A .NET Apache Druid client written in C#

Supports .NET 4.5 and above, .NET Standard 1.6 and 2.0

1. Add a reference to druid4net from Nuget or download and reference the dll from releases
2. Add your favorite JSON parser (if you don't already have one referenced)
3. Implement the IJsonSerializer interface
4. Create a DruidClient and start querying

To query druid, create an instance of the DruidClient using code similar to the following:

var options = new ConfigurationOptions()
{
  JsonSerializer = new JilSerializer(),
  QueryApiBaseAddress = new Uri("http://localhost:8082")
};
new DruidClient(options);

Note the JilSerializer implementation can be found in the Integration tests project along with sample queries of all supported query types.

See Apache Druid Timeseries query documentation for more details on this type of query.

The following example query is performing a timeseries query against the sample wikiticker datasource. It filters the data where the country code is 'US' and the data timestamp is within the specified date interval. It then returns the total pages added by hour in a descending order.

var response = _druidClient.Timeseries<T>(q => q
  .Descending(true)
  .Aggregations(new LongSumAggregator("totalAdded", "added"))
  .Filter(new SelectorFilter("countryIsoCode", "US"))
  .DataSource("wikiticker")
  .Interval(FromDate, ToDate)
  .Granularity(Granularities.Hour)
);

See Apache Druid TopN query documentation for more details on this type of query.

The following example query is performing a topN query against the sample wikiticker datasource. It filters the data where the country code is 'US' and the user was anonymous and the data timestamp is within the specified date interval. It then returns the top 5 pages by count.

var response = _druidClient.TopN<T>(q => q
  .Metric("totalCount")
  .Dimension("page")
  .Threshold(5)
  .Aggregations(new LongSumAggregator("totalCount", "count"))
  .Filter(new AndFilter(
    new SelectorFilter("isAnonymous", "true"),
    new SelectorFilter("countryIsoCode", "US")
  ))
  .DataSource("wikiticker")
  .Interval(FromDate, ToDate)
  .Granularity(Granularities.All)
);

See Apache Druid GroupBy query documentation for more details on this type of query.

The following example query is performing a groupBy query against the sample wikiticker datasource. It returns the sum of page count grouped by Country name, then by city name and finally by page name.

var response = _druidClient.GroupBy<T>(q => q
  .Dimensions("countryName", "cityName", "page")
  .Aggregations(new LongSumAggregator("totalCount", "count"))
  .DataSource("wikiticker")
  .Interval(FromDate, ToDate)
  .Granularity(Granularities.All)
);

See Apache Druid Select query documentation for more details on this type of query.

The following example query is performing a select query against the sample wikiticker datasource. It selects the country name, city name, page, added and deleted values, filtered to anonymous users and limited to 10 records.

var response = _druidClient.Select<T>(q => q
  .Dimensions("countryName", "cityName", "page")
  .Metrics("added", "deleted")
  .Paging(new PagingSpec(10))
  .Filter(new SelectorFilter("isAnonymous", "true"))
  .DataSource("wikiticker")
  .Interval(FromDate, ToDate)
);

See Apache Druid Search query documentation for more details on this type of query.

The following example query is performing a search query against the sample wikiticker datasource. It searches for pages that contain the term "Dragon" and returns the page dimension value limited to the top 10 records.

var response = _druidClient.Search(q => q
  .DataSource("wikiticker")
  .Granularity(Granularities.All)
  .SearchDimensions("page")
  .Query(new ContainsSearchQuery("Dragon"))
  .Limit(10)
  .Interval(FromDate, ToDate)
);

See Apache Druid TimeBoundary query documentation for more details on this type of query.

The following example query is performing a timeBoundary query against the sample wikiticker datasource. It finds the minimum and maximum data points filtered to anonymous users.

var response = _druidClient.TimeBoundary(q => q
  .DataSource("wikiticker")
  .Filter(new SelectorFilter("isAnonymous", "true"))
);

See Apache Druid TimeBoundary query documentation for more details on this type of query.

The following example query is performing a scan query against the sample wikiticker datasource. It returns druid records in streaming mode, filtered to anonymous users and limited to the first 10 results.

var response = _druidClient.Scan<T>(q => q
  .DataSource("wikiticker")
  .Interval(FromDate, ToDate)
  .Filter(new SelectorFilter("isAnonymous", "true"))
  .Limit(10)
);

All query types have both synchronous and asynchronous methods available.

For example:

var response = _druidClient.Timeseries<T>(q => q...);

var response = await _druidClient.TimeseriesAsync<T>(q => q...);

The short answer is we wanted no dependencies. We also didn't want to implement our own JSON serialization as there are already so many good libraries out there that do this. Most projects already have a library included in their solution that can be used by implementing the interface in a simple pass-through class.

• Union data source
• Extraction filter