tor-browser

The Tor Browser
git clone https://git.dasho.dev/tor-browser.git
Log | Files | Refs | README | LICENSE

locdspnm.cpp (41456B)


      1 // © 2016 and later: Unicode, Inc. and others.
      2 // License & terms of use: http://www.unicode.org/copyright.html
      3 /*
      4 *******************************************************************************
      5 * Copyright (C) 2010-2016, International Business Machines Corporation and
      6 * others. All Rights Reserved.
      7 *******************************************************************************
      8 */
      9 
     10 #include "unicode/utypes.h"
     11 
     12 #if !UCONFIG_NO_FORMATTING
     13 
     14 #include "unicode/locdspnm.h"
     15 #include "unicode/simpleformatter.h"
     16 #include "unicode/ucasemap.h"
     17 #include "unicode/ures.h"
     18 #include "unicode/udisplaycontext.h"
     19 #include "unicode/brkiter.h"
     20 #include "unicode/ucurr.h"
     21 #include "bytesinkutil.h"
     22 #include "charstr.h"
     23 #include "cmemory.h"
     24 #include "cstring.h"
     25 #include "mutex.h"
     26 #include "uassert.h"
     27 #include "ulocimp.h"
     28 #include "umutex.h"
     29 #include "ureslocs.h"
     30 #include "uresimp.h"
     31 
     32 U_NAMESPACE_BEGIN
     33 
     34 ////////////////////////////////////////////////////////////////////////////////////////////////////
     35 
     36 // Access resource data for locale components.
     37 // Wrap code in uloc.c for now.
     38 class ICUDataTable {
     39    const char* const path;
     40    Locale locale;
     41 
     42 public:
     43    // Note: path should be a pointer to a statically allocated string.
     44    ICUDataTable(const char* path, const Locale& locale);
     45    ~ICUDataTable() = default;
     46 
     47    const Locale& getLocale();
     48 
     49    UnicodeString& get(const char* tableKey, const char* itemKey,
     50                        UnicodeString& result) const;
     51    UnicodeString& get(const char* tableKey, const char* subTableKey, const char* itemKey,
     52                        UnicodeString& result) const;
     53 
     54    UnicodeString& getNoFallback(const char* tableKey, const char* itemKey,
     55                                UnicodeString &result) const;
     56    UnicodeString& getNoFallback(const char* tableKey, const char* subTableKey, const char* itemKey,
     57                                UnicodeString &result) const;
     58 };
     59 
     60 inline UnicodeString &
     61 ICUDataTable::get(const char* tableKey, const char* itemKey, UnicodeString& result) const {
     62    return get(tableKey, nullptr, itemKey, result);
     63 }
     64 
     65 inline UnicodeString &
     66 ICUDataTable::getNoFallback(const char* tableKey, const char* itemKey, UnicodeString& result) const {
     67    return getNoFallback(tableKey, nullptr, itemKey, result);
     68 }
     69 
     70 ICUDataTable::ICUDataTable(const char* path, const Locale& locale)
     71    : path(path), locale(locale)
     72 {
     73    U_ASSERT(path != nullptr);
     74 }
     75 
     76 const Locale&
     77 ICUDataTable::getLocale() {
     78  return locale;
     79 }
     80 
     81 UnicodeString &
     82 ICUDataTable::get(const char* tableKey, const char* subTableKey, const char* itemKey,
     83                  UnicodeString &result) const {
     84  UErrorCode status = U_ZERO_ERROR;
     85  int32_t len = 0;
     86 
     87  const char16_t *s = uloc_getTableStringWithFallback(path, locale.getName(),
     88                                                   tableKey, subTableKey, itemKey,
     89                                                   &len, &status);
     90  if (U_SUCCESS(status) && len > 0) {
     91    return result.setTo(s, len);
     92  }
     93  return result.setTo(UnicodeString(itemKey, -1, US_INV));
     94 }
     95 
     96 UnicodeString &
     97 ICUDataTable::getNoFallback(const char* tableKey, const char* subTableKey, const char* itemKey,
     98                            UnicodeString& result) const {
     99  UErrorCode status = U_ZERO_ERROR;
    100  int32_t len = 0;
    101 
    102  const char16_t *s = uloc_getTableStringWithFallback(path, locale.getName(),
    103                                                   tableKey, subTableKey, itemKey,
    104                                                   &len, &status);
    105  if (U_SUCCESS(status)) {
    106    return result.setTo(s, len);
    107  }
    108 
    109  result.setToBogus();
    110  return result;
    111 }
    112 
    113 ////////////////////////////////////////////////////////////////////////////////////////////////////
    114 
    115 LocaleDisplayNames::~LocaleDisplayNames() {}
    116 
    117 ////////////////////////////////////////////////////////////////////////////////////////////////////
    118 
    119 #if 0  // currently unused
    120 
    121 class DefaultLocaleDisplayNames : public LocaleDisplayNames {
    122  UDialectHandling dialectHandling;
    123 
    124 public:
    125  // constructor
    126  DefaultLocaleDisplayNames(UDialectHandling dialectHandling);
    127 
    128  virtual ~DefaultLocaleDisplayNames();
    129 
    130  virtual const Locale& getLocale() const;
    131  virtual UDialectHandling getDialectHandling() const;
    132 
    133  virtual UnicodeString& localeDisplayName(const Locale& locale,
    134                                           UnicodeString& result) const;
    135  virtual UnicodeString& localeDisplayName(const char* localeId,
    136                                           UnicodeString& result) const;
    137  virtual UnicodeString& languageDisplayName(const char* lang,
    138                                             UnicodeString& result) const;
    139  virtual UnicodeString& scriptDisplayName(const char* script,
    140                                           UnicodeString& result) const;
    141  virtual UnicodeString& scriptDisplayName(UScriptCode scriptCode,
    142                                           UnicodeString& result) const;
    143  virtual UnicodeString& regionDisplayName(const char* region,
    144                                           UnicodeString& result) const;
    145  virtual UnicodeString& variantDisplayName(const char* variant,
    146                                            UnicodeString& result) const;
    147  virtual UnicodeString& keyDisplayName(const char* key,
    148                                        UnicodeString& result) const;
    149  virtual UnicodeString& keyValueDisplayName(const char* key,
    150                                             const char* value,
    151                                             UnicodeString& result) const;
    152 };
    153 
    154 DefaultLocaleDisplayNames::DefaultLocaleDisplayNames(UDialectHandling dialectHandling)
    155    : dialectHandling(dialectHandling) {
    156 }
    157 
    158 DefaultLocaleDisplayNames::~DefaultLocaleDisplayNames() {
    159 }
    160 
    161 const Locale&
    162 DefaultLocaleDisplayNames::getLocale() const {
    163  return Locale::getRoot();
    164 }
    165 
    166 UDialectHandling
    167 DefaultLocaleDisplayNames::getDialectHandling() const {
    168  return dialectHandling;
    169 }
    170 
    171 UnicodeString&
    172 DefaultLocaleDisplayNames::localeDisplayName(const Locale& locale,
    173                                             UnicodeString& result) const {
    174  return result = UnicodeString(locale.getName(), -1, US_INV);
    175 }
    176 
    177 UnicodeString&
    178 DefaultLocaleDisplayNames::localeDisplayName(const char* localeId,
    179                                             UnicodeString& result) const {
    180  return result = UnicodeString(localeId, -1, US_INV);
    181 }
    182 
    183 UnicodeString&
    184 DefaultLocaleDisplayNames::languageDisplayName(const char* lang,
    185                                               UnicodeString& result) const {
    186  return result = UnicodeString(lang, -1, US_INV);
    187 }
    188 
    189 UnicodeString&
    190 DefaultLocaleDisplayNames::scriptDisplayName(const char* script,
    191                                             UnicodeString& result) const {
    192  return result = UnicodeString(script, -1, US_INV);
    193 }
    194 
    195 UnicodeString&
    196 DefaultLocaleDisplayNames::scriptDisplayName(UScriptCode scriptCode,
    197                                             UnicodeString& result) const {
    198  const char* name = uscript_getName(scriptCode);
    199  if (name) {
    200    return result = UnicodeString(name, -1, US_INV);
    201  }
    202  return result.remove();
    203 }
    204 
    205 UnicodeString&
    206 DefaultLocaleDisplayNames::regionDisplayName(const char* region,
    207                                             UnicodeString& result) const {
    208  return result = UnicodeString(region, -1, US_INV);
    209 }
    210 
    211 UnicodeString&
    212 DefaultLocaleDisplayNames::variantDisplayName(const char* variant,
    213                                              UnicodeString& result) const {
    214  return result = UnicodeString(variant, -1, US_INV);
    215 }
    216 
    217 UnicodeString&
    218 DefaultLocaleDisplayNames::keyDisplayName(const char* key,
    219                                          UnicodeString& result) const {
    220  return result = UnicodeString(key, -1, US_INV);
    221 }
    222 
    223 UnicodeString&
    224 DefaultLocaleDisplayNames::keyValueDisplayName(const char* /* key */,
    225                                               const char* value,
    226                                               UnicodeString& result) const {
    227  return result = UnicodeString(value, -1, US_INV);
    228 }
    229 
    230 #endif  // currently unused class DefaultLocaleDisplayNames
    231 
    232 ////////////////////////////////////////////////////////////////////////////////////////////////////
    233 
    234 class LocaleDisplayNamesImpl : public LocaleDisplayNames {
    235    Locale locale;
    236    UDialectHandling dialectHandling;
    237    ICUDataTable langData;
    238    ICUDataTable regionData;
    239    SimpleFormatter separatorFormat;
    240    SimpleFormatter format;
    241    SimpleFormatter keyTypeFormat;
    242    UDisplayContext capitalizationContext;
    243 #if !UCONFIG_NO_BREAK_ITERATION
    244    BreakIterator* capitalizationBrkIter;
    245 #else
    246    UObject* capitalizationBrkIter;
    247 #endif
    248    UnicodeString formatOpenParen;
    249    UnicodeString formatReplaceOpenParen;
    250    UnicodeString formatCloseParen;
    251    UnicodeString formatReplaceCloseParen;
    252    UDisplayContext nameLength;
    253    UDisplayContext substitute;
    254 
    255    // Constants for capitalization context usage types.
    256    enum CapContextUsage {
    257        kCapContextUsageLanguage,
    258        kCapContextUsageScript,
    259        kCapContextUsageTerritory,
    260        kCapContextUsageVariant,
    261        kCapContextUsageKey,
    262        kCapContextUsageKeyValue,
    263        kCapContextUsageCount
    264    };
    265    // Capitalization transforms. For each usage type, indicates whether to titlecase for
    266    // the context specified in capitalizationContext (which we know at construction time)
    267     bool fCapitalization[kCapContextUsageCount];
    268 
    269 public:
    270    // constructor
    271    LocaleDisplayNamesImpl(const Locale& locale, UDialectHandling dialectHandling);
    272    LocaleDisplayNamesImpl(const Locale& locale, UDisplayContext *contexts, int32_t length);
    273    virtual ~LocaleDisplayNamesImpl();
    274 
    275    virtual const Locale& getLocale() const override;
    276    virtual UDialectHandling getDialectHandling() const override;
    277    virtual UDisplayContext getContext(UDisplayContextType type) const override;
    278 
    279    virtual UnicodeString& localeDisplayName(const Locale& locale,
    280                                                UnicodeString& result) const override;
    281    virtual UnicodeString& localeDisplayName(const char* localeId,
    282                                                UnicodeString& result) const override;
    283    virtual UnicodeString& languageDisplayName(const char* lang,
    284                                               UnicodeString& result) const override;
    285    virtual UnicodeString& scriptDisplayName(const char* script,
    286                                                UnicodeString& result) const override;
    287    virtual UnicodeString& scriptDisplayName(UScriptCode scriptCode,
    288                                                UnicodeString& result) const override;
    289    virtual UnicodeString& regionDisplayName(const char* region,
    290                                                UnicodeString& result) const override;
    291    virtual UnicodeString& variantDisplayName(const char* variant,
    292                                                UnicodeString& result) const override;
    293    virtual UnicodeString& keyDisplayName(const char* key,
    294                                                UnicodeString& result) const override;
    295    virtual UnicodeString& keyValueDisplayName(const char* key,
    296                                                const char* value,
    297                                                UnicodeString& result) const override;
    298 private:
    299    UnicodeString& localeIdName(const char* localeId,
    300                                UnicodeString& result, bool substitute) const;
    301    UnicodeString& appendWithSep(UnicodeString& buffer, const UnicodeString& src) const;
    302    UnicodeString& adjustForUsageAndContext(CapContextUsage usage, UnicodeString& result) const;
    303    UnicodeString& scriptDisplayName(const char* script, UnicodeString& result, bool skipAdjust) const;
    304    UnicodeString& regionDisplayName(const char* region, UnicodeString& result, bool skipAdjust) const;
    305    UnicodeString& variantDisplayName(const char* variant, UnicodeString& result, bool skipAdjust) const;
    306    UnicodeString& keyDisplayName(const char* key, UnicodeString& result, bool skipAdjust) const;
    307    UnicodeString& keyValueDisplayName(const char* key, const char* value,
    308                                        UnicodeString& result, bool skipAdjust) const;
    309    void initialize();
    310 
    311    struct CapitalizationContextSink;
    312 };
    313 
    314 LocaleDisplayNamesImpl::LocaleDisplayNamesImpl(const Locale& locale,
    315                                               UDialectHandling dialectHandling)
    316    : dialectHandling(dialectHandling)
    317    , langData(U_ICUDATA_LANG, locale)
    318    , regionData(U_ICUDATA_REGION, locale)
    319    , capitalizationContext(UDISPCTX_CAPITALIZATION_NONE)
    320    , capitalizationBrkIter(nullptr)
    321    , nameLength(UDISPCTX_LENGTH_FULL)
    322    , substitute(UDISPCTX_SUBSTITUTE)
    323 {
    324    initialize();
    325 }
    326 
    327 LocaleDisplayNamesImpl::LocaleDisplayNamesImpl(const Locale& locale,
    328                                               UDisplayContext *contexts, int32_t length)
    329    : dialectHandling(ULDN_STANDARD_NAMES)
    330    , langData(U_ICUDATA_LANG, locale)
    331    , regionData(U_ICUDATA_REGION, locale)
    332    , capitalizationContext(UDISPCTX_CAPITALIZATION_NONE)
    333    , capitalizationBrkIter(nullptr)
    334    , nameLength(UDISPCTX_LENGTH_FULL)
    335    , substitute(UDISPCTX_SUBSTITUTE)
    336 {
    337    while (length-- > 0) {
    338        UDisplayContext value = *contexts++;
    339        UDisplayContextType selector =
    340            static_cast<UDisplayContextType>(static_cast<uint32_t>(value) >> 8);
    341        switch (selector) {
    342            case UDISPCTX_TYPE_DIALECT_HANDLING:
    343                dialectHandling = static_cast<UDialectHandling>(value);
    344                break;
    345            case UDISPCTX_TYPE_CAPITALIZATION:
    346                capitalizationContext = value;
    347                break;
    348            case UDISPCTX_TYPE_DISPLAY_LENGTH:
    349                nameLength = value;
    350                break;
    351            case UDISPCTX_TYPE_SUBSTITUTE_HANDLING:
    352                substitute = value;
    353                break;
    354            default:
    355                break;
    356        }
    357    }
    358    initialize();
    359 }
    360 
    361 struct LocaleDisplayNamesImpl::CapitalizationContextSink : public ResourceSink {
    362    bool hasCapitalizationUsage;
    363    LocaleDisplayNamesImpl& parent;
    364 
    365    CapitalizationContextSink(LocaleDisplayNamesImpl& _parent)
    366      : hasCapitalizationUsage(false), parent(_parent) {}
    367    virtual ~CapitalizationContextSink();
    368 
    369    virtual void put(const char *key, ResourceValue &value, UBool /*noFallback*/,
    370            UErrorCode &errorCode) override {
    371        ResourceTable contexts = value.getTable(errorCode);
    372        if (U_FAILURE(errorCode)) { return; }
    373        for (int i = 0; contexts.getKeyAndValue(i, key, value); ++i) {
    374 
    375            CapContextUsage usageEnum;
    376            if (uprv_strcmp(key, "key") == 0) {
    377                usageEnum = kCapContextUsageKey;
    378            } else if (uprv_strcmp(key, "keyValue") == 0) {
    379                usageEnum = kCapContextUsageKeyValue;
    380            } else if (uprv_strcmp(key, "languages") == 0) {
    381                usageEnum = kCapContextUsageLanguage;
    382            } else if (uprv_strcmp(key, "script") == 0) {
    383                usageEnum = kCapContextUsageScript;
    384            } else if (uprv_strcmp(key, "territory") == 0) {
    385                usageEnum = kCapContextUsageTerritory;
    386            } else if (uprv_strcmp(key, "variant") == 0) {
    387                usageEnum = kCapContextUsageVariant;
    388            } else {
    389                continue;
    390            }
    391 
    392            int32_t len = 0;
    393            const int32_t* intVector = value.getIntVector(len, errorCode);
    394            if (U_FAILURE(errorCode)) { return; }
    395            if (len < 2) { continue; }
    396 
    397            int32_t titlecaseInt = (parent.capitalizationContext == UDISPCTX_CAPITALIZATION_FOR_UI_LIST_OR_MENU) ? intVector[0] : intVector[1];
    398            if (titlecaseInt == 0) { continue; }
    399 
    400            parent.fCapitalization[usageEnum] = true;
    401            hasCapitalizationUsage = true;
    402        }
    403    }
    404 };
    405 
    406 // Virtual destructors must be defined out of line.
    407 LocaleDisplayNamesImpl::CapitalizationContextSink::~CapitalizationContextSink() {}
    408 
    409 void
    410 LocaleDisplayNamesImpl::initialize() {
    411    LocaleDisplayNamesImpl* nonConstThis = this;
    412    nonConstThis->locale = langData.getLocale() == Locale::getRoot()
    413        ? regionData.getLocale()
    414        : langData.getLocale();
    415 
    416    UnicodeString sep;
    417    langData.getNoFallback("localeDisplayPattern", "separator", sep);
    418    if (sep.isBogus()) {
    419        sep = UnicodeString("{0}, {1}", -1, US_INV);
    420    }
    421    UErrorCode status = U_ZERO_ERROR;
    422    separatorFormat.applyPatternMinMaxArguments(sep, 2, 2, status);
    423 
    424    UnicodeString pattern;
    425    langData.getNoFallback("localeDisplayPattern", "pattern", pattern);
    426    if (pattern.isBogus()) {
    427        pattern = UnicodeString("{0} ({1})", -1, US_INV);
    428    }
    429    format.applyPatternMinMaxArguments(pattern, 2, 2, status);
    430    if (pattern.indexOf(static_cast<char16_t>(0xFF08)) >= 0) {
    431        formatOpenParen.setTo(static_cast<char16_t>(0xFF08));         // fullwidth (
    432        formatReplaceOpenParen.setTo(static_cast<char16_t>(0xFF3B));  // fullwidth [
    433        formatCloseParen.setTo(static_cast<char16_t>(0xFF09));        // fullwidth )
    434        formatReplaceCloseParen.setTo(static_cast<char16_t>(0xFF3D)); // fullwidth ]
    435    } else {
    436        formatOpenParen.setTo(static_cast<char16_t>(0x0028));         // (
    437        formatReplaceOpenParen.setTo(static_cast<char16_t>(0x005B));  // [
    438        formatCloseParen.setTo(static_cast<char16_t>(0x0029));        // )
    439        formatReplaceCloseParen.setTo(static_cast<char16_t>(0x005D)); // ]
    440    }
    441 
    442    UnicodeString ktPattern;
    443    langData.get("localeDisplayPattern", "keyTypePattern", ktPattern);
    444    if (ktPattern.isBogus()) {
    445        ktPattern = UnicodeString("{0}={1}", -1, US_INV);
    446    }
    447    keyTypeFormat.applyPatternMinMaxArguments(ktPattern, 2, 2, status);
    448 
    449    uprv_memset(fCapitalization, 0, sizeof(fCapitalization));
    450 #if !UCONFIG_NO_BREAK_ITERATION
    451    // Only get the context data if we need it! This is a const object so we know now...
    452    // Also check whether we will need a break iterator (depends on the data)
    453    bool needBrkIter = false;
    454    if (capitalizationContext == UDISPCTX_CAPITALIZATION_FOR_UI_LIST_OR_MENU || capitalizationContext == UDISPCTX_CAPITALIZATION_FOR_STANDALONE) {
    455        LocalUResourceBundlePointer resource(ures_open(nullptr, locale.getName(), &status));
    456        if (U_FAILURE(status)) { return; }
    457        CapitalizationContextSink sink(*this);
    458        ures_getAllItemsWithFallback(resource.getAlias(), "contextTransforms", sink, status);
    459        if (status == U_MISSING_RESOURCE_ERROR) {
    460            // Silently ignore.  Not every locale has contextTransforms.
    461            status = U_ZERO_ERROR;
    462        } else if (U_FAILURE(status)) {
    463            return;
    464        }
    465        needBrkIter = sink.hasCapitalizationUsage;
    466    }
    467    // Get a sentence break iterator if we will need it
    468    if (needBrkIter || capitalizationContext == UDISPCTX_CAPITALIZATION_FOR_BEGINNING_OF_SENTENCE) {
    469        status = U_ZERO_ERROR;
    470        capitalizationBrkIter = BreakIterator::createSentenceInstance(locale, status);
    471        if (U_FAILURE(status)) {
    472            delete capitalizationBrkIter;
    473            capitalizationBrkIter = nullptr;
    474        }
    475    }
    476 #endif
    477 }
    478 
    479 LocaleDisplayNamesImpl::~LocaleDisplayNamesImpl() {
    480 #if !UCONFIG_NO_BREAK_ITERATION
    481    delete capitalizationBrkIter;
    482 #endif
    483 }
    484 
    485 const Locale&
    486 LocaleDisplayNamesImpl::getLocale() const {
    487    return locale;
    488 }
    489 
    490 UDialectHandling
    491 LocaleDisplayNamesImpl::getDialectHandling() const {
    492    return dialectHandling;
    493 }
    494 
    495 UDisplayContext
    496 LocaleDisplayNamesImpl::getContext(UDisplayContextType type) const {
    497    switch (type) {
    498        case UDISPCTX_TYPE_DIALECT_HANDLING:
    499            return static_cast<UDisplayContext>(dialectHandling);
    500        case UDISPCTX_TYPE_CAPITALIZATION:
    501            return capitalizationContext;
    502        case UDISPCTX_TYPE_DISPLAY_LENGTH:
    503            return nameLength;
    504        case UDISPCTX_TYPE_SUBSTITUTE_HANDLING:
    505            return substitute;
    506        default:
    507            break;
    508    }
    509    return static_cast<UDisplayContext>(0);
    510 }
    511 
    512 UnicodeString&
    513 LocaleDisplayNamesImpl::adjustForUsageAndContext(CapContextUsage usage,
    514                                                UnicodeString& result) const {
    515 #if !UCONFIG_NO_BREAK_ITERATION
    516    // check to see whether we need to titlecase result
    517    if ( result.length() > 0 && u_islower(result.char32At(0)) && capitalizationBrkIter!= nullptr &&
    518          ( capitalizationContext==UDISPCTX_CAPITALIZATION_FOR_BEGINNING_OF_SENTENCE || fCapitalization[usage] ) ) {
    519        // note fCapitalization[usage] won't be set unless capitalizationContext is UI_LIST_OR_MENU or STANDALONE
    520        static UMutex capitalizationBrkIterLock;
    521        Mutex lock(&capitalizationBrkIterLock);
    522        result.toTitle(capitalizationBrkIter, locale, U_TITLECASE_NO_LOWERCASE | U_TITLECASE_NO_BREAK_ADJUSTMENT);
    523    }
    524 #endif
    525    return result;
    526 }
    527 
    528 UnicodeString&
    529 LocaleDisplayNamesImpl::localeDisplayName(const Locale& loc,
    530                                          UnicodeString& result) const {
    531  if (loc.isBogus()) {
    532    result.setToBogus();
    533    return result;
    534  }
    535  UnicodeString resultName;
    536 
    537  const char* lang = loc.getLanguage();
    538  if (uprv_strlen(lang) == 0) {
    539    lang = "root";
    540  }
    541  const char* script = loc.getScript();
    542  const char* country = loc.getCountry();
    543  const char* variant = loc.getVariant();
    544 
    545  bool hasScript = uprv_strlen(script) > 0;
    546  bool hasCountry = uprv_strlen(country) > 0;
    547  bool hasVariant = uprv_strlen(variant) > 0;
    548 
    549  if (dialectHandling == ULDN_DIALECT_NAMES) {
    550    UErrorCode status = U_ZERO_ERROR;
    551    CharString buffer;
    552    do { // loop construct is so we can break early out of search
    553      if (hasScript && hasCountry) {
    554        buffer.append(lang, status)
    555              .append('_', status)
    556              .append(script, status)
    557              .append('_', status)
    558              .append(country, status);
    559        if (U_SUCCESS(status)) {
    560          localeIdName(buffer.data(), resultName, false);
    561          if (!resultName.isBogus()) {
    562            hasScript = false;
    563            hasCountry = false;
    564            break;
    565          }
    566        }
    567      }
    568      if (hasScript) {
    569        buffer.append(lang, status)
    570              .append('_', status)
    571              .append(script, status);
    572        if (U_SUCCESS(status)) {
    573          localeIdName(buffer.data(), resultName, false);
    574          if (!resultName.isBogus()) {
    575            hasScript = false;
    576            break;
    577          }
    578        }
    579      }
    580      if (hasCountry) {
    581        buffer.append(lang, status)
    582              .append('_', status)
    583              .append(country, status);
    584        if (U_SUCCESS(status)) {
    585          localeIdName(buffer.data(), resultName, false);
    586          if (!resultName.isBogus()) {
    587            hasCountry = false;
    588            break;
    589          }
    590        }
    591      }
    592    } while (false);
    593  }
    594  if (resultName.isBogus() || resultName.isEmpty()) {
    595    localeIdName(lang, resultName, substitute == UDISPCTX_SUBSTITUTE);
    596    if (resultName.isBogus()) {
    597      result.setToBogus();
    598      return result;
    599    }
    600  }
    601 
    602  UnicodeString resultRemainder;
    603  UnicodeString temp;
    604  UErrorCode status = U_ZERO_ERROR;
    605 
    606  if (hasScript) {
    607    UnicodeString script_str = scriptDisplayName(script, temp, true);
    608    if (script_str.isBogus()) {
    609      result.setToBogus();
    610      return result;
    611    }
    612    resultRemainder.append(script_str);
    613  }
    614  if (hasCountry) {
    615    UnicodeString region_str = regionDisplayName(country, temp, true);
    616    if (region_str.isBogus()) {
    617      result.setToBogus();
    618      return result;
    619    }
    620    appendWithSep(resultRemainder, region_str);
    621  }
    622  if (hasVariant) {
    623    UnicodeString variant_str = variantDisplayName(variant, temp, true);
    624    if (variant_str.isBogus()) {
    625      result.setToBogus();
    626      return result;
    627    }
    628    appendWithSep(resultRemainder, variant_str);
    629  }
    630  resultRemainder.findAndReplace(formatOpenParen, formatReplaceOpenParen);
    631  resultRemainder.findAndReplace(formatCloseParen, formatReplaceCloseParen);
    632 
    633  LocalPointer<StringEnumeration> e(loc.createKeywords(status));
    634  if (e.isValid() && U_SUCCESS(status)) {
    635    UnicodeString temp2;
    636    const char* key;
    637    while ((key = e->next((int32_t*)nullptr, status)) != nullptr) {
    638        auto value = loc.getKeywordValue<CharString>(key, status);
    639        if (U_FAILURE(status)) {
    640            return result;
    641      }
    642      keyDisplayName(key, temp, true);
    643      temp.findAndReplace(formatOpenParen, formatReplaceOpenParen);
    644      temp.findAndReplace(formatCloseParen, formatReplaceCloseParen);
    645      keyValueDisplayName(key, value.data(), temp2, true);
    646      temp2.findAndReplace(formatOpenParen, formatReplaceOpenParen);
    647      temp2.findAndReplace(formatCloseParen, formatReplaceCloseParen);
    648      if (temp2 != UnicodeString(value.data(), -1, US_INV)) {
    649        appendWithSep(resultRemainder, temp2);
    650      } else if (temp != UnicodeString(key, -1, US_INV)) {
    651        UnicodeString temp3;
    652        keyTypeFormat.format(temp, temp2, temp3, status);
    653        appendWithSep(resultRemainder, temp3);
    654      } else {
    655        appendWithSep(resultRemainder, temp)
    656          .append(static_cast<char16_t>(0x3d) /* = */)
    657          .append(temp2);
    658      }
    659    }
    660  }
    661 
    662  if (!resultRemainder.isEmpty()) {
    663    format.format(resultName, resultRemainder, result.remove(), status);
    664    return adjustForUsageAndContext(kCapContextUsageLanguage, result);
    665  }
    666 
    667  result = resultName;
    668  return adjustForUsageAndContext(kCapContextUsageLanguage, result);
    669 }
    670 
    671 UnicodeString&
    672 LocaleDisplayNamesImpl::appendWithSep(UnicodeString& buffer, const UnicodeString& src) const {
    673    if (buffer.isEmpty()) {
    674        buffer.setTo(src);
    675    } else {
    676        const UnicodeString *values[2] = { &buffer, &src };
    677        UErrorCode status = U_ZERO_ERROR;
    678        separatorFormat.formatAndReplace(values, 2, buffer, nullptr, 0, status);
    679    }
    680    return buffer;
    681 }
    682 
    683 UnicodeString&
    684 LocaleDisplayNamesImpl::localeDisplayName(const char* localeId,
    685                                          UnicodeString& result) const {
    686    return localeDisplayName(Locale(localeId), result);
    687 }
    688 
    689 // private
    690 UnicodeString&
    691 LocaleDisplayNamesImpl::localeIdName(const char* localeId,
    692                                     UnicodeString& result, bool substitute) const {
    693    if (nameLength == UDISPCTX_LENGTH_SHORT) {
    694        langData.getNoFallback("Languages%short", localeId, result);
    695        if (!result.isBogus()) {
    696            return result;
    697        }
    698    }
    699    langData.getNoFallback("Languages", localeId, result);
    700    if (result.isBogus() && uprv_strchr(localeId, '_') == nullptr) {
    701        // Canonicalize lang and try again, ICU-20870
    702        // (only for language codes without script or region)
    703        Locale canonLocale = Locale::createCanonical(localeId);
    704        const char* canonLocId = canonLocale.getName();
    705        if (nameLength == UDISPCTX_LENGTH_SHORT) {
    706            langData.getNoFallback("Languages%short", canonLocId, result);
    707            if (!result.isBogus()) {
    708                return result;
    709            }
    710        }
    711        langData.getNoFallback("Languages", canonLocId, result);
    712    }
    713    if (result.isBogus() && substitute) {
    714        // use key, this is what langData.get (with fallback) falls back to.
    715        result.setTo(UnicodeString(localeId, -1, US_INV)); // use key (
    716    }
    717    return result;
    718 }
    719 
    720 UnicodeString&
    721 LocaleDisplayNamesImpl::languageDisplayName(const char* lang,
    722                                            UnicodeString& result) const {
    723    if (uprv_strcmp("root", lang) == 0 || uprv_strchr(lang, '_') != nullptr) {
    724        return result = UnicodeString(lang, -1, US_INV);
    725    }
    726    if (nameLength == UDISPCTX_LENGTH_SHORT) {
    727        langData.getNoFallback("Languages%short", lang, result);
    728        if (!result.isBogus()) {
    729            return adjustForUsageAndContext(kCapContextUsageLanguage, result);
    730        }
    731    }
    732    langData.getNoFallback("Languages", lang, result);
    733    if (result.isBogus()) {
    734        // Canonicalize lang and try again, ICU-20870
    735        Locale canonLocale = Locale::createCanonical(lang);
    736        const char* canonLocId = canonLocale.getName();
    737        if (nameLength == UDISPCTX_LENGTH_SHORT) {
    738            langData.getNoFallback("Languages%short", canonLocId, result);
    739            if (!result.isBogus()) {
    740                return adjustForUsageAndContext(kCapContextUsageLanguage, result);
    741            }
    742        }
    743        langData.getNoFallback("Languages", canonLocId, result);
    744    }
    745    if (result.isBogus() && substitute == UDISPCTX_SUBSTITUTE) {
    746        // use key, this is what langData.get (with fallback) falls back to.
    747        result.setTo(UnicodeString(lang, -1, US_INV)); // use key (
    748    }
    749    return adjustForUsageAndContext(kCapContextUsageLanguage, result);
    750 }
    751 
    752 UnicodeString&
    753 LocaleDisplayNamesImpl::scriptDisplayName(const char* script,
    754                                          UnicodeString& result,
    755                                          bool skipAdjust) const {
    756    if (nameLength == UDISPCTX_LENGTH_SHORT) {
    757        langData.getNoFallback("Scripts%short", script, result);
    758        if (!result.isBogus()) {
    759            return skipAdjust? result: adjustForUsageAndContext(kCapContextUsageScript, result);
    760        }
    761    }
    762    if (substitute == UDISPCTX_SUBSTITUTE) {
    763        langData.get("Scripts", script, result);
    764    } else {
    765        langData.getNoFallback("Scripts", script, result);
    766    }
    767    return skipAdjust? result: adjustForUsageAndContext(kCapContextUsageScript, result);
    768 }
    769 
    770 UnicodeString&
    771 LocaleDisplayNamesImpl::scriptDisplayName(const char* script,
    772                                          UnicodeString& result) const {
    773    return scriptDisplayName(script, result, false);
    774 }
    775 
    776 UnicodeString&
    777 LocaleDisplayNamesImpl::scriptDisplayName(UScriptCode scriptCode,
    778                                          UnicodeString& result) const {
    779    return scriptDisplayName(uscript_getName(scriptCode), result, false);
    780 }
    781 
    782 UnicodeString&
    783 LocaleDisplayNamesImpl::regionDisplayName(const char* region,
    784                                          UnicodeString& result,
    785                                          bool skipAdjust) const {
    786    if (nameLength == UDISPCTX_LENGTH_SHORT) {
    787         regionData.getNoFallback("Countries%short", region, result);
    788        if (!result.isBogus()) {
    789            return skipAdjust? result: adjustForUsageAndContext(kCapContextUsageTerritory, result);
    790        }
    791    }
    792    if (substitute == UDISPCTX_SUBSTITUTE) {
    793        regionData.get("Countries", region, result);
    794    } else {
    795        regionData.getNoFallback("Countries", region, result);
    796    }
    797    return skipAdjust? result: adjustForUsageAndContext(kCapContextUsageTerritory, result);
    798 }
    799 
    800 UnicodeString&
    801 LocaleDisplayNamesImpl::regionDisplayName(const char* region,
    802                                          UnicodeString& result) const {
    803    return regionDisplayName(region, result, false);
    804 }
    805 
    806 
    807 UnicodeString&
    808 LocaleDisplayNamesImpl::variantDisplayName(const char* variant,
    809                                           UnicodeString& result,
    810                                           bool skipAdjust) const {
    811    // don't have a resource for short variant names
    812    if (substitute == UDISPCTX_SUBSTITUTE) {
    813        langData.get("Variants", variant, result);
    814    } else {
    815        langData.getNoFallback("Variants", variant, result);
    816    }
    817    return skipAdjust? result: adjustForUsageAndContext(kCapContextUsageVariant, result);
    818 }
    819 
    820 UnicodeString&
    821 LocaleDisplayNamesImpl::variantDisplayName(const char* variant,
    822                                           UnicodeString& result) const {
    823    return variantDisplayName(variant, result, false);
    824 }
    825 
    826 UnicodeString&
    827 LocaleDisplayNamesImpl::keyDisplayName(const char* key,
    828                                       UnicodeString& result,
    829                                       bool skipAdjust) const {
    830    // don't have a resource for short key names
    831    if (substitute == UDISPCTX_SUBSTITUTE) {
    832        langData.get("Keys", key, result);
    833    } else {
    834        langData.getNoFallback("Keys", key, result);
    835    }
    836    return skipAdjust? result: adjustForUsageAndContext(kCapContextUsageKey, result);
    837 }
    838 
    839 UnicodeString&
    840 LocaleDisplayNamesImpl::keyDisplayName(const char* key,
    841                                       UnicodeString& result) const {
    842    return keyDisplayName(key, result, false);
    843 }
    844 
    845 UnicodeString&
    846 LocaleDisplayNamesImpl::keyValueDisplayName(const char* key,
    847                                            const char* value,
    848                                            UnicodeString& result,
    849                                            bool skipAdjust) const {
    850    if (uprv_strcmp(key, "currency") == 0) {
    851        // ICU4C does not have ICU4J CurrencyDisplayInfo equivalent for now.
    852        UErrorCode sts = U_ZERO_ERROR;
    853        UnicodeString ustrValue(value, -1, US_INV);
    854        int32_t len;
    855        const char16_t *currencyName = ucurr_getName(ustrValue.getTerminatedBuffer(),
    856            locale.getBaseName(), UCURR_LONG_NAME, nullptr /* isChoiceFormat */, &len, &sts);
    857        if (U_FAILURE(sts)) {
    858            // Return the value as is on failure
    859            result = ustrValue;
    860            return result;
    861        }
    862        result.setTo(currencyName, len);
    863        return skipAdjust? result: adjustForUsageAndContext(kCapContextUsageKeyValue, result);
    864    }
    865 
    866    if (nameLength == UDISPCTX_LENGTH_SHORT) {
    867        langData.getNoFallback("Types%short", key, value, result);
    868        if (!result.isBogus()) {
    869            return skipAdjust? result: adjustForUsageAndContext(kCapContextUsageKeyValue, result);
    870        }
    871    }
    872    if (substitute == UDISPCTX_SUBSTITUTE) {
    873        langData.get("Types", key, value, result);
    874    } else {
    875        langData.getNoFallback("Types", key, value, result);
    876    }
    877    return skipAdjust? result: adjustForUsageAndContext(kCapContextUsageKeyValue, result);
    878 }
    879 
    880 UnicodeString&
    881 LocaleDisplayNamesImpl::keyValueDisplayName(const char* key,
    882                                            const char* value,
    883                                            UnicodeString& result) const {
    884    return keyValueDisplayName(key, value, result, false);
    885 }
    886 
    887 ////////////////////////////////////////////////////////////////////////////////////////////////////
    888 
    889 LocaleDisplayNames*
    890 LocaleDisplayNames::createInstance(const Locale& locale,
    891                                   UDialectHandling dialectHandling) {
    892    return new LocaleDisplayNamesImpl(locale, dialectHandling);
    893 }
    894 
    895 LocaleDisplayNames*
    896 LocaleDisplayNames::createInstance(const Locale& locale,
    897                                   UDisplayContext *contexts, int32_t length) {
    898    if (contexts == nullptr) {
    899        length = 0;
    900    }
    901    return new LocaleDisplayNamesImpl(locale, contexts, length);
    902 }
    903 
    904 U_NAMESPACE_END
    905 
    906 ////////////////////////////////////////////////////////////////////////////////////////////////////
    907 
    908 U_NAMESPACE_USE
    909 
    910 U_CAPI ULocaleDisplayNames * U_EXPORT2
    911 uldn_open(const char * locale,
    912          UDialectHandling dialectHandling,
    913          UErrorCode *pErrorCode) {
    914  if (U_FAILURE(*pErrorCode)) {
    915    return nullptr;
    916  }
    917  if (locale == nullptr) {
    918    locale = uloc_getDefault();
    919  }
    920  return (ULocaleDisplayNames *)LocaleDisplayNames::createInstance(Locale(locale), dialectHandling);
    921 }
    922 
    923 U_CAPI ULocaleDisplayNames * U_EXPORT2
    924 uldn_openForContext(const char * locale,
    925                    UDisplayContext *contexts, int32_t length,
    926                    UErrorCode *pErrorCode) {
    927  if (U_FAILURE(*pErrorCode)) {
    928    return nullptr;
    929  }
    930  if (locale == nullptr) {
    931    locale = uloc_getDefault();
    932  }
    933  return (ULocaleDisplayNames *)LocaleDisplayNames::createInstance(Locale(locale), contexts, length);
    934 }
    935 
    936 
    937 U_CAPI void U_EXPORT2
    938 uldn_close(ULocaleDisplayNames *ldn) {
    939  delete (LocaleDisplayNames *)ldn;
    940 }
    941 
    942 U_CAPI const char * U_EXPORT2
    943 uldn_getLocale(const ULocaleDisplayNames *ldn) {
    944  if (ldn) {
    945    return ((const LocaleDisplayNames *)ldn)->getLocale().getName();
    946  }
    947  return nullptr;
    948 }
    949 
    950 U_CAPI UDialectHandling U_EXPORT2
    951 uldn_getDialectHandling(const ULocaleDisplayNames *ldn) {
    952  if (ldn) {
    953    return ((const LocaleDisplayNames *)ldn)->getDialectHandling();
    954  }
    955  return ULDN_STANDARD_NAMES;
    956 }
    957 
    958 U_CAPI UDisplayContext U_EXPORT2
    959 uldn_getContext(const ULocaleDisplayNames *ldn,
    960              UDisplayContextType type,
    961              UErrorCode *pErrorCode) {
    962  if (U_FAILURE(*pErrorCode)) {
    963    return (UDisplayContext)0;
    964  }
    965  return ((const LocaleDisplayNames *)ldn)->getContext(type);
    966 }
    967 
    968 U_CAPI int32_t U_EXPORT2
    969 uldn_localeDisplayName(const ULocaleDisplayNames *ldn,
    970                       const char *locale,
    971                       char16_t *result,
    972                       int32_t maxResultSize,
    973                       UErrorCode *pErrorCode) {
    974  if (U_FAILURE(*pErrorCode)) {
    975    return 0;
    976  }
    977  if (ldn == nullptr || locale == nullptr || (result == nullptr && maxResultSize > 0) || maxResultSize < 0) {
    978    *pErrorCode = U_ILLEGAL_ARGUMENT_ERROR;
    979    return 0;
    980  }
    981  UnicodeString temp(result, 0, maxResultSize);
    982  ((const LocaleDisplayNames *)ldn)->localeDisplayName(locale, temp);
    983  if (temp.isBogus()) {
    984    *pErrorCode = U_ILLEGAL_ARGUMENT_ERROR;
    985    return 0;
    986  }
    987  return temp.extract(result, maxResultSize, *pErrorCode);
    988 }
    989 
    990 U_CAPI int32_t U_EXPORT2
    991 uldn_languageDisplayName(const ULocaleDisplayNames *ldn,
    992                         const char *lang,
    993                         char16_t *result,
    994                         int32_t maxResultSize,
    995                         UErrorCode *pErrorCode) {
    996  if (U_FAILURE(*pErrorCode)) {
    997    return 0;
    998  }
    999  if (ldn == nullptr || lang == nullptr || (result == nullptr && maxResultSize > 0) || maxResultSize < 0) {
   1000    *pErrorCode = U_ILLEGAL_ARGUMENT_ERROR;
   1001    return 0;
   1002  }
   1003  UnicodeString temp(result, 0, maxResultSize);
   1004  ((const LocaleDisplayNames *)ldn)->languageDisplayName(lang, temp);
   1005  return temp.extract(result, maxResultSize, *pErrorCode);
   1006 }
   1007 
   1008 U_CAPI int32_t U_EXPORT2
   1009 uldn_scriptDisplayName(const ULocaleDisplayNames *ldn,
   1010                       const char *script,
   1011                       char16_t *result,
   1012                       int32_t maxResultSize,
   1013                       UErrorCode *pErrorCode) {
   1014  if (U_FAILURE(*pErrorCode)) {
   1015    return 0;
   1016  }
   1017  if (ldn == nullptr || script == nullptr || (result == nullptr && maxResultSize > 0) || maxResultSize < 0) {
   1018    *pErrorCode = U_ILLEGAL_ARGUMENT_ERROR;
   1019    return 0;
   1020  }
   1021  UnicodeString temp(result, 0, maxResultSize);
   1022  ((const LocaleDisplayNames *)ldn)->scriptDisplayName(script, temp);
   1023  return temp.extract(result, maxResultSize, *pErrorCode);
   1024 }
   1025 
   1026 U_CAPI int32_t U_EXPORT2
   1027 uldn_scriptCodeDisplayName(const ULocaleDisplayNames *ldn,
   1028                           UScriptCode scriptCode,
   1029                           char16_t *result,
   1030                           int32_t maxResultSize,
   1031                           UErrorCode *pErrorCode) {
   1032  return uldn_scriptDisplayName(ldn, uscript_getName(scriptCode), result, maxResultSize, pErrorCode);
   1033 }
   1034 
   1035 U_CAPI int32_t U_EXPORT2
   1036 uldn_regionDisplayName(const ULocaleDisplayNames *ldn,
   1037                       const char *region,
   1038                       char16_t *result,
   1039                       int32_t maxResultSize,
   1040                       UErrorCode *pErrorCode) {
   1041  if (U_FAILURE(*pErrorCode)) {
   1042    return 0;
   1043  }
   1044  if (ldn == nullptr || region == nullptr || (result == nullptr && maxResultSize > 0) || maxResultSize < 0) {
   1045    *pErrorCode = U_ILLEGAL_ARGUMENT_ERROR;
   1046    return 0;
   1047  }
   1048  UnicodeString temp(result, 0, maxResultSize);
   1049  ((const LocaleDisplayNames *)ldn)->regionDisplayName(region, temp);
   1050  return temp.extract(result, maxResultSize, *pErrorCode);
   1051 }
   1052 
   1053 U_CAPI int32_t U_EXPORT2
   1054 uldn_variantDisplayName(const ULocaleDisplayNames *ldn,
   1055                        const char *variant,
   1056                        char16_t *result,
   1057                        int32_t maxResultSize,
   1058                        UErrorCode *pErrorCode) {
   1059  if (U_FAILURE(*pErrorCode)) {
   1060    return 0;
   1061  }
   1062  if (ldn == nullptr || variant == nullptr || (result == nullptr && maxResultSize > 0) || maxResultSize < 0) {
   1063    *pErrorCode = U_ILLEGAL_ARGUMENT_ERROR;
   1064    return 0;
   1065  }
   1066  UnicodeString temp(result, 0, maxResultSize);
   1067  ((const LocaleDisplayNames *)ldn)->variantDisplayName(variant, temp);
   1068  return temp.extract(result, maxResultSize, *pErrorCode);
   1069 }
   1070 
   1071 U_CAPI int32_t U_EXPORT2
   1072 uldn_keyDisplayName(const ULocaleDisplayNames *ldn,
   1073                    const char *key,
   1074                    char16_t *result,
   1075                    int32_t maxResultSize,
   1076                    UErrorCode *pErrorCode) {
   1077  if (U_FAILURE(*pErrorCode)) {
   1078    return 0;
   1079  }
   1080  if (ldn == nullptr || key == nullptr || (result == nullptr && maxResultSize > 0) || maxResultSize < 0) {
   1081    *pErrorCode = U_ILLEGAL_ARGUMENT_ERROR;
   1082    return 0;
   1083  }
   1084  UnicodeString temp(result, 0, maxResultSize);
   1085  ((const LocaleDisplayNames *)ldn)->keyDisplayName(key, temp);
   1086  return temp.extract(result, maxResultSize, *pErrorCode);
   1087 }
   1088 
   1089 U_CAPI int32_t U_EXPORT2
   1090 uldn_keyValueDisplayName(const ULocaleDisplayNames *ldn,
   1091                         const char *key,
   1092                         const char *value,
   1093                         char16_t *result,
   1094                         int32_t maxResultSize,
   1095                         UErrorCode *pErrorCode) {
   1096  if (U_FAILURE(*pErrorCode)) {
   1097    return 0;
   1098  }
   1099  if (ldn == nullptr || key == nullptr || value == nullptr || (result == nullptr && maxResultSize > 0)
   1100      || maxResultSize < 0) {
   1101    *pErrorCode = U_ILLEGAL_ARGUMENT_ERROR;
   1102    return 0;
   1103  }
   1104  UnicodeString temp(result, 0, maxResultSize);
   1105  ((const LocaleDisplayNames *)ldn)->keyValueDisplayName(key, value, temp);
   1106  return temp.extract(result, maxResultSize, *pErrorCode);
   1107 }
   1108 
   1109 #endif