{-|
Description:    

Copyright:      (c) 2020 Samuel May
License:        MPL-2.0
Maintainer:     ag.eitilt@gmail.com

Stability:      experimental
Portability:    portable
-}
module Test.Willow.Unit.Encoding.Utf8
    ( tests
    ) where


import qualified Data.ByteString.Builder as BS.B
import qualified Data.ByteString.Lazy as BS.L
import qualified Data.Either as E

import qualified Test.HUnit as U

import Test.HUnit ( (~:), (~?=) )

import Web.Willow.Common.Encoding

import Test.Willow.Unit.Encoding.Common


tests :: U.Test
tests = "UTF-8 encoding" ~: U.TestList
    [ "decode" ~: U.TestList $
        [ "ASCII" ~: U.TestList $ map (decodeAscii Utf8) ['\NUL'..'\DEL']
        , "failures" ~: U.TestList $ map (\(c, b) -> decodeTest Utf8 c b Nothing)
            [ ([toEnum i], [fromIntegral i])
            | i <- [0x80..0xC1] ++ [0xF5..0xFF]
            ] ++ ["surrogates" ~: U.TestList $
            [ surrogatesInvalid c
            | c <- ['\xD800'..'\xDFFF']
            ]]
        ]
    , "encode" ~: U.TestList $
        [ "ASCII" ~: U.TestList $ map (encodeAscii Utf8) ['\NUL'..'\DEL']
        ]
    ]


surrogatesInvalid :: Char -> U.Test
surrogatesInvalid surrogate = show surrogate ~:
    filter E.isRight (fst . decodeUtf8NoBom . BS.L.toStrict . BS.B.toLazyByteString $ BS.B.charUtf8 surrogate) ~?= []