cloud-fan commented on a change in pull request #32940:
URL: https://github.com/apache/spark/pull/32940#discussion_r656867600
##########
File path:
sql/catalyst/src/main/scala/org/apache/spark/sql/catalyst/util/IntervalUtils.scala
##########
@@ -105,25 +105,46 @@ object IntervalUtils {
private val yearMonthRegex = (s"^$yearMonthPatternString$$").r
private val yearMonthLiteralRegex =
(s"(?i)^INTERVAL\\s+([+|-])?'$yearMonthPatternString'\\s+YEAR\\s+TO\\s+MONTH$$").r
+ private val yearMonthIndividualLiteralRegex =
+ (s"(?i)^INTERVAL\\s+([+|-])?'([+|-])?(\\d+)'\\s+(YEAR|MONTH)$$").r
def castStringToYMInterval(
input: UTF8String,
- // TODO(SPARK-35768): Take into account year-month interval fields in
cast
startField: Byte,
endField: Byte): Int = {
+
+ def truncatedMonth(month: String) : String = {
+ if (endField == YM.YEAR) "0" else month
+ }
+
+ def getSigh(firstSign: String, secondSign: String): Int = {
+ (firstSign, secondSign) match {
+ case ("-", "-") => 1
+ case ("-", _) => -1
+ case (_, "-") => -1
+ case (_, _) => 1
+ }
+ }
+
input.trimAll().toString match {
- case yearMonthRegex("-", year, month) => toYMInterval(year, month, -1)
- case yearMonthRegex(_, year, month) => toYMInterval(year, month, 1)
+ case yearMonthRegex("-", year, month) => toYMInterval(year,
truncatedMonth(month), -1)
+ case yearMonthRegex(_, year, month) => toYMInterval(year,
truncatedMonth(month), 1)
case yearMonthLiteralRegex(firstSign, secondSign, year, month) =>
- (firstSign, secondSign) match {
- case ("-", "-") => toYMInterval(year, month, 1)
- case ("-", _) => toYMInterval(year, month, -1)
- case (_, "-") => toYMInterval(year, month, -1)
- case (_, _) => toYMInterval(year, month, 1)
+ toYMInterval(year, truncatedMonth(month), getSigh(firstSign,
secondSign))
+ case yearMonthIndividualLiteralRegex(firstSign, secondSign, value,
suffix) =>
+ suffix match {
+ case "YEAR" => toYMInterval(value, "0", getSigh(firstSign,
secondSign))
+ case _ =>
+ val months = value.toLong
+ val year = (months / 12).toString
+ val month = (months % 12).toString
Review comment:
This looks a bit inefficient. Under the hood, we just need the number of
months for YM interval, and it's a waste to calculate year and month and then
go back to num months.
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
[email protected]
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
