如何将雪花时间戳转换为日期

Question

can anyone help me suggest how can i convert snowflake ltz timestamp to given date using java?

Sample in snowflake ltz format: 1658443006.285000000

Answer 1

Use smart objects, not dumb text

While I do not use Snowflake, a quick look at the documentation shows that their TIMESTAMP_LTZ type is akin to the SQL standard type TIMESTAMP WITH TIME ZONE .

java.time.OffsetDateTime

So the appropriate class in Java is OffsetDateTime .

Assuming the Snowflake JDBC driver complies with JDBC 4.2 and later, you can extract an object rather than mere text from the database.

OffsetDateTime odt = myResultSet.getObject( … , OffsetDateTime.class ) ;

When writing to the database:

myPreparedStatement.setObject( … , odt ) ;

If you want to view this moment as seen in a particular time zone, apply a ZoneId to get a ZonedDateTime .

ZoneId z = ZoneId.of( "Asia/Tokyo" ) ;
ZonedDateTime zdt = odt.atZoneSameInstant( z ) ;

Parsing text

You said:

Sample in snowflake ltz format: 1658443006.285000000

Can you cite the documentation on that? I see no such format in their page on TIMESTAMP_LTZ data type.

At any rate, I imagine the text you give represents a number of whole seconds since an epoch reference, along with a fractional second with nanosecond resolution.

Assuming the epoch reference is the first moment of 1970 as seen with an offset of zero hours-minutes-seconds from UTC (1970-01-01T00:00Z), we can easily parse that as a java.time.Instant object in Java. But I strongly urge you to use objects instead of this approach, as discussed above.

Split the input into two parts, two strings. Parse each a 64-bit long integer number.

To do the split, we must escape the FULL STOP character with a pair of REVERSE SOLIDUS characters.

String[] parts = "1658443006.285000000".split( "\\." ) ;
long seconds = Long.parseLong( parts [ 0 ] ) ;
long nanos = Long.parseLong( parts [ 1 ] ) ;

Use a factory method to instantiate a Instant object.

Instant instant = Instant.ofEpochSecond( seconds , nanos ) ;

See this code run live at Ideone.com .

instant.toString() → 2022-07-21T22:36:46.285Z

An Instant is in UTC , always in UTC, meaning an offset of zero.

If you want to view this moment as seen in a particular time zone, apply a ZoneId to get a ZonedDateTime .

ZoneId z = ZoneId.of( "Asia/Tokyo" ) ;
ZonedDateTime zdt = instant.atZone( z ) ;

zit.toString() → 2022-07-22T07:36:46.285+09:00[Asia/Tokyo]