BUG Fix select normalization

The code is slightly hacky now in terms of internal API, but externally, it works well.
ngless-toolkit · Nov 9, 2018 · 71b842e · 71b842e
1 parent 912df97
commit 71b842e
Show file tree

Hide file tree

Showing 2 changed files with 13 additions and 6 deletions.
diff --git a/NGLess/Data/Sam.hs b/NGLess/Data/Sam.hs
@@ -193,16 +193,23 @@ X sequence mismatch.
 --}
 
 matchSize :: SamLine -> Either NGError Int
-matchSize = matchSize' . samCigar
-matchSize' cigar
+matchSize = matchSize' False . samCigar
+
+matchSize' includeSoft cigar
     | B8.null cigar = return 0
     | otherwise = case B8.readInt cigar of
         Nothing -> throwDataError ("could not parse cigar '"++B8.unpack cigar ++"'")
         Just (n,code_rest) -> do
             let code = B8.head code_rest
                 rest = B8.tail code_rest
-                n' = if code `elem` ("M=XS" :: String) then n else 0
-            r <- matchSize' rest
+                n' = case code of
+                            'M' -> n
+                            '=' -> n
+                            'X' -> n
+                            'S'
+                                | includeSoft -> n
+                            _ -> 0
+            r <- matchSize' includeSoft rest
             return (n' + r)
 
 matchIdentity :: SamLine -> Either NGError Double

diff --git a/NGLess/Interpretation/Select.hs b/NGLess/Interpretation/Select.hs
@@ -88,12 +88,12 @@ matchConditions doReinject conds sg = reinjectSequences doReinject (matchConditi
 
 _fixCigar :: B.ByteString -> Int -> Either NGError B.ByteString
 _fixCigar prev n = do
-    prevM <- matchSize' prev
+    prevM <- matchSize' True prev
     if prevM == n
         then return prev
         else do
             let prev' = B8.map (\c -> if c == 'H' then 'S' else c) prev
-            prevM' <- matchSize' prev'
+            prevM' <- matchSize' True prev'
             if prevM' == n
                 then return prev'
                 else throwDataError ("Cannot fix CIGAR string \"" ++ B8.unpack prev ++ "\" to represent a sequence of length " ++ show n)