28 #ifndef UTF8_FOR_CPP_UNCHECKED_H_2675DCD0_9480_4c0c_B92A_CC14C027B731
29 #define UTF8_FOR_CPP_UNCHECKED_H_2675DCD0_9480_4c0c_B92A_CC14C027B731
37 template <
typename octet_iterator>
38 octet_iterator append(uint32_t cp, octet_iterator result)
41 *(result++) = static_cast<uint8_t>(cp);
42 else if (cp < 0x800) {
43 *(result++) = static_cast<uint8_t>((cp >> 6) | 0xc0);
44 *(result++) = static_cast<uint8_t>((cp & 0x3f) | 0x80);
46 else if (cp < 0x10000) {
47 *(result++) = static_cast<uint8_t>((cp >> 12) | 0xe0);
48 *(result++) = static_cast<uint8_t>((cp >> 6) & 0x3f | 0x80);
49 *(result++) = static_cast<uint8_t>((cp & 0x3f) | 0x80);
52 *(result++) = static_cast<uint8_t>((cp >> 18) | 0xf0);
53 *(result++) = static_cast<uint8_t>((cp >> 12)& 0x3f | 0x80);
54 *(result++) = static_cast<uint8_t>((cp >> 6) & 0x3f | 0x80);
55 *(result++) = static_cast<uint8_t>((cp & 0x3f) | 0x80);
59 template <
typename octet_iterator>
60 uint32_t next(octet_iterator& it)
62 uint32_t cp = internal::mask8(*it);
63 typename std::iterator_traits<octet_iterator>::difference_type length = utf8::internal::sequence_length(it);
69 cp = ((cp << 6) & 0x7ff) + ((*it) & 0x3f);
73 cp = ((cp << 12) & 0xffff) + ((internal::mask8(*it) << 6) & 0xfff);
79 cp = ((cp << 18) & 0x1fffff) + ((internal::mask8(*it) << 12) & 0x3ffff);
81 cp += (internal::mask8(*it) << 6) & 0xfff;
90 template <
typename octet_iterator>
91 uint32_t prior(octet_iterator& it)
93 while (internal::is_trail(*(--it))) ;
94 octet_iterator temp = it;
99 template <
typename octet_iterator>
100 inline uint32_t previous(octet_iterator& it)
105 template <
typename octet_iterator,
typename distance_type>
106 void advance (octet_iterator& it, distance_type n)
108 for (distance_type i = 0; i < n; ++i)
112 template <
typename octet_iterator>
113 typename std::iterator_traits<octet_iterator>::difference_type
114 distance (octet_iterator first, octet_iterator last)
116 typename std::iterator_traits<octet_iterator>::difference_type dist;
117 for (dist = 0; first < last; ++dist)
122 template <
typename u16bit_iterator,
typename octet_iterator>
123 octet_iterator utf16to8 (u16bit_iterator start, u16bit_iterator end, octet_iterator result)
125 while (start != end) {
126 uint32_t cp = internal::mask16(*start++);
128 if (internal::is_surrogate(cp)) {
129 uint32_t trail_surrogate = internal::mask16(*start++);
130 cp = (cp << 10) + trail_surrogate + internal::SURROGATE_OFFSET;
132 result = append(cp, result);
137 template <
typename u16bit_iterator,
typename octet_iterator>
138 u16bit_iterator utf8to16 (octet_iterator start, octet_iterator end, u16bit_iterator result)
140 while (start != end) {
141 uint32_t cp = next(start);
143 *result++ =
static_cast<uint16_t
>((cp >> 10) + internal::LEAD_OFFSET);
144 *result++ =
static_cast<uint16_t
>((cp & 0x3ff) + internal::TRAIL_SURROGATE_MIN);
147 *result++ =
static_cast<uint16_t
>(cp);
152 template <
typename octet_iterator,
typename u32bit_iterator>
153 octet_iterator utf32to8 (u32bit_iterator start, u32bit_iterator end, octet_iterator result)
156 result = append(*(start++), result);
161 template <
typename octet_iterator,
typename u32bit_iterator>
162 u32bit_iterator utf8to32 (octet_iterator start, octet_iterator end, u32bit_iterator result)
165 (*result++) = next(start);
171 template <
typename octet_iterator>
172 class iterator :
public std::iterator <std::bidirectional_iterator_tag, uint32_t> {
176 explicit iterator (
const octet_iterator& octet_it): it(octet_it) {}
178 octet_iterator base ()
const {
return it; }
179 uint32_t operator * ()
const
181 octet_iterator temp = it;
184 bool operator == (
const iterator& rhs)
const
186 return (it == rhs.it);
188 bool operator != (
const iterator& rhs)
const
190 return !(operator == (rhs));
192 iterator& operator ++ ()
194 std::advance(it, internal::sequence_length(it));
197 iterator operator ++ (
int)
199 iterator temp = *
this;
200 std::advance(it, internal::sequence_length(it));
203 iterator& operator -- ()
208 iterator operator -- (
int)
210 iterator temp = *
this;
220 #endif // header guard