@0: 
michael@0:   size_type rfind(const BasicStringPiece& s, size_type pos = npos) const {
michael@0:     return internal::rfind(*this, s, pos);
michael@0:   }
michael@0: 
michael@0:   size_type rfind(char c, size_type pos = npos) const {
michael@0:     return internal::rfind(*this, c, pos);
michael@0:   }
michael@0: 
michael@0:   size_type find_first_of(const BasicStringPiece& s, size_type pos = 0) const {
michael@0:     return internal::find_first_of(*this, s, pos);
michael@0:   }
michael@0: 
michael@0:   size_type find_first_of(char c, size_type pos = 0) const {
michael@0:     return find(c, pos);
michael@0:   }
michael@0: 
michael@0:   size_type find_first_not_of(const BasicStringPiece& s,
michael@0:                               size_type pos = 0) const {
michael@0:     return internal::find_first_not_of(*this, s, pos);
michael@0:   }
michael@0: 
michael@0:   size_type find_first_not_of(char c, size_type pos = 0) const {
michael@0:     return internal::find_first_not_of(*this, c, pos);
michael@0:   }
michael@0: 
michael@0:   size_type find_last_of(const BasicStringPiece& s,
michael@0:                          size_type pos = npos) const {
michael@0:     return internal::find_last_of(*this, s, pos);
michael@0:   }
michael@0: 
michael@0:   size_type find_last_of(char c, size_type pos = npos) const {
michael@0:     return rfind(c, pos);
michael@0:   }
michael@0: 
michael@0:   size_type find_last_not_of(const BasicStringPiece& s,
michael@0:                              size_type pos = npos) const {
michael@0:     return internal::find_last_not_of(*this, s, pos);
michael@0:   }
michael@0: 
michael@0:   size_type find_last_not_of(char c, size_type pos = npos) const {
michael@0:     return internal::find_last_not_of(*this, c, pos);
michael@0:   }
michael@0: 
michael@0:   BasicStringPiece substr(size_type pos, size_type n = npos) const {
michael@0:     return internal::substr(*this, pos, n);
michael@0:   }
michael@0: };
michael@0: 
michael@0: // MSVC doesn't like complex extern templates and DLLs.
michael@0: #if !defined(COMPILER_MSVC)
michael@0: // We can't explicitly declare the std::string instantiation here because it was
michael@0: // already instantiated when specialized, above. Not only is it a no-op, but
michael@0: // currently it also crashes Clang (see http://crbug.com/107412).
michael@0: extern template class BASE_EXPORT BasicStringPiece<string16>;
michael@0: #endif
michael@0: 
michael@0: BASE_EXPORT bool operator==(const StringPiece& x, const StringPiece& y);
michael@0: 
michael@0: inline bool operator!=(const StringPiece& x, const StringPiece& y) {
michael@0:   return !(x == y);
michael@0: }
michael@0: 
michael@0: inline bool operator<(const StringPiece& x, const StringPiece& y) {
michael@0:   const int r = StringPiece::wordmemcmp(
michael@0:       x.data(), y.data(), (x.size() < y.size() ? x.size() : y.size()));
michael@0:   return ((r < 0) || ((r == 0) && (x.size() < y.size())));
michael@0: }
michael@0: 
michael@0: inline bool operator>(const StringPiece& x, const StringPiece& y) {
michael@0:   return y < x;
michael@0: }
michael@0: 
michael@0: inline bool operator<=(const StringPiece& x, const StringPiece& y) {
michael@0:   return !(x > y);
michael@0: }
michael@0: 
michael@0: inline bool operator>=(const StringPiece& x, const StringPiece& y) {
michael@0:   return !(x < y);
michael@0: }
michael@0: 
michael@0: inline bool operator==(const StringPiece16& x, const StringPiece16& y) {
michael@0:   if (x.size() != y.size())
michael@0:     return false;
michael@0: 
michael@0:   return StringPiece16::wordmemcmp(x.data(), y.data(), x.size()) == 0;
michael@0: }
michael@0: 
michael@0: inline bool operator!=(const StringPiece16& x, const StringPiece16& y) {
michael@0:   return !(x == y);
michael@0: }
michael@0: 
michael@0: inline bool operator<(const StringPiece16& x, const StringPiece16& y) {
michael@0:   const int r = StringPiece16::wordmemcmp(
michael@0:       x.data(), y.data(), (x.size() < y.size() ? x.size() : y.size()));
michael@0:   return ((r < 0) || ((r == 0) && (x.size() < y.size())));
michael@0: }
michael@0: 
michael@0: inline bool operator>(const StringPiece16& x, const StringPiece16& y) {
michael@0:   return y < x;
michael@0: }
michael@0: 
michael@0: inline bool operator<=(const StringPiece16& x, const StringPiece16& y) {
michael@0:   return !(x > y);
michael@0: }
michael@0: 
michael@0: inline bool operator>=(const StringPiece16& x, const StringPiece16& y) {
michael@0:   return !(x < y);
michael@0: }
michael@0: 
michael@0: BASE_EXPORT std::ostream& operator<<(std::ostream& o,
michael@0:                                      const StringPiece& piece);
michael@0: 
michael@0: }  // namespace base
michael@0: 
michael@0: // We provide appropriate hash functions so StringPiece and StringPiece16 can
michael@0: // be used as keys in hash sets and maps.
michael@0: 
michael@0: // This hash function is copied from base/containers/hash_tables.h. We don't
michael@0: // use the ones already defined for string and string16 directly because it
michael@0: // would require the string constructors to be called, which we don't want.
michael@0: #define HASH_STRING_PIECE(StringPieceType, string_piece)                \
michael@0:   std::size_t result = 0;                                               \
michael@0:   for (StringPieceType::const_iterator i = string_piece.begin();        \
michael@0:        i != string_piece.end(); ++i)                                    \
michael@0:     result = (result * 131) + *i;                                       \
michael@0:   return result;                                                        \
michael@0: 
michael@0: namespace BASE_HASH_NAMESPACE {
michael@0: #if defined(COMPILER_GCC)
michael@0: 
michael@0: template<>
michael@0: struct hash<base::StringPiece> {
michael@0:   std::size_t operator()(const base::StringPiece& sp) const {
michael@0:     HASH_STRING_PIECE(base::StringPiece, sp);
michael@0:   }
michael@0: };
michael@0: template<>
michael@0: struct hash<base::StringPiece16> {
michael@0:   std::size_t operator()(const base::StringPiece16& sp16) const {
michael@0:     HASH_STRING_PIECE(base::StringPiece16, sp16);
michael@0:   }
michael@0: };
michael@0: 
michael@0: #elif defined(COMPILER_MSVC)
michael@0: 
michael@0: inline size_t hash_value(const base::StringPiece& sp) {
michael@0:   HASH_STRING_PIECE(base::StringPiece, sp);
michael@0: }
michael@0: inline size_t hash_value(const base::StringPiece16& sp16) {
michael@0:   HASH_STRING_PIECE(base::StringPiece16, sp16);
michael@0: }
michael@0: 
michael@0: #endif  // COMPILER
michael@0: 
michael@0: }  // namespace BASE_HASH_NAMESPACE
michael@0: 
michael@0: #endif  // BASE_STRINGS_STRING_PIECE_H_