1 {-# LANGUAGE NoImplicitPrelude, PatternGuards #-}
2 -----------------------------------------------------------------------------
4 -- Module : GHC.IO.Encoding.Failure
5 -- Copyright : (c) The University of Glasgow, 2008-2011
6 -- License : see libraries/base/LICENSE
8 -- Maintainer : libraries@haskell.org
9 -- Stability : internal
10 -- Portability : non-portable
12 -- Types for specifying how text encoding/decoding fails
14 -----------------------------------------------------------------------------
16 module GHC.IO.Encoding.Failure (
17 CodingFailureMode(..), codingFailureModeSuffix,
19 surrogatifyRoundtripCharacter, desurrogatifyRoundtripCharacter,
20 recoverDecode, recoverEncode
25 import GHC.IO.Exception
31 import GHC.Real ( fromIntegral )
33 --import System.Posix.Internals
37 -- | The 'CodingFailureMode' is used to construct 'TextEncoding's, and specifies
38 -- how they handle illegal sequences.
39 data CodingFailureMode = ErrorOnCodingFailure -- ^ Throw an error when an illegal sequence is encountered
40 | IgnoreCodingFailure -- ^ Attempt to ignore and recover if an illegal sequence is encountered
41 | TransliterateCodingFailure -- ^ Replace with the closest visual match upon an illegal sequence
42 | RoundtripFailure -- ^ Use the private-use escape mechanism to attempt to allow illegal sequences to be roundtripped.
43 deriving (Show) -- This will only work properly for those encodings which are strict supersets of ASCII in the sense
44 -- that valid ASCII data is also valid in that encoding. This is not true for e.g. UTF-16, because
45 -- ASCII characters must be padded to two bytes to retain their meaning.
47 -- Note [Roundtripping]
48 -- ~~~~~~~~~~~~~~~~~~~~
50 -- Roundtripping is based on the ideas of PEP383. However, unlike PEP383 we do not wish to use lone surrogate codepoints
51 -- to escape undecodable bytes, because that may confuse Unicode processing software written in Haskell. Instead, we use
52 -- the range of private-use characters from 0xF1E00 to 0xF1EFF.
54 -- This introduces a technical problem when it comes to encoding back to bytes using iconv. The iconv code will not fail when
55 -- it tries to encode a private-use character (as it would if trying to encode a surrogate), which means that we won't get a
56 -- chance to replace it with the byte we originally escaped.
58 -- To work around this, when filling the buffer to be encoded (in writeBlocks/withEncodedCString/newEncodedCString), we replace
59 -- the private-use characters with lone surrogates again! Likewise, when reading from a buffer (unpack/unpack_nl/peekEncodedCString)
60 -- we have to do the inverse process.
62 -- The user of String should never see these lone surrogates, but it ensures that iconv will throw an error when encountering them.
63 -- We use lone surrogates in the range 0xDC00 to 0xDCFF for this purpose.
65 codingFailureModeSuffix :: CodingFailureMode -> String
66 codingFailureModeSuffix ErrorOnCodingFailure = ""
67 codingFailureModeSuffix IgnoreCodingFailure = "//IGNORE"
68 codingFailureModeSuffix TransliterateCodingFailure = "//TRANSLIT"
69 codingFailureModeSuffix RoundtripFailure = "//ROUNDTRIP"
71 -- | In transliterate mode, we use this character when decoding unknown bytes.
73 -- This is the defined Unicode replacement character: <http://www.fileformat.info/info/unicode/char/0fffd/index.htm>
74 unrepresentableChar :: Char
75 unrepresentableChar = '\xFFFD'
77 -- | Some characters are actually "surrogate" codepoints defined for use in UTF-16. We need to signal an
78 -- invalid character if we detect them when encoding a sequence of 'Char's into 'Word8's because they won't
79 -- give valid Unicode.
81 -- We may also need to signal an invalid character if we detect them when encoding a sequence of 'Char's into 'Word8's
82 -- because the 'RoundtripFailure' mode creates these to round-trip bytes through our internal UTF-16 encoding.
83 isSurrogate :: Char -> Bool
84 isSurrogate c = (0xD800 <= x && x <= 0xDBFF) || (0xDC00 <= x && x <= 0xDFFF)
87 -- | We use some private-use characters for roundtripping unknown bytes through a String
88 isRoundtripEscapeChar :: Char -> Bool
89 isRoundtripEscapeChar c = 0xF1E00 <= x && x < 0xF1F00
92 -- | We use some surrogate characters for roundtripping unknown bytes through a String
93 isRoundtripEscapeSurrogateChar :: Char -> Bool
94 isRoundtripEscapeSurrogateChar c = 0xDC00 <= x && x < 0xDD00
97 -- Private use characters (in Strings) --> lone surrogates (in Buffer CharBufElem)
98 surrogatifyRoundtripCharacter :: Char -> Char
99 surrogatifyRoundtripCharacter c | isRoundtripEscapeChar c = chr (ord c - 0xF1E00 + 0xDC00)
102 -- Lone surrogates (in Buffer CharBufElem) --> private use characters (in Strings)
103 desurrogatifyRoundtripCharacter :: Char -> Char
104 desurrogatifyRoundtripCharacter c | isRoundtripEscapeSurrogateChar c = chr (ord c - 0xDC00 + 0xF1E00)
107 -- Bytes (in Buffer Word8) --> lone surrogates (in Buffer CharBufElem)
108 escapeToRoundtripCharacterSurrogate :: Word8 -> Char
109 escapeToRoundtripCharacterSurrogate b
110 | b < 128 = chr (fromIntegral b) -- Disallow 'smuggling' of ASCII bytes. For roundtripping to work, this assumes encoding is ASCII-superset.
111 | otherwise = chr (0xDC00 + fromIntegral b)
113 -- Lone surrogates (in Buffer CharBufElem) --> bytes (in Buffer Word8)
114 unescapeRoundtripCharacterSurrogate :: Char -> Maybe Word8
115 unescapeRoundtripCharacterSurrogate c
116 | 0xDC80 <= x && x < 0xDD00 = Just (fromIntegral x) -- Discard high byte
117 | otherwise = Nothing
120 recoverDecode :: CodingFailureMode -> Buffer Word8 -> Buffer Char -> IO (Buffer Word8, Buffer Char)
121 recoverDecode cfm input@Buffer{ bufRaw=iraw, bufL=ir, bufR=_ }
122 output@Buffer{ bufRaw=oraw, bufL=_, bufR=ow } = do
123 --puts $ "recoverDecode " ++ show ir
125 ErrorOnCodingFailure -> ioe_decodingError
126 IgnoreCodingFailure -> return (input { bufL=ir+1 }, output)
127 TransliterateCodingFailure -> do
128 ow' <- writeCharBuf oraw ow unrepresentableChar
129 return (input { bufL=ir+1 }, output { bufR=ow' })
130 RoundtripFailure -> do
131 b <- readWord8Buf iraw ir
132 ow' <- writeCharBuf oraw ow (escapeToRoundtripCharacterSurrogate b)
133 return (input { bufL=ir+1 }, output { bufR=ow' })
135 recoverEncode :: CodingFailureMode -> Buffer Char -> Buffer Word8 -> IO (Buffer Char, Buffer Word8)
136 recoverEncode cfm input@Buffer{ bufRaw=iraw, bufL=ir, bufR=_ }
137 output@Buffer{ bufRaw=oraw, bufL=_, bufR=ow } = do
138 (c,ir') <- readCharBuf iraw ir
139 --puts $ "recoverEncode " ++ show ir ++ " " ++ show ir'
141 IgnoreCodingFailure -> return (input { bufL=ir' }, output)
142 TransliterateCodingFailure -> do
144 then return (input { bufL=ir' }, output)
146 -- XXX: evil hack! To implement transliteration, we just poke an
147 -- ASCII ? into the input buffer and tell the caller to try and decode
148 -- again. This is *probably* safe given current uses of TextEncoding.
150 -- The "if" test above ensures we skip if the encoding fails to deal with
151 -- the ?, though this should never happen in practice as all encodings are
152 -- in fact capable of reperesenting all ASCII characters.
153 _ir' <- writeCharBuf iraw ir '?'
154 return (input, output)
156 -- This implementation does not work because e.g. UTF-16 requires 2 bytes to
157 -- encode a simple ASCII value
158 --writeWord8Buf oraw ow unrepresentableByte
159 --return (input { bufL=ir' }, output { bufR=ow+1 })
160 RoundtripFailure | Just x <- unescapeRoundtripCharacterSurrogate c -> do
161 writeWord8Buf oraw ow x
162 return (input { bufL=ir' }, output { bufR=ow+1 })
163 _ -> ioe_encodingError
165 ioe_decodingError :: IO a
166 ioe_decodingError = ioException
167 (IOError Nothing InvalidArgument "recoverDecode"
168 "invalid byte sequence" Nothing Nothing)
170 ioe_encodingError :: IO a
171 ioe_encodingError = ioException
172 (IOError Nothing InvalidArgument "recoverEncode"
173 "invalid character" Nothing Nothing)