ICU Parsing/Formatting ISO 8601 date times

icu-iso8601-test.cpp

#include <windows.h>
#include <icu.h>
#include <memory>
#include <functional>

// A unique pointer with a custom deleter method for wrapping the opaque ICU objects.
template<typename T>
using deleted_unique_ptr = std::unique_ptr<T, std::function<void(T*)>>;

int wmain()
{
    wprintf(L"Hello World!\n");
    
    UErrorCode status = U_ZERO_ERROR;
    
    /*
        From: https://www.w3.org/TR/NOTE-datetime

        Complete date plus hours, minutes and seconds:
          YYYY-MM-DDThh:mm:ssTZD (eg 1997-07-16T19:20:30+01:00)

        Complete date plus hours, minutes, seconds and a decimal fraction of a second
          YYYY-MM-DDThh:mm:ss.sTZD (eg 1997-07-16T19:20:30.45+01:00)
    */
    
    // In order to handle both styles, we create two parsers and fallback to the second one if the first one fails.
    // Note: Alternatively we could also check the string length of the input date/time as well.

    UChar patternWithMilliseconds[] = u"yyyy-MM-dd'T'HH:mm:ss.SSSZ";
    UChar pattern[] = u"yyyy-MM-dd'T'HH:mm:ssZ";

    deleted_unique_ptr<UDateFormat> parser1(
        udat_open(UDAT_PATTERN, UDAT_PATTERN, "", u"UTC", -1, patternWithMilliseconds, -1, &status),
        [](auto* p) { udat_close(p); }
    );
    if (U_FAILURE(status)) {
        wprintf(L"Failed to create parser1.\n");
        return -1;
    }

    deleted_unique_ptr<UDateFormat> parser2(
        udat_open(UDAT_PATTERN, UDAT_PATTERN, "", u"UTC", -1, pattern, -1, &status),
        [](auto* p) { udat_close(p); }
    );
    if (U_FAILURE(status)) {
        wprintf(L"Failed to create parser2.\n");
        return -1;
    }

    // array of ICU parsers
    auto parsers = {
            parser1.get(),
            parser2.get()
    };

    // dates to parse
    auto dates = {
            u"1994-11-05T13:15:30Z",
            u"2016-03-31T15:04:32.049Z",
            u"2017-10-30T15:05:33Z",
            u"1985-04-12T23:20:30.1234567-03",
            u"2012-01-01T12:34:56+09:00"
    };

    // for testing we'll output the date/time again.
    deleted_unique_ptr<UDateFormat> formatter(
        udat_open(UDAT_PATTERN, UDAT_PATTERN, "", u"UTC", -1, patternWithMilliseconds, -1, &status),
        [](auto* p) { udat_close(p); }
    );
    if (U_FAILURE(status)) {
        wprintf(L"Failed to create formatter.\n");
        return -1;
    }

    for (const auto& date : dates) {
        // Note: ICU's UChar is char16_t which is the same as wchar_t on Windows.
        wprintf(L"\nparsing date: %s\n", reinterpret_cast<const wchar_t*>(date));

        UDate parsedDate = {};
        
        for (const auto& parser : parsers) {
            status = U_ZERO_ERROR;
            parsedDate = udat_parse(parser, date, -1, 0, &status);

            if (U_FAILURE(status)) {
                wprintf(L" Note: falling back to second parser.\n");
            } else {
                break;
            }
        }
        if (U_FAILURE(status)) {
            wprintf(L"ERROR: Could not parse the date! -- Error: %hs \n", u_errorName(status));
            continue;
        }

        UChar buffer[200] = {};
        int32_t len = udat_format(formatter.get(), parsedDate, buffer, ARRAYSIZE(buffer), 0, &status);
        if (U_FAILURE(status)) {
            wprintf(L"Failed to format the date! -- Error: %hs \n", u_errorName(status));
        }

        if (len >= ARRAYSIZE(buffer)) {
            wprintf(L"The formatted date is too big to fit!\n");
        }

        wprintf(L"Formatted: %s \n", reinterpret_cast<const wchar_t*>(buffer));
    }

    return 0;
}


/*
Output:
Hello World!

parsing date: 1994-11-05T13:15:30Z
 Note: falling back to second parser.
Formatted: 1994-11-05T13:15:30.000+0000

parsing date: 2016-03-31T15:04:32.049Z
Formatted: 2016-03-31T15:04:32.049+0000

parsing date: 2017-10-30T15:05:33Z
 Note: falling back to second parser.
Formatted: 2017-10-30T15:05:33.000+0000

parsing date: 1985-04-12T23:20:30.1234567-03
Formatted: 1985-04-13T02:20:30.123+0000

parsing date: 2012-01-01T12:34:56+09:00
 Note: falling back to second parser.
Formatted: 2012-01-01T03:34:56.000+0000
*/