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 "unicode/simpleformatter.h"
9 #include "unicode/ures.h"
10 #include "ureslocs.h"
11 #include "charstr.h"
12 #include "uresimp.h"
13 #include "number_longnames.h"
14 #include "number_microprops.h"
15 #include <algorithm>
16 #include "cstring.h"
17 #include "util.h"
18 
19 using namespace icu;
20 using namespace icu::number;
21 using namespace icu::number::impl;
22 
23 namespace {
24 
25 constexpr int32_t DNAM_INDEX = StandardPlural::Form::COUNT;
26 constexpr int32_t PER_INDEX = StandardPlural::Form::COUNT + 1;
27 constexpr int32_t ARRAY_LENGTH = StandardPlural::Form::COUNT + 2;
28 
getIndex(const char * pluralKeyword,UErrorCode & status)29 static int32_t getIndex(const char* pluralKeyword, UErrorCode& status) {
30     // pluralKeyword can also be "dnam" or "per"
31     if (uprv_strcmp(pluralKeyword, "dnam") == 0) {
32         return DNAM_INDEX;
33     } else if (uprv_strcmp(pluralKeyword, "per") == 0) {
34         return PER_INDEX;
35     } else {
36         StandardPlural::Form plural = StandardPlural::fromString(pluralKeyword, status);
37         return plural;
38     }
39 }
40 
getWithPlural(const UnicodeString * strings,StandardPlural::Form plural,UErrorCode & status)41 static UnicodeString getWithPlural(
42         const UnicodeString* strings,
43         StandardPlural::Form plural,
44         UErrorCode& status) {
45     UnicodeString result = strings[plural];
46     if (result.isBogus()) {
47         result = strings[StandardPlural::Form::OTHER];
48     }
49     if (result.isBogus()) {
50         // There should always be data in the "other" plural variant.
51         status = U_INTERNAL_PROGRAM_ERROR;
52     }
53     return result;
54 }
55 
56 
57 //////////////////////////
58 /// BEGIN DATA LOADING ///
59 //////////////////////////
60 
61 class PluralTableSink : public ResourceSink {
62   public:
PluralTableSink(UnicodeString * outArray)63     explicit PluralTableSink(UnicodeString *outArray) : outArray(outArray) {
64         // Initialize the array to bogus strings.
65         for (int32_t i = 0; i < ARRAY_LENGTH; i++) {
66             outArray[i].setToBogus();
67         }
68     }
69 
put(const char * key,ResourceValue & value,UBool,UErrorCode & status)70     void put(const char *key, ResourceValue &value, UBool /*noFallback*/, UErrorCode &status) U_OVERRIDE {
71         ResourceTable pluralsTable = value.getTable(status);
72         if (U_FAILURE(status)) { return; }
73         for (int32_t i = 0; pluralsTable.getKeyAndValue(i, key, value); ++i) {
74             int32_t index = getIndex(key, status);
75             if (U_FAILURE(status)) { return; }
76             if (!outArray[index].isBogus()) {
77                 continue;
78             }
79             outArray[index] = value.getUnicodeString(status);
80             if (U_FAILURE(status)) { return; }
81         }
82     }
83 
84   private:
85     UnicodeString *outArray;
86 };
87 
88 // NOTE: outArray MUST have room for all StandardPlural values.  No bounds checking is performed.
89 
getMeasureData(const Locale & locale,const MeasureUnit & unit,const UNumberUnitWidth & width,UnicodeString * outArray,UErrorCode & status)90 void getMeasureData(const Locale &locale, const MeasureUnit &unit, const UNumberUnitWidth &width,
91                     UnicodeString *outArray, UErrorCode &status) {
92     PluralTableSink sink(outArray);
93     LocalUResourceBundlePointer unitsBundle(ures_open(U_ICUDATA_UNIT, locale.getName(), &status));
94     if (U_FAILURE(status)) { return; }
95 
96     // Map duration-year-person, duration-week-person, etc. to duration-year, duration-week, ...
97     // TODO(ICU-20400): Get duration-*-person data properly with aliases.
98     StringPiece subtypeForResource;
99     int32_t subtypeLen = static_cast<int32_t>(uprv_strlen(unit.getSubtype()));
100     if (subtypeLen > 7 && uprv_strcmp(unit.getSubtype() + subtypeLen - 7, "-person") == 0) {
101         subtypeForResource = {unit.getSubtype(), subtypeLen - 7};
102     } else {
103         subtypeForResource = unit.getSubtype();
104     }
105 
106     CharString key;
107     key.append("units", status);
108     if (width == UNUM_UNIT_WIDTH_NARROW) {
109         key.append("Narrow", status);
110     } else if (width == UNUM_UNIT_WIDTH_SHORT) {
111         key.append("Short", status);
112     }
113     key.append("/", status);
114     key.append(unit.getType(), status);
115     key.append("/", status);
116     key.append(subtypeForResource, status);
117 
118     UErrorCode localStatus = U_ZERO_ERROR;
119     ures_getAllItemsWithFallback(unitsBundle.getAlias(), key.data(), sink, localStatus);
120     if (width == UNUM_UNIT_WIDTH_SHORT) {
121         if (U_FAILURE(localStatus)) {
122             status = localStatus;
123         }
124         return;
125     }
126 
127     // TODO(ICU-13353): The fallback to short does not work in ICU4C.
128     // Manually fall back to short (this is done automatically in Java).
129     key.clear();
130     key.append("unitsShort/", status);
131     key.append(unit.getType(), status);
132     key.append("/", status);
133     key.append(subtypeForResource, status);
134     ures_getAllItemsWithFallback(unitsBundle.getAlias(), key.data(), sink, status);
135 }
136 
getCurrencyLongNameData(const Locale & locale,const CurrencyUnit & currency,UnicodeString * outArray,UErrorCode & status)137 void getCurrencyLongNameData(const Locale &locale, const CurrencyUnit &currency, UnicodeString *outArray,
138                              UErrorCode &status) {
139     // In ICU4J, this method gets a CurrencyData from CurrencyData.provider.
140     // TODO(ICU4J): Implement this without going through CurrencyData, like in ICU4C?
141     PluralTableSink sink(outArray);
142     LocalUResourceBundlePointer unitsBundle(ures_open(U_ICUDATA_CURR, locale.getName(), &status));
143     if (U_FAILURE(status)) { return; }
144     ures_getAllItemsWithFallback(unitsBundle.getAlias(), "CurrencyUnitPatterns", sink, status);
145     if (U_FAILURE(status)) { return; }
146     for (int32_t i = 0; i < StandardPlural::Form::COUNT; i++) {
147         UnicodeString &pattern = outArray[i];
148         if (pattern.isBogus()) {
149             continue;
150         }
151         int32_t longNameLen = 0;
152         const char16_t *longName = ucurr_getPluralName(
153                 currency.getISOCurrency(),
154                 locale.getName(),
155                 nullptr /* isChoiceFormat */,
156                 StandardPlural::getKeyword(static_cast<StandardPlural::Form>(i)),
157                 &longNameLen,
158                 &status);
159         // Example pattern from data: "{0} {1}"
160         // Example output after find-and-replace: "{0} US dollars"
161         pattern.findAndReplace(UnicodeString(u"{1}"), UnicodeString(longName, longNameLen));
162     }
163 }
164 
getPerUnitFormat(const Locale & locale,const UNumberUnitWidth & width,UErrorCode & status)165 UnicodeString getPerUnitFormat(const Locale& locale, const UNumberUnitWidth &width, UErrorCode& status) {
166     LocalUResourceBundlePointer unitsBundle(ures_open(U_ICUDATA_UNIT, locale.getName(), &status));
167     if (U_FAILURE(status)) { return {}; }
168     CharString key;
169     key.append("units", status);
170     if (width == UNUM_UNIT_WIDTH_NARROW) {
171         key.append("Narrow", status);
172     } else if (width == UNUM_UNIT_WIDTH_SHORT) {
173         key.append("Short", status);
174     }
175     key.append("/compound/per", status);
176     int32_t len = 0;
177     const UChar* ptr = ures_getStringByKeyWithFallback(unitsBundle.getAlias(), key.data(), &len, &status);
178     return UnicodeString(ptr, len);
179 }
180 
181 ////////////////////////
182 /// END DATA LOADING ///
183 ////////////////////////
184 
185 } // namespace
186 
187 LongNameHandler*
forMeasureUnit(const Locale & loc,const MeasureUnit & unitRef,const MeasureUnit & perUnit,const UNumberUnitWidth & width,const PluralRules * rules,const MicroPropsGenerator * parent,UErrorCode & status)188 LongNameHandler::forMeasureUnit(const Locale &loc, const MeasureUnit &unitRef, const MeasureUnit &perUnit,
189                                 const UNumberUnitWidth &width, const PluralRules *rules,
190                                 const MicroPropsGenerator *parent, UErrorCode &status) {
191     MeasureUnit unit = unitRef;
192     if (uprv_strcmp(perUnit.getType(), "none") != 0) {
193         // Compound unit: first try to simplify (e.g., meters per second is its own unit).
194         bool isResolved = false;
195         MeasureUnit resolved = MeasureUnit::resolveUnitPerUnit(unit, perUnit, &isResolved);
196         if (isResolved) {
197             unit = resolved;
198         } else {
199             // No simplified form is available.
200             return forCompoundUnit(loc, unit, perUnit, width, rules, parent, status);
201         }
202     }
203 
204     auto* result = new LongNameHandler(rules, parent);
205     if (result == nullptr) {
206         status = U_MEMORY_ALLOCATION_ERROR;
207         return nullptr;
208     }
209     UnicodeString simpleFormats[ARRAY_LENGTH];
210     getMeasureData(loc, unit, width, simpleFormats, status);
211     if (U_FAILURE(status)) { return result; }
212     result->simpleFormatsToModifiers(simpleFormats, UNUM_MEASURE_UNIT_FIELD, status);
213     return result;
214 }
215 
216 LongNameHandler*
forCompoundUnit(const Locale & loc,const MeasureUnit & unit,const MeasureUnit & perUnit,const UNumberUnitWidth & width,const PluralRules * rules,const MicroPropsGenerator * parent,UErrorCode & status)217 LongNameHandler::forCompoundUnit(const Locale &loc, const MeasureUnit &unit, const MeasureUnit &perUnit,
218                                  const UNumberUnitWidth &width, const PluralRules *rules,
219                                  const MicroPropsGenerator *parent, UErrorCode &status) {
220     auto* result = new LongNameHandler(rules, parent);
221     if (result == nullptr) {
222         status = U_MEMORY_ALLOCATION_ERROR;
223         return nullptr;
224     }
225     UnicodeString primaryData[ARRAY_LENGTH];
226     getMeasureData(loc, unit, width, primaryData, status);
227     if (U_FAILURE(status)) { return result; }
228     UnicodeString secondaryData[ARRAY_LENGTH];
229     getMeasureData(loc, perUnit, width, secondaryData, status);
230     if (U_FAILURE(status)) { return result; }
231 
232     UnicodeString perUnitFormat;
233     if (!secondaryData[PER_INDEX].isBogus()) {
234         perUnitFormat = secondaryData[PER_INDEX];
235     } else {
236         UnicodeString rawPerUnitFormat = getPerUnitFormat(loc, width, status);
237         if (U_FAILURE(status)) { return result; }
238         // rawPerUnitFormat is something like "{0}/{1}"; we need to substitute in the secondary unit.
239         SimpleFormatter compiled(rawPerUnitFormat, 2, 2, status);
240         if (U_FAILURE(status)) { return result; }
241         UnicodeString secondaryFormat = getWithPlural(secondaryData, StandardPlural::Form::ONE, status);
242         if (U_FAILURE(status)) { return result; }
243         SimpleFormatter secondaryCompiled(secondaryFormat, 1, 1, status);
244         if (U_FAILURE(status)) { return result; }
245         UnicodeString secondaryString = secondaryCompiled.getTextWithNoArguments().trim();
246         // TODO: Why does UnicodeString need to be explicit in the following line?
247         compiled.format(UnicodeString(u"{0}"), secondaryString, perUnitFormat, status);
248         if (U_FAILURE(status)) { return result; }
249     }
250     result->multiSimpleFormatsToModifiers(primaryData, perUnitFormat, UNUM_MEASURE_UNIT_FIELD, status);
251     return result;
252 }
253 
getUnitDisplayName(const Locale & loc,const MeasureUnit & unit,UNumberUnitWidth width,UErrorCode & status)254 UnicodeString LongNameHandler::getUnitDisplayName(
255         const Locale& loc,
256         const MeasureUnit& unit,
257         UNumberUnitWidth width,
258         UErrorCode& status) {
259     if (U_FAILURE(status)) {
260         return ICU_Utility::makeBogusString();
261     }
262     UnicodeString simpleFormats[ARRAY_LENGTH];
263     getMeasureData(loc, unit, width, simpleFormats, status);
264     return simpleFormats[DNAM_INDEX];
265 }
266 
getUnitPattern(const Locale & loc,const MeasureUnit & unit,UNumberUnitWidth width,StandardPlural::Form pluralForm,UErrorCode & status)267 UnicodeString LongNameHandler::getUnitPattern(
268         const Locale& loc,
269         const MeasureUnit& unit,
270         UNumberUnitWidth width,
271         StandardPlural::Form pluralForm,
272         UErrorCode& status) {
273     if (U_FAILURE(status)) {
274         return ICU_Utility::makeBogusString();
275     }
276     UnicodeString simpleFormats[ARRAY_LENGTH];
277     getMeasureData(loc, unit, width, simpleFormats, status);
278     // The above already handles fallback from other widths to short
279     if (U_FAILURE(status)) {
280         return ICU_Utility::makeBogusString();
281     }
282     // Now handle fallback from other plural forms to OTHER
283     return (!(simpleFormats[pluralForm]).isBogus())? simpleFormats[pluralForm]:
284             simpleFormats[StandardPlural::Form::OTHER];
285 }
286 
forCurrencyLongNames(const Locale & loc,const CurrencyUnit & currency,const PluralRules * rules,const MicroPropsGenerator * parent,UErrorCode & status)287 LongNameHandler* LongNameHandler::forCurrencyLongNames(const Locale &loc, const CurrencyUnit &currency,
288                                                       const PluralRules *rules,
289                                                       const MicroPropsGenerator *parent,
290                                                       UErrorCode &status) {
291     auto* result = new LongNameHandler(rules, parent);
292     if (result == nullptr) {
293         status = U_MEMORY_ALLOCATION_ERROR;
294         return nullptr;
295     }
296     UnicodeString simpleFormats[ARRAY_LENGTH];
297     getCurrencyLongNameData(loc, currency, simpleFormats, status);
298     if (U_FAILURE(status)) { return nullptr; }
299     result->simpleFormatsToModifiers(simpleFormats, UNUM_CURRENCY_FIELD, status);
300     return result;
301 }
302 
simpleFormatsToModifiers(const UnicodeString * simpleFormats,Field field,UErrorCode & status)303 void LongNameHandler::simpleFormatsToModifiers(const UnicodeString *simpleFormats, Field field,
304                                                UErrorCode &status) {
305     for (int32_t i = 0; i < StandardPlural::Form::COUNT; i++) {
306         StandardPlural::Form plural = static_cast<StandardPlural::Form>(i);
307         UnicodeString simpleFormat = getWithPlural(simpleFormats, plural, status);
308         if (U_FAILURE(status)) { return; }
309         SimpleFormatter compiledFormatter(simpleFormat, 0, 1, status);
310         if (U_FAILURE(status)) { return; }
311         fModifiers[i] = SimpleModifier(compiledFormatter, field, false, {this, SIGNUM_ZERO, plural});
312     }
313 }
314 
multiSimpleFormatsToModifiers(const UnicodeString * leadFormats,UnicodeString trailFormat,Field field,UErrorCode & status)315 void LongNameHandler::multiSimpleFormatsToModifiers(const UnicodeString *leadFormats, UnicodeString trailFormat,
316                                                     Field field, UErrorCode &status) {
317     SimpleFormatter trailCompiled(trailFormat, 1, 1, status);
318     if (U_FAILURE(status)) { return; }
319     for (int32_t i = 0; i < StandardPlural::Form::COUNT; i++) {
320         StandardPlural::Form plural = static_cast<StandardPlural::Form>(i);
321         UnicodeString leadFormat = getWithPlural(leadFormats, plural, status);
322         if (U_FAILURE(status)) { return; }
323         UnicodeString compoundFormat;
324         trailCompiled.format(leadFormat, compoundFormat, status);
325         if (U_FAILURE(status)) { return; }
326         SimpleFormatter compoundCompiled(compoundFormat, 0, 1, status);
327         if (U_FAILURE(status)) { return; }
328         fModifiers[i] = SimpleModifier(compoundCompiled, field, false, {this, SIGNUM_ZERO, plural});
329     }
330 }
331 
processQuantity(DecimalQuantity & quantity,MicroProps & micros,UErrorCode & status) const332 void LongNameHandler::processQuantity(DecimalQuantity &quantity, MicroProps &micros,
333                                       UErrorCode &status) const {
334     parent->processQuantity(quantity, micros, status);
335     StandardPlural::Form pluralForm = utils::getPluralSafe(micros.rounder, rules, quantity, status);
336     micros.modOuter = &fModifiers[pluralForm];
337 }
338 
getModifier(Signum,StandardPlural::Form plural) const339 const Modifier* LongNameHandler::getModifier(Signum /*signum*/, StandardPlural::Form plural) const {
340     return &fModifiers[plural];
341 }
342 
343 #endif /* #if !UCONFIG_NO_FORMATTING */
344