463fdc8 · typeable/bloodhound

git.iko.soy/typeable/bloodhound

on master help sign in

Add value_count aggregation support

463fdc8Michael Xavier11 years ago

src/Database/Bloodhound/Types.hs+16−1

compare with master

unifiedsplit

@@ -103,6 +103,7 @@ module Database.Bloodhound.Types

103 , RegexpFlags(..)

104 , RegexpFlag(..)

105 , FieldName(..)

106+ , Script(..)

106 , IndexName(..)

107 , IndexName(..)

107 , TemplateName(..)

108 , TemplateName(..)

108 , TemplatePattern(..)

109 , TemplatePattern(..)

@@ -201,6 +202,7 @@ module Database.Bloodhound.Types

201 , Bucket(..)

202 , Bucket(..)

202 , BucketAggregation(..)

203 , BucketAggregation(..)

203 , TermsAggregation(..)

204 , TermsAggregation(..)

205+ , ValueCountAggregation(..)

204 , DateHistogramAggregation(..)

206 , DateHistogramAggregation(..)

205

207

206 , Highlights(..)

208 , Highlights(..)

@@ -607,6 +609,12 @@ newtype QueryString = QueryString Text deriving (Eq, Generic, Show)

607-}

609-}

608newtype FieldName = FieldName Text deriving (Eq, Show)

610newtype FieldName = FieldName Text deriving (Eq, Show)

609

611

612+

613+{-| 'Script' is often used in place of 'FieldName' to specify more

614+complex ways of extracting a value from a document.

615+-}

616+newtype Script = Script { scriptText :: Text } deriving (Eq, Show)

617+

610{-| 'CacheName' is used in 'RegexpFilter' for describing the

618{-| 'CacheName' is used in 'RegexpFilter' for describing the

611 'CacheKey' keyed caching behavior.

619 'CacheKey' keyed caching behavior.

612-}

620-}

@@ -1310,7 +1318,8 @@ data Interval = Year

1310 | FractionalInterval Float TimeInterval deriving (Eq, Show)

1318 | FractionalInterval Float TimeInterval deriving (Eq, Show)

1311

1319

1312data Aggregation = TermsAgg TermsAggregation

1320data Aggregation = TermsAgg TermsAggregation

1313− | DateHistogramAgg DateHistogramAggregation deriving (Eq, Show)

1321+ | DateHistogramAgg DateHistogramAggregation

1322+ | ValueCountAgg ValueCountAggregation deriving (Eq, Show)

1314

1323

1315

1324

1316data TermsAggregation = TermsAggregation { term :: Either Text Text

1325data TermsAggregation = TermsAggregation { term :: Either Text Text

@@ -1336,6 +1345,8 @@ data DateHistogramAggregation = DateHistogramAggregation { dateField :: Fie

1336 , dateAggs :: Maybe Aggregations

1345 , dateAggs :: Maybe Aggregations

1337 } deriving (Eq, Show)

1346 } deriving (Eq, Show)

1338

1347

1348+data ValueCountAggregation = FieldValueCount FieldName

1349+ | ScriptValueCount Script deriving (Eq, Show)

1339

1350

1340mkTermsAggregation :: Text -> TermsAggregation

1351mkTermsAggregation :: Text -> TermsAggregation

1341mkTermsAggregation t = TermsAggregation (Left t) Nothing Nothing Nothing Nothing Nothing Nothing Nothing Nothing Nothing

1352mkTermsAggregation t = TermsAggregation (Left t) Nothing Nothing Nothing Nothing Nothing Nothing Nothing Nothing Nothing

@@ -1409,6 +1420,10 @@ instance ToJSON Aggregation where

1409 "post_offset" .= postOffset

1420 "post_offset" .= postOffset

1410 ],

1421 ],

1411 "aggs" .= dateHistoAggs ]

1422 "aggs" .= dateHistoAggs ]

1423+ toJSON (ValueCountAgg a) = object ["value_count" .= v]

1424+ where v = case a of

1425+ (FieldValueCount (FieldName n)) -> object ["field" .= n]

1426+ (ScriptValueCount (Script s)) -> object ["script" .= s]

1412

1427

1413type AggregationResults = M.Map Text Value

1428type AggregationResults = M.Map Text Value

1414

1429

tests/tests.hs+21−8

@@ -14,6 +14,7 @@ import Data.List (nub)

14import Data.List.NonEmpty (NonEmpty (..))

15import qualified Data.List.NonEmpty as NE

16import qualified Data.Map.Strict as M

17+import Data.Monoid

17import Data.Text (Text)

18import Data.Text (Text)

18import qualified Data.Text as T

19import qualified Data.Text as T

19import Data.Time.Calendar (Day (..))

20import Data.Time.Calendar (Day (..))

@@ -174,15 +175,16 @@ insertOther = do

174

175

175searchTweet :: Search -> BH IO (Either String Tweet)

176searchTweet :: Search -> BH IO (Either String Tweet)

176searchTweet search = do

177searchTweet search = do

177− reply <- searchByIndex testIndex search

178+ result <- searchTweets search

178− let result = eitherDecode (responseBody reply) :: Either String (SearchResult Tweet)

179 let myTweet = fmap (hitSource . head . hits . searchHits) result

180 return myTweet

181

182+searchTweets :: Search -> BH IO (Either String (SearchResult Tweet))

183+searchTweets search = eitherDecode . responseBody <$> searchByIndex testIndex search

184+

182searchExpectNoResults :: Search -> BH IO ()

185searchExpectNoResults :: Search -> BH IO ()

183searchExpectNoResults search = do

186searchExpectNoResults search = do

184− reply <- searchByIndex testIndex search

187+ result <- searchTweets search

185− let result = eitherDecode (responseBody reply) :: Either String (SearchResult Tweet)

186 let emptyHits = fmap (hits . searchHits) result

188 let emptyHits = fmap (hits . searchHits) result

187 liftIO $

189 liftIO $

188 emptyHits `shouldBe` Right []

190 emptyHits `shouldBe` Right []

@@ -214,8 +216,7 @@ searchTermsAggHint hints = do

214

216

215searchTweetHighlight :: Search -> BH IO (Either String (Maybe HitHighlight))

217searchTweetHighlight :: Search -> BH IO (Either String (Maybe HitHighlight))

216searchTweetHighlight search = do

218searchTweetHighlight search = do

217− reply <- searchByIndex testIndex search

219+ result <- searchTweets search

218− let result = eitherDecode (responseBody reply) :: Either String (SearchResult Tweet)

219 let myHighlight = fmap (hitHighlight . head . hits . searchHits) result

220 let myHighlight = fmap (hitHighlight . head . hits . searchHits) result

220 return myHighlight

221 return myHighlight

221

222

@@ -435,8 +436,7 @@ main = hspec $ do

435 let search = Search Nothing

436 let search = Search Nothing

436 (Just IdentityFilter) (Just [sortSpec]) Nothing Nothing

437 (Just IdentityFilter) (Just [sortSpec]) Nothing Nothing

437 False (From 0) (Size 10) Nothing

438 False (From 0) (Size 10) Nothing

438− reply <- searchByIndex testIndex search

439+ result <- searchTweets search

439− let result = eitherDecode (responseBody reply) :: Either String (SearchResult Tweet)

440 let myTweet = fmap (hitSource . head . hits . searchHits) result

441 liftIO $

442 myTweet `shouldBe` Right otherTweet

@@ -628,6 +628,19 @@ main = hspec $ do

628 _ <- insertData

629 searchTermsAggHint [GlobalOrdinals, GlobalOrdinalsHash, GlobalOrdinalsLowCardinality, Map]

630

631+

632+ it "can execute value_count aggregations" $ withTestEnv $ do

633+ _ <- insertData

634+ _ <- insertOther

635+ let ags = mkAggregations "user_count" (ValueCountAgg (FieldValueCount (FieldName "user"))) <>

636+ mkAggregations "bogus_count" (ValueCountAgg (FieldValueCount (FieldName "bogus")))

637+ let search = mkAggregateSearch Nothing ags

638+ let countPair k n = (k, object ["value" .= Number n])

639+ res <- searchTweets search

640+ liftIO $

641+ fmap aggregations res `shouldBe` Right (Just (M.fromList [ countPair "user_count" 2

642+ , countPair "bogus_count" 0

643+ ]))

631 -- Interaction of date serialization and date histogram aggregation is broken.

644 -- Interaction of date serialization and date histogram aggregation is broken.

632 -- it "returns date histogram aggregation results" $ withTestEnv $ do

645 -- it "returns date histogram aggregation results" $ withTestEnv $ do

633 -- _ <- insertData

646 -- _ <- insertData