/src/mozilla-central/intl/icu/source/i18n/number_modifiers.cpp
Line | Count | Source (jump to first uncovered line) |
1 | | // © 2017 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 | | #include "umutex.h" |
9 | | #include "ucln_cmn.h" |
10 | | #include "ucln_in.h" |
11 | | #include "number_modifiers.h" |
12 | | |
13 | | using namespace icu; |
14 | | using namespace icu::number; |
15 | | using namespace icu::number::impl; |
16 | | |
17 | | namespace { |
18 | | |
19 | | // TODO: This is copied from simpleformatter.cpp |
20 | | const int32_t ARG_NUM_LIMIT = 0x100; |
21 | | |
22 | | // These are the default currency spacing UnicodeSets in CLDR. |
23 | | // Pre-compute them for performance. |
24 | | // The Java unit test testCurrencySpacingPatternStability() will start failing if these change in CLDR. |
25 | | icu::UInitOnce gDefaultCurrencySpacingInitOnce = U_INITONCE_INITIALIZER; |
26 | | |
27 | | UnicodeSet *UNISET_DIGIT = nullptr; |
28 | | UnicodeSet *UNISET_NOTS = nullptr; |
29 | | |
30 | 0 | UBool U_CALLCONV cleanupDefaultCurrencySpacing() { |
31 | 0 | delete UNISET_DIGIT; |
32 | 0 | UNISET_DIGIT = nullptr; |
33 | 0 | delete UNISET_NOTS; |
34 | 0 | UNISET_NOTS = nullptr; |
35 | 0 | gDefaultCurrencySpacingInitOnce.reset(); |
36 | 0 | return TRUE; |
37 | 0 | } |
38 | | |
39 | 0 | void U_CALLCONV initDefaultCurrencySpacing(UErrorCode &status) { |
40 | 0 | ucln_i18n_registerCleanup(UCLN_I18N_CURRENCY_SPACING, cleanupDefaultCurrencySpacing); |
41 | 0 | UNISET_DIGIT = new UnicodeSet(UnicodeString(u"[:digit:]"), status); |
42 | 0 | UNISET_NOTS = new UnicodeSet(UnicodeString(u"[:^S:]"), status); |
43 | 0 | if (UNISET_DIGIT == nullptr || UNISET_NOTS == nullptr) { |
44 | 0 | status = U_MEMORY_ALLOCATION_ERROR; |
45 | 0 | return; |
46 | 0 | } |
47 | 0 | UNISET_DIGIT->freeze(); |
48 | 0 | UNISET_NOTS->freeze(); |
49 | 0 | } |
50 | | |
51 | | } // namespace |
52 | | |
53 | | |
54 | 0 | Modifier::~Modifier() = default; |
55 | | |
56 | | |
57 | | int32_t ConstantAffixModifier::apply(NumberStringBuilder &output, int leftIndex, int rightIndex, |
58 | 0 | UErrorCode &status) const { |
59 | 0 | // Insert the suffix first since inserting the prefix will change the rightIndex |
60 | 0 | int length = output.insert(rightIndex, fSuffix, fField, status); |
61 | 0 | length += output.insert(leftIndex, fPrefix, fField, status); |
62 | 0 | return length; |
63 | 0 | } |
64 | | |
65 | 0 | int32_t ConstantAffixModifier::getPrefixLength(UErrorCode &status) const { |
66 | 0 | (void)status; |
67 | 0 | return fPrefix.length(); |
68 | 0 | } |
69 | | |
70 | 0 | int32_t ConstantAffixModifier::getCodePointCount(UErrorCode &status) const { |
71 | 0 | (void)status; |
72 | 0 | return fPrefix.countChar32() + fSuffix.countChar32(); |
73 | 0 | } |
74 | | |
75 | 0 | bool ConstantAffixModifier::isStrong() const { |
76 | 0 | return fStrong; |
77 | 0 | } |
78 | | |
79 | | SimpleModifier::SimpleModifier(const SimpleFormatter &simpleFormatter, Field field, bool strong) |
80 | 0 | : fCompiledPattern(simpleFormatter.compiledPattern), fField(field), fStrong(strong) { |
81 | 0 | int32_t argLimit = SimpleFormatter::getArgumentLimit( |
82 | 0 | fCompiledPattern.getBuffer(), fCompiledPattern.length()); |
83 | 0 | if (argLimit == 0) { |
84 | 0 | // No arguments in compiled pattern |
85 | 0 | fPrefixLength = fCompiledPattern.charAt(1) - ARG_NUM_LIMIT; |
86 | 0 | U_ASSERT(2 + fPrefixLength == fCompiledPattern.length()); |
87 | 0 | // Set suffixOffset = -1 to indicate no arguments in compiled pattern. |
88 | 0 | fSuffixOffset = -1; |
89 | 0 | fSuffixLength = 0; |
90 | 0 | } else { |
91 | 0 | U_ASSERT(argLimit == 1); |
92 | 0 | if (fCompiledPattern.charAt(1) != 0) { |
93 | 0 | fPrefixLength = fCompiledPattern.charAt(1) - ARG_NUM_LIMIT; |
94 | 0 | fSuffixOffset = 3 + fPrefixLength; |
95 | 0 | } else { |
96 | 0 | fPrefixLength = 0; |
97 | 0 | fSuffixOffset = 2; |
98 | 0 | } |
99 | 0 | if (3 + fPrefixLength < fCompiledPattern.length()) { |
100 | 0 | fSuffixLength = fCompiledPattern.charAt(fSuffixOffset) - ARG_NUM_LIMIT; |
101 | 0 | } else { |
102 | 0 | fSuffixLength = 0; |
103 | 0 | } |
104 | 0 | } |
105 | 0 | } |
106 | | |
107 | | SimpleModifier::SimpleModifier() |
108 | 0 | : fField(UNUM_FIELD_COUNT), fStrong(false), fPrefixLength(0), fSuffixLength(0) { |
109 | 0 | } |
110 | | |
111 | | int32_t SimpleModifier::apply(NumberStringBuilder &output, int leftIndex, int rightIndex, |
112 | 0 | UErrorCode &status) const { |
113 | 0 | return formatAsPrefixSuffix(output, leftIndex, rightIndex, fField, status); |
114 | 0 | } |
115 | | |
116 | 0 | int32_t SimpleModifier::getPrefixLength(UErrorCode &status) const { |
117 | 0 | (void)status; |
118 | 0 | return fPrefixLength; |
119 | 0 | } |
120 | | |
121 | 0 | int32_t SimpleModifier::getCodePointCount(UErrorCode &status) const { |
122 | 0 | (void)status; |
123 | 0 | int32_t count = 0; |
124 | 0 | if (fPrefixLength > 0) { |
125 | 0 | count += fCompiledPattern.countChar32(2, fPrefixLength); |
126 | 0 | } |
127 | 0 | if (fSuffixLength > 0) { |
128 | 0 | count += fCompiledPattern.countChar32(1 + fSuffixOffset, fSuffixLength); |
129 | 0 | } |
130 | 0 | return count; |
131 | 0 | } |
132 | | |
133 | 0 | bool SimpleModifier::isStrong() const { |
134 | 0 | return fStrong; |
135 | 0 | } |
136 | | |
137 | | int32_t |
138 | | SimpleModifier::formatAsPrefixSuffix(NumberStringBuilder &result, int32_t startIndex, int32_t endIndex, |
139 | 0 | Field field, UErrorCode &status) const { |
140 | 0 | if (fSuffixOffset == -1) { |
141 | 0 | // There is no argument for the inner number; overwrite the entire segment with our string. |
142 | 0 | return result.splice(startIndex, endIndex, fCompiledPattern, 2, 2 + fPrefixLength, field, status); |
143 | 0 | } else { |
144 | 0 | if (fPrefixLength > 0) { |
145 | 0 | result.insert(startIndex, fCompiledPattern, 2, 2 + fPrefixLength, field, status); |
146 | 0 | } |
147 | 0 | if (fSuffixLength > 0) { |
148 | 0 | result.insert( |
149 | 0 | endIndex + fPrefixLength, |
150 | 0 | fCompiledPattern, |
151 | 0 | 1 + fSuffixOffset, |
152 | 0 | 1 + fSuffixOffset + fSuffixLength, |
153 | 0 | field, |
154 | 0 | status); |
155 | 0 | } |
156 | 0 | return fPrefixLength + fSuffixLength; |
157 | 0 | } |
158 | 0 | } |
159 | | |
160 | | int32_t ConstantMultiFieldModifier::apply(NumberStringBuilder &output, int leftIndex, int rightIndex, |
161 | 0 | UErrorCode &status) const { |
162 | 0 | int32_t length = output.insert(leftIndex, fPrefix, status); |
163 | 0 | if (fOverwrite) { |
164 | 0 | length += output.splice( |
165 | 0 | leftIndex + length, |
166 | 0 | rightIndex + length, |
167 | 0 | UnicodeString(), 0, 0, |
168 | 0 | UNUM_FIELD_COUNT, status); |
169 | 0 | } |
170 | 0 | length += output.insert(rightIndex + length, fSuffix, status); |
171 | 0 | return length; |
172 | 0 | } |
173 | | |
174 | 0 | int32_t ConstantMultiFieldModifier::getPrefixLength(UErrorCode &status) const { |
175 | 0 | (void)status; |
176 | 0 | return fPrefix.length(); |
177 | 0 | } |
178 | | |
179 | 0 | int32_t ConstantMultiFieldModifier::getCodePointCount(UErrorCode &status) const { |
180 | 0 | (void)status; |
181 | 0 | return fPrefix.codePointCount() + fSuffix.codePointCount(); |
182 | 0 | } |
183 | | |
184 | 0 | bool ConstantMultiFieldModifier::isStrong() const { |
185 | 0 | return fStrong; |
186 | 0 | } |
187 | | |
188 | | CurrencySpacingEnabledModifier::CurrencySpacingEnabledModifier(const NumberStringBuilder &prefix, |
189 | | const NumberStringBuilder &suffix, |
190 | | bool overwrite, |
191 | | bool strong, |
192 | | const DecimalFormatSymbols &symbols, |
193 | | UErrorCode &status) |
194 | 0 | : ConstantMultiFieldModifier(prefix, suffix, overwrite, strong) { |
195 | 0 | // Check for currency spacing. Do not build the UnicodeSets unless there is |
196 | 0 | // a currency code point at a boundary. |
197 | 0 | if (prefix.length() > 0 && prefix.fieldAt(prefix.length() - 1) == UNUM_CURRENCY_FIELD) { |
198 | 0 | int prefixCp = prefix.getLastCodePoint(); |
199 | 0 | UnicodeSet prefixUnicodeSet = getUnicodeSet(symbols, IN_CURRENCY, PREFIX, status); |
200 | 0 | if (prefixUnicodeSet.contains(prefixCp)) { |
201 | 0 | fAfterPrefixUnicodeSet = getUnicodeSet(symbols, IN_NUMBER, PREFIX, status); |
202 | 0 | fAfterPrefixUnicodeSet.freeze(); |
203 | 0 | fAfterPrefixInsert = getInsertString(symbols, PREFIX, status); |
204 | 0 | } else { |
205 | 0 | fAfterPrefixUnicodeSet.setToBogus(); |
206 | 0 | fAfterPrefixInsert.setToBogus(); |
207 | 0 | } |
208 | 0 | } else { |
209 | 0 | fAfterPrefixUnicodeSet.setToBogus(); |
210 | 0 | fAfterPrefixInsert.setToBogus(); |
211 | 0 | } |
212 | 0 | if (suffix.length() > 0 && suffix.fieldAt(0) == UNUM_CURRENCY_FIELD) { |
213 | 0 | int suffixCp = suffix.getLastCodePoint(); |
214 | 0 | UnicodeSet suffixUnicodeSet = getUnicodeSet(symbols, IN_CURRENCY, SUFFIX, status); |
215 | 0 | if (suffixUnicodeSet.contains(suffixCp)) { |
216 | 0 | fBeforeSuffixUnicodeSet = getUnicodeSet(symbols, IN_NUMBER, SUFFIX, status); |
217 | 0 | fBeforeSuffixUnicodeSet.freeze(); |
218 | 0 | fBeforeSuffixInsert = getInsertString(symbols, SUFFIX, status); |
219 | 0 | } else { |
220 | 0 | fBeforeSuffixUnicodeSet.setToBogus(); |
221 | 0 | fBeforeSuffixInsert.setToBogus(); |
222 | 0 | } |
223 | 0 | } else { |
224 | 0 | fBeforeSuffixUnicodeSet.setToBogus(); |
225 | 0 | fBeforeSuffixInsert.setToBogus(); |
226 | 0 | } |
227 | 0 | } |
228 | | |
229 | | int32_t CurrencySpacingEnabledModifier::apply(NumberStringBuilder &output, int leftIndex, int rightIndex, |
230 | 0 | UErrorCode &status) const { |
231 | 0 | // Currency spacing logic |
232 | 0 | int length = 0; |
233 | 0 | if (rightIndex - leftIndex > 0 && !fAfterPrefixUnicodeSet.isBogus() && |
234 | 0 | fAfterPrefixUnicodeSet.contains(output.codePointAt(leftIndex))) { |
235 | 0 | // TODO: Should we use the CURRENCY field here? |
236 | 0 | length += output.insert(leftIndex, fAfterPrefixInsert, UNUM_FIELD_COUNT, status); |
237 | 0 | } |
238 | 0 | if (rightIndex - leftIndex > 0 && !fBeforeSuffixUnicodeSet.isBogus() && |
239 | 0 | fBeforeSuffixUnicodeSet.contains(output.codePointBefore(rightIndex))) { |
240 | 0 | // TODO: Should we use the CURRENCY field here? |
241 | 0 | length += output.insert(rightIndex + length, fBeforeSuffixInsert, UNUM_FIELD_COUNT, status); |
242 | 0 | } |
243 | 0 |
|
244 | 0 | // Call super for the remaining logic |
245 | 0 | length += ConstantMultiFieldModifier::apply(output, leftIndex, rightIndex + length, status); |
246 | 0 | return length; |
247 | 0 | } |
248 | | |
249 | | int32_t |
250 | | CurrencySpacingEnabledModifier::applyCurrencySpacing(NumberStringBuilder &output, int32_t prefixStart, |
251 | | int32_t prefixLen, int32_t suffixStart, |
252 | | int32_t suffixLen, |
253 | | const DecimalFormatSymbols &symbols, |
254 | 0 | UErrorCode &status) { |
255 | 0 | int length = 0; |
256 | 0 | bool hasPrefix = (prefixLen > 0); |
257 | 0 | bool hasSuffix = (suffixLen > 0); |
258 | 0 | bool hasNumber = (suffixStart - prefixStart - prefixLen > 0); // could be empty string |
259 | 0 | if (hasPrefix && hasNumber) { |
260 | 0 | length += applyCurrencySpacingAffix(output, prefixStart + prefixLen, PREFIX, symbols, status); |
261 | 0 | } |
262 | 0 | if (hasSuffix && hasNumber) { |
263 | 0 | length += applyCurrencySpacingAffix(output, suffixStart + length, SUFFIX, symbols, status); |
264 | 0 | } |
265 | 0 | return length; |
266 | 0 | } |
267 | | |
268 | | int32_t |
269 | | CurrencySpacingEnabledModifier::applyCurrencySpacingAffix(NumberStringBuilder &output, int32_t index, |
270 | | EAffix affix, |
271 | | const DecimalFormatSymbols &symbols, |
272 | 0 | UErrorCode &status) { |
273 | 0 | // NOTE: For prefix, output.fieldAt(index-1) gets the last field type in the prefix. |
274 | 0 | // This works even if the last code point in the prefix is 2 code units because the |
275 | 0 | // field value gets populated to both indices in the field array. |
276 | 0 | Field affixField = (affix == PREFIX) ? output.fieldAt(index - 1) : output.fieldAt(index); |
277 | 0 | if (affixField != UNUM_CURRENCY_FIELD) { |
278 | 0 | return 0; |
279 | 0 | } |
280 | 0 | int affixCp = (affix == PREFIX) ? output.codePointBefore(index) : output.codePointAt(index); |
281 | 0 | UnicodeSet affixUniset = getUnicodeSet(symbols, IN_CURRENCY, affix, status); |
282 | 0 | if (!affixUniset.contains(affixCp)) { |
283 | 0 | return 0; |
284 | 0 | } |
285 | 0 | int numberCp = (affix == PREFIX) ? output.codePointAt(index) : output.codePointBefore(index); |
286 | 0 | UnicodeSet numberUniset = getUnicodeSet(symbols, IN_NUMBER, affix, status); |
287 | 0 | if (!numberUniset.contains(numberCp)) { |
288 | 0 | return 0; |
289 | 0 | } |
290 | 0 | UnicodeString spacingString = getInsertString(symbols, affix, status); |
291 | 0 |
|
292 | 0 | // NOTE: This next line *inserts* the spacing string, triggering an arraycopy. |
293 | 0 | // It would be more efficient if this could be done before affixes were attached, |
294 | 0 | // so that it could be prepended/appended instead of inserted. |
295 | 0 | // However, the build code path is more efficient, and this is the most natural |
296 | 0 | // place to put currency spacing in the non-build code path. |
297 | 0 | // TODO: Should we use the CURRENCY field here? |
298 | 0 | return output.insert(index, spacingString, UNUM_FIELD_COUNT, status); |
299 | 0 | } |
300 | | |
301 | | UnicodeSet |
302 | | CurrencySpacingEnabledModifier::getUnicodeSet(const DecimalFormatSymbols &symbols, EPosition position, |
303 | 0 | EAffix affix, UErrorCode &status) { |
304 | 0 | // Ensure the static defaults are initialized: |
305 | 0 | umtx_initOnce(gDefaultCurrencySpacingInitOnce, &initDefaultCurrencySpacing, status); |
306 | 0 | if (U_FAILURE(status)) { |
307 | 0 | return UnicodeSet(); |
308 | 0 | } |
309 | 0 | |
310 | 0 | const UnicodeString& pattern = symbols.getPatternForCurrencySpacing( |
311 | 0 | position == IN_CURRENCY ? UNUM_CURRENCY_MATCH : UNUM_CURRENCY_SURROUNDING_MATCH, |
312 | 0 | affix == SUFFIX, |
313 | 0 | status); |
314 | 0 | if (pattern.compare(u"[:digit:]", -1) == 0) { |
315 | 0 | return *UNISET_DIGIT; |
316 | 0 | } else if (pattern.compare(u"[:^S:]", -1) == 0) { |
317 | 0 | return *UNISET_NOTS; |
318 | 0 | } else { |
319 | 0 | return UnicodeSet(pattern, status); |
320 | 0 | } |
321 | 0 | } |
322 | | |
323 | | UnicodeString |
324 | | CurrencySpacingEnabledModifier::getInsertString(const DecimalFormatSymbols &symbols, EAffix affix, |
325 | 0 | UErrorCode &status) { |
326 | 0 | return symbols.getPatternForCurrencySpacing(UNUM_CURRENCY_INSERT, affix == SUFFIX, status); |
327 | 0 | } |
328 | | |
329 | | #endif /* #if !UCONFIG_NO_FORMATTING */ |