-
Notifications
You must be signed in to change notification settings - Fork 29k
[SPARK-20754][SQL] Support TRUNC (number) #18106
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Changes from 13 commits
a5ade70
c63856b
224c867
e7e6e5b
7157820
c1019c9
3d92a48
b391b6a
5456e61
d40a46f
88d1e38
3fd4189
7fee61b
f8b1f44
3d40c36
931f07d
ea72fe0
b59a2df
679ff98
File filter
Filter by extension
Conversations
Jump to
Diff view
Diff view
There are no files selected for viewing
| Original file line number | Diff line number | Diff line change |
|---|---|---|
|
|
@@ -1028,20 +1028,28 @@ def to_timestamp(col, format=None): | |
|
|
||
|
|
||
| @since(1.5) | ||
| def trunc(date, format): | ||
| def trunc(data, format): | ||
| """ | ||
| Returns date truncated to the unit specified by the format. | ||
| Returns date truncated to the unit specified by the format or | ||
| number truncated by specified decimal places. | ||
|
||
|
|
||
| :param format: 'year', 'YYYY', 'yy' or 'month', 'mon', 'mm' | ||
|
||
|
|
||
| >>> df = spark.createDataFrame([('1997-02-28',)], ['d']) | ||
| >>> df.select(trunc(df.d, 'year').alias('year')).collect() | ||
| >>> df.select(trunc(to_date(df.d), 'year').alias('year')).collect() | ||
| [Row(year=datetime.date(1997, 1, 1))] | ||
| >>> df.select(trunc(df.d, 'mon').alias('month')).collect() | ||
| >>> df.select(trunc(to_date(df.d), 'mon').alias('month')).collect() | ||
|
||
| [Row(month=datetime.date(1997, 2, 1))] | ||
| >>> df = spark.createDataFrame([(1234567891.1234567891,)], ['d']) | ||
| >>> df.select(trunc(df.d, 4).alias('positive')).collect() | ||
| [Row(positive=1234567891.1234)] | ||
| >>> df.select(trunc(df.d, -4).alias('negative')).collect() | ||
| [Row(negative=1234560000.0)] | ||
| >>> df.select(trunc(df.d, 0).alias('zero')).collect() | ||
| [Row(zero=1234567891.0)] | ||
| """ | ||
| sc = SparkContext._active_spark_context | ||
| return Column(sc._jvm.functions.trunc(_to_java_column(date), format)) | ||
| return Column(sc._jvm.functions.trunc(_to_java_column(data), format)) | ||
|
|
||
|
|
||
| @since(1.5) | ||
|
|
||
| Original file line number | Diff line number | Diff line change |
|---|---|---|
|
|
@@ -21,6 +21,7 @@ import java.util.UUID | |
|
|
||
| import org.apache.spark.sql.catalyst.InternalRow | ||
| import org.apache.spark.sql.catalyst.expressions.codegen._ | ||
| import org.apache.spark.sql.catalyst.util.{BigDecimalUtils, DateTimeUtils} | ||
| import org.apache.spark.sql.types._ | ||
| import org.apache.spark.unsafe.types.UTF8String | ||
|
|
||
|
|
@@ -132,3 +133,154 @@ case class Uuid() extends LeafExpression { | |
| s"UTF8String.fromString(java.util.UUID.randomUUID().toString());", isNull = "false") | ||
| } | ||
| } | ||
|
|
||
| /** | ||
| * Returns date truncated to the unit specified by the format or | ||
| * numeric truncated to scale decimal places. | ||
| */ | ||
| // scalastyle:off line.size.limit | ||
| @ExpressionDescription( | ||
| usage = """ | ||
| _FUNC_(data[, fmt]) - Returns `data` truncated by the format model `fmt`. | ||
|
||
| If `data` is DateType, returns `data` with the time portion of the day truncated to the unit specified by the format model `fmt`. | ||
| If `data` is DecimalType/DoubleType, returns `data` truncated to `fmt` decimal places. | ||
|
||
| """, | ||
| extended = """ | ||
| Examples: | ||
| > SELECT _FUNC_('2009-02-12', 'MM'); | ||
|
Member
There was a problem hiding this comment. Choose a reason for hiding this commentThe reason will be displayed to describe this comment to others. Learn more. As it doesn't extends
Member
There was a problem hiding this comment. Choose a reason for hiding this commentThe reason will be displayed to describe this comment to others. Learn more. And I don't think we should drop this support.
Member
There was a problem hiding this comment. Choose a reason for hiding this commentThe reason will be displayed to describe this comment to others. Learn more. I guess this also drops the support of other types (e.g., timestamp) basically as we don't allow implicit cast (e.g,
Member
Author
There was a problem hiding this comment. Choose a reason for hiding this commentThe reason will be displayed to describe this comment to others. Learn more. Yes, This is what I worry about, because |
||
| 2009-02-01. | ||
| > SELECT _FUNC_('2015-10-27', 'YEAR'); | ||
| 2015-01-01 | ||
| > SELECT _FUNC_('1989-03-13'); | ||
| 1989-03-01 | ||
| > SELECT _FUNC_(1234567891.1234567891, 4); | ||
| 1234567891.1234 | ||
| > SELECT _FUNC_(1234567891.1234567891, -4); | ||
| 1234560000 | ||
| > SELECT _FUNC_(1234567891.1234567891); | ||
| 1234567891 | ||
| """) | ||
| // scalastyle:on line.size.limit | ||
| case class Trunc(data: Expression, format: Expression) | ||
|
||
| extends BinaryExpression with ExpectsInputTypes { | ||
|
|
||
| def this(data: Expression) = { | ||
| this(data, Literal(if (data.dataType.isInstanceOf[DateType]) "MM" else 0)) | ||
| } | ||
|
|
||
| override def left: Expression = data | ||
| override def right: Expression = format | ||
|
|
||
| override def dataType: DataType = data.dataType | ||
|
|
||
| override def inputTypes: Seq[AbstractDataType] = dataType match { | ||
| case NullType => Seq(dataType, TypeCollection(StringType, IntegerType)) | ||
| case DateType => Seq(dataType, StringType) | ||
| case DoubleType | DecimalType.Fixed(_, _) => Seq(dataType, IntegerType) | ||
| case _ => Seq(TypeCollection(DateType, DoubleType, DecimalType), | ||
|
||
| TypeCollection(StringType, IntegerType)) | ||
|
||
| } | ||
|
|
||
| override def nullable: Boolean = true | ||
|
|
||
| override def prettyName: String = "trunc" | ||
|
|
||
| private val isTruncNumber = | ||
| (dataType.isInstanceOf[DoubleType] || dataType.isInstanceOf[DecimalType]) && | ||
| format.dataType.isInstanceOf[IntegerType] | ||
|
||
| private val isTruncDate = | ||
| dataType.isInstanceOf[DateType] && format.dataType.isInstanceOf[StringType] | ||
|
|
||
| private lazy val truncFormat: Int = if (isTruncNumber) { | ||
| format.eval().asInstanceOf[Int] | ||
| } else if (isTruncDate) { | ||
| DateTimeUtils.parseTruncLevel(format.eval().asInstanceOf[UTF8String]) | ||
| } else { | ||
| 0 | ||
| } | ||
|
|
||
| override def eval(input: InternalRow): Any = { | ||
|
Member
There was a problem hiding this comment. Choose a reason for hiding this commentThe reason will be displayed to describe this comment to others. Learn more. override |
||
| val d = data.eval(input) | ||
| val form = format.eval() | ||
| if (null == d || null == form) { | ||
| null | ||
| } else { | ||
| if (isTruncNumber) { | ||
| val scale = if (format.foldable) truncFormat else format.eval().asInstanceOf[Int] | ||
| data.dataType match { | ||
| case DoubleType => BigDecimalUtils.trunc(d.asInstanceOf[Double], scale) | ||
| case DecimalType.Fixed(_, _) => | ||
| BigDecimalUtils.trunc(d.asInstanceOf[Decimal].toJavaBigDecimal, scale) | ||
| } | ||
| } else if (isTruncDate) { | ||
| val level = if (format.foldable) { | ||
| truncFormat | ||
| } else { | ||
| DateTimeUtils.parseTruncLevel(format.eval().asInstanceOf[UTF8String]) | ||
| } | ||
| if (level == -1) { | ||
| // unknown format | ||
| null | ||
| } else { | ||
| DateTimeUtils.truncDate(d.asInstanceOf[Int], level) | ||
| } | ||
| } else { | ||
| null | ||
| } | ||
| } | ||
| } | ||
|
|
||
| override def doGenCode(ctx: CodegenContext, ev: ExprCode): ExprCode = { | ||
|
|
||
| if (isTruncNumber) { | ||
| val bdu = BigDecimalUtils.getClass.getName.stripSuffix("$") | ||
|
|
||
| if (format.foldable) { | ||
| val d = data.genCode(ctx) | ||
| ev.copy(code = s""" | ||
| ${d.code} | ||
| boolean ${ev.isNull} = ${d.isNull}; | ||
| ${ctx.javaType(dataType)} ${ev.value} = ${ctx.defaultValue(dataType)}; | ||
| if (!${ev.isNull}) { | ||
| ${ev.value} = $bdu.trunc(${d.value}, $truncFormat); | ||
| }""") | ||
| } else { | ||
| nullSafeCodeGen(ctx, ev, (doubleVal, fmt) => s"${ev.value} = $bdu.trunc($doubleVal, $fmt);") | ||
| } | ||
| } else if (isTruncDate) { | ||
| val dtu = DateTimeUtils.getClass.getName.stripSuffix("$") | ||
|
|
||
| if (format.foldable) { | ||
| if (truncFormat == -1) { | ||
| ev.copy(code = s""" | ||
| boolean ${ev.isNull} = true; | ||
| ${ctx.javaType(dataType)} ${ev.value} = ${ctx.defaultValue(dataType)}; | ||
| """) | ||
| } else { | ||
| val d = data.genCode(ctx) | ||
| ev.copy(code = s""" | ||
| ${d.code} | ||
| boolean ${ev.isNull} = ${d.isNull}; | ||
| ${ctx.javaType(dataType)} ${ev.value} = ${ctx.defaultValue(dataType)}; | ||
| if (!${ev.isNull}) { | ||
| ${ev.value} = $dtu.truncDate(${d.value}, $truncFormat); | ||
| }""") | ||
| } | ||
| } else { | ||
| nullSafeCodeGen(ctx, ev, (dateVal, fmt) => { | ||
| val form = ctx.freshName("form") | ||
| s""" | ||
| int $form = $dtu.parseTruncLevel($fmt); | ||
| if ($form == -1) { | ||
| ${ev.isNull} = true; | ||
| } else { | ||
| ${ev.value} = $dtu.truncDate($dateVal, $form); | ||
| } | ||
| """ | ||
| }) | ||
| } | ||
| } else { | ||
| nullSafeCodeGen(ctx, ev, (dateVal, fmt) => s"${ev.isNull} = true;") | ||
| } | ||
| } | ||
| } | ||
| Original file line number | Diff line number | Diff line change |
|---|---|---|
| @@ -0,0 +1,57 @@ | ||
| /* | ||
| * Licensed to the Apache Software Foundation (ASF) under one or more | ||
| * contributor license agreements. See the NOTICE file distributed with | ||
| * this work for additional information regarding copyright ownership. | ||
| * The ASF licenses this file to You under the Apache License, Version 2.0 | ||
| * (the "License"); you may not use this file except in compliance with | ||
| * the License. You may obtain a copy of the License at | ||
| * | ||
| * http://www.apache.org/licenses/LICENSE-2.0 | ||
| * | ||
| * Unless required by applicable law or agreed to in writing, software | ||
| * distributed under the License is distributed on an "AS IS" BASIS, | ||
| * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. | ||
| * See the License for the specific language governing permissions and | ||
| * limitations under the License. | ||
| */ | ||
|
|
||
| package org.apache.spark.sql.catalyst.util | ||
|
|
||
| import java.math.{BigDecimal => JBigDecimal} | ||
|
|
||
| /** | ||
| * Helper functions for BigDecimal. | ||
| */ | ||
| object BigDecimalUtils { | ||
|
|
||
| /** | ||
| * Returns double type input truncated to scale decimal places. | ||
| */ | ||
| def trunc(input: Double, scale: Int): Double = { | ||
| trunc(JBigDecimal.valueOf(input), scale).doubleValue() | ||
| } | ||
|
|
||
| /** | ||
| * Returns BigDecimal type input truncated to scale decimal places. | ||
| */ | ||
| def trunc(input: JBigDecimal, scale: Int): JBigDecimal = { | ||
| // Copy from (https://github.com/apache/hive/blob/release-2.3.0-rc0 | ||
| // /ql/src/java/org/apache/hadoop/hive/ql/udf/generic/GenericUDFTrunc.java#L471-L487) | ||
| val pow = if (scale >= 0) { | ||
| JBigDecimal.valueOf(Math.pow(10, scale)) | ||
| } else { | ||
| JBigDecimal.valueOf(Math.pow(10, Math.abs(scale))) | ||
| } | ||
|
|
||
| val out = if (scale > 0) { | ||
| val longValue = input.multiply(pow).longValue() | ||
| JBigDecimal.valueOf(longValue).divide(pow) | ||
| } else if (scale == 0) { | ||
| JBigDecimal.valueOf(input.longValue()) | ||
| } else { | ||
| val longValue = input.divide(pow).longValue() | ||
| JBigDecimal.valueOf(longValue).multiply(pow) | ||
| } | ||
| out | ||
| } | ||
| } |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Isn't this actually behaviour change?
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
y. I don't think we should do this.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
R should support both date and number too
Uh oh!
There was an error while loading. Please reload this page.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Ah, just to be clear, is this change needed to pass the test? I am just worried of backward compatibility.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
that's a good point. fortunately (?) trunc was only added to R in 2.3.0, so I think we need to make sure (manually, add unit test) that trunc works on date columns and numeric columns