Lexer: report indentation warnings at column 1

We use a Latin1 generated parser with Alex, but we also parses Unicode
BOM, unbreakable spaces, etc. In recent Alex, the reported column isn't
expressed in Unicode chars anymore but in bytes/ASCII chars (probably
due to
haskell/alex@ae525e3
but I haven't checked), which broke our tests (see
haskell#8896).

To work around this we report indentation warnings at token start position,
instead of token end position (i.e. always 1). Otherwise position makes
no sense anymore for the user.

Cabal-syntax/src/Distribution/Fields/Lexer.hs

@@ -195,17 +195,17 @@ toki t pos  len  input = return $! L pos (t (B.take len input))
 tok :: Token -> Position -> Int -> ByteString -> Lex LToken
 tok  t pos _len _input = return $! L pos t
 
-checkLeadingWhitespace :: Int -> ByteString -> Lex Int
-checkLeadingWhitespace len bs
+checkLeadingWhitespace :: Position -> Int -> ByteString -> Lex Int
+checkLeadingWhitespace pos len bs
     | B.any (== 9) (B.take len bs) = do
-        addWarning LexWarningTab
-        checkWhitespace len bs
-    | otherwise = checkWhitespace len bs
+        addWarningAt pos LexWarningTab
+        checkWhitespace pos len bs
+    | otherwise = checkWhitespace pos len bs
 
-checkWhitespace :: Int -> ByteString -> Lex Int
-checkWhitespace len bs
+checkWhitespace :: Position -> Int -> ByteString -> Lex Int
+checkWhitespace pos len bs
     | B.any (== 194) (B.take len bs) = do
-        addWarning LexWarningNBSP
+        addWarningAt pos LexWarningNBSP
         return $ len - B.count 194 (B.take len bs)
     | otherwise = return len
 
@@ -313,12 +313,12 @@ bol_section = 3
 in_field_braces = 4
 in_field_layout = 5
 in_section = 6
-alex_action_0 = \_ len _ -> do
-              when (len /= 0) $ addWarning LexWarningBOM
+alex_action_0 = \pos len _ -> do
+              when (len /= 0) $ addWarningAt pos LexWarningBOM
               setStartCode bol_section
               lexToken
-alex_action_1 = \_pos len inp -> checkWhitespace len inp >> adjustPos retPos >> lexToken
-alex_action_3 = \pos len inp -> checkLeadingWhitespace len inp >>
+alex_action_1 = \pos len inp -> checkWhitespace pos len inp >> adjustPos retPos >> lexToken
+alex_action_3 = \pos len inp -> checkLeadingWhitespace pos len inp >>
                                      if B.length inp == len
                                        then return (L pos EOF)
                                        else setStartCode in_section
@@ -333,7 +333,7 @@ alex_action_12 = tok  Colon
 alex_action_13 = tok  OpenBrace
 alex_action_14 = tok  CloseBrace
 alex_action_15 = \_ _ _ -> adjustPos retPos >> setStartCode bol_section >> lexToken
-alex_action_16 = \pos len inp -> checkLeadingWhitespace len inp >>= \len' ->
+alex_action_16 = \pos len inp -> checkLeadingWhitespace pos len inp >>= \len' ->
                                   if B.length inp == len
                                     then return (L pos EOF)
                                     else setStartCode in_field_layout

Cabal-syntax/src/Distribution/Fields/LexerMonad.hs

@@ -27,6 +27,7 @@ module Distribution.Fields.LexerMonad (
     LexWarning(..),
     LexWarningType(..),
     addWarning,
+    addWarningAt,
     toPWarnings,
 
   ) where
@@ -153,3 +154,8 @@ setStartCode c = Lex $ \s -> LexResult s{ curCode = c } ()
 addWarning :: LexWarningType -> Lex ()
 addWarning wt = Lex $ \s@LexState{ curPos = pos, warnings = ws  } ->
     LexResult s{ warnings = LexWarning wt pos : ws } ()
+
+-- | Add warning at specific position
+addWarningAt :: Position -> LexWarningType -> Lex ()
+addWarningAt pos wt = Lex $ \s@LexState{ warnings = ws  } ->
+    LexResult s{ warnings = LexWarning wt pos : ws } ()

Cabal-tests/tests/ParserTests/regressions/Octree-0.5.format

@@ -1,4 +1,4 @@
-Octree-0.5.cabal:39:5: Non breaking spaces at 39:5, 41:5, 43:5
+Octree-0.5.cabal:39:1: Non breaking spaces at 39:1, 41:1, 43:1
 cabal-version: >=1.8
 name:          Octree
 version:       0.5

Cabal-tests/tests/ParserTests/regressions/monad-param.format

@@ -1,4 +1,4 @@
-monad-param.cabal:19:3: Tabs used as indentation at 19:3, 20:3
+monad-param.cabal:19:1: Tabs used as indentation at 19:1, 20:1
 name:         monad-param
 version:      0.0.1
 license:      BSD3

Cabal-tests/tests/ParserTests/regressions/th-lift-instances.format

@@ -1,4 +1,4 @@
-th-lift-instances.cabal:15:9: Tabs used as indentation at 15:9
+th-lift-instances.cabal:15:1: Tabs used as indentation at 15:1
 cabal-version:      >=1.10
 name:               th-lift-instances
 version:            0.1.4

templates/Lexer.x

@@ -83,22 +83,22 @@ $instresc        = $printable
 tokens :-
 
 <0> {
-  @bom?  { \_ len _ -> do
-              when (len /= 0) $ addWarning LexWarningBOM
+  @bom?  { \pos len _ -> do
+              when (len /= 0) $ addWarningAt pos LexWarningBOM
               setStartCode bol_section
               lexToken
          }
 }
 
 <bol_section, bol_field_layout, bol_field_braces> {
-  @nbspspacetab* @nl         { \_pos len inp -> checkWhitespace len inp >> adjustPos retPos >> lexToken }
+  @nbspspacetab* @nl         { \pos len inp -> checkWhitespace pos len inp >> adjustPos retPos >> lexToken }
   -- no @nl here to allow for comments on last line of the file with no trailing \n
   $spacetab* "--" $comment*  ;  -- TODO: check the lack of @nl works here
                                 -- including counting line numbers
 }
 
 <bol_section> {
-  @nbspspacetab*   { \pos len inp -> checkLeadingWhitespace len inp >>
+  @nbspspacetab*   { \pos len inp -> checkLeadingWhitespace pos len inp >>
                                      if B.length inp == len
                                        then return (L pos EOF)
                                        else setStartCode in_section
@@ -123,7 +123,7 @@ tokens :-
 }
 
 <bol_field_layout> {
-  @nbspspacetab* { \pos len inp -> checkLeadingWhitespace len inp >>= \len' ->
+  @nbspspacetab* { \pos len inp -> checkLeadingWhitespace pos len inp >>= \len' ->
                                   if B.length inp == len
                                     then return (L pos EOF)
                                     else setStartCode in_field_layout
@@ -172,17 +172,17 @@ toki t pos  len  input = return $! L pos (t (B.take len input))
 tok :: Token -> Position -> Int -> ByteString -> Lex LToken
 tok  t pos _len _input = return $! L pos t
 
-checkLeadingWhitespace :: Int -> ByteString -> Lex Int
-checkLeadingWhitespace len bs
+checkLeadingWhitespace :: Position -> Int -> ByteString -> Lex Int
+checkLeadingWhitespace pos len bs
     | B.any (== 9) (B.take len bs) = do
-        addWarning LexWarningTab
-        checkWhitespace len bs
-    | otherwise = checkWhitespace len bs
+        addWarningAt pos LexWarningTab
+        checkWhitespace pos len bs
+    | otherwise = checkWhitespace pos len bs
 
-checkWhitespace :: Int -> ByteString -> Lex Int
-checkWhitespace len bs
+checkWhitespace :: Position -> Int -> ByteString -> Lex Int
+checkWhitespace pos len bs
     | B.any (== 194) (B.take len bs) = do
-        addWarning LexWarningNBSP
+        addWarningAt pos LexWarningNBSP
         return $ len - B.count 194 (B.take len bs)
     | otherwise = return len