[FEA] Support parsing dates and timestamps with timezone IDs in the string

res-life commented 10 months ago

Is your feature request related to a problem? Please describe. This is from https://github.com/NVIDIA/spark-rapids/issues/6846 Needs a kernel to parse dates and timestamps with timezone IDs in the string

[ ] Parse to timestamp
[ ] Parse Just time https://github.com/NVIDIA/spark-rapids/issues/10221
[ ] Parse to date

[ ] Check boundary/overflow of long value. Spark saves ts as long. Spark code uses Math.addExact to check overflow:

def instantToMicros(instant: Instant): Long = {
val secs = instant.getEpochSecond
if (secs == MIN_SECONDS) {
  val us = Math.multiplyExact(secs + 1, MICROS_PER_SECOND)
  Math.addExact(us, NANOSECONDS.toMicros(instant.getNano) - MICROS_PER_SECOND)
} else {
  val us = Math.multiplyExact(secs, MICROS_PER_SECOND)
  Math.addExact(us, NANOSECONDS.toMicros(instant.getNano))
}
}

res-life commented 10 months ago

Alfred is co-working on this.

res-life commented 7 months ago

I'm not planing work on this for release 24.04 @sameerz can we move it to next release?

NVIDIA / spark-rapids-jni

[FEA] Support parsing dates and timestamps with timezone IDs in the string #1654