Druid.jl

Apache Druid querying library.

Installation

pkg> add Druid

Usage

Native Query

Druid native queries
documentation

using Druid
client = Client("http://localhost:8888")
timeseries_query = Timeseries(
    dataSource=Table("wikipedia"),
    intervals=[Interval("2015-09-12","2015-09-13")],
    granularity=SimpleGranularity("hour"),
    aggregations=[Count("total_rows"), SingleField("longSum", "added", "documents_added")]
)
println(execute(client, timeseries_query))

SQL Query

Druid SQL
documentation

using Druid
client = Client("http://localhost:8888")
sql_query = Sql(query="""
    SELECT FLOOR(__time TO HOUR) AS "timestamp", COUNT(*) AS "total_rows", SUM("added") AS "documents_added"
    FROM wikipedia
    WHERE __time >= TIMESTAMP '2015-09-12' AND __time < TIMESTAMP '2015-09-13'
    GROUP BY FLOOR(__time TO HOUR)
    ORDER BY "timestamp" ASC
""")
println(execute(client, sql_query))

Tables.jl compatibility

Most queries return the query response as an object compatible with the
Tables.jl interface. So it is quite easy to convert the result into another
compatible type, like DataFrame.

result = execute(client, query)
df = DataFrame(result)

Compatible queries: Timeseries, TopN, GroupBy, Scan, Search, Sql.

Sql query returns the result as either a Druid.SqlResult{ResultFormat} or a
CSV.File depending on the resultFormat provided in the SQL query. Both are
compatible with the Tables.jl interface.

TimeBoundary, SegmentMetadata and DatasourceMetadata return their results
as Dicts.