gather

Converts several columns into two columns key and value. key column will contain names of original columns, value column will contain values from original columns.

This operation is reverse to pivot

gather { columns }
    [.explodeLists()]
    [.cast<Type>()]
    [.notNull()]
    [.where { valueFilter }]
    [.mapKeys { keyTransform }]
    [.mapValues { valueTransform }]
    .into(keyColumn, valueColumn) | .keysInto(keyColumn) | .valuesInto(valueColumn)

valueFilter: (value) -> Boolean
keyTransform: (columnName: String) -> K
valueTransform: (value) -> R 

See column selectors

Configuration options:

explodeLists — gathered values of type List will be exploded into their elements, so where, cast, notNull and mapValues will be applied to list elements instead of lists themselves
cast — inform compiler about the expected type of gathered elements. This type will be passed to where and mapKeys lambdas
notNull — skip gathered null values
where — filter gathered values
mapKeys — transform gathered column names (keys)
mapValues — transform gathered column values

Storage options:

into(keyColumn, valueColumn) — store gathered key-value pairs in two new columns with names keyColumn and valueColumn
keysInto(keyColumn) — store only gathered keys (column names) in a new column keyColumn
valuesInto(valueColumn) — store only gathered values in a new column valueColumn

pivoted.gather { "London".."Tokyo" }.into("city", "population")

pivoted.gather { "London".."Tokyo" }
    .cast<Int>()
    .where { it > 10 }
    .mapKeys { it.lowercase() }
    .mapValues { 1.0 / it }
    .into("city", "density")

Last modified: 27 September 2024