/src/icu/icu4c/source/i18n/formattedval_sbimpl.cpp
Line | Count | Source |
1 | | // © 2018 and later: Unicode, Inc. and others. |
2 | | // License & terms of use: http://www.unicode.org/copyright.html |
3 | | |
4 | | #include "unicode/utypes.h" |
5 | | |
6 | | #if !UCONFIG_NO_FORMATTING |
7 | | |
8 | | // This file contains one implementation of FormattedValue. |
9 | | // Other independent implementations should go into their own cpp file for |
10 | | // better dependency modularization. |
11 | | |
12 | | #include "unicode/ustring.h" |
13 | | #include "formattedval_impl.h" |
14 | | #include "number_types.h" |
15 | | #include "formatted_string_builder.h" |
16 | | #include "number_utils.h" |
17 | | #include "static_unicode_sets.h" |
18 | | #include "unicode/listformatter.h" |
19 | | |
20 | | U_NAMESPACE_BEGIN |
21 | | |
22 | | |
23 | | typedef FormattedStringBuilder::Field Field; |
24 | | |
25 | | |
26 | | FormattedValueStringBuilderImpl::FormattedValueStringBuilderImpl(Field numericField) |
27 | 10.8M | : fNumericField(numericField) { |
28 | 10.8M | } |
29 | | |
30 | 10.8M | FormattedValueStringBuilderImpl::~FormattedValueStringBuilderImpl() { |
31 | 10.8M | } |
32 | | |
33 | | |
34 | 0 | UnicodeString FormattedValueStringBuilderImpl::toString(UErrorCode&) const { |
35 | 0 | return fString.toUnicodeString(); |
36 | 0 | } |
37 | | |
38 | 0 | UnicodeString FormattedValueStringBuilderImpl::toTempString(UErrorCode&) const { |
39 | 0 | return fString.toTempUnicodeString(); |
40 | 0 | } |
41 | | |
42 | 102k | Appendable& FormattedValueStringBuilderImpl::appendTo(Appendable& appendable, UErrorCode&) const { |
43 | 102k | appendable.appendString(fString.chars(), fString.length()); |
44 | 102k | return appendable; |
45 | 102k | } |
46 | | |
47 | 0 | UBool FormattedValueStringBuilderImpl::nextPosition(ConstrainedFieldPosition& cfpos, UErrorCode& status) const { |
48 | | // NOTE: MSVC sometimes complains when implicitly converting between bool and UBool |
49 | 0 | return nextPositionImpl(cfpos, fNumericField, status) ? true : false; |
50 | 0 | } |
51 | | |
52 | 0 | UBool FormattedValueStringBuilderImpl::nextFieldPosition(FieldPosition& fp, UErrorCode& status) const { |
53 | 0 | int32_t rawField = fp.getField(); |
54 | |
|
55 | 0 | if (rawField == FieldPosition::DONT_CARE) { |
56 | 0 | return false; |
57 | 0 | } |
58 | | |
59 | 0 | if (rawField < 0 || rawField >= UNUM_FIELD_COUNT) { |
60 | 0 | status = U_ILLEGAL_ARGUMENT_ERROR; |
61 | 0 | return false; |
62 | 0 | } |
63 | | |
64 | 0 | ConstrainedFieldPosition cfpos; |
65 | 0 | cfpos.constrainField(UFIELD_CATEGORY_NUMBER, rawField); |
66 | 0 | cfpos.setState(UFIELD_CATEGORY_NUMBER, rawField, fp.getBeginIndex(), fp.getEndIndex()); |
67 | 0 | if (nextPositionImpl(cfpos, kUndefinedField, status)) { |
68 | 0 | fp.setBeginIndex(cfpos.getStart()); |
69 | 0 | fp.setEndIndex(cfpos.getLimit()); |
70 | 0 | return true; |
71 | 0 | } |
72 | | |
73 | | // Special case: fraction should start after integer if fraction is not present |
74 | 0 | if (rawField == UNUM_FRACTION_FIELD && fp.getEndIndex() == 0) { |
75 | 0 | bool inside = false; |
76 | 0 | int32_t i = fString.fZero; |
77 | 0 | for (; i < fString.fZero + fString.fLength; i++) { |
78 | 0 | if (isIntOrGroup(fString.getFieldPtr()[i]) || fString.getFieldPtr()[i] == Field(UFIELD_CATEGORY_NUMBER, UNUM_DECIMAL_SEPARATOR_FIELD)) { |
79 | 0 | inside = true; |
80 | 0 | } else if (inside) { |
81 | 0 | break; |
82 | 0 | } |
83 | 0 | } |
84 | 0 | fp.setBeginIndex(i - fString.fZero); |
85 | 0 | fp.setEndIndex(i - fString.fZero); |
86 | 0 | } |
87 | |
|
88 | 0 | return false; |
89 | 0 | } |
90 | | |
91 | | void FormattedValueStringBuilderImpl::getAllFieldPositions(FieldPositionIteratorHandler& fpih, |
92 | 0 | UErrorCode& status) const { |
93 | 0 | ConstrainedFieldPosition cfpos; |
94 | 0 | while (nextPositionImpl(cfpos, kUndefinedField, status)) { |
95 | 0 | fpih.addAttribute(cfpos.getField(), cfpos.getStart(), cfpos.getLimit()); |
96 | 0 | } |
97 | 0 | } |
98 | | |
99 | 0 | void FormattedValueStringBuilderImpl::resetString() { |
100 | 0 | fString.clear(); |
101 | 0 | spanIndicesCount = 0; |
102 | 0 | } |
103 | | |
104 | | // Signal the end of the string using a field that doesn't exist and that is |
105 | | // different from kUndefinedField, which is used for "null field". |
106 | | static constexpr Field kEndField = Field(0xf, 0xf); |
107 | | |
108 | 0 | bool FormattedValueStringBuilderImpl::nextPositionImpl(ConstrainedFieldPosition& cfpos, Field numericField, UErrorCode& /*status*/) const { |
109 | 0 | int32_t fieldStart = -1; |
110 | 0 | Field currField = kUndefinedField; |
111 | 0 | bool prevIsSpan = false; |
112 | 0 | int32_t nextSpanStart = -1; |
113 | 0 | if (spanIndicesCount > 0) { |
114 | 0 | int64_t si = cfpos.getInt64IterationContext(); |
115 | 0 | U_ASSERT(si <= spanIndicesCount); |
116 | 0 | if (si < spanIndicesCount) { |
117 | 0 | nextSpanStart = spanIndices[si].start; |
118 | 0 | } |
119 | 0 | if (si > 0) { |
120 | 0 | prevIsSpan = cfpos.getCategory() == spanIndices[si-1].category |
121 | 0 | && cfpos.getField() == spanIndices[si-1].spanValue; |
122 | 0 | } |
123 | 0 | } |
124 | 0 | bool prevIsNumeric = false; |
125 | 0 | if (numericField != kUndefinedField) { |
126 | 0 | prevIsNumeric = cfpos.getCategory() == numericField.getCategory() |
127 | 0 | && cfpos.getField() == numericField.getField(); |
128 | 0 | } |
129 | 0 | bool prevIsInteger = cfpos.getCategory() == UFIELD_CATEGORY_NUMBER |
130 | 0 | && cfpos.getField() == UNUM_INTEGER_FIELD; |
131 | |
|
132 | 0 | for (int32_t i = fString.fZero + cfpos.getLimit(); i <= fString.fZero + fString.fLength; i++) { |
133 | 0 | Field _field = (i < fString.fZero + fString.fLength) ? fString.getFieldPtr()[i] : kEndField; |
134 | | // Case 1: currently scanning a field. |
135 | 0 | if (currField != kUndefinedField) { |
136 | 0 | if (currField != _field) { |
137 | 0 | int32_t end = i - fString.fZero; |
138 | | // Grouping separators can be whitespace; don't throw them out! |
139 | 0 | if (isTrimmable(currField)) { |
140 | 0 | end = trimBack(i - fString.fZero); |
141 | 0 | } |
142 | 0 | if (end <= fieldStart) { |
143 | | // Entire field position is ignorable; skip. |
144 | 0 | fieldStart = -1; |
145 | 0 | currField = kUndefinedField; |
146 | 0 | i--; // look at this index again |
147 | 0 | continue; |
148 | 0 | } |
149 | 0 | int32_t start = fieldStart; |
150 | 0 | if (isTrimmable(currField)) { |
151 | 0 | start = trimFront(start); |
152 | 0 | } |
153 | 0 | cfpos.setState(currField.getCategory(), currField.getField(), start, end); |
154 | 0 | return true; |
155 | 0 | } |
156 | 0 | continue; |
157 | 0 | } |
158 | | // Special case: emit normalField if we are pointing at the end of spanField. |
159 | 0 | if (i > fString.fZero && prevIsSpan) { |
160 | 0 | int64_t si = cfpos.getInt64IterationContext() - 1; |
161 | 0 | U_ASSERT(si >= 0); |
162 | 0 | int32_t previ = i - spanIndices[si].length; |
163 | 0 | U_ASSERT(previ >= fString.fZero); |
164 | 0 | Field prevField = fString.getFieldPtr()[previ]; |
165 | 0 | if (prevField == Field(UFIELD_CATEGORY_LIST, ULISTFMT_ELEMENT_FIELD)) { |
166 | | // Special handling for ULISTFMT_ELEMENT_FIELD |
167 | 0 | if (cfpos.matchesField(UFIELD_CATEGORY_LIST, ULISTFMT_ELEMENT_FIELD)) { |
168 | 0 | fieldStart = i - fString.fZero - spanIndices[si].length; |
169 | 0 | int32_t end = fieldStart + spanIndices[si].length; |
170 | 0 | cfpos.setState( |
171 | 0 | UFIELD_CATEGORY_LIST, |
172 | 0 | ULISTFMT_ELEMENT_FIELD, |
173 | 0 | fieldStart, |
174 | 0 | end); |
175 | 0 | return true; |
176 | 0 | } else { |
177 | 0 | prevIsSpan = false; |
178 | 0 | } |
179 | 0 | } else { |
180 | | // Re-wind, since there may be multiple fields in the span. |
181 | 0 | i = previ; |
182 | 0 | _field = prevField; |
183 | 0 | } |
184 | 0 | } |
185 | | // Special case: coalesce the INTEGER if we are pointing at the end of the INTEGER. |
186 | 0 | if (cfpos.matchesField(UFIELD_CATEGORY_NUMBER, UNUM_INTEGER_FIELD) |
187 | 0 | && i > fString.fZero |
188 | 0 | && !prevIsInteger |
189 | 0 | && !prevIsNumeric |
190 | 0 | && isIntOrGroup(fString.getFieldPtr()[i - 1]) |
191 | 0 | && !isIntOrGroup(_field)) { |
192 | 0 | int j = i - 1; |
193 | 0 | for (; j >= fString.fZero && isIntOrGroup(fString.getFieldPtr()[j]); j--) {} |
194 | 0 | cfpos.setState( |
195 | 0 | UFIELD_CATEGORY_NUMBER, |
196 | 0 | UNUM_INTEGER_FIELD, |
197 | 0 | j - fString.fZero + 1, |
198 | 0 | i - fString.fZero); |
199 | 0 | return true; |
200 | 0 | } |
201 | | // Special case: coalesce NUMERIC if we are pointing at the end of the NUMERIC. |
202 | 0 | if (numericField != kUndefinedField |
203 | 0 | && cfpos.matchesField(numericField.getCategory(), numericField.getField()) |
204 | 0 | && i > fString.fZero |
205 | 0 | && !prevIsNumeric |
206 | 0 | && fString.getFieldPtr()[i - 1].isNumeric() |
207 | 0 | && !_field.isNumeric()) { |
208 | | // Re-wind to the beginning of the field and then emit it |
209 | 0 | int32_t j = i - 1; |
210 | 0 | for (; j >= fString.fZero && fString.getFieldPtr()[j].isNumeric(); j--) {} |
211 | 0 | cfpos.setState( |
212 | 0 | numericField.getCategory(), |
213 | 0 | numericField.getField(), |
214 | 0 | j - fString.fZero + 1, |
215 | 0 | i - fString.fZero); |
216 | 0 | return true; |
217 | 0 | } |
218 | | // Check for span field |
219 | 0 | if (!prevIsSpan && ( |
220 | 0 | _field == Field(UFIELD_CATEGORY_LIST, ULISTFMT_ELEMENT_FIELD) || |
221 | 0 | i - fString.fZero == nextSpanStart)) { |
222 | 0 | int64_t si = cfpos.getInt64IterationContext(); |
223 | 0 | if (si >= spanIndicesCount) { |
224 | 0 | break; |
225 | 0 | } |
226 | 0 | UFieldCategory spanCategory = spanIndices[si].category; |
227 | 0 | int32_t spanValue = spanIndices[si].spanValue; |
228 | 0 | int32_t length = spanIndices[si].length; |
229 | 0 | cfpos.setInt64IterationContext(si + 1); |
230 | 0 | if (si + 1 < spanIndicesCount) { |
231 | 0 | nextSpanStart = spanIndices[si + 1].start; |
232 | 0 | } |
233 | 0 | if (length == 0) { |
234 | | // ICU-21871: Don't return fields on empty spans |
235 | 0 | i--; |
236 | 0 | continue; |
237 | 0 | } |
238 | 0 | if (cfpos.matchesField(spanCategory, spanValue)) { |
239 | 0 | fieldStart = i - fString.fZero; |
240 | 0 | int32_t end = fieldStart + length; |
241 | 0 | cfpos.setState( |
242 | 0 | spanCategory, |
243 | 0 | spanValue, |
244 | 0 | fieldStart, |
245 | 0 | end); |
246 | 0 | return true; |
247 | 0 | } else if (_field == Field(UFIELD_CATEGORY_LIST, ULISTFMT_ELEMENT_FIELD)) { |
248 | | // Special handling for ULISTFMT_ELEMENT_FIELD |
249 | 0 | if (cfpos.matchesField(UFIELD_CATEGORY_LIST, ULISTFMT_ELEMENT_FIELD)) { |
250 | 0 | fieldStart = i - fString.fZero; |
251 | 0 | int32_t end = fieldStart + length; |
252 | 0 | cfpos.setState( |
253 | 0 | UFIELD_CATEGORY_LIST, |
254 | 0 | ULISTFMT_ELEMENT_FIELD, |
255 | 0 | fieldStart, |
256 | 0 | end); |
257 | 0 | return true; |
258 | 0 | } else { |
259 | | // Failed to match; jump ahead |
260 | 0 | i += length - 1; |
261 | | // goto loopend |
262 | 0 | } |
263 | 0 | } |
264 | 0 | } |
265 | | // Special case: skip over INTEGER; will be coalesced later. |
266 | 0 | else if (_field == Field(UFIELD_CATEGORY_NUMBER, UNUM_INTEGER_FIELD)) { |
267 | 0 | _field = kUndefinedField; |
268 | 0 | } |
269 | | // No field starting at this position. |
270 | 0 | else if (_field.isUndefined() || _field == kEndField) { |
271 | | // goto loopend |
272 | 0 | } |
273 | | // No SpanField |
274 | 0 | else if (cfpos.matchesField(_field.getCategory(), _field.getField())) { |
275 | 0 | fieldStart = i - fString.fZero; |
276 | 0 | currField = _field; |
277 | 0 | } |
278 | | // loopend: |
279 | 0 | prevIsSpan = false; |
280 | 0 | prevIsNumeric = false; |
281 | 0 | prevIsInteger = false; |
282 | 0 | } |
283 | | |
284 | 0 | U_ASSERT(currField == kUndefinedField); |
285 | | // Always set the position to the end so that we don't revisit previous sections |
286 | 0 | cfpos.setState( |
287 | 0 | cfpos.getCategory(), |
288 | 0 | cfpos.getField(), |
289 | 0 | fString.fLength, |
290 | 0 | fString.fLength); |
291 | 0 | return false; |
292 | 0 | } |
293 | | |
294 | 45.3k | void FormattedValueStringBuilderImpl::appendSpanInfo(UFieldCategory category, int32_t spanValue, int32_t start, int32_t length, UErrorCode& status) { |
295 | 45.3k | if (U_FAILURE(status)) { return; } |
296 | 45.3k | U_ASSERT(spanIndices.getCapacity() >= spanIndicesCount); |
297 | 45.3k | if (spanIndices.getCapacity() == spanIndicesCount) { |
298 | 0 | if (!spanIndices.resize(spanIndicesCount * 2, spanIndicesCount)) { |
299 | 0 | status = U_MEMORY_ALLOCATION_ERROR; |
300 | 0 | return; |
301 | 0 | } |
302 | 0 | } |
303 | 45.3k | spanIndices[spanIndicesCount] = {category, spanValue, start, length}; |
304 | 45.3k | spanIndicesCount++; |
305 | 45.3k | } |
306 | | |
307 | 0 | void FormattedValueStringBuilderImpl::prependSpanInfo(UFieldCategory category, int32_t spanValue, int32_t start, int32_t length, UErrorCode& status) { |
308 | 0 | if (U_FAILURE(status)) { return; } |
309 | 0 | U_ASSERT(spanIndices.getCapacity() >= spanIndicesCount); |
310 | 0 | if (spanIndices.getCapacity() == spanIndicesCount) { |
311 | 0 | if (!spanIndices.resize(spanIndicesCount * 2, spanIndicesCount)) { |
312 | 0 | status = U_MEMORY_ALLOCATION_ERROR; |
313 | 0 | return; |
314 | 0 | } |
315 | 0 | } |
316 | 0 | for (int32_t i = spanIndicesCount - 1; i >= 0; i--) { |
317 | 0 | spanIndices[i+1] = spanIndices[i]; |
318 | 0 | } |
319 | 0 | spanIndices[0] = {category, spanValue, start, length}; |
320 | 0 | spanIndicesCount++; |
321 | 0 | } |
322 | | |
323 | 0 | bool FormattedValueStringBuilderImpl::isIntOrGroup(Field field) { |
324 | 0 | return field == Field(UFIELD_CATEGORY_NUMBER, UNUM_INTEGER_FIELD) |
325 | 0 | || field == Field(UFIELD_CATEGORY_NUMBER, UNUM_GROUPING_SEPARATOR_FIELD); |
326 | 0 | } |
327 | | |
328 | 0 | bool FormattedValueStringBuilderImpl::isTrimmable(Field field) { |
329 | 0 | return field != Field(UFIELD_CATEGORY_NUMBER, UNUM_GROUPING_SEPARATOR_FIELD) |
330 | 0 | && field.getCategory() != UFIELD_CATEGORY_LIST; |
331 | 0 | } |
332 | | |
333 | 0 | int32_t FormattedValueStringBuilderImpl::trimBack(int32_t limit) const { |
334 | 0 | return unisets::get(unisets::DEFAULT_IGNORABLES)->spanBack( |
335 | 0 | fString.getCharPtr() + fString.fZero, |
336 | 0 | limit, |
337 | 0 | USET_SPAN_CONTAINED); |
338 | 0 | } |
339 | | |
340 | 0 | int32_t FormattedValueStringBuilderImpl::trimFront(int32_t start) const { |
341 | 0 | return start + unisets::get(unisets::DEFAULT_IGNORABLES)->span( |
342 | 0 | fString.getCharPtr() + fString.fZero + start, |
343 | 0 | fString.fLength - start, |
344 | 0 | USET_SPAN_CONTAINED); |
345 | 0 | } |
346 | | |
347 | | |
348 | | U_NAMESPACE_END |
349 | | |
350 | | #endif /* #if !UCONFIG_NO_FORMATTING */ |