C++ UnicodeString类代码示例

 virtual int32_t
 spanQuickCheckYes(const UnicodeString &s, UErrorCode &) const {
     return s.length();
 }

Transliterator* TransliteratorAlias::create(UParseError& pe,
                                            UErrorCode& ec) {
    if (U_FAILURE(ec)) {
        return 0;
    }
    Transliterator *t = NULL;
    switch (type) {
    case SIMPLE:
        t = Transliterator::createInstance(aliasesOrRules, UTRANS_FORWARD, pe, ec);
        if(U_FAILURE(ec)){
            return 0;
        }
        if (compoundFilter != 0)
            t->adoptFilter((UnicodeSet*)compoundFilter->clone());
        break;
    case COMPOUND:
        {
            // the total number of transliterators in the compound is the total number of anonymous transliterators
            // plus the total number of ID blocks-- we start by assuming the list begins and ends with an ID
            // block and that each pair anonymous transliterators has an ID block between them.  Then we go back
            // to see whether there really are ID blocks at the beginning and end (by looking for U+FFFF, which
            // marks the position where an anonymous transliterator goes) and adjust accordingly
            int32_t anonymousRBTs = transes->size();
            int32_t transCount = anonymousRBTs * 2 + 1;
            if (!aliasesOrRules.isEmpty() && aliasesOrRules[0] == (UChar)(0xffff))
                --transCount;
            if (aliasesOrRules.length() >= 2 && aliasesOrRules[aliasesOrRules.length() - 1] == (UChar)(0xffff))
                --transCount;
            UnicodeString noIDBlock((UChar)(0xffff));
            noIDBlock += ((UChar)(0xffff));
            int32_t pos = aliasesOrRules.indexOf(noIDBlock);
            while (pos >= 0) {
                --transCount;
                pos = aliasesOrRules.indexOf(noIDBlock, pos + 1);
            }

            UVector transliterators(ec);
            UnicodeString idBlock;
            int32_t blockSeparatorPos = aliasesOrRules.indexOf((UChar)(0xffff));
            while (blockSeparatorPos >= 0) {
                aliasesOrRules.extract(0, blockSeparatorPos, idBlock);
                aliasesOrRules.remove(0, blockSeparatorPos + 1);
                if (!idBlock.isEmpty())
                    transliterators.addElement(Transliterator::createInstance(idBlock, UTRANS_FORWARD, pe, ec), ec);
                if (!transes->isEmpty())
                    transliterators.addElement(transes->orphanElementAt(0), ec);
                blockSeparatorPos = aliasesOrRules.indexOf((UChar)(0xffff));
            }
            if (!aliasesOrRules.isEmpty())
                transliterators.addElement(Transliterator::createInstance(aliasesOrRules, UTRANS_FORWARD, pe, ec), ec);
            while (!transes->isEmpty())
                transliterators.addElement(transes->orphanElementAt(0), ec);

            if (U_SUCCESS(ec)) {
                t = new CompoundTransliterator(ID, transliterators,
                    (compoundFilter ? (UnicodeSet*)(compoundFilter->clone()) : 0),
                    anonymousRBTs, pe, ec);
                if (t == 0) {
                    ec = U_MEMORY_ALLOCATION_ERROR;
                    return 0;
                }
            } else {
                for (int32_t i = 0; i < transliterators.size(); i++)
                    delete (Transliterator*)(transliterators.elementAt(i));
            }
        }
        break;
    case RULES:
        U_ASSERT(FALSE); // don't call create() if isRuleBased() returns TRUE!
        break;
    }
    return t;
}

/**
 * Parse the pattern from the given RuleCharacterIterator.  The
 * iterator is advanced over the parsed pattern.
 * @param chars iterator over the pattern characters.  Upon return
 * it will be advanced to the first character after the parsed
 * pattern, or the end of the iteration if all characters are
 * parsed.
 * @param symbols symbol table to use to parse and dereference
 * variables, or null if none.
 * @param rebuiltPat the pattern that was parsed, rebuilt or
 * copied from the input pattern, as appropriate.
 * @param options a bit mask of zero or more of the following:
 * IGNORE_SPACE, CASE.
 */
void UnicodeSet::applyPattern(RuleCharacterIterator& chars,
                              const SymbolTable* symbols,
                              UnicodeString& rebuiltPat,
                              uint32_t options,
                              UnicodeSet& (UnicodeSet::*caseClosure)(int32_t attribute),
                              UErrorCode& ec) {
    if (U_FAILURE(ec)) return;

    // Syntax characters: [ ] ^ - & { }

    // Recognized special forms for chars, sets: c-c s-s s&s

    int32_t opts = RuleCharacterIterator::PARSE_VARIABLES |
                   RuleCharacterIterator::PARSE_ESCAPES;
    if ((options & USET_IGNORE_SPACE) != 0) {
        opts |= RuleCharacterIterator::SKIP_WHITESPACE;
    }

    UnicodeString patLocal, buf;
    UBool usePat = FALSE;
    UnicodeSetPointer scratch;
    RuleCharacterIterator::Pos backup;

    // mode: 0=before [, 1=between [...], 2=after ]
    // lastItem: 0=none, 1=char, 2=set
    int8_t lastItem = 0, mode = 0;
    UChar32 lastChar = 0;
    UChar op = 0;

    UBool invert = FALSE;

    clear();

    while (mode != 2 && !chars.atEnd()) {
        U_ASSERT((lastItem == 0 && op == 0) ||
                 (lastItem == 1 && (op == 0 || op == HYPHEN /*'-'*/)) ||
                 (lastItem == 2 && (op == 0 || op == HYPHEN /*'-'*/ ||
                                    op == INTERSECTION /*'&'*/)));

        UChar32 c = 0;
        UBool literal = FALSE;
        UnicodeSet* nested = 0; // alias - do not delete

        // -------- Check for property pattern

        // setMode: 0=none, 1=unicodeset, 2=propertypat, 3=preparsed
        int8_t setMode = 0;
        if (resemblesPropertyPattern(chars, opts)) {
            setMode = 2;
        }

        // -------- Parse '[' of opening delimiter OR nested set.
        // If there is a nested set, use `setMode' to define how
        // the set should be parsed.  If the '[' is part of the
        // opening delimiter for this pattern, parse special
        // strings "[", "[^", "[-", and "[^-".  Check for stand-in
        // characters representing a nested set in the symbol
        // table.

        else {
            // Prepare to backup if necessary
            chars.getPos(backup);
            c = chars.next(opts, literal, ec);
            if (U_FAILURE(ec)) return;

            if (c == 0x5B /*'['*/ && !literal) {
                if (mode == 1) {
                    chars.setPos(backup); // backup
                    setMode = 1;
                } else {
                    // Handle opening '[' delimiter
                    mode = 1;
                    patLocal.append((UChar) 0x5B /*'['*/);
                    chars.getPos(backup); // prepare to backup
                    c = chars.next(opts, literal, ec);
                    if (U_FAILURE(ec)) return;
                    if (c == 0x5E /*'^'*/ && !literal) {
                        invert = TRUE;
                        patLocal.append((UChar) 0x5E /*'^'*/);
                        chars.getPos(backup); // prepare to backup
                        c = chars.next(opts, literal, ec);
                        if (U_FAILURE(ec)) return;
                    }
                    // Fall through to handle special leading '-';
                    // otherwise restart loop for nested [], \p{}, etc.
                    if (c == HYPHEN /*'-'*/) {
//.........这里部分代码省略.........

/**
 * If in "by digits" mode, fills in the substitution one decimal digit
 * at a time using the rule set containing this substitution.
 * Otherwise, uses the superclass function.
 * @param number The number being formatted
 * @param toInsertInto The string to insert the result of formatting
 * the substitution into
 * @param pos The position of the owning rule's rule text in
 * toInsertInto
 */
void
FractionalPartSubstitution::doSubstitution(double number, UnicodeString& toInsertInto, int32_t _pos) const
{
  // if we're not in "byDigits" mode, just use the inherited
  // doSubstitution() routine
  if (!byDigits) {
    NFSubstitution::doSubstitution(number, toInsertInto, _pos);

    // if we're in "byDigits" mode, transform the value into an integer
    // by moving the decimal point eight places to the right and
    // pulling digits off the right one at a time, formatting each digit
    // as an integer using this substitution's owning rule set
    // (this is slower, but more accurate, than doing it from the
    // other end)
  } else {
    //          int32_t numberToFormat = (int32_t)uprv_round(transformNumber(number) * uprv_pow(10, kMaxDecimalDigits));
    //          // this flag keeps us from formatting trailing zeros.  It starts
    //          // out false because we're pulling from the right, and switches
    //          // to true the first time we encounter a non-zero digit
    //          UBool doZeros = FALSE;
    //          for (int32_t i = 0; i < kMaxDecimalDigits; i++) {
    //              int64_t digit = numberToFormat % 10;
    //              if (digit != 0 || doZeros) {
    //                  if (doZeros && useSpaces) {
    //                      toInsertInto.insert(_pos + getPos(), gSpace);
    //                  }
    //                  doZeros = TRUE;
    //                  getRuleSet()->format(digit, toInsertInto, _pos + getPos());
    //              }
    //              numberToFormat /= 10;
    //          }

    DigitList dl;
    dl.set(number, 20, TRUE);
    
    UBool pad = FALSE;
    while (dl.fCount > (dl.fDecimalAt <= 0 ? 0 : dl.fDecimalAt)) {
      if (pad && useSpaces) {
        toInsertInto.insert(_pos + getPos(), gSpace);
      } else {
        pad = TRUE;
      }
      getRuleSet()->format((int64_t)(dl.fDigits[--dl.fCount] - '0'), toInsertInto, _pos + getPos());
    }
    while (dl.fDecimalAt < 0) {
      if (pad && useSpaces) {
        toInsertInto.insert(_pos + getPos(), gSpace);
      } else {
        pad = TRUE;
      }
      getRuleSet()->format((int64_t)0, toInsertInto, _pos + getPos());
      ++dl.fDecimalAt;
    }

    if (!pad) {
      // hack around lack of precision in digitlist. if we would end up with
      // "foo point" make sure we add a " zero" to the end.
      getRuleSet()->format((int64_t)0, toInsertInto, _pos + getPos());
    }
  }
}

static inline UBool
isPerlOpen(const UnicodeString &pattern, int32_t pos) {
    UChar c;
    return pattern.charAt(pos)==BACKSLASH && ((c=pattern.charAt(pos+1))==LOWER_P || c==UPPER_P);
}

static inline UBool
isPOSIXOpen(const UnicodeString &pattern, int32_t pos) {
    return pattern.charAt(pos)==SET_OPEN && pattern.charAt(pos+1)==COLON;
}

void printUnicodeString(const UnicodeString &s) {
    char charBuf[1000];
    s.extract(0, s.length(), charBuf, sizeof(charBuf)-1, 0);   
    charBuf[sizeof(charBuf)-1] = 0;          
    printf("%s", charBuf);
}

/**
 * Append c to buf, unless buf is empty or buf already ends in c.
 */
static void _smartAppend(UnicodeString& buf, UChar c) {
    if (buf.length() != 0 &&
        buf.charAt(buf.length() - 1) != c) {
        buf.append(c);
    }
}

BOOST_FIXTURE_TEST_CASE(test11, base_fixture_t)
{
    // Тесты на ::FmtLoadStr FMTLOAD ::Format ::LoadStr ::LoadStrPart ::CutToChar ::TrimLeft ::TrimRight
    {
        UnicodeString str = ::FmtLoadStr(CONST_TEST_STRING, L"lalala", 42);
        // BOOST_TEST_MESSAGE("str = " << W2MB(str.c_str()));
        // BOOST_TEST_MESSAGE("length = " << str.size());
        BOOST_CHECK(W2MB(str.c_str()) == "test string: \"lalala\" 42");
    }
    {
        UnicodeString str2 = FMTLOAD(CONST_TEST_STRING, L"lalala", 42);
        // BOOST_TEST_MESSAGE("str2 = " << W2MB(str2.c_str()));
        BOOST_CHECK(W2MB(str2.c_str()) == "test string: \"lalala\" 42");
    }
    {
        UnicodeString str2 = ::Format(L"test: %s %d", L"lalala", 42);
        BOOST_TEST_MESSAGE("str2 = " << W2MB(str2.c_str()));
        BOOST_CHECK_EQUAL(0, wcscmp(str2.c_str(), L"test: lalala 42"));
    }
    {
        UnicodeString str3 = FORMAT(L"test: %s %d", L"lalala", 42);
        BOOST_TEST_MESSAGE("str3 = " << W2MB(str3.c_str()));
        BOOST_CHECK_EQUAL(0, wcscmp(str3.c_str(), L"test: lalala 42"));
    }
    {
        UnicodeString str = ::TrimLeft(L"");
        BOOST_TEST_MESSAGE("str = " << W2MB(str.c_str()));
        BOOST_CHECK_EQUAL(0, wcscmp(str.c_str(), L""));
    }
    {
        UnicodeString str = ::TrimLeft(L"1");
        BOOST_TEST_MESSAGE("str = " << W2MB(str.c_str()));
        BOOST_CHECK_EQUAL(0, wcscmp(str.c_str(), L"1"));
    }
    {
        UnicodeString str = ::TrimLeft(L" 1");
        BOOST_TEST_MESSAGE("str = " << W2MB(str.c_str()));
        BOOST_CHECK_EQUAL(0, wcscmp(str.c_str(), L"1"));
    }
    {
        UnicodeString str = ::TrimRight(L"");
        BOOST_TEST_MESSAGE("str = " << W2MB(str.c_str()));
        BOOST_CHECK_EQUAL(0, wcscmp(str.c_str(), L""));
    }
    {
        UnicodeString str = ::TrimRight(L"1");
        BOOST_TEST_MESSAGE("str = " << W2MB(str.c_str()));
        BOOST_CHECK_EQUAL(0, wcscmp(str.c_str(), L"1"));
    }
    {
        UnicodeString str = ::TrimRight(L"1 ");
        BOOST_TEST_MESSAGE("str = " << W2MB(str.c_str()));
        BOOST_CHECK_EQUAL(0, wcscmp(str.c_str(), L"1"));
    }
    {
        // UnicodeString CutToChar(UnicodeString &Str, char Ch, bool Trim)
        UnicodeString Str1 = L" part 1 | part 2 ";
        UnicodeString str1 = ::CutToChar(Str1, '|', false);
        BOOST_TEST_MESSAGE("str1 = '" << W2MB(str1.c_str()) << "'");
        BOOST_TEST_MESSAGE("Str1 = '" << W2MB(Str1.c_str()) << "'");
        // BOOST_TEST_MESSAGE("Str1 = '" << W2MB(Str1.c_str()) << "'");
        // DEBUG_PRINTF(L"str1 = \"%s\"", str1.c_str());
        BOOST_CHECK_EQUAL(0, wcscmp(str1.c_str(), L" part 1 "));

        UnicodeString str2 = ::CutToChar(Str1, '|', true);
        BOOST_TEST_MESSAGE("str2 = '" << W2MB(str2.c_str()) << "'");
        BOOST_TEST_MESSAGE("Str1 = '" << W2MB(Str1.c_str()) << "'");
        BOOST_CHECK_EQUAL(0, wcscmp(str2.c_str(), L" part 2"));
    }
    {
        UnicodeString str = ::LoadStr(CONST_TEST_STRING);
        BOOST_TEST_MESSAGE("str = " << W2MB(str.c_str()));
        BOOST_CHECK_EQUAL(0, wcscmp(str.c_str(), L"test string: \"%s\" %d"));
    }
    {
        UnicodeString str = ::LoadStrPart(CONST_TEST_STRING2, 1);
        BOOST_TEST_MESSAGE("str = " << W2MB(str.c_str()));
        BOOST_CHECK_EQUAL(0, wcscmp(str.c_str(), L"test string part 1"));
    }
    {
        UnicodeString str = ::LoadStrPart(CONST_TEST_STRING2, 2);
        BOOST_TEST_MESSAGE("str = " << W2MB(str.c_str()));
        BOOST_CHECK_EQUAL(0, wcscmp(str.c_str(), L"part 2"));
    }
}

//.........这里部分代码省略.........
                continue;
            }
            // Create canonical map entry for this alias
            entry = (CanonicalMapEntry*)uprv_malloc(sizeof(CanonicalMapEntry));
            if (entry == NULL) {
                status = U_MEMORY_ALLOCATION_ERROR;
                goto error_cleanup;
            }
            entry->id = canonical;
            if (territory  == NULL || u_strcmp(territory, gWorld) == 0) {
                entry->country = NULL;
            } else {
                entry->country = territory;
            }

            // Put this entry in the hashtable. Since this hashtable has no key deleter,
            // key is treated as const, but must be passed as non-const.
            uhash_put(canonicalMap, (UChar*)alias, entry, &status);
            if (U_FAILURE(status)) {
                goto error_cleanup;
            }
        }
    }

    // Some available Olson zones are not included in CLDR data (such as Asia/Riyadh87).
    // Also, when we update Olson tzdata, new zones may be added.
    // This code scans all available zones in zoneinfo.res, and if any of them are
    // missing, add them to the map.
    tzenum = TimeZone::createEnumeration();
    numZones = tzenum->count(status);
    if (U_SUCCESS(status)) {
        int32_t i;
        for (i = 0; i < numZones; i++) {
            const UnicodeString *zone = tzenum->snext(status);
            if (U_FAILURE(status)) {
                // We should not get here.
                status = U_ZERO_ERROR;
                continue;
            }
            UChar zoneUChars[ZID_KEY_MAX];
            int32_t zoneUCharsLen = zone->extract(zoneUChars, ZID_KEY_MAX, status) + 1; // Add one for NUL termination
            if (U_FAILURE(status) || status==U_STRING_NOT_TERMINATED_WARNING) {
                status = U_ZERO_ERROR;
                continue; // zone id is too long to extract
            }
            CanonicalMapEntry *entry = (CanonicalMapEntry*)uhash_get(canonicalMap, zoneUChars);
            if (entry) {
                // Already included in CLDR data
                continue;
            }
            // Not in CLDR data, but it could be new one whose alias is available
            // in CLDR.
            int32_t nTzdataEquivalent = TimeZone::countEquivalentIDs(*zone);
            int32_t j;
            for (j = 0; j < nTzdataEquivalent; j++) {
                UnicodeString alias = TimeZone::getEquivalentID(*zone, j);
                if (alias == *zone) {
                    continue;
                }
                UChar aliasUChars[ZID_KEY_MAX];
                alias.extract(aliasUChars, ZID_KEY_MAX, status);
                if (U_FAILURE(status) || status==U_STRING_NOT_TERMINATED_WARNING) {
                    status = U_ZERO_ERROR;
                    continue; // zone id is too long to extract
                }
                entry = (CanonicalMapEntry*)uhash_get(canonicalMap, aliasUChars);

void DataDrivenFormatTest::testConvertDate(TestData *testData,
        const DataMap * /* settings */, UBool fmt) {
    UnicodeString kPATTERN("PATTERN="); // TODO: static
    UnicodeString kMILLIS("MILLIS="); // TODO: static
    UnicodeString kRELATIVE_MILLIS("RELATIVE_MILLIS="); // TODO: static
    UnicodeString kRELATIVE_ADD("RELATIVE_ADD:"); // TODO: static

    UErrorCode status = U_ZERO_ERROR;
    SimpleDateFormat basicFmt(UnicodeString("EEE MMM dd yyyy / YYYY'-W'ww-ee"),
                              status);
    if (U_FAILURE(status)) {
        dataerrln("FAIL: Couldn't create basic SimpleDateFormat: %s",
                  u_errorName(status));
        return;
    }

    const DataMap *currentCase= NULL;
    // Start the processing
    int n = 0;
    while (testData->nextCase(currentCase, status)) {
        char calLoc[256] = "";
        DateTimeStyleSet styleSet;
        UnicodeString pattern;
        UBool usePattern = FALSE;
        (void)usePattern;   // Suppress unused warning.
        CalendarFieldsSet fromSet;
        UDate fromDate = 0;
        UBool useDate = FALSE;

        UDate now = Calendar::getNow();

        ++n;

        char theCase[200];
        sprintf(theCase, "case %d:", n);
        UnicodeString caseString(theCase, "");

        // load params
        UnicodeString locale = currentCase->getString("locale", status);
        if (U_FAILURE(status)) {
            errln("case %d: No 'locale' line.", n);
            continue;
        }
        UnicodeString zone = currentCase->getString("zone", status);
        if (U_FAILURE(status)) {
            errln("case %d: No 'zone' line.", n);
            continue;
        }
        UnicodeString spec = currentCase->getString("spec", status);
        if(U_FAILURE(status)) {
            errln("case %d: No 'spec' line.", n);
            continue;
        }
        UnicodeString date = currentCase->getString("date", status);
        if(U_FAILURE(status)) {
            errln("case %d: No 'date' line.", n);
            continue;
        }
        UnicodeString expectStr= currentCase->getString("str", status);
        if(U_FAILURE(status)) {
            errln("case %d: No 'str' line.", n);
            continue;
        }

        DateFormat *format = NULL;

        // Process: 'locale'
        locale.extract(0, locale.length(), calLoc, (const char*)0); // default codepage.  Invariant codepage doesn't have '@'!
        Locale loc(calLoc);
        if(spec.startsWith(kPATTERN)) {
            pattern = UnicodeString(spec,kPATTERN.length());
            usePattern = TRUE;
            format = new SimpleDateFormat(pattern, loc, status);
            if(U_FAILURE(status)) {
                errln("case %d: could not create SimpleDateFormat from pattern: %s", n, u_errorName(status));
                continue;
            }
        } else {
            if(styleSet.parseFrom(spec, status)<0 || U_FAILURE(status)) {
                errln("case %d: could not parse spec as style fields: %s", n, u_errorName(status));
                continue;
            }
            format = DateFormat::createDateTimeInstance((DateFormat::EStyle)styleSet.getDateStyle(), (DateFormat::EStyle)styleSet.getTimeStyle(), loc);
            if(format == NULL ) {
                errln("case %d: could not create SimpleDateFormat from styles.", n);
                continue;
            }
        }

        Calendar *cal = Calendar::createInstance(loc, status);
        if(U_FAILURE(status)) {
            errln("case %d: could not create calendar from %s", n, calLoc);
        }

        if (zone.length() > 0) {
            TimeZone * tz = TimeZone::createTimeZone(zone);
            cal->setTimeZone(*tz);
            format->setTimeZone(*tz);
            delete tz;
        }
//.........这里部分代码省略.........

void QuantityFormatterTest::TestBasic() {
    UErrorCode status = U_ZERO_ERROR;
#if !UCONFIG_NO_FORMATTING
    QuantityFormatter fmt;
    assertFalse(
            "adding bad variant",
            fmt.addIfAbsent("a bad variant", "{0} pounds", status));
    assertEquals("adding bad variant status", (int32_t)U_ILLEGAL_ARGUMENT_ERROR, status);
    status = U_ZERO_ERROR;
    assertFalse(
            "Adding bad pattern",
            fmt.addIfAbsent("other", "{0} {1} too many placeholders", status));
    assertEquals("adding bad pattern status", (int32_t)U_ILLEGAL_ARGUMENT_ERROR, status);
    status = U_ZERO_ERROR;
    assertFalse("isValid with no patterns", fmt.isValid());
    assertTrue(
            "Adding good pattern with no placeholders",
            fmt.addIfAbsent("zero", "no placeholder", status));
    assertTrue(
            "Adding good pattern",
            fmt.addIfAbsent("other", "{0} pounds", status));
    assertTrue("isValid with other", fmt.isValid());
    assertTrue(
            "Adding good pattern",
            fmt.addIfAbsent("one", "{0} pound", status));

    assertEquals(
            "getByVariant",
            fmt.getByVariant("bad variant")->getTextWithNoArguments(),
            " pounds");
    assertEquals(
            "getByVariant",
            fmt.getByVariant("other")->getTextWithNoArguments(),
            " pounds");
    assertEquals(
            "getByVariant",
            fmt.getByVariant("one")->getTextWithNoArguments(),
            " pound");
    assertEquals(
            "getByVariant",
            fmt.getByVariant("few")->getTextWithNoArguments(),
            " pounds");

    // Test copy constructor
    {
        QuantityFormatter copied(fmt);
        assertEquals(
                "copied getByVariant",
                copied.getByVariant("other")->getTextWithNoArguments(),
                " pounds");
        assertEquals(
                "copied getByVariant",
                copied.getByVariant("one")->getTextWithNoArguments(),
                " pound");
        assertEquals(
                "copied getByVariant",
                copied.getByVariant("few")->getTextWithNoArguments(),
                " pounds");
    }
        
    // Test assignment
    {
        QuantityFormatter assigned;
        assigned = fmt;
        assertEquals(
                "assigned getByVariant",
                assigned.getByVariant("other")->getTextWithNoArguments(),
                " pounds");
        assertEquals(
                "assigned getByVariant",
                assigned.getByVariant("one")->getTextWithNoArguments(),
                " pound");
        assertEquals(
                "assigned getByVariant",
                assigned.getByVariant("few")->getTextWithNoArguments(),
                " pounds");
    }

    // Test format.
    {
        LocalPointer<NumberFormat> numfmt(
                NumberFormat::createInstance(Locale::getEnglish(), status));
        LocalPointer<PluralRules> plurrule(
                PluralRules::forLocale("en", status));
        FieldPosition pos(FieldPosition::DONT_CARE);
        UnicodeString appendTo;
        assertEquals(
                "format singular",
                UnicodeString("1 pound"),
                fmt.format(
                        1.0,
                        *numfmt,
                        *plurrule,
                        appendTo,
                        pos,
                        status), TRUE);
        appendTo.remove();
        assertEquals(
                "format plural",
                UnicodeString("2 pounds"),
//.........这里部分代码省略.........

int main(int argc, const char *argv[]) {
    UErrorCode errorCode = U_ZERO_ERROR;

    // Get the unsafeBackwardsSet
    const CollationCacheEntry *rootEntry = CollationRoot::getRootCacheEntry(errorCode);
    if(U_FAILURE(errorCode)) {
      fprintf(stderr, "Err: %s getting root cache entry\n", u_errorName(errorCode));
      return 1;
    }
    const UVersionInfo &version = rootEntry->tailoring->version;
    const UnicodeSet *unsafeBackwardSet = rootEntry->tailoring->unsafeBackwardSet;
    char verString[20];
    u_versionToString(version, verString);
    fprintf(stderr, "Generating data for ICU %s, Collation %s\n", U_ICU_VERSION, verString);
    int32_t rangeCount = unsafeBackwardSet->getRangeCount();
    
#if SERIALIZE
    fprintf(stderr, ".. serializing\n");
    // UnicodeSet serialization
    
    UErrorCode preflightCode = U_ZERO_ERROR;
    // preflight
    int32_t serializedCount = unsafeBackwardSet->serialize(NULL,0,preflightCode);
    if(U_FAILURE(preflightCode) && preflightCode != U_BUFFER_OVERFLOW_ERROR) {
      fprintf(stderr, "Err: %s preflighting unicode set\n", u_errorName(preflightCode));
      return 1;
    }
    uint16_t *serializedData = new uint16_t[serializedCount];
    // serialize
    unsafeBackwardSet->serialize(serializedData, serializedCount, errorCode);
    if(U_FAILURE(errorCode)) {
      delete [] serializedData;
      fprintf(stderr, "Err: %s serializing unicodeset\n", u_errorName(errorCode));
      return 1;
    }
#endif
    
#if PATTERN
    fprintf(stderr,".. pattern. (Note: collationdatareader.cpp does not support this form also see #11891)\n");
    // attempt to use pattern
    
    UnicodeString pattern;
    UnicodeSet set(*unsafeBackwardSet);
    set.compact();
    set.toPattern(pattern, FALSE);

    if(U_SUCCESS(errorCode)) {
      // This fails (bug# ?) - which is why this method was abandoned.
      
      // UnicodeSet usA(pattern, errorCode);
      // fprintf(stderr, "\n%s:%d: err creating set A %s\n", __FILE__, __LINE__, u_errorName(errorCode));
      // return 1;
    }


    const UChar *buf = pattern.getBuffer();
    int32_t needed = pattern.length();

    // print
    {
      char buf2[2048];
      int32_t len2 = pattern.extract(0, pattern.length(), buf2, "utf-8");
      buf2[len2]=0;
      fprintf(stderr,"===\n%s\n===\n", buf2);
    }

    const UnicodeString unsafeBackwardPattern(FALSE, buf, needed);
  if(U_SUCCESS(errorCode)) {
    //UnicodeSet us(unsafeBackwardPattern, errorCode);
    //    fprintf(stderr, "\n%s:%d: err creating set %s\n", __FILE__, __LINE__, u_errorName(errorCode));
  } else {
    fprintf(stderr, "Uset OK - \n");
  }
#endif


  // Generate the output file.

  printf("// collunsafe.h\n");
  printf("// %s\n", U_COPYRIGHT_STRING);
  printf("\n");
  printf("// To be included by collationdatareader.cpp, and generated by gencolusb.\n");
  printf("// Machine generated, do not edit.\n");
  printf("\n");
  printf("#ifndef COLLUNSAFE_H\n"
         "#define COLLUNSAFE_H\n"
         "\n"
         "#include \"unicode/utypes.h\"\n"
         "\n"
         "#define COLLUNSAFE_ICU_VERSION \"" U_ICU_VERSION "\"\n");
  printf("#define COLLUNSAFE_COLL_VERSION \"%s\"\n", verString);

  
  
#if PATTERN
  printf("#define COLLUNSAFE_PATTERN 1\n");
  printf("static const int32_t collunsafe_len = %d;\n", needed);
  printf("static const UChar collunsafe_pattern[collunsafe_len] = {\n");
  for(int i=0;i<needed;i++) {
    if( (i>0) && (i%8 == 0) ) {
//.........这里部分代码省略.........

std::string GlobalizationNDK::numberToString(const std::string& args)
{
    if (args.empty()) {
        slog2f(0, ID_G11N, SLOG2_ERROR, "GlobalizationNDK::numberToString: no arguments provided!");
        return errorInJson(UNKNOWN_ERROR, "No arguments provided!");
    }

    Json::Reader reader;
    Json::Value root;
    bool parse = reader.parse(args, root);

    if (!parse) {
        slog2f(0, ID_G11N, SLOG2_ERROR, "GlobalizationNDK::numberToString: invalid json data: %s",
                args.c_str());
        return errorInJson(PARSING_ERROR, "Invalid json data!");
    }

    Json::Value nv = root["number"];
    if (nv.isNull()) {
        slog2f(0, ID_G11N, SLOG2_ERROR, "GlobalizationNDK::numberToString: no number provided!");
        return errorInJson(FORMATTING_ERROR, "No number provided!");
    }

    if (!nv.isNumeric()) {
        slog2f(0, ID_G11N, SLOG2_ERROR, "GlobalizationNDK::numberToString: invalid number type: %d!",
                nv.type());
        return errorInJson(FORMATTING_ERROR, "Invalid number type!");
    }

    // This is the default value when no options provided.
    ENumberType type = kNumberDecimal;

    Json::Value options = root["options"];
    std::string error;
    if (!handleNumberOptions(options, type, error))
        return errorInJson(PARSING_ERROR, error);

    UErrorCode status = U_ZERO_ERROR;
    NumberFormat* nf;
    switch (type) {
    case kNumberDecimal:
    default:
        nf = NumberFormat::createInstance(status);
        break;
    case kNumberCurrency:
        nf = NumberFormat::createCurrencyInstance(status);
        break;
    case kNumberPercent:
        nf = NumberFormat::createPercentInstance(status);
        break;
    }

    if (!nf) {
        slog2f(0, ID_G11N, SLOG2_ERROR, "GlobalizationNDK::numberToString: failed to create NumberFormat instance for type %d: %d",
                status, type);
        return errorInJson(UNKNOWN_ERROR, "Failed to create NumberFormat instance!");
    }
    std::auto_ptr<NumberFormat> deleter(nf);

    UnicodeString result;
    nf->format(nv.asDouble(), result);
    std::string utf8;
    result.toUTF8String(utf8);

    return resultInJson(utf8);
}

//------------------------------------------------------------------------------
//
//   findSetFor    given a UnicodeString,
//                  - find the corresponding Unicode Set  (uset node)
//                         (create one if necessary)
//                  - Set fLeftChild of the caller's node (should be a setRef node)
//                         to the uset node
//                 Maintain a hash table of uset nodes, so the same one is always used
//                    for the same string.
//                 If a "to adopt" set is provided and we haven't seen this key before,
//                    add the provided set to the hash table.
//                 If the string is one (32 bit) char in length, the set contains
//                    just one element which is the char in question.
//                 If the string is "any", return a set containing all chars.
//
//------------------------------------------------------------------------------
void RBBIRuleScanner::findSetFor(const UnicodeString &s, RBBINode *node, UnicodeSet *setToAdopt) {

    RBBISetTableEl   *el;

    // First check whether we've already cached a set for this string.
    // If so, just use the cached set in the new node.
    //   delete any set provided by the caller, since we own it.
    el = (RBBISetTableEl *)uhash_get(fSetTable, &s);
    if (el != NULL) {
        delete setToAdopt;
        node->fLeftChild = el->val;
        U_ASSERT(node->fLeftChild->fType == RBBINode::uset);
        return;
    }

    // Haven't seen this set before.
    // If the caller didn't provide us with a prebuilt set,
    //   create a new UnicodeSet now.
    if (setToAdopt == NULL) {
        if (s.compare(kAny, -1) == 0) {
            setToAdopt = new UnicodeSet(0x000000, 0x10ffff);
        } else {
            UChar32 c;
            c = s.char32At(0);
            setToAdopt = new UnicodeSet(c, c);
        }
    }

    //
    // Make a new uset node to refer to this UnicodeSet
    // This new uset node becomes the child of the caller's setReference node.
    //
    RBBINode *usetNode    = new RBBINode(RBBINode::uset);
    if (usetNode == NULL) {
        error(U_MEMORY_ALLOCATION_ERROR);
        return;
    }
    usetNode->fInputSet   = setToAdopt;
    usetNode->fParent     = node;
    node->fLeftChild      = usetNode;
    usetNode->fText = s;


    //
    // Add the new uset node to the list of all uset nodes.
    //
    fRB->fUSetNodes->addElement(usetNode, *fRB->fStatus);


    //
    // Add the new set to the set hash table.
    //
    el      = (RBBISetTableEl *)uprv_malloc(sizeof(RBBISetTableEl));
    UnicodeString *tkey = new UnicodeString(s);
    if (tkey == NULL || el == NULL || setToAdopt == NULL) {
        // Delete to avoid memory leak
        delete tkey;
        tkey = NULL;
        uprv_free(el);
        el = NULL;
        delete setToAdopt;
        setToAdopt = NULL;

        error(U_MEMORY_ALLOCATION_ERROR);
        return;
    }
    el->key = tkey;
    el->val = usetNode;
    uhash_put(fSetTable, el->key, el, fRB->fStatus);

    return;
}