Parsing Instant using OpenCsv

Question

I am trying to parse Instant from a CSV using OpenCsv this way:

@CsvDate("yyyy-MM-dd hh:mm:ss")
@CsvBindByName(column = "date")
private Instant date;

I know that OpenCsv is supposed to support java.time.

But when trying to use it I am getting the following exception:

Error parsing CSV line: 8.

...

Caused by: java.time.format.DateTimeParseException: Text '2022-04-21 00:00:00' could not be parsed: Unable to obtain Instant from TemporalAccessor: {HourOfAmPm=0, MicroOfSecond=0, NanoOfSecond=0, MilliOfSecond=0, MinuteOfHour=0, SecondOfMinute=0},ISO resolved to 2022-04-21 of type java.time.format.Parsed

...

Caused by: java.time.DateTimeException: Unable to obtain Instant from TemporalAccessor: {HourOfAmPm=0, MicroOfSecond=0, NanoOfSecond=0, MilliOfSecond=0, MinuteOfHour=0, SecondOfMinute=0},ISO resolved to 2022-04-21 of type java.time.format.Parsed

...

Caused by: java.time.temporal.UnsupportedTemporalTypeException: Unsupported field: InstantSeconds

What comes from my investigation, is that the root cause is the lack of a time zone, but my question is how do I specify the time zone using the annotation only?

An Instant is a point in time independent of time zone and therefore neither defines year, month, day-of-month. hour, minute nor second. So trying to parse it using a format holding those elements does not make sense, at least not if you don’t specify the time zone or offset from UTC. — Anonymous
– Anonymous, Commented Apr 6, 2022 at 18:49
That is a good point, but as I was saying in the question I understand that the issue here is the lack of a time zone. So, you can say that my question is more about - is it possible to specify the time zone in OpenCsv + @CsvDate. Also, I'd assume that it would get a default UTC time zone if not supplied (though it seems like it isn't). — Tom Carmi
– Tom Carmi, Commented Apr 7, 2022 at 11:26

Joop Eggen · Accepted Answer · 2022-04-06 19:39:43Z

2

See DateTimeFormatter. It should be HH for 24 hours time (0-23) i.o. hh for 12 hours time (1-12). The error is caused by hour "00". Fortunately, otherwise you would have only the first half of the day.

See h, k, H and K.

The class Instant does not use units like seconds, is more a long internally. Use LocalDateTime for your format.

edited Apr 6, 2022 at 19:39

answered Apr 6, 2022 at 17:58

Joop Eggen

110k8 gold badges89 silver badges142 bronze badges

Sign up to request clarification or add additional context in comments.

4 Comments

Tom Carmi Over a year ago

I've changed it to HH, but still having the same error

Joop Eggen Over a year ago

Would you try it with private LocalDateTime date; i.o. Instant?

Tom Carmi Over a year ago

This one did the trick, I wonder why Instant does not work... Thx a lot!!

Joop Eggen Over a year ago

Extended the answer. Already thought that Instant is a too amorph class. The same as adding time HH:mm for a LocalDate (no time).

nyg · Accepted Answer · 2023-06-03 00:03:01Z

What comes from my investigation, is that the root cause is the lack of a time zone, but my question is how do I specify the time zone using the annotation only?

I haven't find a way to change the timezone of the DateTimeFormatter used by OpenCSV. You could, in theory, use @CsvCustomBindByName and provide a custom converter in which you can use your own DateTimeFormatter with the timezone you want.

However, if you also control how your CSV is generated, you can simply use the following datetime patterns to make it work with Instant:

@CsvBindByName(column = "date")
@CsvDate(
        value = "yyyy-MM-dd'T'HH:mm:ssX",
        writeFormat = "yyyy-MM-dd'T'HH:mm:ss'Z'",
        writeFormatEqualsReadFormat = false)
private Instant date;

value is the pattern used when reading a bean from a CSV file. X is a pattern letter which will corresponds to the zone offset, i.e. when reading the 'Z' character, it will assume the time to be in UTC+00:00. See doc.

When writing a bean to a CSV file, the writeFormat pattern will be used and needs to be different, you need to print the 'Z' char manually at the end, hence the escaped 'Z'. The reason is that internally OpenCSV does the following when writing an Instant:

LocalDateTime ldt = LocalDateTime.ofInstant((Instant) value, ZoneId.of("UTC"));
return writeDtf.format(ldt);

LocalDateTime does not have a timezone, so the X cannot be used. Arguably, they could instead do the following:

return writeDtf.withZone(ZoneOffset.UTC).format(value);

Which would then allow us to use the X. They could also do the same when parsing a date to an Instant, in order to be consistent.

Collectives™ on Stack Overflow

Parsing Instant using OpenCsv

2 Answers 2

4 Comments

Comments

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

2 Answers 2

4 Comments

Comments

Your Answer

Sign up or log in

Post as a guest

Related