C++ CharString::data方法代码示例

void
PNamesBuilderImpl::writeCSourceFile(const char *path, UErrorCode &errorCode) {
    if(U_FAILURE(errorCode)) { return; }
    FILE *f=usrc_create(path, "propname_data.h",
                        "icu/tools/unicode/c/genprops/pnamesbuilder.cpp");
    if(f==NULL) {
        errorCode=U_FILE_ACCESS_ERROR;
        return;
    }

    fputs("#ifndef INCLUDED_FROM_PROPNAME_CPP\n"
          "#   error This file must be #included from propname.cpp only.\n"
          "#endif\n\n", f);

    fputs("U_NAMESPACE_BEGIN\n\n", f);

    usrc_writeArray(f, "const int32_t PropNameData::indexes[%ld]={",
                    indexes, 32, PropNameData::IX_COUNT,
                    "};\n\n");
    usrc_writeArray(f, "const int32_t PropNameData::valueMaps[%ld]={\n",
                    valueMaps.getBuffer(), 32, valueMaps.size(),
                    "\n};\n\n");
    usrc_writeArray(f, "const uint8_t PropNameData::bytesTries[%ld]={\n",
                    bytesTries.data(), 8, bytesTries.length(),
                    "\n};\n\n");
    usrc_writeArrayOfMostlyInvChars(
        f, "const char PropNameData::nameGroups[%ld]={\n",
        nameGroups.data(), nameGroups.length(),
        "\n};\n\n");

    fputs("U_NAMESPACE_END\n", f);

    fclose(f);
}

void CompactData::populate(const Locale &locale, const char *nsName, CompactStyle compactStyle,
                           CompactType compactType, UErrorCode &status) {
    CompactDataSink sink(*this);
    LocalUResourceBundlePointer rb(ures_open(nullptr, locale.getName(), &status));
    if (U_FAILURE(status)) { return; }

    bool nsIsLatn = strcmp(nsName, "latn") == 0;
    bool compactIsShort = compactStyle == CompactStyle::UNUM_SHORT;

    // Fall back to latn numbering system and/or short compact style.
    CharString resourceKey;
    getResourceBundleKey(nsName, compactStyle, compactType, resourceKey, status);
    UErrorCode localStatus = U_ZERO_ERROR;
    ures_getAllItemsWithFallback(rb.getAlias(), resourceKey.data(), sink, localStatus);
    if (isEmpty && !nsIsLatn) {
        getResourceBundleKey("latn", compactStyle, compactType, resourceKey, status);
        localStatus = U_ZERO_ERROR;
        ures_getAllItemsWithFallback(rb.getAlias(), resourceKey.data(), sink, localStatus);
    }
    if (isEmpty && !compactIsShort) {
        getResourceBundleKey(nsName, CompactStyle::UNUM_SHORT, compactType, resourceKey, status);
        localStatus = U_ZERO_ERROR;
        ures_getAllItemsWithFallback(rb.getAlias(), resourceKey.data(), sink, localStatus);
    }
    if (isEmpty && !nsIsLatn && !compactIsShort) {
        getResourceBundleKey("latn", CompactStyle::UNUM_SHORT, compactType, resourceKey, status);
        localStatus = U_ZERO_ERROR;
        ures_getAllItemsWithFallback(rb.getAlias(), resourceKey.data(), sink, localStatus);
    }

    // The last fallback should be guaranteed to return data.
    if (isEmpty) {
        status = U_INTERNAL_PROGRAM_ERROR;
    }
}

DictionaryMatcher *
ICULanguageBreakFactory::loadDictionaryMatcherFor(UScriptCode script, int32_t /* brkType */) {
    UErrorCode status = U_ZERO_ERROR;
    // open root from brkitr tree.
    UResourceBundle *b = ures_open(U_ICUDATA_BRKITR, "", &status);
    b = ures_getByKeyWithFallback(b, "dictionaries", b, &status);
    int32_t dictnlength = 0;
    const UChar *dictfname =
        ures_getStringByKeyWithFallback(b, uscript_getShortName(script), &dictnlength, &status);
    if (U_FAILURE(status)) {
        ures_close(b);
        return NULL;
    }
    CharString dictnbuf;
    CharString ext;
    const UChar *extStart = u_memrchr(dictfname, 0x002e, dictnlength);  // last dot
    if (extStart != NULL) {
        int32_t len = (int32_t)(extStart - dictfname);
        ext.appendInvariantChars(UnicodeString(FALSE, extStart + 1, dictnlength - len - 1), status);
        dictnlength = len;
    }
    dictnbuf.appendInvariantChars(UnicodeString(FALSE, dictfname, dictnlength), status);
    ures_close(b);

    UDataMemory *file = udata_open(U_ICUDATA_BRKITR, ext.data(), dictnbuf.data(), &status);
    if (U_SUCCESS(status)) {
        // build trie
        const uint8_t *data = (const uint8_t *)udata_getMemory(file);
        const int32_t *indexes = (const int32_t *)data;
        const int32_t offset = indexes[DictionaryData::IX_STRING_TRIE_OFFSET];
        const int32_t trieType = indexes[DictionaryData::IX_TRIE_TYPE] & DictionaryData::TRIE_TYPE_MASK;
        DictionaryMatcher *m = NULL;
        if (trieType == DictionaryData::TRIE_TYPE_BYTES) {
            const int32_t transform = indexes[DictionaryData::IX_TRANSFORM];
            const char *characters = (const char *)(data + offset);
            m = new BytesDictionaryMatcher(characters, transform, file);
        }
        else if (trieType == DictionaryData::TRIE_TYPE_UCHARS) {
            const UChar *characters = (const UChar *)(data + offset);
            m = new UCharsDictionaryMatcher(characters, file);
        }
        if (m == NULL) {
            // no matcher exists to take ownership - either we are an invalid
            // type or memory allocation failed
            udata_close(file);
        }
        return m;
    } else if (dictfname != NULL) {
        // we don't have a dictionary matcher.
        // returning NULL here will cause us to fail to find a dictionary break engine, as expected
        status = U_ZERO_ERROR;
        return NULL;
    }
    return NULL;
}

PluralMapBase::Category
PluralMapBase::toCategory(const UnicodeString &pluralForm) {
    CharString cCategory;
    UErrorCode status = U_ZERO_ERROR;
    cCategory.appendInvariantChars(pluralForm, status);
    return U_FAILURE(status) ? NONE : toCategory(cCategory.data());
}

static UnicodeString loadNumericDateFormatterPattern(
        const UResourceBundle *resource,
        const char *pattern,
        UErrorCode &status) {
    UnicodeString result;
    if (U_FAILURE(status)) {
        return result;
    }
    CharString chs;
    chs.append("durationUnits", status)
            .append("/", status).append(pattern, status);
    LocalUResourceBundlePointer patternBundle(
            ures_getByKeyWithFallback(
                resource,
                chs.data(),
                NULL,
                &status));
    if (U_FAILURE(status)) {
        return result;
    }
    getString(patternBundle.getAlias(), result, status);
    // Replace 'h' with 'H'
    int32_t len = result.length();
    UChar *buffer = result.getBuffer(len);
    for (int32_t i = 0; i < len; ++i) {
        if (buffer[i] == 0x68) { // 'h'
            buffer[i] = 0x48; // 'H'
        }
    }
    result.releaseBuffer(len);
    return result;
}

const Hashtable*
LocaleUtility::getAvailableLocaleNames(const UnicodeString& bundleID)
{
    // LocaleUtility_cache is a hash-of-hashes.  The top-level keys
    // are path strings ('bundleID') passed to
    // ures_openAvailableLocales.  The top-level values are
    // second-level hashes.  The second-level keys are result strings
    // from ures_openAvailableLocales.  The second-level values are
    // garbage ((void*)1 or other random pointer).

    UErrorCode status = U_ZERO_ERROR;
    umtx_initOnce(LocaleUtilityInitOnce, locale_utility_init, status);
    Hashtable *cache = LocaleUtility_cache;
    if (cache == NULL) {
        // Catastrophic failure.
        return NULL;
    }

    Hashtable* htp;
    umtx_lock(NULL);
    htp = (Hashtable*) cache->get(bundleID);
    umtx_unlock(NULL);

    if (htp == NULL) {
        htp = new Hashtable(status);
        if (htp && U_SUCCESS(status)) {
            CharString cbundleID;
            cbundleID.appendInvariantChars(bundleID, status);
            const char* path = cbundleID.isEmpty() ? NULL : cbundleID.data();
            UEnumeration *uenum = ures_openAvailableLocales(path, &status);
            for (;;) {
                const UChar* id = uenum_unext(uenum, NULL, &status);
                if (id == NULL) {
                    break;
                }
                htp->put(UnicodeString(id), (void*)htp, status);
            }
            uenum_close(uenum);
            if (U_FAILURE(status)) {
                delete htp;
                return NULL;
            }
            umtx_lock(NULL);
            Hashtable *t = static_cast<Hashtable *>(cache->get(bundleID));
            if (t != NULL) {
                // Another thread raced through this code, creating the cache entry first.
                // Discard ours and return theirs.
                umtx_unlock(NULL);
                delete htp;
                htp = t;
            } else {
                cache->put(bundleID, (void*)htp, status);
                umtx_unlock(NULL);
            }
        }
    }
    return htp;
}

static void load(const Locale& inLocale, CDFLocaleData* result, UErrorCode& status) {
  LocalPointer<NumberingSystem> ns(NumberingSystem::createInstance(inLocale, status));
  if (U_FAILURE(status)) {
    return;
  }
  const char* nsName = ns->getName();

  LocalUResourceBundlePointer resource(ures_open(NULL, inLocale.getName(), &status));
  if (U_FAILURE(status)) {
    return;
  }
  CmptDecDataSink sink(*result);
  sink.isFallback = FALSE;

  // First load the number elements data if nsName is not Latin.
  if (uprv_strcmp(nsName, gLatnTag) != 0) {
    sink.isLatin = FALSE;
    CharString path;
    path.append(gNumberElementsTag, status)
        .append('/', status)
        .append(nsName, status);
    ures_getAllItemsWithFallback(resource.getAlias(), path.data(), sink, status);
    if (status == U_MISSING_RESOURCE_ERROR) {
      // Silently ignore and use Latin
      status = U_ZERO_ERROR;
    } else if  (U_FAILURE(status)) {
      return;
    }
    sink.isFallback = TRUE;
  }

  // Now load Latin.
  sink.isLatin = TRUE;
  ures_getAllItemsWithFallback(resource.getAlias(), gLatnPath, sink, status);
  if (U_FAILURE(status)) return;

  // If longData is empty, default it to be equal to shortData
  if (result->longData.isEmpty()) {
    result->longData.setToBogus();
  }

  // Check for "other" variants in each of the three data classes, and resolve missing elements.

  if (!result->longData.isBogus()) {
    checkForOtherVariants(&result->longData, status);
    if (U_FAILURE(status)) return;
    fillInMissing(&result->longData);
  }

  checkForOtherVariants(&result->shortData, status);
  if (U_FAILURE(status)) return;
  fillInMissing(&result->shortData);

  // TODO: Enable this statement when currency support is added
  // checkForOtherVariants(&result->shortCurrencyData, status);
  // if (U_FAILURE(status)) return;
  // fillInMissing(&result->shortCurrencyData);
}

void
TimeUnitFormat::checkConsistency(UTimeUnitFormatStyle style, const char* key, UErrorCode& err) {
    if (U_FAILURE(err)) {
        return;
    }
    // there should be patterns for each plural rule in each time unit.
    // For each time unit,
    //     for each plural rule, following is unit pattern fall-back rule:
    //         ( for example: "one" hour )
    //         look for its unit pattern in its locale tree.
    //         if pattern is not found in its own locale, such as de_DE,
    //         look for the pattern in its parent, such as de,
    //         keep looking till found or till root.
    //         if the pattern is not found in root either,
    //         fallback to plural count "other",
    //         look for the pattern of "other" in the locale tree:
    //         "de_DE" to "de" to "root".
    //         If not found, fall back to value of
    //         static variable DEFAULT_PATTERN_FOR_xxx, such as "{0} h".
    //
    // Following is consistency check to create pattern for each
    // plural rule in each time unit using above fall-back rule.
    //
    StringEnumeration* keywords = getPluralRules().getKeywords(err);
    if (U_SUCCESS(err)) {
        const UnicodeString* pluralCount;
        while ((pluralCount = keywords->snext(err)) != NULL) {
            if ( U_SUCCESS(err) ) {
                for (int32_t i = 0; i < TimeUnit::UTIMEUNIT_FIELD_COUNT; ++i) {
                    // for each time unit,
                    // get all the patterns for each plural rule in this locale.
                    Hashtable* countToPatterns = fTimeUnitToCountToPatterns[i];
                    if ( countToPatterns == NULL ) {
                        countToPatterns = initHash(err);
                        if (U_FAILURE(err)) {
                            delete countToPatterns;
                            return;
                        }
                        fTimeUnitToCountToPatterns[i] = countToPatterns;
                    }
                    MessageFormat** formatters = (MessageFormat**)countToPatterns->get(*pluralCount);
                    if( formatters == NULL || formatters[style] == NULL ) {
                        // look through parents
                        const char* localeName = getLocaleID(err);
                        CharString pluralCountChars;
                        pluralCountChars.appendInvariantChars(*pluralCount, err);
                        searchInLocaleChain(style, key, localeName,
                                            (TimeUnit::UTimeUnitFields)i,
                                            *pluralCount, pluralCountChars.data(),
                                            countToPatterns, err);
                    }
                }
            }
        }
    }
    delete keywords;
}

static void strToLocale(
        const UnicodeString &str, void *localePtr, UErrorCode &status) {
    if (U_FAILURE(status)) {
        return;
    }
    CharString localeStr;
    localeStr.appendInvariantChars(str, status);
    *static_cast<Locale *>(localePtr) = Locale(localeStr.data());
}

void
CurrencyAffixInfo::set(
        const char *locale,
        const PluralRules *rules,
        const UChar *currency,
        UErrorCode &status) {
    if (U_FAILURE(status)) {
        return;
    }
    fIsDefault = FALSE;
    if (currency == NULL) {
        fSymbol.setTo(gDefaultSymbols, 1);
        fISO.setTo(gDefaultSymbols, 2);
        fLong.remove();
        fLong.append(gDefaultSymbols, 3);
        fIsDefault = TRUE;
        return;
    }
    int32_t len;
    UBool unusedIsChoice;
    const UChar *symbol = ucurr_getName(
            currency, locale, UCURR_SYMBOL_NAME, &unusedIsChoice,
            &len, &status);
    if (U_FAILURE(status)) {
        return;
    }
    fSymbol.setTo(symbol, len);
    fISO.setTo(currency, u_strlen(currency));
    fLong.remove();
    StringEnumeration* keywords = rules->getKeywords(status);
    if (U_FAILURE(status)) {
        return;
    }
    const UnicodeString* pluralCount;
    while ((pluralCount = keywords->snext(status)) != NULL) {
        CharString pCount;
        pCount.appendInvariantChars(*pluralCount, status);
        const UChar *pluralName = ucurr_getPluralName(
            currency, locale, &unusedIsChoice, pCount.data(),
            &len, &status);
        fLong.setVariant(pCount.data(), UnicodeString(pluralName, len), status);
    }
    delete keywords;
}

static void strToDouble(
        const UnicodeString &str, void *doublePtr, UErrorCode &status) {
    if (U_FAILURE(status)) {
        return;
    }
    CharString buffer;
    buffer.appendInvariantChars(str, status);
    if (U_FAILURE(status)) {
        return;
    }
    *static_cast<double *>(doublePtr) = atof(buffer.data());
}

// getCDFUnitFallback returns a pointer to the prefix-suffix pair for a given
// variant and log10 value within table. If the given variant doesn't exist, it
// falls back to the OTHER variant. Therefore, this method will always return
// some non-NULL value.
static const CDFUnit* getCDFUnitFallback(const UHashtable* table, const UnicodeString& variant, int32_t log10Value) {
  CharString cvariant;
  UErrorCode status = U_ZERO_ERROR;
  const CDFUnit *cdfUnit = NULL;
  cvariant.appendInvariantChars(variant, status);
  if (!U_FAILURE(status)) {
    cdfUnit = (const CDFUnit*) uhash_get(table, cvariant.data());
  }
  if (cdfUnit == NULL) {
    cdfUnit = (const CDFUnit*) uhash_get(table, gOther);
  }
  return &cdfUnit[log10Value];
}

void
PNamesBuilderImpl::writeBinaryData(const char *path, UBool withCopyright, UErrorCode &errorCode) {
    if(U_FAILURE(errorCode)) { return; }
    UNewDataMemory *pdata=udata_create(path, PNAME_DATA_TYPE, PNAME_DATA_NAME, &dataInfo,
                                       withCopyright ? U_COPYRIGHT_STRING : 0, &errorCode);
    if(U_FAILURE(errorCode)) {
        fprintf(stderr, "genprops: udata_create(%s, pnames.icu) failed - %s\n",
                path, u_errorName(errorCode));
        return;
    }

    udata_writeBlock(pdata, indexes, PropNameData::IX_COUNT*4);
    udata_writeBlock(pdata, valueMaps.getBuffer(), valueMaps.size()*4);
    udata_writeBlock(pdata, bytesTries.data(), bytesTries.length());
    udata_writeBlock(pdata, nameGroups.data(), nameGroups.length());

    int32_t dataLength=(int32_t)udata_finish(pdata, &errorCode);
    if(dataLength!=indexes[PropNameData::IX_TOTAL_SIZE]) {
        fprintf(stderr,
                "udata_finish(pnames.icu) reports %ld bytes written but should be %ld\n",
                (long)dataLength, (long)indexes[PropNameData::IX_TOTAL_SIZE]);
        errorCode=U_INTERNAL_PROGRAM_ERROR;
    }
}

void
NamesPropsBuilder::setProps(const UniProps &props, const UnicodeSet &newValues,
                            UErrorCode &errorCode) {
    if(U_FAILURE(errorCode)) { return; }
    if(!newValues.contains(UCHAR_NAME) && !newValues.contains(PPUCD_NAME_ALIAS)) {
        return;
    }

    U_ASSERT(props.start==props.end);

    const char *names[4]={ NULL, NULL, NULL, NULL };
    int16_t lengths[4]={ 0, 0, 0, 0 };

    /* get the character name */
    if(props.name!=NULL) {
        names[0]=props.name;
        lengths[0]=(int16_t)uprv_strlen(props.name);
        parseName(names[0], lengths[0]);
    }

    CharString buffer;
    if(props.nameAlias!=NULL) {
        /*
         * Only use "correction" aliases for now, from Unicode 6.1 NameAliases.txt with 3 fields per line.
         * TODO: Work on ticket #8963 to deal with multiple type:alias pairs per character.
         */
        const char *corr=uprv_strstr(props.nameAlias, "correction=");
        if(corr!=NULL) {
            corr+=11;  // skip "correction="
            const char *limit=uprv_strchr(corr, ',');
            if(limit!=NULL) {
                buffer.append(corr, limit-corr, errorCode);
                names[3]=buffer.data();
                lengths[3]=(int16_t)(limit-corr);
            } else {
                names[3]=corr;
                lengths[3]=(int16_t)uprv_strlen(corr);
            }
            parseName(names[3], lengths[3]);
        }
    }

    addLine(props.start, names, lengths, LENGTHOF(names));
}

U_DRAFT const char * U_EXPORT2
ufmt_getDecNumChars(UFormattable *fmt, int32_t *len, UErrorCode *status) {
  if(U_FAILURE(*status)) {
    return "";
  }
  Formattable *obj = Formattable::fromUFormattable(fmt);
  CharString *charString = obj->internalGetCharString(*status);
  if(U_FAILURE(*status)) {
    return "";
  }
  if(charString == NULL) {
    *status = U_MEMORY_ALLOCATION_ERROR;
    return "";
  } else {
    if(len!=NULL) {
      *len = charString->length();
    }
    return charString->data();
  }
}