regression: NBSP can occur in other place too

Author: phadej

Cabal/Cabal.cabal

@@ -32,6 +32,7 @@ extra-source-files:
   -- Generated with 'misc/gen-extra-source-files.sh'
   -- Do NOT edit this section manually; instead, run the script.
   -- BEGIN gen-extra-source-files
+  tests/ParserTests/regressions/Octree-0.5.cabal
   tests/ParserTests/regressions/encoding-0.8.cabal
   tests/ParserTests/warnings/bom.cabal
   tests/ParserTests/warnings/bool.cabal

Cabal/Distribution/PackageDescription/Check.hs

@@ -55,6 +55,7 @@ import Distribution.Simple.Utils hiding (findPackageDesc, notice)
 import Distribution.Version
 import Distribution.Package
 import Distribution.Text
+import Distribution.Utils.Generic (isAscii)
 import Language.Haskell.Extension
 
 import Control.Monad (mapM)
@@ -146,6 +147,7 @@ checkPackage gpkg mpkg =
   ++ checkConditionals gpkg
   ++ checkPackageVersions gpkg
   ++ checkDevelopmentOnlyFlags gpkg
+  ++ checkFlagNames gpkg
   where
     pkg = fromMaybe (flattenPackageDescription gpkg) mpkg
 
@@ -1584,6 +1586,27 @@ checkConditionals pkg =
       COr  c1 c2 -> condfv c1 ++ condfv c2
       CAnd c1 c2 -> condfv c1 ++ condfv c2
 
+checkFlagNames ::GenericPackageDescription -> [PackageCheck]
+checkFlagNames gpd
+    | null invalidFlagNames = []
+    | otherwise             = [ PackageDistSuspicious
+        $ "Suspicious flag names: " ++ unwords invalidFlagNames ++ ". "
+        ++ "To avoid ambiguity in command line interfaces, flag shouldn't "
+        ++ "start with a dash. Also for better compatibility, flag names "
+        ++ "shouldn't contain non-ascii characters."
+        ]
+  where
+    invalidFlagNames =
+        [ fn
+        | flag <- genPackageFlags gpd
+        , let fn = unFlagName (flagName flag)
+        , invalidFlagName fn
+        ]
+    -- starts with dash
+    invalidFlagName ('-':_) = True
+    -- mon ascii letter
+    invalidFlagName cs = any (\c -> not (isAscii c) && isAlpha c) cs
+
 checkDevelopmentOnlyFlagsBuildInfo :: BuildInfo -> [PackageCheck]
 checkDevelopmentOnlyFlagsBuildInfo bi =
   catMaybes [

Cabal/Distribution/Parsec/Lexer.hs

@@ -197,6 +197,8 @@ problematicFiles =
     , eq "ixset/1.0.4/ixset.cabal"
     -- comments in braces
     , isPrefixOf "hint/"
+    -- other-modules:\n  .
+    , eq "unicode-transforms/0.3.3/unicode-transforms.cabal"
     ]
   where
     eq = (==)

Cabal/tests/ParserHackageTests.hs

@@ -16,8 +16,13 @@ import qualified Data.ByteString as BS
 tests :: TestTree
 tests = testGroup "parsec tests"
     [ warningTests
+    , regressionTests
     ]
 
+-------------------------------------------------------------------------------
+-- Warnings
+-------------------------------------------------------------------------------
+
 -- Verify that we trigger warnings
 warningTests :: TestTree
 warningTests = testGroup "warnings triggered"
@@ -52,5 +57,28 @@ warningTest wt fp = testCase (show wt) $ do
         []                 -> assertFailure "got no warnings"
         _                  -> assertFailure $ "got multiple warnings: " ++ show warns
 
+-------------------------------------------------------------------------------
+-- Regressions
+-------------------------------------------------------------------------------
+
+regressionTests :: TestTree
+regressionTests = testGroup "regressions"
+    [ regressionTest "encoding-0.8.cabal"
+    , regressionTest "Octree-0.5.cabal"
+    ]
+
+regressionTest :: FilePath -> TestTree
+regressionTest fp = testCase fp $ do
+    contents <- BS.readFile $ "tests" </> "ParserTests" </> "regressions" </> fp
+    let res =  parseGenericPackageDescription contents
+    let (_, errs, x) = runParseResult res
+
+    assertBool ("parses successfully: " ++ show errs) $ isJust x
+    assertBool ("parses without errors: " ++ show errs) $ null errs
+
+-------------------------------------------------------------------------------
+-- Main
+-------------------------------------------------------------------------------
+
 main :: IO ()
 main = defaultMain tests

Cabal/tests/ParserTests.hs

@@ -0,0 +1,44 @@
+name:                Octree
+version:             0.5
+stability:           beta
+homepage:            https://github.com/mgajda/octree
+package-url:         http://hackage.haskell.org/package/octree
+synopsis:            Simple unbalanced Octree for storing data about 3D points
+description:         Octree data structure is relatively shallow data structure for space partitioning.
+category:            Data
+license:             BSD3
+license-file:        LICENSE
+
+author:              Michal J. Gajda
+copyright:           Copyright by Michal J. Gajda '2012
+maintainer:          [email protected]
+bug-reports:         mailto:[email protected]
+
+
+build-type:          Simple
+cabal-version:       >=1.8
+tested-with:         GHC==7.0.4,GHC==7.4.1,GHC==7.4.2,GHC==7.6.0
+
+source-repository head
+  type:     git
+  location: [email protected]:mgajda/octree.git
+
+Library
+   build-depends:    base>=4.0 && < 4.7, AC-Vector >= 2.3.0, QuickCheck >= 2.4.0
+   exposed-modules:  Data.Octree
+   other-modules:    Data.Octree.Internal
+   exposed:          True
+   extensions:       ScopedTypeVariables
+
+Test-suite test_Octree
+  Type:              exitcode-stdio-1.0
+  Build-depends:     base>=4.0 && < 4.7, AC-Vector >= 2.3.0, QuickCheck >= 2.4.0
+  Main-is:           tests/test_Octree.hs
+
+Test-suite readme
+  type:           exitcode-stdio-1.0
+  -- We have a symlink: README.lhs -> README.md
+  main-is:        README.lhs
+  Build-depends:  base>=4.0 && < 4.7, AC-Vector >= 2.3.0, QuickCheck >= 2.4.0, markdown-unlit
+  ghc-options:    -pgmL markdown-unlit
+

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

@@ -0,0 +1,12 @@
+Name:   encoding
+Version:  0.8
+
+custom-setup
+  setup-depends:
+    base < 5,
+    ghc-prim
+
+Library
+  build-depends: base
+  Exposed-Modules:
+    Data.Encoding

Cabal/tests/ParserTests/regressions/encoding-0.8.cabal

@@ -125,11 +125,12 @@ tokens :-
 }
 
 <bol_field_layout> {
-  $spacetab*   --TODO prevent or record leading tabs
-                { \pos len inp -> if B.length inp == len
+  $nbspspacetab*   --TODO prevent or record leading tabs
+                { \pos len inp -> checkWhitespace len inp >>= \len' ->
+                                  if B.length inp == len
                                     then return (L pos EOF)
                                     else setStartCode in_field_layout
-                                      >> return (L pos (Indent len)) }
+                                      >> return (L pos (Indent len')) }
 }
 
 <in_field_layout> {
@@ -175,11 +176,12 @@ toki t pos  len  input = return $! L pos (t (B.take len input))
 tok :: Monad m => Token -> Position -> t -> t1 -> m LToken
 tok  t pos _len _input = return $! L pos t
 
-checkWhitespace :: Int -> ByteString -> Lex ()
+checkWhitespace :: Int -> ByteString -> Lex Int
 checkWhitespace len bs
-    | B.any (== 194) (B.take len bs) =
-          addWarning LexWarningNBSP "Non-breaking space found"
-    | otherwise = return ()
+    | B.any (== 194) (B.take len bs) = do
+        addWarning LexWarningNBSP "Non-breaking space found"
+        return $ len - B.count 194 (B.take len bs)
+    | otherwise = return len
 
 -- -----------------------------------------------------------------------------
 -- The input type