node package manager

druid-query

druid-query

NPM version Build Status

Simple querying for Druid (http://druid.io) in Node.js. Inspired by ruby-druid.

Table of Contents:

Installation

npm install druid-query --save

Example (simple)

var Druid = require('druid-query')
  , Client = Druid.Client
  , Query = Druid.Query
  , client = new Client('http://127.0.0.1:8080')
 
var q1 = client.groupBy()
q1.dataSource('randSeq')
q1.granularity('all')
q1
  .dimensions([])
  .aggregation('count', 'rows')
  .aggregation('doubleSum', 'e', 'events')
  .aggregation('doubleSum', 'randomNumberSum', 'outColumn')
  .postAggregation('arithmetic', 'avg_random', '/', [
    Query.postAggregation('fieldAccess', null, 'randomNumberSum')
    Query.postAggregation('fieldAccess', null, 'rows')
  ])
  .interval(Date.UTC(2012, 9, 1), Date.UTC(2020, 0, 1))
 
q1.exec(function(err, result) {
  if (err) {
    // handle error... 
  }
  else {
    beCoolWith(result)
  }
})
 
var q2 = new Druid.TimeBoundaryQuery()
q2.dataSource('wikipedia')
client.exec(q2, function(err, result) {
  // handle results 
})

Example (ZooKeeper)

var Druid = require('druid-query')
  , druid = new Druid('localhost:2181,localhost:2182/druid', '/broker', {preferSSL: true})
 
 
var query = druid.groupBy('myCoolDS')
 
query
  .filter('selector', 'dimension1', 100500)
  .dimensions('dimension2', 'dimension3')
  .granularity('day')
  .aggregation('count', 'howMany')
  .interval(Date.UTC(2012, 0, 1), Date.UTC(2015, 0, 1))
  .exec(function(err, result) {
    // handle error 
    // handle result 
  })
 
 
var anotherQuery = new Druid.SegmentMetadataQuery()
anotherQuery.dataSource('superDS')
anotherQuery.interval('2011-01-01/2012-01-01')
anotherQuery.interval('2013-01-01/2014-01-01')
 
druid.exec(anotherQuery, function(err, results) {
  if (err) {
    // error reasons: 
    // 1. data source is not served by any known node 
    // 2. query validation error 
    // 3. error from Druid node after executing query 
  }
  else {
    // handle results 
  }
})
 
druid.once('ready', function() {
  // Do what you want with this event :-) 
})
 
 
druid.on('error', function(err) {
  // handle client error here 
})
 
 
// Call .end() when finished working with Druid 
druid.end()

API


Druid

Client which uses ZooKeeper to get data about Druid nodes and then gets data sources served by each node.

Events

  • ready - emitted when client finished (re)loading of nodes data (so it's ready to use). If client occasionally looses connection to ZooKeeper it's re-establed and client loads node data again.
  • error - emitted when client receives any kind of error.

Druid(connectionString, discoveryPath, [options])

Create client instance.

Arguments

  • connectionString string - ZooKeeper connection string.
  • discoveryPath string - Service discovery path.
  • options object - Client options.
    • zookeeper - Options passed to node-zookeeper-client createClient() function.
    • preferSSL - Use SSL port of Druid node if available. Default: false.

void cancel(query, callback)

Cancel query. Works same way as #exec(query, callback).

Arguments

  • query Query - Query (or descendant class) instance.
  • callback function - Callback function with following signature: (err).

void end()

End working with client.


void exec(query, callback)

Run query on suitable node.

If client is not ready (read Events section above) method will wait for ready or error event to continue.

If query data source is not among served by found nodes callback will be called with corresponding error.

Once node with least number of concurrent running queries is choosed query is sent to it.

Arguments

  • query Query - Query (or descendant class) instance.
  • callback function - Callback function with following signature: (err, result).

string[] getDataSources()

Get list of data sources.


DruidNodes[] getNodes()

Get list of Nodes available.

DruidNode extends Druid.Client. It keeps ZooKeeper node data and number of concurrent running queries (used for simple load-balancing).


GroupByQuery groupBy(dataSource, [rawQuery])

SearchQuery search(dataSource, [rawQuery])

SegmentMetadataQuery segmentMetadata(dataSource, [rawQuery])

TimeBoundaryQuery timeBoundary(dataSource, [rawQuery])

TimeseriesQuery timeseries(dataSource, [rawQuery])

TopNQuery topN(dataSource, [rawQuery])

Return query instance with dataSource set. Query is attached to calling Druid instance, so Druid#exec(query, callback) is called to execute query.

Arguments

  • dataSource string - name of data source to create Query for.
  • rawQuery object - passed to Query constructor as second argument.

Client (Druid.Client)

Base client class which uses Druid node URL.

Client(url)

Create client instance.

Arguments

  • url string - Druid node URL.

static void fromZooKeeper(connectionString, discoveryPath, [options], callback)

Lookup Druid services via ZooKeeper using node-zookeper-client and choose random node. For choosed node Client instance is created.

Arguments

  • connectionString string - ZooKeeper connection string.
  • discoveryPath string - service discovery path.
  • options object - Lookup options. We have only one option currently available:
    • preferSSL - Use SSL port of Druid node if available. Default: false.

void cancel(query, callback)

Cancel query.

Arguments

  • query Query - Query object.
  • callback(err) function - The callback function.

void dataSources(callback)

Get list of dataSources.

Arguments

  • callback(err, dataSources) function - The callback function.

void exec(query, callback)

Execute query.

Arguments

  • query Query - Query object.
  • callback(err, result) function - The callback function.

GroupByQuery groupBy([rawQuery])

SearchQuery search([rawQuery])

SegmentMetadataQuery segmentMetadata([rawQuery])

TimeBoundaryQuery timeBoundary([rawQuery])

TimeseriesQuery timeseries([rawQuery])

TopNQuery topN([rawQuery])

Create Query instance and attach it to client.

Arguments

  • rawQuery object - passed as second argument to Query constructor.

Query (Druid.Query)

Note: each field method returns field value if no arguments specified.

Query(client, [rawQuery])

Create query instance

Arguments

  • client Client - Client instance.
  • rawQuery object - Raw query data (so you can call Query#exec(callback) or Druid#exec(query, callback) right after creating Query object. Keep in mind that if constructor is not base Query class (e.g. GroupByQuery) queryType property is first removed from rawQuery object to prevent errors.

void cancel(callback)

Cancel query. context.queryId should be set for this.

Arguments

  • callback(err) function - The callback function.

void exec(callback)

Execute query (only if it's attached to client e.g. created by some client instance).

Arguments

  • callback(err, result) function - The callback function.

object toJSON()

Returns query data.


static object aggregation(type, name, [args...])

Create aggregation spec.

Arguments

  • type string | object - Aggregation type: cardinality, count, doubleSum, hyperUnique, javascript, longSum, max, min. Also you can specify aggregation as object in this argument.
  • name string - Aggregation output name.
  • args ...* - Aggregation-specific arguments.

Query.aggregation('cardinality', name, fieldNames, byRow)

  • fieldNames string[] - Fields to compute cardinality over.
  • byRow boolean - If we should compute cardinality over distinct combinations. Default: false.

Query.aggregation('count', name)

  • No arguments here

Query.aggregation('doubleSum', name, fieldName)

  • fieldName string - Name of the metric column to sum over.

Query.aggregation('hyperUnique', name, fieldName)

  • fieldName string - Dimension name.

Query.aggregation('javascript', name, fieldNames, aggregateFn, combineFn, resetFn)

  • fieldNames string[] - Names of fields which are passed to aggregate function.
  • aggregateFn string | function - Aggregation function.
  • combineFn string | function - Combines partials.
  • resetFn string | function - Initialization function.

Query.aggregation('longSum', name, fieldName)

  • fieldName string - Name of the metric column to sum over.

Query.aggregation('max', name, fieldName)

  • fieldName string - Name of the metric column.

Query.aggregation('min', name, fieldName)

  • fieldName string - Name of the metric column.

static object[] aggregations(list...)

Return array of aggregations.

Arguments

  • list object[] | object... - Array of aggregation specs. Specs can be returned by Query.aggregation() or raw JavaScript objects.

static object extractionFunction(type, [args...])

Create DimExtractionFn spec.

Arguments

  • type string | object - Spec type: javascript, partial, regex, searchQuery, time - or DimExtractionFn spec object.
  • args ...* - Function-specific arguments.

Query.extractionFunction('javascript', fn)

  • fn string | function - JavaScript function.

Query.extractionFunction('partial', regex)

  • regex string | RegExp - Regular expression to match.

Query.extractionFunction('regex', regex)

  • regex string | RegExp - Regular expression to match.

Query.extractionFunction('searchQuery`, query...)

  • query object | ...* - If one argument is specified we treat it as SearchQuerySpec object. Otherwise Query.query() is called for all the passed arguments.

Query.extractionFunction('time', input, output)

  • input string - Input time format.
  • output string - Output time format.

static object filter(type, [args...])

Create filter spec.

Arguments

  • type string | object - Filter type: and, javascript, not, or, regex, selector, search - or raw filter object.
  • args ...* - Filter-specific arguments. Described below.

Query.filter('and', filters...)

  • filters object[] | ...object - List of filters for AND.

Query.filter('javascript', dimension, fn)

  • dimension string - Dimension to which filter is applied.
  • fn string | function - Function to apply (should return boolean value).

Query.filter('not', filter...)

  • filter string | ...* - If this argument is object we use it as filter spec. Otherwise all arguments are passed again to Query.filter().

Query.filter('or', filters...)

  • filters object[] | ...object - List of filters for OR.

Query.filter('regex', dimension, pattern)

  • dimension string - Dimension to which filter is applied.
  • pattern string - Java regular expression.

Query.filter('selector', dimension, value)

  • dimensions string - Dimension to which filter is applied.
  • value * - Value to match.

Query.filter('search', dimension, query)

  • dimensions string - Dimension to which filter is applied.
  • query * - SearchQuerySpec object

static object having(type, [args...])

Create having spec.

Arguments

  • type string | object - HavingSpec object or type: and, equalTo, greaterThan, lessThan, not, or.
  • args ...* - Arguments specific to spec type.

Query.having('and', specs...)

  • specs object[] | ...object - List of specs for AND operation.

Query.having('equalTo', aggregation, value)

  • aggregation string - Aggregation name.
  • value * - Value to match.

Query.having('greaterThan', aggregation, value)

  • aggregation string - Aggregation name.
  • value * - Value to compare.

Query.having('lessThan', aggregation, value)

  • aggregation string - Aggregation name.
  • value * - Value to compare.

Query.having('not', spec...)

  • spec object | ...* - If first argument is object we use it as filter spec. Otherwise all arguments are passed again to Query.having().

Query.having('or', specs...)

  • specs object[] | ...object - List of specs for OR operation.

static object interval(start, [end])

Create interval string.

Of one argument specified it's treated as interval string.

Arguments

  • start string | number | Date - Interval string or start time as timestamp, date string or Date object.
  • end string | number | Date - End time.

static object orderBy(dimension, [direction])

Create OrderBy spec.

Arguments

  • dimension string - Dimension to sort by.
  • direction string - Sorting direction. Default: ASCENDING.

static object postAggregation(type, name, [args...])

Create post-aggregation spec.

Arguments

  • type string | object - Post-aggregation type: arithmetic, constant, fieldAccess, hyperUniqueCardinality, javascript. Or it can be ready-to-use post-aggregation object (no need in other arguments in this case, of course).
  • name string - Post-aggregation output name.
  • args ...* - Post-aggregation specific arguments. Read about arguments below.

Query.postAggregation('arithmetic', name, op, fields)

  • op string - Arithmetic operation: +, -, * or /.
  • fields object[] | ...object - List of Post-Aggregation specs: raw objects or Query.postAggregation() results.

Query.postAggregation('constant', name, value)

  • value * - Constant value.

Query.postAggregation('fieldAccess', name, fieldName)

  • fieldName string - Name of aggregator field. If not specified postAggregation() second argument (name) is used as fieldName instead.

Query.postAggregation('hyperUniqueCardinality', name, fieldName)

  • fieldName string - Name of hyperUnique aggregator. If not specified postAggregation() second argument (name) is used as fieldName instead.

Query.postAggregation('javascript', name, fieldNames, fn)

  • fieldNames string[] - List of aggregator names - passed as arguments to function.
  • fn string | function - Post-aggregator function.

static object[] postAggregations(list...)

Return array of post-aggregation specs.

Arguments

  • list object[] | object... - Array of aggregation specs. They can be ones returned by Query.postAggregation() or raw JavaScript objects.

static object query(type, value...)

Create SearchQuery spec.

Arguments

  • type string | object - SearchQuery type: insensitive_contains, fragment. Or ready SearchQuerySpec object.
  • value string | string[] | ...string - Value(s) to match. If type is fragment value (or all the values) is treated as array. If type is insensitive_contains value is used as string.

Query aggregation(type, name, [args...])

Add aggregation spec to aggregations.

Arguments

  • type string | object - Aggregation type: cardinality, count, doubleSum, hyperUnique, javascript, longSum, max, min. Or aggregation spec as JS object.
  • name string - Aggregation output name.
  • args ...* - Aggregation specific arguments. Read above about arguments in Query.aggregation() description.

Query aggregations(list...)

Set aggregations field.

Arguments

  • list object[] | object... - Array of aggregation specs. Specs can be returned by Query.aggregation() or raw JavaScript objects.

Query bound(value)

Set bound field for TimeBoundaryQuery.

Arguments

  • value string - Must be either "minTime" or "maxTime". Otherwise error is thrown.

Query context(data)

Set context field. Read more about it here.

Arguments

  • data object
    • timeout number
    • priority number
    • queryId string
    • useCache boolean
    • populateCache boolean
    • bySegment boolean
    • finalize boolean

Query dataSource(type, args...)

Set dataSource field

Arguments

  • type string | object - Data source type. Or data source as string or as object (DataSource structure).
  • args ...* - Arguments specific to each data source type.

Query#dataSource('table', name)

  • name string - Name of data source.

Query#dataSource('query', subQuery)

  • subQuery object | Query - Sub-query as Query instance or raw query object.

Query dimension(dimension, [outputName], [extractFn])

Set DimensionSpec.

If first argument is an object, then just use it as DimensionSpec.

If not depending on arguments length creates default or extraction dimension spec.

If second or third argument is object ExtractionDimensionSpec is created.

In other cases DefaultDimensionSpec is created.

Arguments

  • dimension string | object - Dimension to operate on. Or dimension definition as object.
  • outputName string - Dimension output name.
  • extractFn object - Extraction function spec created by Query.extractionFunction() or raw JavaScript object.

Query dimensions(list...)

Set dimensions.

Arguments

  • list string[] | ...string - Dimensions list.

Query filter(type, [args...])

Set filter spec.

Arguments

  • type string | object - Filter type: and, javascript, not, or, regex, selector. Otherwise whole filter object can be specified as first argument.
  • args ...* - Filter-specific arguments. They are described in Query.filter() method description.

Query granularity(type, [args...])

Set granularity of query.

Arguments

  • value string | object - Granularity as string or object. If value is string it must be one of those: all, none, minute, fifteen_minute, thirty_minute, hour, day plus duration and period which mean granularity spec object is created.
  • args ...* - Specific arguments (in case if value is period or duration).

Query#granularity('duration', duration, [origin])

  • duration string | number - Duration value in ms.
  • origin string | number | Date - Start time (optional).

Query#granularity('period', period, [timeZone], [origin])

  • period string - ISO-8601 duration format.
  • timeZone string - Timezone. Default: UTC (optional).
  • origin string | number | Date - Start time (optional).

Query having(type, [args...])

Set having field.

Arguments

  • type string | object - HavingSpec object or type: and, equalTo, greaterThan, lessThan, not, or.
  • args ...* - Arguments specific to spec type. They are described in Query.having().

Query interval(start, [end])

Add interval string to intervals field.

Arguments

  • start number | string | Date - Start time or interval string.
  • end number | string | Date - End time.

Query intervals(intervals...)

Set intervals.

Arguments

  • list string[] | ...string - List of interval strings.

Query limitSpec(type, limit, orderByColumns)

Set LimitSpec field.

Arguments

  • type string | object - raw LimitSpec object or LimitSpec type.
  • limit number - Limit of records returned.
  • orderByColumns object[] | string[] - OrderBy specs array. Specs can be strings or results of Query.orderBy()

Query merge(value)

Set merge field value.

Arguments

  • value boolean - Merge all individual segment metadata results into a single result.

Query metric(type, [args...])

Set TopNMetricSpec identified by metric value.

Arguments

  • type string | object - TopNMetricSpec object or spec type: alphaNumeric, lexicographic, numeric.
  • args ...* - Arguments specific to spec type. They are described below.

Query#metric('alphaNumeric', [previousStop])

  • previousStop string - The starting point of the lexicographic sort (optional).

Query#metric('lexicographic', [previousStop])

  • previousStop string - The starting point of the alpha-numeric sort (optional).

Query#metric('numeric', metric)

  • metric string - The actual metric field in which results will be sorted by.

Query postAggregation(type, name, [args...])

Add post-aggregation spec to postAggregations array.

Arguments

  • type string | object - Post-aggregation type: arithmetic, constant, fieldAccess, hyperUniqueCardinality, javascript. It can be post-aggregation object itself.
  • name string - Post-aggregation output name.
  • args ...* - Post-aggregation specific arguments. Read above about arguments in Query.postAggregation() method description.

Query postAggregations(list...)

Set postAggregations field.

Arguments

  • list object[] | object... - Array of aggregation specs. They can be ones returned by Query.postAggregation() or raw JavaScript objects.

Query query([type], value, caseSensitive)

Set SearchQuery spec (query field).

Arguments

  • type string | object - SearchQuery type: insensitive_contains, fragment, contains. Or it can be SearchQuerySpec object.
  • value string | string[] - Value(s) to match. If type is fragment value is treated as array. If type is insensitive_contains value is used as string.
  • caseSensitive boolean - Whether strings should be compared as case sensitive or not. Has no effect for type insensitive_contains.

Query queryType(type)

Set type of query. This method should be used only if you're using Query base class. All the Query descendants have queryType field set automatically.

Arguments

  • type string - Valid query type: groupBy, search, segmentMetadata, timeBoundary, timeseries, topN.

Query searchDimensions(list...)

Set searchDimensions field.

Arguments

  • list string[] | ...string - Dimensions list.

Query sort(type)

Set sort field.

Arguments

  • type string - Sorting type: lexicographic or strlen.

Query threshold(value)

Set threshold value.

Arguments

  • value number - Threshold number value.

Query toInclude(value)

Set toInclude field - columns which should be returned in result.

Arguments

  • value string | string[] | object - all, none or array of column names (list) or toInclude raw spec data as object.

Queries

GroupBy (Druid.GroupByQuery)

http://druid.io/docs/0.6.121/GroupByQuery.html

client
  .groupBy()
  .dataSource('sample_datasource')
  .granularity('day')
  .dimensions('dim1', 'dim2')
  .limitSpec('default', 5000, ['dim1', 'metric1'])
  .filter('and', [
    Query.filter('selector', 'sample_dimension1', 'sample_value1'),
    Query.filter('or', [
      Query.filter('selector', 'sample_dimension2', 'sample_value2'),
      Query.filter('selector', 'sample_dimension3', 'sample_value3')
    ])
  ])
  .aggregation('longSum', 'sample_name1', 'sample_fieldName1')
  .aggregation('doubleSum', 'sample_name2', 'sample_fieldName2')
  .postAggregation('arithmetic', 'sample_divide', '/', [
    Query.postAggregation('fieldAccess', 'sample_name1', 'sample_fieldName1'),
    Query.postAggregation('fieldAccess', 'sample_name2', 'sample_fieldName2')
  ])
  .intervals(new Date('2012-01-01T00:00:00.00'), new Date('2012-01-03T00:00:00.000'))
  .having('greaterThan', 'sample_name1', 0)
  .exec(/* result callback */)

Search (Druid.SearchQuery)

http://druid.io/docs/0.6.121/SearchQuery.html

client
  .search()
  .dataSource('sample_datasource')
  .granularity('day')
  .searchDimensions('dim1', 'dim2')
  .query('insensitive_contains', 'Ke')
  .sort('lexicographic')
  .intervals(new Date('2013-01-01T00:00:00.000'), new Date('2013-01-03T00:00:00.000'))
  .exec(/* result callback */)

Segment Metadata (Druid.SegmentMetadataQuery)

http://druid.io/docs/0.6.121/SegmentMetadataQuery.html

client
  .segmentMetadata()
  .dataSource('sample_datasource')
  .intervals(new Date('2013-01-01'), new Date('2014-01-01'))
  .exec(/* result callback */)

Time Boundary (Druid.TimeBoundaryQuery)

http://druid.io/docs/0.6.121/TimeBoundaryQuery.html

client
  .timeBoundary()
  .dataSource('sample_datasource')
  .exec(/* result callback */)

Timeseries (Druid.TimeseriesQuery)

http://druid.io/docs/0.6.121/TimeseriesQuery.html

client
  .timeseries()
  .dataSource('sample_datasource')
  .granularity('day')
  .filter('and', [
    Query.filter('selector', 'sample_dimension1', 'sample_value1'),
    Query.filter('or', [
      Query.filter('selector', 'sample_dimension2', 'sample_value2'),
      Query.filter('selector', 'sample_dimension3', 'sample_value3')
    ])
  ])
  .aggregation('longSum', 'sample_name1', 'sample_fieldName1')
  .aggregation('doubleSum', 'sample_name2', 'sample_fieldName2')
  .postAggregation('arithmetic', 'sample_divide', '/', [
    Query.postAggregation('fieldAccess', 'sample_name1', 'sample_fieldName1'),
    Query.postAggregation('fieldAccess', 'sample_name2', 'sample_fieldName2')
  ])
  .intervals(new Date('2013-01-01T00:00:00.000'), new Date('2013-01-03T00:00:00.000'))
  .exec(/* result callback */)

TopN (Druid.TopNQuery)

http://druid.io/docs/0.6.121/TopNQuery.html

client
  .topN()
  .dataSource('sample_data')
  .dimension('sample_dim')
  .threshold(5)
  .metric('count')
  .granularity('all')
  .filter('and', [
    Query.filter('selector', 'dim1', 'some_value'),
    Query.filter('selector', 'dim2', 'some_other_val')
  ])
  .aggregation('longSum', 'count', 'count')
  .aggregation('doubleSum', 'some_metric', 'some_metric')
  .postAggregation('arithmetic', 'sample_divide', '/', [
    Query.postAggregation('fieldAccess', 'some_metric', 'some_metric'),
    Query.postAggregation('fieldAccess', 'count', 'count')
  ])
  .intervals(new Date('2013-08-31T00:00:00.000'), new Date('2013-09-03T00:00:00.000'))
  .exec(/* result callback */)

TODO

  • More tests

LICENSE

MIT