31 #ifndef UTF8_FOR_CPP_UNCHECKED_H_2675DCD0_9480_4c0c_B92A_CC14C027B731
32 #define UTF8_FOR_CPP_UNCHECKED_H_2675DCD0_9480_4c0c_B92A_CC14C027B731
40 template <
typename octet_iterator>
41 octet_iterator append(uint32_t cp, octet_iterator result)
44 *(result++) = static_cast<uint8_t>(cp);
45 else if (cp < 0x800) {
46 *(result++) = static_cast<uint8_t>((cp >> 6) | 0xc0);
47 *(result++) = static_cast<uint8_t>((cp & 0x3f) | 0x80);
49 else if (cp < 0x10000) {
50 *(result++) = static_cast<uint8_t>((cp >> 12) | 0xe0);
51 *(result++) = static_cast<uint8_t>((cp >> 6) & 0x3f | 0x80);
52 *(result++) = static_cast<uint8_t>((cp & 0x3f) | 0x80);
55 *(result++) = static_cast<uint8_t>((cp >> 18) | 0xf0);
56 *(result++) = static_cast<uint8_t>((cp >> 12)& 0x3f | 0x80);
57 *(result++) = static_cast<uint8_t>((cp >> 6) & 0x3f | 0x80);
58 *(result++) = static_cast<uint8_t>((cp & 0x3f) | 0x80);
63 template <
typename octet_iterator>
64 uint32_t next(octet_iterator& it)
66 uint32_t cp = internal::mask8(*it);
67 typename std::iterator_traits<octet_iterator>::difference_type length = utf8::internal::sequence_length(it);
73 cp = ((cp << 6) & 0x7ff) + ((*it) & 0x3f);
77 cp = ((cp << 12) & 0xffff) + ((internal::mask8(*it) << 6) & 0xfff);
83 cp = ((cp << 18) & 0x1fffff) + ((internal::mask8(*it) << 12) & 0x3ffff);
85 cp += (internal::mask8(*it) << 6) & 0xfff;
94 template <
typename octet_iterator>
95 uint32_t peek_next(octet_iterator it)
100 template <
typename octet_iterator>
101 uint32_t prior(octet_iterator& it)
103 while (internal::is_trail(*(--it))) ;
104 octet_iterator temp = it;
109 template <
typename octet_iterator>
110 inline uint32_t previous(octet_iterator& it)
115 template <
typename octet_iterator,
typename distance_type>
116 void advance (octet_iterator& it, distance_type n)
118 for (distance_type i = 0; i < n; ++i)
122 template <
typename octet_iterator>
123 typename std::iterator_traits<octet_iterator>::difference_type
124 distance (octet_iterator first, octet_iterator last)
126 typename std::iterator_traits<octet_iterator>::difference_type dist;
127 for (dist = 0; first < last; ++dist)
132 template <
typename u16bit_iterator,
typename octet_iterator>
133 octet_iterator utf16to8 (u16bit_iterator start, u16bit_iterator end, octet_iterator result)
135 while (start != end) {
136 uint32_t cp = internal::mask16(*start++);
138 if (internal::is_surrogate(cp)) {
139 uint32_t trail_surrogate = internal::mask16(*start++);
140 cp = (cp << 10) + trail_surrogate + internal::SURROGATE_OFFSET;
142 result = append(cp, result);
147 template <
typename u16bit_iterator,
typename octet_iterator>
148 u16bit_iterator utf8to16 (octet_iterator start, octet_iterator end, u16bit_iterator result)
150 while (start != end) {
151 uint32_t cp = next(start);
153 *result++ =
static_cast<uint16_t
>((cp >> 10) + internal::LEAD_OFFSET);
154 *result++ =
static_cast<uint16_t
>((cp & 0x3ff) + internal::TRAIL_SURROGATE_MIN);
157 *result++ =
static_cast<uint16_t
>(cp);
162 template <
typename octet_iterator,
typename u32bit_iterator>
163 octet_iterator utf32to8 (u32bit_iterator start, u32bit_iterator end, octet_iterator result)
166 result = append(*(start++), result);
171 template <
typename octet_iterator,
typename u32bit_iterator>
172 u32bit_iterator utf8to32 (octet_iterator start, octet_iterator end, u32bit_iterator result)
175 (*result++) = next(start);
181 template <
typename octet_iterator>
182 class iterator :
public std::iterator <std::bidirectional_iterator_tag, uint32_t> {
186 explicit iterator (
const octet_iterator& octet_it): it(octet_it) {}
188 octet_iterator base ()
const {
return it; }
189 uint32_t operator * ()
const
191 octet_iterator temp = it;
196 return (it == rhs.it);
202 iterator& operator ++ ()
204 std::advance(it, internal::sequence_length(it));
207 iterator operator ++ (
int)
209 iterator temp = *
this;
210 std::advance(it, internal::sequence_length(it));
213 iterator& operator -- ()
218 iterator operator -- (
int)
220 iterator temp = *
this;
230 #endif // header guard