ApiSurrogatesUtf8.hs

{-|
Description:    Emulating tests from @encoding/api-surrogates-utf8.any.js@

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

Stability:      experimental
Portability:    portable
-}
module Test.Willow.WebPlatformTests.Manual.Encoding.ApiSurrogatesUtf8
    ( tests
    ) where


import qualified Data.ByteString as BS
import qualified Data.Text as T

import qualified Test.HUnit as U

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

import Web.Willow.Common.Encoding


tests :: U.Test
tests = "api-surrogates-utf8.any.js" ~: U.TestList
    [ "Invalid surrogates encoded into UTF-8" ~: U.TestList (map surrogates badStrings)
    ]


surrogates :: SurrogateTest -> U.Test
surrogates s = name s ~: U.TestCase $ do
    fst (encodeUtf8 $ T.pack $ input s) @?= expected s
    fst (decodeUtf8' $ expected s) @?= decoded s


data SurrogateTest = SurrogateTest
    { input :: String
    , expected :: BS.ByteString
    , decoded :: T.Text
    , name :: String
    }

badStrings :: [SurrogateTest]
badStrings =
    [ SurrogateTest
        { input = "abc123"
        , expected = BS.pack [0x61, 0x62, 0x63, 0x31, 0x32, 0x33]
        , decoded = T.pack "abc123"
        , name = "Sanity check"
        }
    , SurrogateTest
        { input = "\xD800"
        , expected = BS.pack [0xef, 0xbf, 0xbd]
        , decoded = T.pack "\xFFFD"
        , name = "Surrogate half (low)"
        }
    , SurrogateTest
        { input = "\xDC00"
        , expected = BS.pack [0xef, 0xbf, 0xbd]
        , decoded = T.pack "\xFFFD"
        , name = "Surrogate half (high)"
        }
    , SurrogateTest
        { input = "abc\xD800\&123"
        , expected = BS.pack [0x61, 0x62, 0x63, 0xef, 0xbf, 0xbd, 0x31, 0x32, 0x33]
        , decoded = T.pack "abc\xFFFD\&123"
        , name = "Surrogate half (low), in a string"
        }
    , SurrogateTest
        { input = "abc\xDC00\&123"
        , expected = BS.pack [0x61, 0x62, 0x63, 0xef, 0xbf, 0xbd, 0x31, 0x32, 0x33]
        , decoded = T.pack "abc\xFFFD\&123"
        , name = "Surrogate half (high), in a string"
        }
    , SurrogateTest
        { input = "\xDC00\xD800"
        , expected = BS.pack [0xef, 0xbf, 0xbd, 0xef, 0xbf, 0xbd]
        , decoded = T.pack "\xFFFD\xFFFD"
        , name = "Wrong order"
        }
    ]