diff --git a/Data/ByteString/ShellEscape.hs b/Data/ByteString/ShellEscape.hs new file mode 100644 --- /dev/null +++ b/Data/ByteString/ShellEscape.hs @@ -0,0 +1,16 @@ + +{-| Typed shell escaping for Bourne Shell and Bash. + -} + +module Data.ByteString.ShellEscape + ( Data.ByteString.ShellEscape.Escape.Escape(..) + , Data.ByteString.ShellEscape.Sh.Sh() + , Data.ByteString.ShellEscape.Sh.sh + , Data.ByteString.ShellEscape.Bash.Bash() + , Data.ByteString.ShellEscape.Bash.bash + ) where + +import Data.ByteString.ShellEscape.Escape +import Data.ByteString.ShellEscape.Sh +import Data.ByteString.ShellEscape.Bash + diff --git a/Data/ByteString/ShellEscape/Bash.hs b/Data/ByteString/ShellEscape/Bash.hs new file mode 100644 --- /dev/null +++ b/Data/ByteString/ShellEscape/Bash.hs @@ -0,0 +1,78 @@ + +module Data.ByteString.ShellEscape.Bash where + +import Data.Maybe +import Data.Char +import Text.Printf +import Data.ByteString (ByteString) + +import qualified Data.Vector as Vector + +import Data.ByteString.ShellEscape.Escape +import qualified Data.ByteString.ShellEscape.Put as Put +import Data.ByteString.ShellEscape.EscapeVector + +{-| A Bash escaped 'ByteString'. The strings are wrapped in @$\'...\'@ if any + bytes within them must be escaped; otherwise, they are left as is. + Newlines and other control characters are represented as ANSI escape + sequences. High bytes are represented as hex codes. Thus Bash escaped + strings will always fit on one line and never contain non-ASCII bytes. + -} +newtype Bash = Bash (EscapeVector EscapingMode) + deriving (Eq, Ord, Show) + +{-| Construct a Bash escaped intermediate form. + -} +bash :: ByteString -> Bash +bash = escape + +instance Escape Bash where + escape = Bash . escWith classify + unescape (Bash v) = stripEsc v + bytes (Bash v) | literal v = stripEsc v + | otherwise = interpretEsc v renderANSI' end (begin, Literal) + where + literal = isNothing . Vector.find ((/= Literal) . snd) + begin = [ Put.putString "$'"] + end = const (Put.putChar '\'') + renderANSI' _ (c, e) = (renderANSI c, e) + + +{-| Bash escaping modes. + -} +data EscapingMode = ANSIHex | ANSIBackslash | Literal | Quoted + deriving (Eq, Ord, Show) + +renderANSI c = + case classify c of + Literal -> Put.putChar c + Quoted -> Put.putChar c + ANSIHex -> Put.putString $ hexify c + ANSIBackslash -> Put.putString $ backslashify c + +backslashify :: Char -> String +backslashify c = (take 2 . drop 1 . show) c + +hexify :: Char -> String +hexify = printf "\\x%02X" . ord + +classify :: Char -> EscapingMode +classify c | c <= '\ACK' = ANSIHex + | c <= '\r' = ANSIBackslash + | c <= '\US' = ANSIHex + | c <= '&' = Quoted + | c == '\'' = ANSIBackslash + | c <= '+' = Quoted + | c <= '9' = Literal + | c <= '?' = Quoted + | c <= 'Z' = Literal + | c == '[' = Quoted + | c == '\\' = ANSIBackslash + | c <= ']' = Quoted + | c == '_' = Literal + | c <= '`' = Quoted + | c <= 'z' = Literal + | c <= '~' = Quoted + | c == '\DEL' = ANSIHex + | otherwise = ANSIHex + diff --git a/Data/ByteString/ShellEscape/Escape.hs b/Data/ByteString/ShellEscape/Escape.hs new file mode 100644 --- /dev/null +++ b/Data/ByteString/ShellEscape/Escape.hs @@ -0,0 +1,21 @@ + +module Data.ByteString.ShellEscape.Escape where + +import Data.ByteString + +{-| A type class for objects that represent an intermediate state of + escaping. + -} +class Escape t where + -- | Transform a 'ByteString' into the escaped intermediate form. + escape :: ByteString -> t + -- | Recover the original 'ByteString'. + unescape :: t -> ByteString + -- | Yield the escaped 'ByteString'. + bytes :: t -> ByteString + +instance Escape ByteString where + escape = id + unescape = id + bytes = id + diff --git a/Data/ByteString/ShellEscape/EscapeVector.hs b/Data/ByteString/ShellEscape/EscapeVector.hs new file mode 100644 --- /dev/null +++ b/Data/ByteString/ShellEscape/EscapeVector.hs @@ -0,0 +1,39 @@ + +module Data.ByteString.ShellEscape.EscapeVector where + +import Data.ByteString.Char8 (ByteString) +import qualified Data.ByteString.Char8 as ByteString + +import Data.Vector (Vector) +import qualified Data.Vector as Vector +import qualified Data.Vector.Mutable as Vector (new, write) + +import qualified Data.ByteString.ShellEscape.Put as Put + + +type EscapeVector escapingMode = Vector (Char, escapingMode) + +escWith :: (Char -> escapingMode) -> ByteString -> EscapeVector escapingMode +escWith cf b = Vector.create $ do + v <- Vector.new (ByteString.length b) + sequence_ . snd $ ByteString.foldl' (f v) (0, []) b + return v + where + f v (i, ops) c = (i + 1, Vector.write v i (c, cf c) : ops) + + +stripEsc :: Vector (Char, escapingMode) -> ByteString +stripEsc v = ByteString.unfoldr f . fst $ Vector.unzip v + where + f v | Vector.null v = Nothing + | otherwise = Just (Vector.unsafeHead v, Vector.unsafeTail v) + +interpretEsc v f finish init = (eval . (finish lastMode :)) instructions + where + eval = Put.runPut' . sequence_ . reverse + (instructions, lastMode) = Vector.foldl' f' init v + where + f' (list, mode) (c, e) = (put:list, mode') + where + (put, mode') = f mode (c, e) + diff --git a/Data/ByteString/ShellEscape/Put.hs b/Data/ByteString/ShellEscape/Put.hs new file mode 100644 --- /dev/null +++ b/Data/ByteString/ShellEscape/Put.hs @@ -0,0 +1,24 @@ + +module Data.ByteString.ShellEscape.Put + ( module Data.Binary.Put + , putChar + , putString + , runPut' + ) where + +import Prelude hiding (putChar) +import Data.Binary.Put +import Data.Char +import Data.ByteString.Lazy +import Data.ByteString +import Data.ByteString.Char8 + +putChar :: Char -> Put +putChar = putWord8 . toEnum . fromEnum + +putString :: String -> Put +putString = putByteString . Data.ByteString.Char8.pack + +runPut' :: Put -> Data.ByteString.ByteString +runPut' = Data.ByteString.concat . Data.ByteString.Lazy.toChunks . runPut + diff --git a/Data/ByteString/ShellEscape/Sh.hs b/Data/ByteString/ShellEscape/Sh.hs new file mode 100644 --- /dev/null +++ b/Data/ByteString/ShellEscape/Sh.hs @@ -0,0 +1,69 @@ + +module Data.ByteString.ShellEscape.Sh where + +import Data.ByteString (ByteString) + +import Data.ByteString.ShellEscape.Escape +import qualified Data.ByteString.ShellEscape.Put as Put +import Data.ByteString.ShellEscape.EscapeVector + + +{-| A Bourne Shell escaped 'ByteString'. An oddity of Bourne shell escaping is + the absence of escape codes for newline and other ASCII control + characters. These bytes are simply placed literally in single quotes; the + effect is that a Bourne Shell escaped string may cover several lines and + contain non-ASCII bytes. Runs of bytes that must be escaped are wrapped in + @\'...\'@; bytes that are acceptable as literals in Bourne Shell are left + as is. + -} +newtype Sh = Sh (EscapeVector EscapingMode) + deriving (Eq, Ord, Show) + +{-| Construct a Bourne Shell escaped intermediate form. + -} +sh :: ByteString -> Sh +sh = escape + +instance Escape Sh where + escape = Sh . escWith classify + unescape (Sh v) = stripEsc v + bytes (Sh v) = interpretEsc v act finish ([], Literal) + where + finish Quote = Put.putChar '\'' + finish Backslash = Put.putChar '\\' + finish Literal = return () + + +{-| Accept the present escaping mode and desired escaping mode and yield an + action and the resulting mode. + -} +act :: EscapingMode -> (Char, EscapingMode) -> (Put.Put, EscapingMode) +act Quote (c, Quote) = (Put.putChar c , Quote) +act Quote (c, Literal) = (Put.putChar c , Quote) +act Quote (c, Backslash) = (Put.putString ['\'', '\\', c] , Literal) +act Backslash (c, Backslash) = (Put.putChar c , Literal) +act Backslash (c, Quote) = (Put.putString ['\\', '\'', c] , Quote) +act Backslash (c, Literal) = (Put.putString ['\\', c] , Literal) +act Literal (c, Literal) = (Put.putChar c , Literal) +act Literal (c, Backslash) = (Put.putString ['\\', c] , Literal) +act Literal (c, Quote) = (Put.putString ['\'', c] , Quote) + +classify :: Char -> EscapingMode +classify c | c <= '&' = Quote -- 0x00..0x26 + | c == '\'' = Backslash -- 0x27 + | c <= ',' = Quote -- 0x28..0x2c + | c <= '9' = Literal -- 0x2d..0x39 + | c <= '?' = Quote -- 0x3a..0x3f + | c <= 'Z' = Literal -- 0x40..0x5a + | c <= '^' = Quote -- 0x5b..0x5e + | c == '_' = Literal -- 0x5f + | c == '`' = Quote -- 0x60 + | c <= 'z' = Literal -- 0x61..0x7a + | c <= '\DEL' = Quote -- 0x7b..0x7f + | otherwise = Quote -- 0x80..0xff + +{-| Bourne Shell escaping modes. + -} +data EscapingMode = Backslash | Literal | Quote + deriving (Eq, Ord, Show) + diff --git a/Text/ShellEscape.hs b/Text/ShellEscape.hs --- a/Text/ShellEscape.hs +++ b/Text/ShellEscape.hs @@ -3,14 +3,14 @@ -} module Text.ShellEscape - ( Text.ShellEscape.Escape.Escape(..) - , Text.ShellEscape.Sh.Sh() - , Text.ShellEscape.Sh.sh - , Text.ShellEscape.Bash.Bash() - , Text.ShellEscape.Bash.bash + ( Data.ByteString.ShellEscape.Escape.Escape(..) + , Data.ByteString.ShellEscape.Sh.Sh() + , Data.ByteString.ShellEscape.Sh.sh + , Data.ByteString.ShellEscape.Bash.Bash() + , Data.ByteString.ShellEscape.Bash.bash ) where -import Text.ShellEscape.Escape -import Text.ShellEscape.Sh -import Text.ShellEscape.Bash +import Data.ByteString.ShellEscape.Escape +import Data.ByteString.ShellEscape.Sh +import Data.ByteString.ShellEscape.Bash diff --git a/Text/ShellEscape/Bash.hs b/Text/ShellEscape/Bash.hs deleted file mode 100644 --- a/Text/ShellEscape/Bash.hs +++ /dev/null @@ -1,78 +0,0 @@ - -module Text.ShellEscape.Bash where - -import Data.Maybe -import Data.Char -import Text.Printf -import Data.ByteString (ByteString) - -import qualified Data.Vector as Vector - -import Text.ShellEscape.Escape -import qualified Text.ShellEscape.Put as Put -import Text.ShellEscape.EscapeVector - -{-| A Bash escaped 'ByteString'. The strings are wrapped in @$\'...\'@ if any - bytes within them must be escaped; otherwise, they are left as is. - Newlines and other control characters are represented as ANSI escape - sequences. High bytes are represented as hex codes. Thus Bash escaped - strings will always fit on one line and never contain non-ASCII bytes. - -} -newtype Bash = Bash (EscapeVector EscapingMode) - deriving (Eq, Ord, Show) - -{-| Construct a Bash escaped intermediate form. - -} -bash :: ByteString -> Bash -bash = escape - -instance Escape Bash where - escape = Bash . escWith classify - unescape (Bash v) = stripEsc v - bytes (Bash v) | literal v = stripEsc v - | otherwise = interpretEsc v renderANSI' end (begin, Literal) - where - literal = isNothing . Vector.find ((/= Literal) . snd) - begin = [ Put.putString "$'"] - end = const (Put.putChar '\'') - renderANSI' _ (c, e) = (renderANSI c, e) - - -{-| Bash escaping modes. - -} -data EscapingMode = ANSIHex | ANSIBackslash | Literal | Quoted - deriving (Eq, Ord, Show) - -renderANSI c = - case classify c of - Literal -> Put.putChar c - Quoted -> Put.putChar c - ANSIHex -> Put.putString $ hexify c - ANSIBackslash -> Put.putString $ backslashify c - -backslashify :: Char -> String -backslashify c = (take 2 . drop 1 . show) c - -hexify :: Char -> String -hexify = printf "\\x%02X" . ord - -classify :: Char -> EscapingMode -classify c | c <= '\ACK' = ANSIHex - | c <= '\r' = ANSIBackslash - | c <= '\US' = ANSIHex - | c <= '&' = Quoted - | c == '\'' = ANSIBackslash - | c <= '+' = Quoted - | c <= '9' = Literal - | c <= '?' = Quoted - | c <= 'Z' = Literal - | c == '[' = Quoted - | c == '\\' = ANSIBackslash - | c <= ']' = Quoted - | c == '_' = Literal - | c <= '`' = Quoted - | c <= 'z' = Literal - | c <= '~' = Quoted - | c == '\DEL' = ANSIHex - | otherwise = ANSIHex - diff --git a/Text/ShellEscape/Escape.hs b/Text/ShellEscape/Escape.hs deleted file mode 100644 --- a/Text/ShellEscape/Escape.hs +++ /dev/null @@ -1,21 +0,0 @@ - -module Text.ShellEscape.Escape where - -import Data.ByteString - -{-| A type class for objects that represent an intermediate state of - escaping. - -} -class Escape t where - -- | Transform a 'ByteString' into the escaped intermediate form. - escape :: ByteString -> t - -- | Recover the original 'ByteString'. - unescape :: t -> ByteString - -- | Yield the escaped 'ByteString'. - bytes :: t -> ByteString - -instance Escape ByteString where - escape = id - unescape = id - bytes = id - diff --git a/Text/ShellEscape/EscapeVector.hs b/Text/ShellEscape/EscapeVector.hs deleted file mode 100644 --- a/Text/ShellEscape/EscapeVector.hs +++ /dev/null @@ -1,39 +0,0 @@ - -module Text.ShellEscape.EscapeVector where - -import Data.ByteString.Char8 (ByteString) -import qualified Data.ByteString.Char8 as ByteString - -import Data.Vector (Vector) -import qualified Data.Vector as Vector -import qualified Data.Vector.Mutable as Vector (new, write) - -import qualified Text.ShellEscape.Put as Put - - -type EscapeVector escapingMode = Vector (Char, escapingMode) - -escWith :: (Char -> escapingMode) -> ByteString -> EscapeVector escapingMode -escWith cf b = Vector.create $ do - v <- Vector.new (ByteString.length b) - sequence_ . snd $ ByteString.foldl' (f v) (0, []) b - return v - where - f v (i, ops) c = (i + 1, Vector.write v i (c, cf c) : ops) - - -stripEsc :: Vector (Char, escapingMode) -> ByteString -stripEsc v = ByteString.unfoldr f . fst $ Vector.unzip v - where - f v | Vector.null v = Nothing - | otherwise = Just (Vector.unsafeHead v, Vector.unsafeTail v) - -interpretEsc v f finish init = (eval . (finish lastMode :)) instructions - where - eval = Put.runPut' . sequence_ . reverse - (instructions, lastMode) = Vector.foldl' f' init v - where - f' (list, mode) (c, e) = (put:list, mode') - where - (put, mode') = f mode (c, e) - diff --git a/Text/ShellEscape/Put.hs b/Text/ShellEscape/Put.hs deleted file mode 100644 --- a/Text/ShellEscape/Put.hs +++ /dev/null @@ -1,24 +0,0 @@ - -module Text.ShellEscape.Put - ( module Data.Binary.Put - , putChar - , putString - , runPut' - ) where - -import Prelude hiding (putChar) -import Data.Binary.Put -import Data.Char -import Data.ByteString.Lazy -import Data.ByteString -import Data.ByteString.Char8 - -putChar :: Char -> Put -putChar = putWord8 . toEnum . fromEnum - -putString :: String -> Put -putString = putByteString . Data.ByteString.Char8.pack - -runPut' :: Put -> Data.ByteString.ByteString -runPut' = Data.ByteString.concat . Data.ByteString.Lazy.toChunks . runPut - diff --git a/Text/ShellEscape/Sh.hs b/Text/ShellEscape/Sh.hs deleted file mode 100644 --- a/Text/ShellEscape/Sh.hs +++ /dev/null @@ -1,69 +0,0 @@ - -module Text.ShellEscape.Sh where - -import Data.ByteString (ByteString) - -import Text.ShellEscape.Escape -import qualified Text.ShellEscape.Put as Put -import Text.ShellEscape.EscapeVector - - -{-| A Bourne Shell escaped 'ByteString'. An oddity of Bourne shell escaping is - the absence of escape codes for newline and other ASCII control - characters. These bytes are simply placed literally in single quotes; the - effect is that a Bourne Shell escaped string may cover several lines and - contain non-ASCII bytes. Runs of bytes that must be escaped are wrapped in - @\'...\'@; bytes that are acceptable as literals in Bourne Shell are left - as is. - -} -newtype Sh = Sh (EscapeVector EscapingMode) - deriving (Eq, Ord, Show) - -{-| Construct a Bourne Shell escaped intermediate form. - -} -sh :: ByteString -> Sh -sh = escape - -instance Escape Sh where - escape = Sh . escWith classify - unescape (Sh v) = stripEsc v - bytes (Sh v) = interpretEsc v act finish ([], Literal) - where - finish Quote = Put.putChar '\'' - finish Backslash = Put.putChar '\\' - finish Literal = return () - - -{-| Accept the present escaping mode and desired escaping mode and yield an - action and the resulting mode. - -} -act :: EscapingMode -> (Char, EscapingMode) -> (Put.Put, EscapingMode) -act Quote (c, Quote) = (Put.putChar c , Quote) -act Quote (c, Literal) = (Put.putChar c , Quote) -act Quote (c, Backslash) = (Put.putString ['\'', '\\', c] , Literal) -act Backslash (c, Backslash) = (Put.putChar c , Literal) -act Backslash (c, Quote) = (Put.putString ['\\', '\'', c] , Quote) -act Backslash (c, Literal) = (Put.putString ['\\', c] , Literal) -act Literal (c, Literal) = (Put.putChar c , Literal) -act Literal (c, Backslash) = (Put.putString ['\\', c] , Literal) -act Literal (c, Quote) = (Put.putString ['\'', c] , Quote) - -classify :: Char -> EscapingMode -classify c | c <= '&' = Quote -- 0x00..0x26 - | c == '\'' = Backslash -- 0x27 - | c <= ',' = Quote -- 0x28..0x2c - | c <= '9' = Literal -- 0x2d..0x39 - | c <= '?' = Quote -- 0x3a..0x3f - | c <= 'Z' = Literal -- 0x40..0x5a - | c <= '^' = Quote -- 0x5b..0x5e - | c == '_' = Literal -- 0x5f - | c == '`' = Quote -- 0x60 - | c <= 'z' = Literal -- 0x61..0x7a - | c <= '\DEL' = Quote -- 0x7b..0x7f - | otherwise = Quote -- 0x80..0xff - -{-| Bourne Shell escaping modes. - -} -data EscapingMode = Backslash | Literal | Quote - deriving (Eq, Ord, Show) - diff --git a/bench/Bench.hs b/bench/Bench.hs --- a/bench/Bench.hs +++ b/bench/Bench.hs @@ -6,7 +6,7 @@ import Criterion.Main -import Text.ShellEscape +import Data.ByteString.ShellEscape strings :: [ByteString] diff --git a/shell-escape.cabal b/shell-escape.cabal --- a/shell-escape.cabal +++ b/shell-escape.cabal @@ -1,5 +1,5 @@ name : shell-escape -version : 0.1.2 +version : 0.2.0 category : Text license : BSD3 license-file : LICENSE @@ -24,10 +24,11 @@ , containers , vector >= 0.6.0.2 , bytestring >= 0.9 - exposed-modules : Text.ShellEscape - other-modules : Text.ShellEscape.Escape - Text.ShellEscape.EscapeVector - Text.ShellEscape.Sh - Text.ShellEscape.Bash - Text.ShellEscape.Put + exposed-modules : Data.ByteString.ShellEscape + Text.ShellEscape + other-modules : Data.ByteString.ShellEscape.Escape + Data.ByteString.ShellEscape.EscapeVector + Data.ByteString.ShellEscape.Sh + Data.ByteString.ShellEscape.Bash + Data.ByteString.ShellEscape.Put diff --git a/test/Echo.hs b/test/Echo.hs --- a/test/Echo.hs +++ b/test/Echo.hs @@ -18,7 +18,7 @@ import Test.QuickCheck import Test.QuickCheck.Monadic -import Text.ShellEscape hiding (sh, bash) +import Data.ByteString.ShellEscape hiding (sh, bash) -- It is best to implement the echo test with `printf':