28 #ifndef UTF8_FOR_CPP_CHECKED_H_2675DCD0_9480_4c0c_B92A_CC14C027B731
29 #define UTF8_FOR_CPP_CHECKED_H_2675DCD0_9480_4c0c_B92A_CC14C027B731
41 virtual const char*
what()
const throw() {
return "Invalid code point"; }
49 virtual const char*
what()
const throw() {
return "Invalid UTF-8"; }
57 virtual const char*
what()
const throw() {
return "Invalid UTF-16"; }
63 virtual const char*
what()
const throw() {
return "Not enough space"; }
68 template <
typename octet_iterator,
typename output_iterator>
71 while (start != end) {
72 octet_iterator sequence_start = start;
76 for (octet_iterator it = sequence_start; it != start; ++it)
99 template <
typename octet_iterator,
typename output_iterator>
100 inline output_iterator
replace_invalid(octet_iterator start, octet_iterator end, output_iterator out)
106 template <
typename octet_iterator>
113 *(result++) = static_cast<uint8_t>(cp);
114 else if (cp < 0x800) {
115 *(result++) = static_cast<uint8_t>((cp >> 6) | 0xc0);
116 *(result++) = static_cast<uint8_t>((cp & 0x3f) | 0x80);
118 else if (cp < 0x10000) {
119 *(result++) = static_cast<uint8_t>((cp >> 12) | 0xe0);
120 *(result++) = static_cast<uint8_t>(((cp >> 6) & 0x3f) | 0x80);
121 *(result++) = static_cast<uint8_t>((cp & 0x3f) | 0x80);
124 *(result++) = static_cast<uint8_t>((cp >> 18) | 0xf0);
125 *(result++) = static_cast<uint8_t>(((cp >> 12)& 0x3f) | 0x80);
126 *(result++) = static_cast<uint8_t>(((cp >> 6) & 0x3f) | 0x80);
127 *(result++) = static_cast<uint8_t>((cp & 0x3f) | 0x80);
135 template <
typename octet_iterator>
155 template <
typename octet_iterator>
158 octet_iterator end = it;
162 octet_iterator temp = it;
163 return next(temp, end);
167 template <
typename octet_iterator>
170 octet_iterator end = it;
172 if (it == pass_start)
174 octet_iterator temp = it;
175 return next(temp, end);
178 template <
typename octet_iterator,
typename distance_type>
179 void advance (octet_iterator& it, distance_type n, octet_iterator end)
181 for (distance_type i = 0; i < n; ++i)
185 template <
typename octet_iterator>
186 typename std::iterator_traits<octet_iterator>::difference_type
187 distance (octet_iterator first, octet_iterator last)
189 typename std::iterator_traits<octet_iterator>::difference_type dist;
190 for (dist = 0; first < last; ++dist)
195 template <
typename u16bit_iterator,
typename octet_iterator>
196 octet_iterator
utf16to8 (u16bit_iterator start, u16bit_iterator end, octet_iterator result)
198 while (start != end) {
213 result =
append(cp, result);
218 template <
typename u16bit_iterator,
typename octet_iterator>
219 u16bit_iterator
utf8to16 (octet_iterator start, octet_iterator end, u16bit_iterator result)
221 while (start != end) {
228 *result++ =
static_cast<uint16_t>(cp);
233 template <
typename octet_iterator,
typename u32bit_iterator>
234 octet_iterator
utf32to8 (u32bit_iterator start, u32bit_iterator end, octet_iterator result)
237 result =
append(*(start++), result);
242 template <
typename octet_iterator,
typename u32bit_iterator>
243 u32bit_iterator
utf8to32 (octet_iterator start, octet_iterator end, u32bit_iterator result)
246 (*result++) =
next(start, end);
252 template <
typename octet_iterator>
253 class iterator :
public std::iterator <std::bidirectional_iterator_tag, uint32_t> {
262 it(octet_it), range_start(range_start), range_end(range_end)
264 if (it < range_start || it > range_end)
265 throw std::out_of_range(
"Invalid utf-8 iterator position");
268 octet_iterator
base ()
const {
return it; }
271 octet_iterator temp =
it;
277 throw std::logic_error(
"Comparing utf-8 iterators defined with different ranges");
278 return (
it == rhs.
it);
310 #endif //header guard
uint16_t utf16_word() const
const uint16_t TRAIL_SURROGATE_MAX
virtual const char * what() const
virtual const char * what() const
u32bit_iterator utf8to32(octet_iterator start, octet_iterator end, u32bit_iterator result)
uint32_t next(octet_iterator &it, octet_iterator end)
const uint16_t TRAIL_SURROGATE_MIN
invalid_utf16(uint16_t u)
const uint32_t CODE_POINT_MAX
octet_iterator base() const
invalid_code_point(uint32_t cp)
octet_iterator utf16to8(u16bit_iterator start, u16bit_iterator end, octet_iterator result)
void advance(octet_iterator &it, distance_type n, octet_iterator end)
octet_iterator utf32to8(u32bit_iterator start, u32bit_iterator end, octet_iterator result)
bool is_code_point_valid(u32 cp)
octet_iterator range_start
const uint32_t SURROGATE_OFFSET
uint16_t mask16(u16_type oc)
uint32_t previous(octet_iterator &it, octet_iterator pass_start)
Deprecated in versions that include "prior".
std::iterator_traits< octet_iterator >::difference_type distance(octet_iterator first, octet_iterator last)
uint32_t prior(octet_iterator &it, octet_iterator start)
uint8_t utf8_octet() const
uint32_t operator*() const
virtual const char * what() const
iterator(const octet_iterator &octet_it, const octet_iterator &range_start, const octet_iterator &range_end)
bool is_surrogate(u16 cp)
u16bit_iterator utf8to16(octet_iterator start, octet_iterator end, u16bit_iterator result)
utf_error validate_next(octet_iterator &it, octet_iterator end, uint32_t *code_point)
output_iterator replace_invalid(octet_iterator start, octet_iterator end, output_iterator out, uint32_t replacement)
The library API - functions intended to be called by the users.
bool is_trail(octet_type oc)
bool operator==(const iterator &rhs) const
const uint16_t LEAD_OFFSET
uint32_t code_point() const
bool operator!=(const iterator &rhs) const
octet_iterator append(uint32_t cp, octet_iterator result)
virtual const char * what() const