auditor

commit a5ec87850ea6e1d5c4ab1d34df162966eab9de9a
parent fb15d345eb83db9a270c3f937552e2a1c2ef7ba1
Author: Jared Tobin <jared@jtobin.io>
Date:   Tue, 10 Feb 2026 14:49:05 +0400

fix: parser fixes for secp256k1.s compatibility

- Fix backtracking for post-index addressing: wrap pPostIndex in try
  to allow backtracking when whitespace is consumed before failing on
  non-comma (e.g., semicolon comment after [sp])

- Parse optional shift amount after extend ops: handle [x9, w8, uxtw #3]
  patterns where extend has a shift amount

- Add --parse flag and parseFile for parse-only mode (useful for
  debugging parser issues)

Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>

Diffstat:
M
app/Main.hs
 | 
38
+++++++++++++++++++++++++++-----------
M
lib/Audit/AArch64.hs
 | 
12
++++++++++++
M
lib/Audit/AArch64/Parser.hs
 | 
18
+++++++++++-------

3 files changed, 50 insertions(+), 18 deletions(-)
diff --git a/app/Main.hs b/app/Main.hs
@@ -3,7 +3,7 @@
 module Main where
 
 import Audit.AArch64 (AuditResult(..), Violation(..), ViolationReason(..),
-                      auditFile, auditFileInterProc, regName)
+                      auditFile, auditFileInterProc, parseFile, regName)
 import Data.Aeson (encode)
 import qualified Data.ByteString.Lazy.Char8 as BL
 import Data.Text (Text)
@@ -17,6 +17,7 @@ data Options = Options
   , optJson      :: !Bool
   , optQuiet     :: !Bool
   , optInterProc :: !Bool
+  , optParseOnly :: !Bool
   } deriving (Eq, Show)
 
 optParser :: Parser Options
@@ -42,6 +43,10 @@ optParser = Options
      <> short 'p'
      <> help "Enable inter-procedural analysis"
       )
+  <*> switch
+      ( long "parse"
+     <> help "Parse only, no analysis"
+      )
 
 optInfo :: ParserInfo Options
 optInfo = info (optParser <**> helper)
@@ -53,16 +58,27 @@ optInfo = info (optParser <**> helper)
 main :: IO ()
 main = do
   opts <- execParser optInfo
-  let auditor = if optInterProc opts then auditFileInterProc else auditFile
-  result <- auditor (optInput opts)
-  case result of
-    Left err -> do
-      TIO.putStrLn $ "Error: " <> err
-      exitFailure
-    Right ar ->
-      if optJson opts
-        then outputJson ar
-        else outputText opts ar
+  if optParseOnly opts
+    then do
+      result <- parseFile (optInput opts)
+      case result of
+        Left err -> do
+          TIO.putStrLn $ "Error: " <> err
+          exitFailure
+        Right n -> do
+          TIO.putStrLn $ "Parsed " <> T.pack (show n) <> " lines"
+          exitSuccess
+    else do
+      let auditor = if optInterProc opts then auditFileInterProc else auditFile
+      result <- auditor (optInput opts)
+      case result of
+        Left err -> do
+          TIO.putStrLn $ "Error: " <> err
+          exitFailure
+        Right ar ->
+          if optJson opts
+            then outputJson ar
+            else outputText opts ar
 
 outputJson :: AuditResult -> IO ()
 outputJson ar = BL.putStrLn (encode (arViolations ar))
diff --git a/lib/Audit/AArch64.hs b/lib/Audit/AArch64.hs
@@ -33,6 +33,7 @@ module Audit.AArch64 (
   , auditInterProc
   , auditFile
   , auditFileInterProc
+  , parseFile
 
     -- * Results
   , AuditResult(..)
@@ -87,3 +88,14 @@ auditFileWith auditor path = do
       case auditor (T.pack path) src of
         Left err -> pure (Left (T.pack (show err)))
         Right result -> pure (Right result)
+
+-- | Parse an assembly file without analysis. Returns line count on success.
+parseFile :: FilePath -> IO (Either Text Int)
+parseFile path = do
+  bs <- BS.readFile path
+  case decodeUtf8' bs of
+    Left err -> pure (Left (T.pack (show err)))
+    Right src ->
+      case parseAsm src of
+        Left err -> pure (Left (T.pack (show err)))
+        Right lns -> pure (Right (length lns))
diff --git a/lib/Audit/AArch64/Parser.hs b/lib/Audit/AArch64/Parser.hs
@@ -433,7 +433,7 @@ pBracketAddr = lexeme $ do
   void (char ']')
   -- Check for pre-index ! or post-index offset
   preIdx <- option False (True <$ char '!')
-  postIdx <- optional pPostIndex
+  postIdx <- optional (try pPostIndex)
   case (innerMode, preIdx, postIdx) of
     (Nothing, False, Nothing) -> pure (BaseImm base 0)
     (Nothing, False, Just imm) -> pure (PostIndex base imm)
@@ -472,12 +472,16 @@ pShiftOrExtend = choice
   ]
 
 pExtend :: Parser Extend
-pExtend = lexeme $ choice
-  [ UXTW <$ string' "uxtw"
-  , SXTW <$ string' "sxtw"
-  , UXTX <$ string' "uxtx"
-  , SXTX <$ string' "sxtx"
-  ]
+pExtend = lexeme $ do
+  ext <- choice
+    [ UXTW <$ string' "uxtw"
+    , SXTW <$ string' "sxtw"
+    , UXTX <$ string' "uxtx"
+    , SXTX <$ string' "sxtx"
+    ]
+  -- Optionally consume shift amount (e.g., #3) - ignored for taint analysis
+  void $ optional (sc *> pImm)
+  pure ext
 
 pPostIndex :: Parser Integer
 pPostIndex = do