/src/icu/source/common/uchriter.cpp
Line | Count | Source (jump to first uncovered line) |
1 | | // © 2016 and later: Unicode, Inc. and others. |
2 | | // License & terms of use: http://www.unicode.org/copyright.html |
3 | | /* |
4 | | ****************************************************************************** |
5 | | * Copyright (C) 1998-2012, International Business Machines Corporation and |
6 | | * others. All Rights Reserved. |
7 | | ****************************************************************************** |
8 | | */ |
9 | | |
10 | | #include "utypeinfo.h" // for 'typeid' to work |
11 | | |
12 | | #include "unicode/uchriter.h" |
13 | | #include "unicode/ustring.h" |
14 | | #include "unicode/utf16.h" |
15 | | #include "ustr_imp.h" |
16 | | |
17 | | U_NAMESPACE_BEGIN |
18 | | |
19 | | UOBJECT_DEFINE_RTTI_IMPLEMENTATION(UCharCharacterIterator) |
20 | | |
21 | | UCharCharacterIterator::UCharCharacterIterator() |
22 | 0 | : CharacterIterator(), |
23 | 0 | text(0) |
24 | 0 | { |
25 | | // never default construct! |
26 | 0 | } |
27 | | |
28 | | UCharCharacterIterator::UCharCharacterIterator(ConstChar16Ptr textPtr, |
29 | | int32_t length) |
30 | 0 | : CharacterIterator(textPtr != 0 ? (length>=0 ? length : u_strlen(textPtr)) : 0), |
31 | 0 | text(textPtr) |
32 | 0 | { |
33 | 0 | } |
34 | | |
35 | | UCharCharacterIterator::UCharCharacterIterator(ConstChar16Ptr textPtr, |
36 | | int32_t length, |
37 | | int32_t position) |
38 | 0 | : CharacterIterator(textPtr != 0 ? (length>=0 ? length : u_strlen(textPtr)) : 0, position), |
39 | 0 | text(textPtr) |
40 | 0 | { |
41 | 0 | } |
42 | | |
43 | | UCharCharacterIterator::UCharCharacterIterator(ConstChar16Ptr textPtr, |
44 | | int32_t length, |
45 | | int32_t textBegin, |
46 | | int32_t textEnd, |
47 | | int32_t position) |
48 | 0 | : CharacterIterator(textPtr != 0 ? (length>=0 ? length : u_strlen(textPtr)) : 0, textBegin, textEnd, position), |
49 | 0 | text(textPtr) |
50 | 0 | { |
51 | 0 | } |
52 | | |
53 | | UCharCharacterIterator::UCharCharacterIterator(const UCharCharacterIterator& that) |
54 | 0 | : CharacterIterator(that), |
55 | 0 | text(that.text) |
56 | 0 | { |
57 | 0 | } |
58 | | |
59 | | UCharCharacterIterator& |
60 | 0 | UCharCharacterIterator::operator=(const UCharCharacterIterator& that) { |
61 | 0 | CharacterIterator::operator=(that); |
62 | 0 | text = that.text; |
63 | 0 | return *this; |
64 | 0 | } |
65 | | |
66 | 0 | UCharCharacterIterator::~UCharCharacterIterator() { |
67 | 0 | } |
68 | | |
69 | | bool |
70 | 0 | UCharCharacterIterator::operator==(const ForwardCharacterIterator& that) const { |
71 | 0 | if (this == &that) { |
72 | 0 | return TRUE; |
73 | 0 | } |
74 | 0 | if (typeid(*this) != typeid(that)) { |
75 | 0 | return FALSE; |
76 | 0 | } |
77 | | |
78 | 0 | UCharCharacterIterator& realThat = (UCharCharacterIterator&)that; |
79 | |
|
80 | 0 | return text == realThat.text |
81 | 0 | && textLength == realThat.textLength |
82 | 0 | && pos == realThat.pos |
83 | 0 | && begin == realThat.begin |
84 | 0 | && end == realThat.end; |
85 | 0 | } |
86 | | |
87 | | int32_t |
88 | 0 | UCharCharacterIterator::hashCode() const { |
89 | 0 | return ustr_hashUCharsN(text, textLength) ^ pos ^ begin ^ end; |
90 | 0 | } |
91 | | |
92 | | UCharCharacterIterator* |
93 | 0 | UCharCharacterIterator::clone() const { |
94 | 0 | return new UCharCharacterIterator(*this); |
95 | 0 | } |
96 | | |
97 | | UChar |
98 | 0 | UCharCharacterIterator::first() { |
99 | 0 | pos = begin; |
100 | 0 | if(pos < end) { |
101 | 0 | return text[pos]; |
102 | 0 | } else { |
103 | 0 | return DONE; |
104 | 0 | } |
105 | 0 | } |
106 | | |
107 | | UChar |
108 | 0 | UCharCharacterIterator::firstPostInc() { |
109 | 0 | pos = begin; |
110 | 0 | if(pos < end) { |
111 | 0 | return text[pos++]; |
112 | 0 | } else { |
113 | 0 | return DONE; |
114 | 0 | } |
115 | 0 | } |
116 | | |
117 | | UChar |
118 | 0 | UCharCharacterIterator::last() { |
119 | 0 | pos = end; |
120 | 0 | if(pos > begin) { |
121 | 0 | return text[--pos]; |
122 | 0 | } else { |
123 | 0 | return DONE; |
124 | 0 | } |
125 | 0 | } |
126 | | |
127 | | UChar |
128 | 0 | UCharCharacterIterator::setIndex(int32_t position) { |
129 | 0 | if(position < begin) { |
130 | 0 | pos = begin; |
131 | 0 | } else if(position > end) { |
132 | 0 | pos = end; |
133 | 0 | } else { |
134 | 0 | pos = position; |
135 | 0 | } |
136 | 0 | if(pos < end) { |
137 | 0 | return text[pos]; |
138 | 0 | } else { |
139 | 0 | return DONE; |
140 | 0 | } |
141 | 0 | } |
142 | | |
143 | | UChar |
144 | 0 | UCharCharacterIterator::current() const { |
145 | 0 | if (pos >= begin && pos < end) { |
146 | 0 | return text[pos]; |
147 | 0 | } else { |
148 | 0 | return DONE; |
149 | 0 | } |
150 | 0 | } |
151 | | |
152 | | UChar |
153 | 0 | UCharCharacterIterator::next() { |
154 | 0 | if (pos + 1 < end) { |
155 | 0 | return text[++pos]; |
156 | 0 | } else { |
157 | | /* make current() return DONE */ |
158 | 0 | pos = end; |
159 | 0 | return DONE; |
160 | 0 | } |
161 | 0 | } |
162 | | |
163 | | UChar |
164 | 0 | UCharCharacterIterator::nextPostInc() { |
165 | 0 | if (pos < end) { |
166 | 0 | return text[pos++]; |
167 | 0 | } else { |
168 | 0 | return DONE; |
169 | 0 | } |
170 | 0 | } |
171 | | |
172 | | UBool |
173 | 0 | UCharCharacterIterator::hasNext() { |
174 | 0 | return (UBool)(pos < end ? TRUE : FALSE); |
175 | 0 | } |
176 | | |
177 | | UChar |
178 | 0 | UCharCharacterIterator::previous() { |
179 | 0 | if (pos > begin) { |
180 | 0 | return text[--pos]; |
181 | 0 | } else { |
182 | 0 | return DONE; |
183 | 0 | } |
184 | 0 | } |
185 | | |
186 | | UBool |
187 | 0 | UCharCharacterIterator::hasPrevious() { |
188 | 0 | return (UBool)(pos > begin ? TRUE : FALSE); |
189 | 0 | } |
190 | | |
191 | | UChar32 |
192 | 0 | UCharCharacterIterator::first32() { |
193 | 0 | pos = begin; |
194 | 0 | if(pos < end) { |
195 | 0 | int32_t i = pos; |
196 | 0 | UChar32 c; |
197 | 0 | U16_NEXT(text, i, end, c); |
198 | 0 | return c; |
199 | 0 | } else { |
200 | 0 | return DONE; |
201 | 0 | } |
202 | 0 | } |
203 | | |
204 | | UChar32 |
205 | 0 | UCharCharacterIterator::first32PostInc() { |
206 | 0 | pos = begin; |
207 | 0 | if(pos < end) { |
208 | 0 | UChar32 c; |
209 | 0 | U16_NEXT(text, pos, end, c); |
210 | 0 | return c; |
211 | 0 | } else { |
212 | 0 | return DONE; |
213 | 0 | } |
214 | 0 | } |
215 | | |
216 | | UChar32 |
217 | 0 | UCharCharacterIterator::last32() { |
218 | 0 | pos = end; |
219 | 0 | if(pos > begin) { |
220 | 0 | UChar32 c; |
221 | 0 | U16_PREV(text, begin, pos, c); |
222 | 0 | return c; |
223 | 0 | } else { |
224 | 0 | return DONE; |
225 | 0 | } |
226 | 0 | } |
227 | | |
228 | | UChar32 |
229 | 0 | UCharCharacterIterator::setIndex32(int32_t position) { |
230 | 0 | if(position < begin) { |
231 | 0 | position = begin; |
232 | 0 | } else if(position > end) { |
233 | 0 | position = end; |
234 | 0 | } |
235 | 0 | if(position < end) { |
236 | 0 | U16_SET_CP_START(text, begin, position); |
237 | 0 | int32_t i = this->pos = position; |
238 | 0 | UChar32 c; |
239 | 0 | U16_NEXT(text, i, end, c); |
240 | 0 | return c; |
241 | 0 | } else { |
242 | 0 | this->pos = position; |
243 | 0 | return DONE; |
244 | 0 | } |
245 | 0 | } |
246 | | |
247 | | UChar32 |
248 | 0 | UCharCharacterIterator::current32() const { |
249 | 0 | if (pos >= begin && pos < end) { |
250 | 0 | UChar32 c; |
251 | 0 | U16_GET(text, begin, pos, end, c); |
252 | 0 | return c; |
253 | 0 | } else { |
254 | 0 | return DONE; |
255 | 0 | } |
256 | 0 | } |
257 | | |
258 | | UChar32 |
259 | 0 | UCharCharacterIterator::next32() { |
260 | 0 | if (pos < end) { |
261 | 0 | U16_FWD_1(text, pos, end); |
262 | 0 | if(pos < end) { |
263 | 0 | int32_t i = pos; |
264 | 0 | UChar32 c; |
265 | 0 | U16_NEXT(text, i, end, c); |
266 | 0 | return c; |
267 | 0 | } |
268 | 0 | } |
269 | | /* make current() return DONE */ |
270 | 0 | pos = end; |
271 | 0 | return DONE; |
272 | 0 | } |
273 | | |
274 | | UChar32 |
275 | 0 | UCharCharacterIterator::next32PostInc() { |
276 | 0 | if (pos < end) { |
277 | 0 | UChar32 c; |
278 | 0 | U16_NEXT(text, pos, end, c); |
279 | 0 | return c; |
280 | 0 | } else { |
281 | 0 | return DONE; |
282 | 0 | } |
283 | 0 | } |
284 | | |
285 | | UChar32 |
286 | 0 | UCharCharacterIterator::previous32() { |
287 | 0 | if (pos > begin) { |
288 | 0 | UChar32 c; |
289 | 0 | U16_PREV(text, begin, pos, c); |
290 | 0 | return c; |
291 | 0 | } else { |
292 | 0 | return DONE; |
293 | 0 | } |
294 | 0 | } |
295 | | |
296 | | int32_t |
297 | 0 | UCharCharacterIterator::move(int32_t delta, CharacterIterator::EOrigin origin) { |
298 | 0 | switch(origin) { |
299 | 0 | case kStart: |
300 | 0 | pos = begin + delta; |
301 | 0 | break; |
302 | 0 | case kCurrent: |
303 | 0 | pos += delta; |
304 | 0 | break; |
305 | 0 | case kEnd: |
306 | 0 | pos = end + delta; |
307 | 0 | break; |
308 | 0 | default: |
309 | 0 | break; |
310 | 0 | } |
311 | | |
312 | 0 | if(pos < begin) { |
313 | 0 | pos = begin; |
314 | 0 | } else if(pos > end) { |
315 | 0 | pos = end; |
316 | 0 | } |
317 | |
|
318 | 0 | return pos; |
319 | 0 | } |
320 | | |
321 | | int32_t |
322 | 0 | UCharCharacterIterator::move32(int32_t delta, CharacterIterator::EOrigin origin) { |
323 | | // this implementation relies on the "safe" version of the UTF macros |
324 | | // (or the trustworthiness of the caller) |
325 | 0 | switch(origin) { |
326 | 0 | case kStart: |
327 | 0 | pos = begin; |
328 | 0 | if(delta > 0) { |
329 | 0 | U16_FWD_N(text, pos, end, delta); |
330 | 0 | } |
331 | 0 | break; |
332 | 0 | case kCurrent: |
333 | 0 | if(delta > 0) { |
334 | 0 | U16_FWD_N(text, pos, end, delta); |
335 | 0 | } else { |
336 | 0 | U16_BACK_N(text, begin, pos, -delta); |
337 | 0 | } |
338 | 0 | break; |
339 | 0 | case kEnd: |
340 | 0 | pos = end; |
341 | 0 | if(delta < 0) { |
342 | 0 | U16_BACK_N(text, begin, pos, -delta); |
343 | 0 | } |
344 | 0 | break; |
345 | 0 | default: |
346 | 0 | break; |
347 | 0 | } |
348 | | |
349 | 0 | return pos; |
350 | 0 | } |
351 | | |
352 | | void UCharCharacterIterator::setText(ConstChar16Ptr newText, |
353 | 0 | int32_t newTextLength) { |
354 | 0 | text = newText; |
355 | 0 | if(newText == 0 || newTextLength < 0) { |
356 | 0 | newTextLength = 0; |
357 | 0 | } |
358 | 0 | end = textLength = newTextLength; |
359 | 0 | pos = begin = 0; |
360 | 0 | } |
361 | | |
362 | | void |
363 | 0 | UCharCharacterIterator::getText(UnicodeString& result) { |
364 | 0 | result = UnicodeString(text, textLength); |
365 | 0 | } |
366 | | |
367 | | U_NAMESPACE_END |