Add options for glean indexer

glean/lang/scip/indexer/Glean/Indexer/SCIP.hs

@@ -25,11 +25,41 @@ import qualified Glean
 import System.Directory (doesFileExist)
 import Util.OptParse (maybeStrOption)
 
--- | A generic SCIP indexer, for existing scip files
-newtype SCIP = SCIP
-  { scipIndexFile :: Maybe FilePath
+data SCIP = SCIP
+  { scipFile :: FilePath -- ^ input file
+  , outputFile :: FilePath -- ^ output file
+  , scipLanguage :: Maybe LanguageId -- ^ a default language if known
+  , inferLanguage :: Bool -- ^ default False, infer language using file suffix
+  , scipPathPrefix :: Maybe FilePath -- ^ optional path to prefix file paths
+  , stripPathPrefix :: Maybe FilePath -- ^ optional prefix to drop from paths
   }
-  -- no options currently
+
+options :: Parser SCIP
+options = do
+  scipFile <- option str $ long "input" <>
+    metavar "PATH" <>
+    help "Path to a specific SCIP file to convert"
+  outputFile <- option str $ long "output"
+    <> metavar "PATH"
+    <> help "Output filepath to write encoded schema info"
+  scipLanguage <- option (Just <$> readLanguage) $ long "language" <>
+    metavar "LANGUAGE" <>
+    value Nothing <>
+    help "Default language of files in the index"
+  inferLanguage <- switch $
+    short 'i' <>
+    long "infer-language" <>
+    help ("Infer symbol language based on file suffix" <>
+     "(when set this takes precedence over --language)")
+  scipPathPrefix <- option (Just <$> str) $ long "root-prefix" <>
+    metavar "PATH" <>
+    value Nothing <>
+    help "Path to prepend to file path data"
+  stripPathPrefix <- option (Just <$> str) $ long "strip-prefix" <>
+    metavar "PATH" <>
+    value Nothing <>
+    help "Path prefix to strip from path data"
+  return SCIP{..}
 
 options :: Parser SCIP
 options =
@@ -59,6 +89,22 @@ indexer = Indexer {
     sendJsonBatches backend repo "scip" val
     derive backend repo
   }
+
+-- If the indexer doesn't set the langauge Id of the files, we
+-- can assert it here. Otherwise Glean won't know what language the
+-- symbols are in
+readLanguage ::  ReadM LanguageId
+readLanguage = do
+  ln <- Text.toLower <$> str
+  case ln of
+    "typescript" -> return TypeScript
+    "rust" -> return Rust
+    "go" -> return Go
+    "java" -> return Java
+    "kotlin" -> return Kotlin
+    "csharp" -> return CSharp
+    "swift" -> return Swift
+    _ -> readerError "Unrecognized SCIP language"
 
 -- | Derive any SCIP stored predicates
 derive :: Glean.Backend b => b -> Glean.Repo -> IO ()