Spark DateTime Library

A library for exposing dateTime functions from the joda time library as SQL functions. Also provide a dsl for dateTime catylst expressions; this utilizes the [scala wrapper library](https://github.com/nscala-time/nscala-time nscala).

Requirements

This library requires Spark 1.4+

Linking

You can link against this library in your program at the following coordiates:

groupId: org.sparklinedata
artifactId: spark-datetime_2.10
version: 0.0.1

Using with Spark shell

This package can be added to Spark using the --jars command line option. For example, to include it when starting the spark shell:

$ bin/spark-shell --packages org.sparklinedata:spark-datetime_2.10:0.0.1

Features

A set of functions from the joda library to operate on dates.
- field access: all functions in the DateTime class are available as sql functions. The first argument is the DateTime object on which the function is to be applied.
- construction: functions are available to convert a String or a epoch value to DateTime
- comparison functions available to compare dates (=, <, <=, >, >=), also compare against now.
- arithmetic: functions available to add/subtract Period from dates.
- intervals: functions available to construct Intervals and compare(contains, overlaps, abuts, gap) intervals and dateTimes.
A dsl for dateTime catylst expressions.
A StringContext to embed date expressions in SQL statements.

Function naming convention

getter functions on the DateTime class are exposed with the same name, in camelCase. So getYear is exposed as year, getMonthOfYear is exposed as monthOfYear etc.

SQL API

Assume you have a table input with a string column called dt

select dt, dateTime(dt), dayOfWeek(dateTime(dt)), dayOfWeekName(dateTime(dt)), dayOfWeekName(dateTimeWithTZ(dt)) 
from input

Date Expressions using the DSL

A basic example

import com.github.nscala_time.time.Imports._
import org.apache.spark.sql.catalyst.dsl.expressions._
import org.sparklinedata.spark.dateTime.dsl.expressions._
import org.sparklinedata.spark.dateTime.Functions


// register all functions 
Functions.register(sqlContext)

val dT = dateTime('dt)
val dOW = dateTime('dt) dayOfWeek
val dOWNm = dateTime('dt) dayOfWeekName
val dOWNm2 = dateTimeWithTZ('dt) dayOfWeekName
val dTFixed = dateTime("2015-05-22T08:52:41.903-07:00")

val t = sql(date"select dt, $dT, $dOW, $dOWNm, $dOWNm2, $dTFixed," +
      " dateTime(\"2015-05-22T08:52:41.903-07:00\") from input")

An example about periods

import com.github.nscala_time.time.Imports._
import org.apache.spark.sql.catalyst.dsl.expressions._
import org.sparklinedata.spark.dateTime.dsl.expressions._

val dT = dateTime('dt)
val dT1 = dateTime('dt) + 3.months
val dT2 = dateTime('dt) - 3.months
val dT3 = dateTime('dt) + 12.week
val cE = dateTime('dt) + 3.months > (dateTime('dt) + 12.week)

val t = sql(date"select dt, $dT, $dT1, $dT2, $dT3, $cE from input")

Weekend filter example

val filter : Expression = ((dateTime('dt) dayOfWeekName) === "Saturday") ||
      ((dateTime('dt) dayOfWeekName) === "Sunday")

val t = sql(date"select dt from input where $filter")

Group By example

val dayOfWeek: Expression = dateTime('dt) dayOfWeekName

val t = sql(date"select $dayOfWeek, count(*) from input group by $dayOfWeek")

Interval example

val i1 = END_DATE - 15.day to END_DATE - 10.day

val isBefore = i1 isBeforeE dateTime('dt)
val isAfter = i1 isAfterE dateTime('dt)
val i2 = dateTime('dt) to (dateTime('dt) + 5.days)
val overlapsE = i1 overlapsE i2
val abutsE = i1 abutsE i2

val t = sql(date"select dt, $isBefore, $isAfter, $overlapsE, $abutsE from input")

Time Bucketing

Use this feature to bucket dates into given Periods. For e.g. 8.hours, 30.mins, 2.days etc.

The following example buckets rows into 3 day periods. The bucket function on a DateExpression takes an origin date and a Period specification. The Period is an iso8061 specification for period.

val start = dateTime("2015-06-23T17:27:43.769-07:00")
val dT = dateTime('dt)
val timeBucket = dateTime('dt) bucket(start, 3.days)

val t = sql(date"select dt, $dT, $timeBucket from input")

Or the direct sql for the above query is:

select dt, dateTime(`dt`), 
      timeBucket(dateTime(`dt`),dateTime("2015-05-24T17:27:43.769-07:00"),period("P3D")) 
from input

Building From Source

This library is built with SBT, which is automatically downloaded by the included shell script. To build a JAR file simply run build/sbt package from the project root. The build configuration includes support for both Scala 2.10 and 2.11.

hbutani / spark-datetime 0.0.3