Jungshik Shin | b318966 | 2017-11-07 11:18:34 -0800 | [diff] [blame^] | 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 && !UPRV_INCOMPLETE_CPP11_SUPPORT |
| 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 | UBool U_CALLCONV cleanupDefaultCurrencySpacing() { |
| 31 | delete UNISET_DIGIT; |
| 32 | UNISET_DIGIT = nullptr; |
| 33 | delete UNISET_NOTS; |
| 34 | UNISET_NOTS = nullptr; |
| 35 | return TRUE; |
| 36 | } |
| 37 | |
| 38 | void U_CALLCONV initDefaultCurrencySpacing(UErrorCode &status) { |
| 39 | ucln_i18n_registerCleanup(UCLN_I18N_CURRENCY_SPACING, cleanupDefaultCurrencySpacing); |
| 40 | UNISET_DIGIT = new UnicodeSet(UnicodeString(u"[:digit:]"), status); |
| 41 | UNISET_NOTS = new UnicodeSet(UnicodeString(u"[:^S:]"), status); |
| 42 | if (UNISET_DIGIT == nullptr || UNISET_NOTS == nullptr) { |
| 43 | status = U_MEMORY_ALLOCATION_ERROR; |
| 44 | return; |
| 45 | } |
| 46 | UNISET_DIGIT->freeze(); |
| 47 | UNISET_NOTS->freeze(); |
| 48 | } |
| 49 | |
| 50 | } // namespace |
| 51 | |
| 52 | |
| 53 | int32_t ConstantAffixModifier::apply(NumberStringBuilder &output, int leftIndex, int rightIndex, |
| 54 | UErrorCode &status) const { |
| 55 | // Insert the suffix first since inserting the prefix will change the rightIndex |
| 56 | int length = output.insert(rightIndex, fSuffix, fField, status); |
| 57 | length += output.insert(leftIndex, fPrefix, fField, status); |
| 58 | return length; |
| 59 | } |
| 60 | |
| 61 | int32_t ConstantAffixModifier::getPrefixLength(UErrorCode &status) const { |
| 62 | (void)status; |
| 63 | return fPrefix.length(); |
| 64 | } |
| 65 | |
| 66 | int32_t ConstantAffixModifier::getCodePointCount(UErrorCode &status) const { |
| 67 | (void)status; |
| 68 | return fPrefix.countChar32() + fSuffix.countChar32(); |
| 69 | } |
| 70 | |
| 71 | bool ConstantAffixModifier::isStrong() const { |
| 72 | return fStrong; |
| 73 | } |
| 74 | |
| 75 | SimpleModifier::SimpleModifier(const SimpleFormatter &simpleFormatter, Field field, bool strong) |
| 76 | : fCompiledPattern(simpleFormatter.compiledPattern), fField(field), fStrong(strong) { |
| 77 | U_ASSERT(1 == |
| 78 | SimpleFormatter::getArgumentLimit(fCompiledPattern.getBuffer(), fCompiledPattern.length())); |
| 79 | if (fCompiledPattern.charAt(1) != 0) { |
| 80 | fPrefixLength = fCompiledPattern.charAt(1) - ARG_NUM_LIMIT; |
| 81 | fSuffixOffset = 3 + fPrefixLength; |
| 82 | } else { |
| 83 | fPrefixLength = 0; |
| 84 | fSuffixOffset = 2; |
| 85 | } |
| 86 | if (3 + fPrefixLength < fCompiledPattern.length()) { |
| 87 | fSuffixLength = fCompiledPattern.charAt(fSuffixOffset) - ARG_NUM_LIMIT; |
| 88 | } else { |
| 89 | fSuffixLength = 0; |
| 90 | } |
| 91 | } |
| 92 | |
| 93 | SimpleModifier::SimpleModifier() |
| 94 | : fField(UNUM_FIELD_COUNT), fStrong(false), fPrefixLength(0), fSuffixLength(0) { |
| 95 | } |
| 96 | |
| 97 | int32_t SimpleModifier::apply(NumberStringBuilder &output, int leftIndex, int rightIndex, |
| 98 | UErrorCode &status) const { |
| 99 | return formatAsPrefixSuffix(output, leftIndex, rightIndex, fField, status); |
| 100 | } |
| 101 | |
| 102 | int32_t SimpleModifier::getPrefixLength(UErrorCode &status) const { |
| 103 | (void)status; |
| 104 | return fPrefixLength; |
| 105 | } |
| 106 | |
| 107 | int32_t SimpleModifier::getCodePointCount(UErrorCode &status) const { |
| 108 | (void)status; |
| 109 | int32_t count = 0; |
| 110 | if (fPrefixLength > 0) { |
| 111 | count += fCompiledPattern.countChar32(2, fPrefixLength); |
| 112 | } |
| 113 | if (fSuffixLength > 0) { |
| 114 | count += fCompiledPattern.countChar32(1 + fSuffixOffset, fSuffixLength); |
| 115 | } |
| 116 | return count; |
| 117 | } |
| 118 | |
| 119 | bool SimpleModifier::isStrong() const { |
| 120 | return fStrong; |
| 121 | } |
| 122 | |
| 123 | int32_t |
| 124 | SimpleModifier::formatAsPrefixSuffix(NumberStringBuilder &result, int32_t startIndex, int32_t endIndex, |
| 125 | Field field, UErrorCode &status) const { |
| 126 | if (fPrefixLength > 0) { |
| 127 | result.insert(startIndex, fCompiledPattern, 2, 2 + fPrefixLength, field, status); |
| 128 | } |
| 129 | if (fSuffixLength > 0) { |
| 130 | result.insert( |
| 131 | endIndex + fPrefixLength, |
| 132 | fCompiledPattern, |
| 133 | 1 + fSuffixOffset, |
| 134 | 1 + fSuffixOffset + fSuffixLength, |
| 135 | field, |
| 136 | status); |
| 137 | } |
| 138 | return fPrefixLength + fSuffixLength; |
| 139 | } |
| 140 | |
| 141 | int32_t ConstantMultiFieldModifier::apply(NumberStringBuilder &output, int leftIndex, int rightIndex, |
| 142 | UErrorCode &status) const { |
| 143 | // Insert the suffix first since inserting the prefix will change the rightIndex |
| 144 | int32_t length = output.insert(rightIndex, fSuffix, status); |
| 145 | length += output.insert(leftIndex, fPrefix, status); |
| 146 | return length; |
| 147 | } |
| 148 | |
| 149 | int32_t ConstantMultiFieldModifier::getPrefixLength(UErrorCode &status) const { |
| 150 | (void)status; |
| 151 | return fPrefix.length(); |
| 152 | } |
| 153 | |
| 154 | int32_t ConstantMultiFieldModifier::getCodePointCount(UErrorCode &status) const { |
| 155 | (void)status; |
| 156 | return fPrefix.codePointCount() + fSuffix.codePointCount(); |
| 157 | } |
| 158 | |
| 159 | bool ConstantMultiFieldModifier::isStrong() const { |
| 160 | return fStrong; |
| 161 | } |
| 162 | |
| 163 | CurrencySpacingEnabledModifier::CurrencySpacingEnabledModifier(const NumberStringBuilder &prefix, |
| 164 | const NumberStringBuilder &suffix, |
| 165 | bool strong, |
| 166 | const DecimalFormatSymbols &symbols, |
| 167 | UErrorCode &status) |
| 168 | : ConstantMultiFieldModifier(prefix, suffix, strong) { |
| 169 | // Check for currency spacing. Do not build the UnicodeSets unless there is |
| 170 | // a currency code point at a boundary. |
| 171 | if (prefix.length() > 0 && prefix.fieldAt(prefix.length() - 1) == UNUM_CURRENCY_FIELD) { |
| 172 | int prefixCp = prefix.getLastCodePoint(); |
| 173 | UnicodeSet prefixUnicodeSet = getUnicodeSet(symbols, IN_CURRENCY, PREFIX, status); |
| 174 | if (prefixUnicodeSet.contains(prefixCp)) { |
| 175 | fAfterPrefixUnicodeSet = getUnicodeSet(symbols, IN_NUMBER, PREFIX, status); |
| 176 | fAfterPrefixUnicodeSet.freeze(); |
| 177 | fAfterPrefixInsert = getInsertString(symbols, PREFIX, status); |
| 178 | } else { |
| 179 | fAfterPrefixUnicodeSet.setToBogus(); |
| 180 | fAfterPrefixInsert.setToBogus(); |
| 181 | } |
| 182 | } else { |
| 183 | fAfterPrefixUnicodeSet.setToBogus(); |
| 184 | fAfterPrefixInsert.setToBogus(); |
| 185 | } |
| 186 | if (suffix.length() > 0 && suffix.fieldAt(0) == UNUM_CURRENCY_FIELD) { |
| 187 | int suffixCp = suffix.getLastCodePoint(); |
| 188 | UnicodeSet suffixUnicodeSet = getUnicodeSet(symbols, IN_CURRENCY, SUFFIX, status); |
| 189 | if (suffixUnicodeSet.contains(suffixCp)) { |
| 190 | fBeforeSuffixUnicodeSet = getUnicodeSet(symbols, IN_NUMBER, SUFFIX, status); |
| 191 | fBeforeSuffixUnicodeSet.freeze(); |
| 192 | fBeforeSuffixInsert = getInsertString(symbols, SUFFIX, status); |
| 193 | } else { |
| 194 | fBeforeSuffixUnicodeSet.setToBogus(); |
| 195 | fBeforeSuffixInsert.setToBogus(); |
| 196 | } |
| 197 | } else { |
| 198 | fBeforeSuffixUnicodeSet.setToBogus(); |
| 199 | fBeforeSuffixInsert.setToBogus(); |
| 200 | } |
| 201 | } |
| 202 | |
| 203 | int32_t CurrencySpacingEnabledModifier::apply(NumberStringBuilder &output, int leftIndex, int rightIndex, |
| 204 | UErrorCode &status) const { |
| 205 | // Currency spacing logic |
| 206 | int length = 0; |
| 207 | if (rightIndex - leftIndex > 0 && !fAfterPrefixUnicodeSet.isBogus() && |
| 208 | fAfterPrefixUnicodeSet.contains(output.codePointAt(leftIndex))) { |
| 209 | // TODO: Should we use the CURRENCY field here? |
| 210 | length += output.insert(leftIndex, fAfterPrefixInsert, UNUM_FIELD_COUNT, status); |
| 211 | } |
| 212 | if (rightIndex - leftIndex > 0 && !fBeforeSuffixUnicodeSet.isBogus() && |
| 213 | fBeforeSuffixUnicodeSet.contains(output.codePointBefore(rightIndex))) { |
| 214 | // TODO: Should we use the CURRENCY field here? |
| 215 | length += output.insert(rightIndex + length, fBeforeSuffixInsert, UNUM_FIELD_COUNT, status); |
| 216 | } |
| 217 | |
| 218 | // Call super for the remaining logic |
| 219 | length += ConstantMultiFieldModifier::apply(output, leftIndex, rightIndex + length, status); |
| 220 | return length; |
| 221 | } |
| 222 | |
| 223 | int32_t |
| 224 | CurrencySpacingEnabledModifier::applyCurrencySpacing(NumberStringBuilder &output, int32_t prefixStart, |
| 225 | int32_t prefixLen, int32_t suffixStart, |
| 226 | int32_t suffixLen, |
| 227 | const DecimalFormatSymbols &symbols, |
| 228 | UErrorCode &status) { |
| 229 | int length = 0; |
| 230 | bool hasPrefix = (prefixLen > 0); |
| 231 | bool hasSuffix = (suffixLen > 0); |
| 232 | bool hasNumber = (suffixStart - prefixStart - prefixLen > 0); // could be empty string |
| 233 | if (hasPrefix && hasNumber) { |
| 234 | length += applyCurrencySpacingAffix(output, prefixStart + prefixLen, PREFIX, symbols, status); |
| 235 | } |
| 236 | if (hasSuffix && hasNumber) { |
| 237 | length += applyCurrencySpacingAffix(output, suffixStart + length, SUFFIX, symbols, status); |
| 238 | } |
| 239 | return length; |
| 240 | } |
| 241 | |
| 242 | int32_t |
| 243 | CurrencySpacingEnabledModifier::applyCurrencySpacingAffix(NumberStringBuilder &output, int32_t index, |
| 244 | EAffix affix, |
| 245 | const DecimalFormatSymbols &symbols, |
| 246 | UErrorCode &status) { |
| 247 | // NOTE: For prefix, output.fieldAt(index-1) gets the last field type in the prefix. |
| 248 | // This works even if the last code point in the prefix is 2 code units because the |
| 249 | // field value gets populated to both indices in the field array. |
| 250 | Field affixField = (affix == PREFIX) ? output.fieldAt(index - 1) : output.fieldAt(index); |
| 251 | if (affixField != UNUM_CURRENCY_FIELD) { |
| 252 | return 0; |
| 253 | } |
| 254 | int affixCp = (affix == PREFIX) ? output.codePointBefore(index) : output.codePointAt(index); |
| 255 | UnicodeSet affixUniset = getUnicodeSet(symbols, IN_CURRENCY, affix, status); |
| 256 | if (!affixUniset.contains(affixCp)) { |
| 257 | return 0; |
| 258 | } |
| 259 | int numberCp = (affix == PREFIX) ? output.codePointAt(index) : output.codePointBefore(index); |
| 260 | UnicodeSet numberUniset = getUnicodeSet(symbols, IN_NUMBER, affix, status); |
| 261 | if (!numberUniset.contains(numberCp)) { |
| 262 | return 0; |
| 263 | } |
| 264 | UnicodeString spacingString = getInsertString(symbols, affix, status); |
| 265 | |
| 266 | // NOTE: This next line *inserts* the spacing string, triggering an arraycopy. |
| 267 | // It would be more efficient if this could be done before affixes were attached, |
| 268 | // so that it could be prepended/appended instead of inserted. |
| 269 | // However, the build code path is more efficient, and this is the most natural |
| 270 | // place to put currency spacing in the non-build code path. |
| 271 | // TODO: Should we use the CURRENCY field here? |
| 272 | return output.insert(index, spacingString, UNUM_FIELD_COUNT, status); |
| 273 | } |
| 274 | |
| 275 | UnicodeSet |
| 276 | CurrencySpacingEnabledModifier::getUnicodeSet(const DecimalFormatSymbols &symbols, EPosition position, |
| 277 | EAffix affix, UErrorCode &status) { |
| 278 | // Ensure the static defaults are initialized: |
| 279 | umtx_initOnce(gDefaultCurrencySpacingInitOnce, &initDefaultCurrencySpacing, status); |
| 280 | if (U_FAILURE(status)) { |
| 281 | return UnicodeSet(); |
| 282 | } |
| 283 | |
| 284 | const UnicodeString& pattern = symbols.getPatternForCurrencySpacing( |
| 285 | position == IN_CURRENCY ? UNUM_CURRENCY_MATCH : UNUM_CURRENCY_SURROUNDING_MATCH, |
| 286 | affix == SUFFIX, |
| 287 | status); |
| 288 | if (pattern.compare(u"[:digit:]", -1) == 0) { |
| 289 | return *UNISET_DIGIT; |
| 290 | } else if (pattern.compare(u"[:^S:]", -1) == 0) { |
| 291 | return *UNISET_NOTS; |
| 292 | } else { |
| 293 | return UnicodeSet(pattern, status); |
| 294 | } |
| 295 | } |
| 296 | |
| 297 | UnicodeString |
| 298 | CurrencySpacingEnabledModifier::getInsertString(const DecimalFormatSymbols &symbols, EAffix affix, |
| 299 | UErrorCode &status) { |
| 300 | return symbols.getPatternForCurrencySpacing(UNUM_CURRENCY_INSERT, affix == SUFFIX, status); |
| 301 | } |
| 302 | |
| 303 | #endif /* #if !UCONFIG_NO_FORMATTING */ |