Merge pull request #63 from bermanjosh/scanType

url query encoding bug

bloodhound.cabal

@@ -1,5 +1,5 @@
 name:                bloodhound
-version:             0.7.1.0
+version:             0.7.2.0
 synopsis:            ElasticSearch client library for Haskell
 description:         ElasticSearch made awesome for Haskell hackers
 homepage:            https://github.com/bitemyapp/bloodhound

src/Database/Bloodhound/Client.hs

@@ -173,16 +173,16 @@ joinPath ps = do
   return $ joinPath' (s:ps)
 
 appendSearchTypeParam :: Text -> SearchType -> Text
-appendSearchTypeParam originalUrl st = addQuery [(keyEq, Just stParams)] originalUrl 
-  where keyEq = "search_type="
-        stParams
-          | st == SearchTypeDfsQueryThenFetch = "dfs_query_then_fetch"
-          | st == SearchTypeCount             = "count"
-          | st == SearchTypeScan              = "scan&scroll=1m"
-          | st == SearchTypeQueryAndFetch     = "query_and_fetch"
-          | st == SearchTypeDfsQueryAndFetch  = "dfs_query_and_fetch"
+appendSearchTypeParam originalUrl st = addQuery params originalUrl 
+  where stText = "search_type"
+        params
+          | st == SearchTypeDfsQueryThenFetch = [(stText, Just "dfs_query_then_fetch")]
+          | st == SearchTypeCount             = [(stText, Just "count")]
+          | st == SearchTypeScan              = [(stText, Just "scan"), ("scroll", Just "1m")]
+          | st == SearchTypeQueryAndFetch     = [(stText, Just "query_and_fetch")]
+          | st == SearchTypeDfsQueryAndFetch  = [(stText, Just "dfs_query_and_fetch")]
         -- used to catch 'SearchTypeQueryThenFetch', which is also the default
-          | otherwise                         = "query_then_fetch" 
+          | otherwise                         = [(stText, Just "query_then_fetch")]
 
 -- | Severely dumbed down query renderer. Assumes your data doesn't
 -- need any encoding
@@ -543,9 +543,9 @@ searchByType (IndexName indexName)
   (MappingName mappingName) = bindM2 dispatchSearch url . return
   where url = joinPath [indexName, mappingName, "_search"]
 
-scanSearch' :: MonadBH m => Search -> m (Maybe ScrollId)
-scanSearch' search = do
-    let url = joinPath ["_search"]
+scanSearch' :: MonadBH m => IndexName -> MappingName -> Search -> m (Maybe ScrollId)
+scanSearch' (IndexName indexName) (MappingName mappingName) search = do
+    let url = joinPath [indexName, mappingName, "_search"]
         search' = search { searchType = SearchTypeScan }
     resp' <- bindM2 dispatchSearch url (return search')
     let msr = decode' $ responseBody resp' :: Maybe (SearchResult ())
@@ -570,9 +570,11 @@ simpleAccumilator oldHits (newHits, msid) = do
     (newHits', msid') <- scroll' msid
     simpleAccumilator (oldHits ++ newHits) (newHits', msid')
 
-scanSearch :: (FromJSON a, MonadBH m) => Search -> m [Hit a]
-scanSearch search = do
-    msid <- scanSearch' search 
+-- | 'scanSearch' uses the 'scan&scroll' API of elastic, 
+-- for a given 'IndexName' and 'MappingName',
+scanSearch :: (FromJSON a, MonadBH m) => IndexName -> MappingName -> Search -> m [Hit a]
+scanSearch indexName mappingName search = do
+    msid <- scanSearch' indexName mappingName search 
     (hits, msid') <- scroll' msid
     (totalHits, _) <- simpleAccumilator [] (hits, msid')
     return totalHits

tests/tests.hs

@@ -762,3 +762,16 @@ main = hspec $ do
       enumFrom (pred maxBound :: DocVersion) `shouldBe` [pred maxBound, maxBound]
       enumFrom (pred maxBound :: DocVersion) `shouldBe` [pred maxBound, maxBound]
       enumFromThen minBound (pred maxBound :: DocVersion) `shouldBe` [minBound, pred maxBound]
+
+  describe "scan&scroll API" $ do
+    it "returns documents using the scan&scroll API" $ withTestEnv $ do
+      _ <- insertData
+      _ <- insertOther
+      let search = (mkSearch (Just $ MatchAllQuery Nothing) Nothing) { size = (Size 1) }
+      regular_search <- searchTweet search
+      scan_search' <- scanSearch testIndex testMapping search :: BH IO [Hit Tweet]
+      let scan_search = map hitSource scan_search'
+      liftIO $
+        regular_search `shouldBe` Right exampleTweet -- Check that the size restrtiction is being honored 
+      liftIO $
+        scan_search `shouldMatchList` [exampleTweet, otherTweet]