873.2.27
by Monty Taylor
Added in utf8cpp. We want this to be an external lib, but it needs packaged. |
1 |
// Copyright 2006 Nemanja Trifunovic
|
2 |
||
3 |
/*
|
|
4 |
Permission is hereby granted, free of charge, to any person or organization
|
|
5 |
obtaining a copy of the software and accompanying documentation covered by
|
|
6 |
this license (the "Software") to use, reproduce, display, distribute,
|
|
7 |
execute, and transmit the Software, and to prepare derivative works of the
|
|
8 |
Software, and to permit third-parties to whom the Software is furnished to
|
|
9 |
do so, all subject to the following:
|
|
10 |
||
11 |
The copyright notices in the Software and this entire statement, including
|
|
12 |
the above license grant, this restriction and the following disclaimer,
|
|
13 |
must be included in all copies of the Software, in whole or in part, and
|
|
14 |
all derivative works of the Software, unless such copies or derivative
|
|
15 |
works are solely in the form of machine-executable object code generated by
|
|
16 |
a source language processor.
|
|
17 |
||
18 |
THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
|
|
19 |
IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
|
|
20 |
FITNESS FOR A PARTICULAR PURPOSE, TITLE AND NON-INFRINGEMENT. IN NO EVENT
|
|
21 |
SHALL THE COPYRIGHT HOLDERS OR ANYONE DISTRIBUTING THE SOFTWARE BE LIABLE
|
|
22 |
FOR ANY DAMAGES OR OTHER LIABILITY, WHETHER IN CONTRACT, TORT OR OTHERWISE,
|
|
23 |
ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER
|
|
24 |
DEALINGS IN THE SOFTWARE.
|
|
25 |
*/
|
|
26 |
||
27 |
||
1122.2.10
by Monty Taylor
Fixed all of the include guards. |
28 |
#ifndef DRIZZLED_UTF8_CHECKED_H
|
29 |
#define DRIZZLED_UTF8_CHECKED_H
|
|
873.2.27
by Monty Taylor
Added in utf8cpp. We want this to be an external lib, but it needs packaged. |
30 |
|
31 |
#include "drizzled/utf8/core.h" |
|
32 |
#include <stdexcept> |
|
33 |
||
34 |
namespace utf8 |
|
35 |
{
|
|
36 |
// Exceptions that may be thrown from the library functions.
|
|
37 |
class invalid_code_point : public std::exception { |
|
38 |
uint32_t cp; |
|
39 |
public: |
|
40 |
invalid_code_point(uint32_t cp) : cp(cp) {} |
|
41 |
virtual const char* what() const throw() { return "Invalid code point"; } |
|
42 |
uint32_t code_point() const {return cp;} |
|
43 |
};
|
|
44 |
||
45 |
class invalid_utf8 : public std::exception { |
|
46 |
uint8_t u8; |
|
47 |
public: |
|
48 |
invalid_utf8 (uint8_t u) : u8(u) {} |
|
49 |
virtual const char* what() const throw() { return "Invalid UTF-8"; } |
|
50 |
uint8_t utf8_octet() const {return u8;} |
|
51 |
};
|
|
52 |
||
53 |
class invalid_utf16 : public std::exception { |
|
54 |
uint16_t u16; |
|
55 |
public: |
|
56 |
invalid_utf16 (uint16_t u) : u16(u) {} |
|
57 |
virtual const char* what() const throw() { return "Invalid UTF-16"; } |
|
58 |
uint16_t utf16_word() const {return u16;} |
|
59 |
};
|
|
60 |
||
61 |
class not_enough_room : public std::exception { |
|
62 |
public: |
|
63 |
virtual const char* what() const throw() { return "Not enough space"; } |
|
64 |
};
|
|
65 |
||
66 |
/// The library API - functions intended to be called by the users
|
|
67 |
||
68 |
template <typename octet_iterator, typename output_iterator> |
|
69 |
output_iterator replace_invalid(octet_iterator start, octet_iterator end, output_iterator out, uint32_t replacement) |
|
70 |
{
|
|
71 |
while (start != end) { |
|
72 |
octet_iterator sequence_start = start; |
|
73 |
internal::utf_error err_code = internal::validate_next(start, end); |
|
74 |
switch (err_code) { |
|
75 |
case internal::OK : |
|
76 |
for (octet_iterator it = sequence_start; it != start; ++it) |
|
77 |
*out++ = *it; |
|
78 |
break; |
|
79 |
case internal::NOT_ENOUGH_ROOM: |
|
80 |
throw not_enough_room(); |
|
81 |
case internal::INVALID_LEAD: |
|
82 |
append (replacement, out); |
|
83 |
++start; |
|
84 |
break; |
|
85 |
case internal::INCOMPLETE_SEQUENCE: |
|
86 |
case internal::OVERLONG_SEQUENCE: |
|
87 |
case internal::INVALID_CODE_POINT: |
|
88 |
append (replacement, out); |
|
89 |
++start; |
|
90 |
// just one replacement mark for the sequence
|
|
91 |
while (internal::is_trail(*start) && start != end) |
|
92 |
++start; |
|
93 |
break; |
|
94 |
}
|
|
95 |
}
|
|
96 |
return out; |
|
97 |
}
|
|
98 |
||
99 |
template <typename octet_iterator, typename output_iterator> |
|
100 |
inline output_iterator replace_invalid(octet_iterator start, octet_iterator end, output_iterator out) |
|
101 |
{
|
|
102 |
static const uint32_t replacement_marker = internal::mask16(0xfffd); |
|
103 |
return replace_invalid(start, end, out, replacement_marker); |
|
104 |
}
|
|
105 |
||
106 |
template <typename octet_iterator> |
|
107 |
octet_iterator append(uint32_t cp, octet_iterator result) |
|
108 |
{
|
|
109 |
if (!internal::is_code_point_valid(cp)) |
|
110 |
throw invalid_code_point(cp); |
|
111 |
||
112 |
if (cp < 0x80) // one octet |
|
113 |
*(result++) = static_cast<uint8_t>(cp); |
|
114 |
else if (cp < 0x800) { // two octets |
|
115 |
*(result++) = static_cast<uint8_t>((cp >> 6) | 0xc0); |
|
116 |
*(result++) = static_cast<uint8_t>((cp & 0x3f) | 0x80); |
|
117 |
}
|
|
118 |
else if (cp < 0x10000) { // three octets |
|
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); |
|
122 |
}
|
|
123 |
else if (cp <= internal::CODE_POINT_MAX) { // four octets |
|
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); |
|
128 |
}
|
|
129 |
else
|
|
130 |
throw invalid_code_point(cp); |
|
131 |
||
132 |
return result; |
|
133 |
}
|
|
134 |
||
135 |
template <typename octet_iterator> |
|
136 |
uint32_t next(octet_iterator& it, octet_iterator end) |
|
137 |
{
|
|
138 |
uint32_t cp = 0; |
|
139 |
internal::utf_error err_code = internal::validate_next(it, end, &cp); |
|
140 |
switch (err_code) { |
|
141 |
case internal::OK : |
|
142 |
break; |
|
143 |
case internal::NOT_ENOUGH_ROOM : |
|
144 |
throw not_enough_room(); |
|
145 |
case internal::INVALID_LEAD : |
|
146 |
case internal::INCOMPLETE_SEQUENCE : |
|
147 |
case internal::OVERLONG_SEQUENCE : |
|
148 |
throw invalid_utf8(*it); |
|
149 |
case internal::INVALID_CODE_POINT : |
|
150 |
throw invalid_code_point(cp); |
|
151 |
}
|
|
152 |
return cp; |
|
153 |
}
|
|
154 |
||
155 |
template <typename octet_iterator> |
|
156 |
uint32_t peek_next(octet_iterator it, octet_iterator end) |
|
157 |
{
|
|
158 |
return next(it, end); |
|
159 |
}
|
|
160 |
||
161 |
template <typename octet_iterator> |
|
162 |
uint32_t prior(octet_iterator& it, octet_iterator start) |
|
163 |
{
|
|
164 |
octet_iterator end = it; |
|
165 |
while (internal::is_trail(*(--it))) |
|
166 |
if (it < start) |
|
167 |
throw invalid_utf8(*it); // error - no lead byte in the sequence |
|
168 |
octet_iterator temp = it; |
|
169 |
return next(temp, end); |
|
170 |
}
|
|
171 |
||
172 |
/// Deprecated in versions that include "prior"
|
|
173 |
template <typename octet_iterator> |
|
174 |
uint32_t previous(octet_iterator& it, octet_iterator pass_start) |
|
175 |
{
|
|
176 |
octet_iterator end = it; |
|
177 |
while (internal::is_trail(*(--it))) |
|
178 |
if (it == pass_start) |
|
179 |
throw invalid_utf8(*it); // error - no lead byte in the sequence |
|
180 |
octet_iterator temp = it; |
|
181 |
return next(temp, end); |
|
182 |
}
|
|
183 |
||
184 |
template <typename octet_iterator, typename distance_type> |
|
185 |
void advance (octet_iterator& it, distance_type n, octet_iterator end) |
|
186 |
{
|
|
187 |
for (distance_type i = 0; i < n; ++i) |
|
188 |
next(it, end); |
|
189 |
}
|
|
190 |
||
191 |
template <typename octet_iterator> |
|
192 |
typename std::iterator_traits<octet_iterator>::difference_type |
|
193 |
distance (octet_iterator first, octet_iterator last) |
|
194 |
{
|
|
195 |
typename std::iterator_traits<octet_iterator>::difference_type dist; |
|
196 |
for (dist = 0; first < last; ++dist) |
|
197 |
next(first, last); |
|
198 |
return dist; |
|
199 |
}
|
|
200 |
||
201 |
template <typename u16bit_iterator, typename octet_iterator> |
|
202 |
octet_iterator utf16to8 (u16bit_iterator start, u16bit_iterator end, octet_iterator result) |
|
203 |
{
|
|
204 |
while (start != end) { |
|
205 |
uint32_t cp = internal::mask16(*start++); |
|
206 |
// Take care of surrogate pairs first
|
|
207 |
if (internal::is_surrogate(cp)) { |
|
208 |
if (start != end) { |
|
209 |
uint32_t trail_surrogate = internal::mask16(*start++); |
|
210 |
if (trail_surrogate >= internal::TRAIL_SURROGATE_MIN && trail_surrogate <= internal::TRAIL_SURROGATE_MAX) |
|
211 |
cp = (cp << 10) + trail_surrogate + internal::SURROGATE_OFFSET; |
|
212 |
else
|
|
213 |
throw invalid_utf16(static_cast<uint16_t>(trail_surrogate)); |
|
214 |
}
|
|
215 |
else
|
|
216 |
throw invalid_utf16(static_cast<uint16_t>(*start)); |
|
217 |
||
218 |
}
|
|
219 |
result = append(cp, result); |
|
220 |
}
|
|
221 |
return result; |
|
222 |
}
|
|
223 |
||
224 |
template <typename u16bit_iterator, typename octet_iterator> |
|
225 |
u16bit_iterator utf8to16 (octet_iterator start, octet_iterator end, u16bit_iterator result) |
|
226 |
{
|
|
227 |
while (start != end) { |
|
228 |
uint32_t cp = next(start, end); |
|
229 |
if (cp > 0xffff) { //make a surrogate pair |
|
230 |
*result++ = static_cast<uint16_t>((cp >> 10) + internal::LEAD_OFFSET); |
|
231 |
*result++ = static_cast<uint16_t>((cp & 0x3ff) + internal::TRAIL_SURROGATE_MIN); |
|
232 |
}
|
|
233 |
else
|
|
234 |
*result++ = static_cast<uint16_t>(cp); |
|
235 |
}
|
|
236 |
return result; |
|
237 |
}
|
|
238 |
||
239 |
template <typename octet_iterator, typename u32bit_iterator> |
|
240 |
octet_iterator utf32to8 (u32bit_iterator start, u32bit_iterator end, octet_iterator result) |
|
241 |
{
|
|
242 |
while (start != end) |
|
243 |
result = append(*(start++), result); |
|
244 |
||
245 |
return result; |
|
246 |
}
|
|
247 |
||
248 |
template <typename octet_iterator, typename u32bit_iterator> |
|
249 |
u32bit_iterator utf8to32 (octet_iterator start, octet_iterator end, u32bit_iterator result) |
|
250 |
{
|
|
251 |
while (start < end) |
|
252 |
(*result++) = next(start, end); |
|
253 |
||
254 |
return result; |
|
255 |
}
|
|
256 |
||
257 |
// The iterator class
|
|
258 |
template <typename octet_iterator> |
|
259 |
class iterator : public std::iterator <std::bidirectional_iterator_tag, uint32_t> { |
|
260 |
octet_iterator it; |
|
261 |
octet_iterator range_start; |
|
262 |
octet_iterator range_end; |
|
263 |
public: |
|
264 |
iterator () {}; |
|
265 |
explicit iterator (const octet_iterator& octet_it, |
|
266 |
const octet_iterator& range_start, |
|
267 |
const octet_iterator& range_end) : |
|
268 |
it(octet_it), range_start(range_start), range_end(range_end) |
|
269 |
{
|
|
270 |
if (it < range_start || it > range_end) |
|
271 |
throw std::out_of_range("Invalid utf-8 iterator position"); |
|
272 |
}
|
|
273 |
// the default "big three" are OK
|
|
274 |
octet_iterator base () const { return it; } |
|
275 |
uint32_t operator * () const |
|
276 |
{
|
|
277 |
octet_iterator temp = it; |
|
278 |
return next(temp, range_end); |
|
279 |
}
|
|
280 |
bool operator == (const iterator& rhs) const |
|
281 |
{
|
|
282 |
if (range_start != rhs.range_start || range_end != rhs.range_end) |
|
283 |
throw std::logic_error("Comparing utf-8 iterators defined with different ranges"); |
|
284 |
return (it == rhs.it); |
|
285 |
}
|
|
286 |
bool operator != (const iterator& rhs) const |
|
287 |
{
|
|
288 |
return !(operator == (rhs)); |
|
289 |
}
|
|
290 |
iterator& operator ++ () |
|
291 |
{
|
|
292 |
next(it, range_end); |
|
293 |
return *this; |
|
294 |
}
|
|
295 |
iterator operator ++ (int) |
|
296 |
{
|
|
297 |
iterator temp = *this; |
|
298 |
next(it, range_end); |
|
299 |
return temp; |
|
300 |
}
|
|
301 |
iterator& operator -- () |
|
302 |
{
|
|
303 |
prior(it, range_start); |
|
304 |
return *this; |
|
305 |
}
|
|
306 |
iterator operator -- (int) |
|
307 |
{
|
|
308 |
iterator temp = *this; |
|
309 |
prior(it, range_start); |
|
310 |
return temp; |
|
311 |
}
|
|
312 |
}; // class iterator |
|
313 |
||
314 |
} // namespace utf8 |
|
315 |
||
1122.2.10
by Monty Taylor
Fixed all of the include guards. |
316 |
#endif /* DRIZZLED_UTF8_CHECKED_H */ |
873.2.27
by Monty Taylor
Added in utf8cpp. We want this to be an external lib, but it needs packaged. |
317 |
|
318 |