Perform a left outer join

Use join.left() to perform an left outer join of two streams of data. Left joins output a row for each row in the left data stream with data matching from the right data stream. If there is no matching data in the right data stream, non-group-key columns with values from the right data stream are null.

{{< svg svg="static/svgs/join-diagram.svg" class="left" >}}

{{< expand-wrapper >}} {{% expand "View table illustration of a left outer join" %}} {{< flex >}} {{% flex-content "third" %}}

left

right

Left outer join result

Use join.left to join your data

1. Import the join package.

2. Define the left and right data streams to join:

   • Each stream must have one or more columns with common values. Column labels do not need to match, but column values do.
   • Each stream should have identical group keys.

   For more information, see join data requirements.

3. Use join.left() to join the two streams together. Provide the following parameters:

   • left: Stream of data representing the left side of the join.
   • right: Stream of data representing the right side of the join.
   • on: Join predicate. For example: (l, r) => l.column == r.column.
   • as: Join output function that returns a record with values from each input stream. For example: (l, r) => ({l with column1: r.column1, column2: r.column2}).

The following example uses a filtered selection from the machineProduction sample data set as the left data stream and an ad-hoc table created with array.from() as the right data stream.

{{% note %}}

Example data grouping

The example below ungroups the left stream to match the grouping of the right stream. After the two streams are joined together, the joined data is grouped by stationID. {{% /note %}}

import "array"
import "influxdata/influxdb/sample"
import "join"

left =
    sample.data(set: "machineProduction")
        |> filter(fn: (r) => r.stationID == "g1" or r.stationID == "g2" or r.stationID == "g3")
        |> filter(fn: (r) => r._field == "oil_temp")
        |> limit(n: 5)

right =
    array.from(
        rows: [
            {station: "g1", opType: "auto", last_maintained: 2021-07-15T00:00:00Z},
            {station: "g2", opType: "manned", last_maintained: 2021-07-02T00:00:00Z},
        ],
    )

join.left(
    left: left |> group(),
    right: right,
    on: (l, r) => l.stationID == r.station,
    as: (l, r) => ({l with opType: r.opType, maintained: r.last_maintained}),
)
    |> group(columns: ["stationID"])

{{< expand-wrapper >}} {{% expand "View example input and output data" %}}

{{% note %}} _start and _stop columns have been omitted from example input and output. {{% /note %}}

Input

left {#left-input}

right {#right-input}

Output {#example-output}

Things to note about the join output

• Because the right stream does not have a row with the g3 station tag, rows from the left stream with the g3 stationID tag include null values in columns that are populated from the right stream (r) in the as parameter. {{% /expand %}} {{< /expand-wrapper >}}