Merge pull request #940 from elasticsearch/feature/aggregations-top-hits

Feature/aggregations top hits

Author: gmarz

src/Nest/DSL/Aggregations/AggregationDescriptor.cs

@@ -84,6 +84,9 @@ public interface IAggregationContainer
 		[JsonProperty("percentile_ranks")]
 		IPercentileRanksAggregaor PercentileRanks { get; set; }
 
+		[JsonProperty("top_hits")]
+		ITopHitsAggregator TopHits { get; set; }
+
 		[JsonProperty("aggs")]
 		[JsonConverter(typeof(DictionaryKeysAreNotPropertyNamesJsonConverter))]
 		IDictionary<string, IAggregationContainer> Aggregations { get; set; }
@@ -111,14 +114,14 @@ public class AggregationContainer : IAggregationContainer
 		private ISignificantTermsAggregator _significantTerms;
 		private IPercentileRanksAggregaor _percentileRanks;
 
+		private ITopHitsAggregator _topHits;
 		public IAverageAggregator Average { get; set; }
 		public IValueCountAggregator ValueCount { get; set; }
 		public IMaxAggregator Max { get; set; }
 		public IMinAggregator Min { get; set; }
 		public IStatsAggregator Stats { get; set; }
 		public ISumAggregator Sum { get; set; }
 		public IExtendedStatsAggregator ExtendedStats { get; set; }
-
 		public IDateHistogramAggregator DateHistogram
 		{
 			get { return _dateHistogram; }
@@ -227,6 +230,12 @@ public IPercentileRanksAggregaor PercentileRanks
 			set { _percentileRanks = value; }
 		}
 
+		public ITopHitsAggregator TopHits
+		{
+			get { return _topHits; }
+			set { _topHits = value; }
+		}
+
 		private void LiftAggregations(IBucketAggregator bucket)
 		{
 			if (bucket == null) return;
@@ -290,7 +299,9 @@ public class AggregationDescriptor<T> : IAggregationContainer
 		IPercentileRanksAggregaor IAggregationContainer.PercentileRanks { get;set; }
 
 		ITermsAggregator IAggregationContainer.Terms { get; set; }
-		
+
+		ITopHitsAggregator IAggregationContainer.TopHits { get; set; }
+
 		public AggregationDescriptor<T> Average(string name, Func<AverageAggregationDescriptor<T>, AverageAggregationDescriptor<T>> selector)
 		{
 			return _SetInnerAggregation(name, selector, (a, d) => a.Average = d);
@@ -429,6 +440,12 @@ public AggregationDescriptor<T> ValueCount(string name,
 			return _SetInnerAggregation(name, selector, (a, d) => a.ValueCount = d);
 		}
 
+		public AggregationDescriptor<T> TopHits(string name,
+			Func<TopHitsAggregationDescriptor<T>, TopHitsAggregationDescriptor<T>> selector)
+		{
+			return _SetInnerAggregation(name, selector, (a, d) => a.TopHits = d);
+		}
+
 		private AggregationDescriptor<T> _SetInnerAggregation<TAggregation>(
 			string key,
 			Func<TAggregation, TAggregation> selector

src/Nest/DSL/Aggregations/TopHitsAggregationDescriptor.cs

@@ -0,0 +1,91 @@
+using Nest.Resolvers.Converters;
+using Newtonsoft.Json;
+using System;
+using System.Collections.Generic;
+using System.Linq;
+using System.Text;
+
+namespace Nest
+{
+	[JsonObject(MemberSerialization = MemberSerialization.OptIn)]
+	[JsonConverter(typeof(ReadAsTypeConverter<TopHitsAggregator>))]
+	public interface ITopHitsAggregator : IMetricAggregator
+	{
+		[JsonProperty("from")]
+		int? From { get; set; }
+
+		[JsonProperty("size")]
+		int? Size { get; set; }
+
+		[JsonProperty("sort")]
+		[JsonConverter(typeof(SortCollectionConverter))]
+		IList<KeyValuePair<PropertyPathMarker, ISort>> Sort { get; set; }
+
+		[JsonProperty("_source")]
+		ISourceFilter Source { get; set; }
+	}
+
+	public class TopHitsAggregator : MetricAggregator, ITopHitsAggregator
+	{
+		public int? From { get; set; }
+		public int? Size { get; set; }
+		public IList<KeyValuePair<PropertyPathMarker, ISort>> Sort { get; set; }
+		public ISourceFilter Source { get; set; }
+	}
+
+	public class TopHitsAggregationDescriptor<T> 
+		: MetricAggregationBaseDescriptor<TopHitsAggregationDescriptor<T>, T>, ITopHitsAggregator
+		where T : class
+	{
+		ITopHitsAggregator Self { get { return this; } }
+
+		int? ITopHitsAggregator.From { get; set; }
+
+		int? ITopHitsAggregator.Size { get; set; }
+
+		IList<KeyValuePair<PropertyPathMarker, ISort>> ITopHitsAggregator.Sort { get; set; }
+
+		ISourceFilter ITopHitsAggregator.Source { get; set; }
+
+		public TopHitsAggregationDescriptor<T> From(int from)
+		{
+			this.Self.From = from;
+			return this;
+		}
+
+		public TopHitsAggregationDescriptor<T> Size(int size)
+		{
+			this.Self.Size = size;
+			return this;
+		}
+
+		public TopHitsAggregationDescriptor<T> Sort(Func<SortFieldDescriptor<T>, IFieldSort> sortSelector)
+		{
+			sortSelector.ThrowIfNull("sortSelector");
+
+			if (Self.Sort == null)
+				Self.Sort = new List<KeyValuePair<PropertyPathMarker, ISort>>();
+			
+			var descriptor = sortSelector(new SortFieldDescriptor<T>());
+			this.Self.Sort.Add(new KeyValuePair<PropertyPathMarker, ISort>(descriptor.Field, descriptor));
+			
+			return this;
+		}
+
+		public TopHitsAggregationDescriptor<T> Source(bool include = true)
+		{
+			if (!include)
+				this.Self.Source = new SourceFilter { Exclude = new PropertyPathMarker[] { "*" } };
+			else
+				this.Self.Source = null;
+
+			return this;
+		}
+
+		public TopHitsAggregationDescriptor<T> Source(Func<SearchSourceDescriptor<T>, SearchSourceDescriptor<T>> sourceSelector)
+		{
+			this.Self.Source = sourceSelector(new SearchSourceDescriptor<T>());
+			return this;
+		}
+	}
+}

src/Nest/Domain/Aggregations/AggregationsHelper.cs

@@ -81,6 +81,11 @@ public PercentilesMetric PercentilesRank(string key)
 			return this.TryGet<PercentilesMetric>(key);
 		}
 
+		public TopHitsMetric TopHitsMetric(string key)
+		{
+			return this.TryGet<TopHitsMetric>(key);
+		}
+
 		public SingleBucket Global(string key)
 		{
 			return this.TryGet<SingleBucket>(key);

src/Nest/Domain/Aggregations/TopHitsMetric.cs

@@ -0,0 +1,41 @@
+using System;
+using System.Collections.Generic;
+using System.Linq;
+using System.Text;
+using Newtonsoft.Json;
+using Newtonsoft.Json.Linq;
+
+namespace Nest
+{
+	public class TopHitsMetric : IMetricAggregation
+	{
+		private IEnumerable<JObject> _hits;
+
+		public TopHitsMetric()
+		{
+		}
+		
+		internal TopHitsMetric(IEnumerable<JObject> hits)
+		{
+			_hits = hits;
+		}
+
+		public long Total { get; set; }
+		public double? MaxScore { get; set; }
+
+		public IEnumerable<Hit<T>> Hits<T>(JsonSerializer serializer = null) where T : class
+		{
+			if (serializer != null)
+				return _hits.Select(h => h.ToObject<Hit<T>>(serializer));
+			return _hits.Select(h => h.ToObject<Hit<T>>());
+		}
+
+		public IEnumerable<T> Documents<T>(JsonSerializer serializer = null) where T : class
+		{
+			return this.Hits<T>(serializer).Select(h => h.Source);
+		}
+
+
+	}
+
+}

src/Nest/Domain/Hit/Hit.cs

@@ -35,8 +35,12 @@ public class Hit<T> : IHit<T>
 		public T Source { get; internal set; }
 		[JsonProperty(PropertyName = "_index")]
 		public string Index { get; internal set; }
+		
+		//TODO in NEST 2.0 make the property itself double?
 		[JsonProperty(PropertyName = "_score")]
-		public double Score { get; internal set; }
+		internal double? _score { get; set; }
+		public double Score { get { return _score.GetValueOrDefault(0); } }
+
 		[JsonProperty(PropertyName = "_type")]
 		public string Type { get; internal set; }
 		[JsonProperty(PropertyName = "_version")]

src/Nest/Nest.csproj

@@ -106,6 +106,7 @@
     <Compile Include="Domain\Aggregations\AggregationsHelper.cs" />
     <Compile Include="Domain\Aggregations\Bucket.cs" />
     <Compile Include="Domain\Aggregations\BucketAggregationBase.cs" />
+    <Compile Include="Domain\Aggregations\TopHitsMetric.cs" />
     <Compile Include="Domain\Aggregations\GeoBoundsMetric.cs" />
     <Compile Include="Domain\Aggregations\HistogramItem.cs" />
     <Compile Include="Domain\Aggregations\ExtendedStatsMetric.cs" />
@@ -211,6 +212,7 @@
     <Compile Include="Domain\Stats\PluginStats.cs" />
     <Compile Include="Domain\Stats\SegmentsStats.cs" />
     <Compile Include="DSL\Aggregations\PercentileRanksAggregationDescriptor.cs" />
+    <Compile Include="DSL\Aggregations\TopHitsAggregationDescriptor.cs" />
     <Compile Include="DSL\Aggregations\GeoBoundsAggregationDescriptor.cs" />
     <Compile Include="DSL\Aggregations\ReverseNestedAggregationDescriptor.cs" />
     <Compile Include="DSL\CatIndicesDescriptor.cs" />

src/Nest/Resolvers/Converters/Aggregations/AggregationConverter.cs

@@ -59,12 +59,27 @@ private IAggregation ReadAggregation(JsonReader reader, JsonSerializer serialize
 					return GetSingleBucketAggregation(reader, serializer);
 				case "bounds":
 					return GetGeoBoundsMetricAggregation(reader, serializer);
+				case "hits":
+					return GetHitsAggregation(reader, serializer);
 				default:
 					return null; 
 
 			}
 		}
 
+		private IAggregation GetHitsAggregation(JsonReader reader, JsonSerializer serializer)
+		{
+			reader.Read();
+			var o = JObject.Load(reader);
+			if (o == null)
+				return null;
+
+			var total = o["total"].ToObject<long>();
+			var maxScore = o["max_score"].ToObject<double?>();
+			var hits = o["hits"].Children().OfType<JObject>().Select(s=>s);
+			return new TopHitsMetric(hits) { Total = total, MaxScore = maxScore };
+		}
+
 		private IAggregation GetGeoBoundsMetricAggregation(JsonReader reader, JsonSerializer serializer)
 		{
 			reader.Read();
@@ -370,15 +385,15 @@ public IAggregation GetRangeAggregation(JsonReader reader, JsonSerializer serial
 						break;
 				}
 			}
-				var bucket = new RangeItem
-				{
-					Key = key,
-					From = fromDouble,
-					To = toDouble,
-					DocCount = docCount.GetValueOrDefault(),
-					FromAsString = fromAsString,
-					ToAsString = toAsString
-				};
+			var bucket = new RangeItem
+			{
+				Key = key,
+				From = fromDouble,
+				To = toDouble,
+				DocCount = docCount.GetValueOrDefault(),
+				FromAsString = fromAsString,
+				ToAsString = toAsString
+			};
 
 			bucket.Aggregations = this.GetNestedAggregations(reader, serializer);
 			return bucket;

src/Nest/Resolvers/Converters/ConcreteTypeConverter.cs

@@ -175,7 +175,7 @@ internal static Type GetConcreteTypeUsingSelector<T>(
 			hitDynamic.Fields = sel;
 			hitDynamic.Source = d._source;
 			hitDynamic.Index = d._index;
-			hitDynamic.Score = (d._score is double) ? d._score : default(double);
+			hitDynamic._score = (d._score is double) ? d._score : default(double);
 			hitDynamic.Type = d._type;
 			hitDynamic.Version = d._version;
 			hitDynamic.Id = d._id;

src/Tests/Nest.Tests.Integration/Aggregations/MetricAggregationTests.cs

@@ -1,4 +1,5 @@
 using System.Linq;
+using Elasticsearch.Net;
 using FluentAssertions;
 using Nest.Tests.MockData.Domain;
 using NUnit.Framework;
@@ -162,5 +163,46 @@ public void GeoBounds()
 			geoBoundsMetric.Bounds.BottomRight.Lat.Should().NotBe(0);
 			geoBoundsMetric.Bounds.BottomRight.Lon.Should().NotBe(0);
 		}
+
+		[Test]
+		public void TopHits()
+		{
+			var results = this.Client.Search<ElasticsearchProject>(s => s
+				.Size(0)
+				.Aggregations(a => a
+					.Terms("top-countries", t => t
+						.Field(p => p.Country)
+						.Size(3)
+						.Aggregations(aa => aa
+							.TopHits("top-country-hits", th => th
+								.Sort(sort => sort
+									.OnField(p => p.StartedOn)
+									.Order(SortOrder.Descending)
+								)
+								.Source(src => src
+									.Include(p => p.Name)
+								)
+								.Size(1)
+							)
+						)
+					)
+				)
+			);
+
+			results.IsValid.Should().BeTrue();
+
+			var topCountries = results.Aggs.Terms("top-countries").Items;
+			foreach(var topCountry in topCountries)
+			{
+				var topHits = topCountry.TopHitsMetric("top-country-hits");
+				topHits.Should().NotBeNull();
+				topHits.Total.Should().BeGreaterThan(0);
+				var hits = topHits.Hits<ElasticsearchProject>();
+				hits.Should().NotBeEmpty().And.NotContain(h=> h.Id.IsNullOrEmpty() || h.Index.IsNullOrEmpty());
+				topHits.Documents<ElasticsearchProject>().Should().NotBeEmpty();
+
+			}
+
+		}
 	}
 }