Quelle  numparse_currency.cpp   Sprache: C

// © 2018 and later: Unicode, Inc. and others.
// License & terms of use: http://www.unicode.org/copyright.html

#include "unicode/utypes.h"

#if !UCONFIG_NO_FORMATTING

// Allow implicit conversion from char16_t* to UnicodeString for this file:
// Helpful in toString methods and elsewhere.
#define UNISTR_FROM_STRING_EXPLICIT

#include "numparse_types.h"
#include "numparse_currency.h"
#include "ucurrimp.h"
#include "unicode/errorcode.h"
#include "numparse_utils.h"
#include "string_segment.h"

using namespace icu;
using namespace icu::numparse;
using namespace icu::numparse::impl;


CombinedCurrencyMatcher::CombinedCurrencyMatcher(const CurrencySymbols& currencySymbols, const DecimalFormatSymbols& dfs,
                                                 parse_flags_t parseFlags, UErrorCode& status)
        : fCurrency1(currencySymbols.getCurrencySymbol(status)),
          fCurrency2(currencySymbols.getIntlCurrencySymbol(status)),
          fUseFullCurrencyData(0 == (parseFlags & PARSE_FLAG_NO_FOREIGN_CURRENCY)),
          afterPrefixInsert(dfs.getPatternForCurrencySpacing(UNUM_CURRENCY_INSERT, false, status)),
          beforeSuffixInsert(dfs.getPatternForCurrencySpacing(UNUM_CURRENCY_INSERT, true, status)),
          fLocaleName(dfs.getLocale().getName(), -1, status) {
    utils::copyCurrencyCode(fCurrencyCode, currencySymbols.getIsoCode());

    // Pre-load the long names for the current locale and currency
    // if we are parsing without the full currency data.
    if (!fUseFullCurrencyData) {
        for (int32_t i=0; i<StandardPlural::COUNT; i++) {
            auto plural = static_cast<StandardPlural::Form>(i);
            fLocalLongNames[i] = currencySymbols.getPluralName(plural, status);
        }
    }

    // TODO: Figure out how to make this faster and re-enable.
    // Computing the "lead code points" set for fastpathing is too slow to use in production.
    // See https://unicode-org.atlassian.net/browse/ICU-13584
//    // Compute the full set of characters that could be the first in a currency to allow for
//    // efficient smoke test.
//    fLeadCodePoints.add(fCurrency1.char32At(0));
//    fLeadCodePoints.add(fCurrency2.char32At(0));
//    fLeadCodePoints.add(beforeSuffixInsert.char32At(0));
//    uprv_currencyLeads(fLocaleName.data(), fLeadCodePoints, status);
//    // Always apply case mapping closure for currencies
//    fLeadCodePoints.closeOver(USET_ADD_CASE_MAPPINGS);
//    fLeadCodePoints.freeze();
}

bool
CombinedCurrencyMatcher::match(StringSegment& segment, ParsedNumber& result, UErrorCode& status) const {
    if (result.currencyCode[0] != 0) {
        return false;
    }

    // Try to match a currency spacing separator.
    int32_t initialOffset = segment.getOffset();
    bool maybeMore = false;
    if (result.seenNumber() && !beforeSuffixInsert.isEmpty()) {
        int32_t overlap = segment.getCommonPrefixLength(beforeSuffixInsert);
        if (overlap == beforeSuffixInsert.length()) {
            segment.adjustOffset(overlap);
            // Note: let currency spacing be a weak match. Don't update chars consumed.
        }
        maybeMore = maybeMore || overlap == segment.length();
    }

    // Match the currency string, and reset if we didn't find one.
    maybeMore = maybeMore || matchCurrency(segment, result, status);
    if (result.currencyCode[0] == 0) {
        segment.setOffset(initialOffset);
        return maybeMore;
    }

    // Try to match a currency spacing separator.
    if (!result.seenNumber() && !afterPrefixInsert.isEmpty()) {
        int32_t overlap = segment.getCommonPrefixLength(afterPrefixInsert);
        if (overlap == afterPrefixInsert.length()) {
            segment.adjustOffset(overlap);
            // Note: let currency spacing be a weak match. Don't update chars consumed.
        }
        maybeMore = maybeMore || overlap == segment.length();
    }

    return maybeMore;
}

bool CombinedCurrencyMatcher::matchCurrency(StringSegment& segment, ParsedNumber& result,
                                            UErrorCode& status) const {
    bool maybeMore = false;

    int32_t overlap1;
    if (!fCurrency1.isEmpty()) {
        overlap1 = segment.getCaseSensitivePrefixLength(fCurrency1);
    } else {
        overlap1 = -1;
    }
    maybeMore = maybeMore || overlap1 == segment.length();
    if (overlap1 == fCurrency1.length()) {
        utils::copyCurrencyCode(result.currencyCode, fCurrencyCode);
        segment.adjustOffset(overlap1);
        result.setCharsConsumed(segment);
        return maybeMore;
    }

    int32_t overlap2;
    if (!fCurrency2.isEmpty()) {
        // ISO codes should be accepted case-insensitive.
        // https://unicode-org.atlassian.net/browse/ICU-13696
        overlap2 = segment.getCommonPrefixLength(fCurrency2);
    } else {
        overlap2 = -1;
    }
    maybeMore = maybeMore || overlap2 == segment.length();
    if (overlap2 == fCurrency2.length()) {
        utils::copyCurrencyCode(result.currencyCode, fCurrencyCode);
        segment.adjustOffset(overlap2);
        result.setCharsConsumed(segment);
        return maybeMore;
    }

    if (fUseFullCurrencyData) {
        // Use the full currency data.
        // NOTE: This call site should be improved with #13584.
        const UnicodeString segmentString = segment.toTempUnicodeString();

        // Try to parse the currency
        ParsePosition ppos(0);
        int32_t partialMatchLen = 0;
        uprv_parseCurrency(
                fLocaleName.data(),
                segmentString,
                ppos,
                UCURR_SYMBOL_NAME, // checks for both UCURR_SYMBOL_NAME and UCURR_LONG_NAME
                &partialMatchLen,
                result.currencyCode,
                status);
        maybeMore = maybeMore || partialMatchLen == segment.length();

        if (U_SUCCESS(status) && ppos.getIndex() != 0) {
            // Complete match.
            // NOTE: The currency code should already be saved in the ParsedNumber.
            segment.adjustOffset(ppos.getIndex());
            result.setCharsConsumed(segment);
            return maybeMore;
        }

    } else {
        // Use the locale long names.
        int32_t longestFullMatch = 0;
        for (int32_t i=0; i<StandardPlural::COUNT; i++) {
            const UnicodeString& name = fLocalLongNames[i];
            int32_t overlap = segment.getCommonPrefixLength(name);
            if (overlap == name.length() && name.length() > longestFullMatch) {
                longestFullMatch = name.length();
            }
            maybeMore = maybeMore || overlap > 0;
        }
        if (longestFullMatch > 0) {
            utils::copyCurrencyCode(result.currencyCode, fCurrencyCode);
            segment.adjustOffset(longestFullMatch);
            result.setCharsConsumed(segment);
            return maybeMore;
        }
    }

    // No match found.
    return maybeMore;
}

bool CombinedCurrencyMatcher::smokeTest(const StringSegment&) const {
    // TODO: See constructor
    return true;
    //return segment.startsWith(fLeadCodePoints);
}

UnicodeString CombinedCurrencyMatcher::toString() const {
    return u"";
}


#endif /* #if !UCONFIG_NO_FORMATTING */