improvements related to page search

[Rakka.git] / Rakka / Storage / Impl.hs

diff --git a/Rakka/Storage/Impl.hs b/Rakka/Storage/Impl.hs
index fed687f3c1266e992efc7973838cf8f41ced0ffb..c9c2de2858320725b75c30396e54e179bcc6b3da 100644 (file)
--- a/Rakka/Storage/Impl.hs
+++ b/Rakka/Storage/Impl.hs
@@ -17,7 +17,9 @@ import           Data.Maybe
 import           Data.Set (Set)
 import qualified Data.Set as S
 import           Network.HTTP.Lucu
+import           Network.HTTP.Lucu.Utils
 import           Network.URI
+import           Prelude hiding (words)
 import           Rakka.Attachment
 import           Rakka.Page
 import           Rakka.Storage.DefaultPage
@@ -29,6 +31,7 @@ import           Subversion.Repository
 import           System.Directory
 import           System.FilePath
 import           System.IO
+import           System.IO.Unsafe
 import           System.Log.Logger
 import           Text.HyperEstraier hiding (WriteLock)
 
@@ -183,16 +186,42 @@ syncIndex' index revFile repos mkDraft
                mapM_ (updateIndex index repos mkDraft newRev) (S.toList pages)
 
 
-searchIndex :: Database -> Condition -> IO [(PageName, RevNum)]
+searchIndex :: Database -> Condition -> IO SearchResult
 searchIndex index cond
-    = searchDatabase index cond >>= mapM fromId
+    = do (ids, hint) <- searchDatabase' index cond
+         let (total, words) = parseHint hint
+         pages <- mapM (fromId words) ids
+         return SearchResult {
+                      srTotal = total
+                    , srPages = pages
+                    }
     where
-      fromId :: DocumentID -> IO (PageName, RevNum)
-      fromId docId
-          = do uri <- getDocURI index docId
-               rev <- getDocAttr index docId "rakka:revision"
-                      >>= return . read . fromJust
-               return (decodePageName $ uriPath uri, rev)
+      parseHint :: [(String, Int)] -> (Int, [String])
+      parseHint xs
+          = let total = fromJust $ lookup "" xs
+                words = filter (/= "") $ map fst xs
+            in
+              (total, words)
+
+      fromId :: [String] -> DocumentID -> IO HitPage
+      fromId words docId
+          = do uri     <- getDocURI index docId
+               rev     <- getDocAttr index docId "rakka:revision"
+                          >>= return . read . fromJust
+               snippet <- unsafeInterleaveIO $
+                          do doc <- getDocument index docId [NoAttributes, NoKeywords]
+                             sn  <- makeSnippet doc words 300 80 80
+                             return (trim (== Boundary) $ map toFragment sn)
+               return HitPage {
+                            hpPageName = decodePageName $ uriPath uri
+                          , hpPageRev  = rev
+                          , hpSnippet  = snippet
+                          }
+
+      toFragment :: Either String (String, String) -> SnippetFragment
+      toFragment (Left "")      = Boundary
+      toFragment (Left t)       = NormalText t
+      toFragment (Right (w, _)) = HighlightedWord w
 
 
 updateIndex :: Database