diff --git a/Encode.cabal b/Encode.cabal --- a/Encode.cabal +++ b/Encode.cabal @@ -1,15 +1,15 @@ name: Encode -version: 1.3.5 +version: 1.3.6 license: GPL license-file: LICENSE extra-source-files: INSTALL, Setup.PL -copyright: 2011 +copyright: 2013 author: Otakar Smrz -maintainer: otakar.smrz mff.cuni.cz -homepage: http://ufal.mff.cuni.cz/~smrz/ +maintainer: otakar-smrz users.sf.net +homepage: http://otakar-smrz.users.sf.net/ package-url: http://sourceforge.net/projects/encode-arabic/ category: Text, Codec -build-depends: Cabal, base < 4.4, mtl, containers +build-depends: Cabal, base < 5.0, mtl, containers build-type: Simple synopsis: Encoding character data description: The "Encode" library provides a unified interface for converting strings @@ -22,9 +22,8 @@ The "Encode" library is being proposed as a Haskell analogy to the /Encode/ extension in Perl, . . - The "Exec.Encode.Main" and "Exec.Decode.Main" programs mimick the - function calls to 'encode' and 'decode', respectively, with the - following usage of the executables: + The "Main.Encode" and "Main.Decode" programs mimick the function calls + to 'encode' and 'decode', respectively, with the following usage: . > decode ArabTeX < decode.d | encode Buckwalter > encode.d > @@ -49,13 +48,14 @@ library developed by Peter Ljunglöf in his licenciate thesis /Pure Functional Parsing – an advanced tutorial/, Göteborg University and Chalmers University of Technology, April 2002, - . + . . The "Parsek" library is included for experimental purposes and subsumes "PureFP.Parsers.Stream". /Parsek/ was developed by Koen Claessen in his functional pearl article /Parallel Parsing Processes/, Journal of Functional Programming, 14(6), 741–757, Cambridge University Press, - 2004, . + 2004, , + . . "Encode" "PureFP" "Parsek" exposed-modules: Encode, @@ -63,6 +63,8 @@ Encode.Arabic.ArabTeX, Encode.Arabic.ArabTeX.ZDMG, Encode.Arabic.Buckwalter, + Encode.Arabic.Parkinson, + Encode.Arabic.Habash, Encode.Arabic.Byte, Encode.Extend, Encode.ExPlus, @@ -93,12 +95,14 @@ hugs-options: -98 +o +u -h96M executable: encode -main-is: Exec/Encode/Main.hs +main-is: Main/Encode.hs other-modules: Encode, Encode.Arabic, Encode.Arabic.ArabTeX, Encode.Arabic.ArabTeX.ZDMG, Encode.Arabic.Buckwalter, + Encode.Arabic.Parkinson, + Encode.Arabic.Habash, Encode.Arabic.Byte, Encode.Extend, Encode.ExPlus, @@ -129,12 +133,14 @@ hugs-options: -98 +o +u -h96M executable: decode -main-is: Exec/Decode/Main.hs +main-is: Main/Decode.hs other-modules: Encode, Encode.Arabic, Encode.Arabic.ArabTeX, Encode.Arabic.ArabTeX.ZDMG, Encode.Arabic.Buckwalter, + Encode.Arabic.Parkinson, + Encode.Arabic.Habash, Encode.Arabic.Byte, Encode.Extend, Encode.ExPlus, diff --git a/Encode.hs b/Encode.hs --- a/Encode.hs +++ b/Encode.hs @@ -1,14 +1,10 @@ --- -------------------------------------------------------------------------- --- $Revision: 1006 $ $Date: 2011-04-21 09:39:00 +0200 (Thu, 21 Apr 2011) $ --- -------------------------------------------------------------------------- - -- | -- -- Module : Encode -- Copyright : Otakar Smrz 2005-2011 -- License : GPL -- --- Maintainer : otakar.smrz mff.cuni.cz +-- Maintainer : otakar-smrz users.sf.net -- Stability : provisional -- Portability : portable -- diff --git a/Encode/Arabic.hs b/Encode/Arabic.hs --- a/Encode/Arabic.hs +++ b/Encode/Arabic.hs @@ -1,14 +1,10 @@ --- -------------------------------------------------------------------------- --- $Revision: 1006 $ $Date: 2011-04-21 09:39:00 +0200 (Thu, 21 Apr 2011) $ --- -------------------------------------------------------------------------- - -- | -- -- Module : Encode.Arabic --- Copyright : Otakar Smrz 2005-2011 +-- Copyright : Otakar Smrz 2005-2012 -- License : GPL -- --- Maintainer : otakar.smrz mff.cuni.cz +-- Maintainer : otakar-smrz users.sf.net -- Stability : provisional -- Portability : portable -- @@ -18,6 +14,9 @@ -- "Encode.Arabic.ArabTeX" -- "Encode.Arabic.ArabTeX.ZDMG" -- "Encode.Arabic.Buckwalter" +-- "Encode.Arabic.Parkinson" +-- "Encode.Arabic.Habash" +-- "Encode.Arabic.Byte" module Encode.Arabic ( @@ -27,6 +26,8 @@ module Encode.Arabic.ArabTeX, module Encode.Arabic.ArabTeX.ZDMG, module Encode.Arabic.Buckwalter, + module Encode.Arabic.Parkinson, + module Encode.Arabic.Habash, module Encode.Arabic.Byte, module Encode.Unicode @@ -39,6 +40,8 @@ import Encode.Arabic.ArabTeX import Encode.Arabic.ArabTeX.ZDMG import Encode.Arabic.Buckwalter +import Encode.Arabic.Parkinson +import Encode.Arabic.Habash import Encode.Arabic.Byte import Encode.Unicode diff --git a/Encode/Arabic/ArabTeX.hs b/Encode/Arabic/ArabTeX.hs --- a/Encode/Arabic/ArabTeX.hs +++ b/Encode/Arabic/ArabTeX.hs @@ -1,14 +1,10 @@ --- -------------------------------------------------------------------------- --- $Revision: 1010 $ $Date: 2011-06-17 01:41:07 +0200 (Fri, 17 Jun 2011) $ --- -------------------------------------------------------------------------- - -- | -- -- Module : Encode.Arabic.ArabTeX -- Copyright : Otakar Smrz 2005-2011 -- License : GPL -- --- Maintainer : otakar.smrz mff.cuni.cz +-- Maintainer : otakar-smrz users.sf.net -- Stability : provisional -- Portability : portable -- diff --git a/Encode/Arabic/ArabTeX/ZDMG.hs b/Encode/Arabic/ArabTeX/ZDMG.hs --- a/Encode/Arabic/ArabTeX/ZDMG.hs +++ b/Encode/Arabic/ArabTeX/ZDMG.hs @@ -1,14 +1,10 @@ --- -------------------------------------------------------------------------- --- $Revision: 1006 $ $Date: 2011-04-21 09:39:00 +0200 (Thu, 21 Apr 2011) $ --- -------------------------------------------------------------------------- - -- | -- -- Module : Encode.Arabic.ArabTeX.ZDMG -- Copyright : Otakar Smrz 2005-2011 -- License : GPL -- --- Maintainer : otakar.smrz mff.cuni.cz +-- Maintainer : otakar-smrz users.sf.net -- Stability : provisional -- Portability : portable -- diff --git a/Encode/Arabic/Buckwalter.hs b/Encode/Arabic/Buckwalter.hs --- a/Encode/Arabic/Buckwalter.hs +++ b/Encode/Arabic/Buckwalter.hs @@ -1,14 +1,10 @@ --- -------------------------------------------------------------------------- --- $Revision: 1010 $ $Date: 2011-06-17 01:41:07 +0200 (Fri, 17 Jun 2011) $ --- -------------------------------------------------------------------------- - -- | -- -- Module : Encode.Arabic.Buckwalter -- Copyright : Otakar Smrz 2005-2011 -- License : GPL -- --- Maintainer : otakar.smrz mff.cuni.cz +-- Maintainer : otakar-smrz users.sf.net -- Stability : provisional -- Portability : portable -- diff --git a/Encode/Arabic/Byte.hs b/Encode/Arabic/Byte.hs --- a/Encode/Arabic/Byte.hs +++ b/Encode/Arabic/Byte.hs @@ -1,14 +1,10 @@ --- -------------------------------------------------------------------------- --- $Revision: 1006 $ $Date: 2011-04-21 09:39:00 +0200 (Thu, 21 Apr 2011) $ --- -------------------------------------------------------------------------- - -- | -- -- Module : Encode.Arabic.Byte -- Copyright : Otakar Smrz 2005-2011 -- License : GPL -- --- Maintainer : otakar.smrz mff.cuni.cz +-- Maintainer : otakar-smrz users.sf.net -- Stability : provisional -- Portability : portable -- diff --git a/Encode/Arabic/Habash.hs b/Encode/Arabic/Habash.hs new file mode 100644 --- /dev/null +++ b/Encode/Arabic/Habash.hs @@ -0,0 +1,97 @@ +-- | +-- +-- Module : Encode.Arabic.Habash +-- Copyright : Otakar Smrz 2005-2012 +-- License : GPL +-- +-- Maintainer : otakar-smrz users.sf.net +-- Stability : provisional +-- Portability : portable +-- +-- Habash-Soudi-Buckwalter notation is a one-to-one transliteration of the +-- graphemes of the Arabic script into various symbols of Unicode defined in +-- . +-- +-- /Encode::Arabic::Habash/ in Perl: +-- + + +module Encode.Arabic.Habash ( + + -- * Types + + Habash (..) + + ) where + + +import Encode + +import Data.Map (Map) +import qualified Data.Map as Map + + +data Habash = Habash | Soudi | HSB + + deriving (Enum, Show) + + +instance Encoding Habash where + + encode _ = recode encoder + + decode _ = recode decoder + + +recode :: (Ord a, Enum b, Enum a) => Map a b -> [a] -> [b] + +recode xry xs = [ Map.findWithDefault ((toEnum . fromEnum) x) x xry | x <- xs ] + + +recoder :: Ord a => [a] -> [b] -> Map a b + +recoder xs ys = Map.fromList (zip xs ys) + + +encoder :: Map UPoint Char + +encoder = recoder decoded encoded + + +decoder :: Map Char UPoint + +decoder = recoder encoded decoded + + +decoded :: [UPoint] + +decoded = map toEnum + + [0x0621, 0x0622, 0x0623, 0x0624, 0x0625, 0x0626, 0x0627, 0x0628, + 0x0629, 0x062A, 0x062B, 0x062C, 0x062D, 0x062E, 0x062F, 0x0630, + 0x0631, 0x0632, 0x0633, 0x0634, 0x0635, 0x0636, 0x0637, 0x0638, + 0x0639, 0x063A, 0x0641, 0x0642, 0x0643, 0x0644, 0x0645, 0x0646, + 0x0647, 0x0648, 0x0649, 0x064A, 0x064E, 0x064F, 0x0650, 0x064B, + 0x064C, 0x064D, 0x0651, 0x0652, 0x0670, 0x0671, 0x067E, 0x0686, + 0x06A4, 0x06AF, 0x0640, + 0x0698, + 0x0660, 0x0661, 0x0662, 0x0663, 0x0664, 0x0665, 0x0666, 0x0667, + 0x0668, 0x0669, + 0x060C, 0x061B, 0x061F] + + +encoded :: [Char] + +encoded = map toEnum + + [0x0027, 0x0100, 0x00C2, 0x0175, 0x01CD, 0x0177, 0x0041, 0x0062, + 0x0127, 0x0074, 0x03B8, 0x006A, 0x0048, 0x0078, 0x0064, 0x00F0, + 0x0072, 0x007A, 0x0073, 0x0161, 0x0053, 0x0044, 0x0054, 0x010E, + 0x03C2, 0x03B3, 0x0066, 0x0071, 0x006B, 0x006C, 0x006D, 0x006E, + 0x0068, 0x0077, 0x00FD, 0x0079, 0x0061, 0x0075, 0x0069, 0x00E3, + 0x0169, 0x0129, 0x007E, 0x00B7, 0x00E1, 0x00C4, 0x0070, 0x0063, + 0x0076, 0x0067, 0x005F, + 0x017E, + 0x0030, 0x0031, 0x0032, 0x0033, 0x0034, 0x0035, 0x0036, 0x0037, + 0x0038, 0x0039, + 0x002C, 0x003B, 0x003F] diff --git a/Encode/Arabic/Parkinson.hs b/Encode/Arabic/Parkinson.hs new file mode 100644 --- /dev/null +++ b/Encode/Arabic/Parkinson.hs @@ -0,0 +1,94 @@ +-- | +-- +-- Module : Encode.Arabic.Parkinson +-- Copyright : Otakar Smrz 2005-2012 +-- License : GPL +-- +-- Maintainer : otakar-smrz users.sf.net +-- Stability : provisional +-- Portability : portable +-- +-- Dil Parkinson's notation is a one-to-one transliteration of the graphemes +-- of the Arabic script using lower ASCII characters only. +-- +-- /Encode::Arabic::Parkinson/ in Perl: +-- + + +module Encode.Arabic.Parkinson ( + + -- * Types + + Parkinson (..) + + ) where + + +import Encode + +import Data.Map (Map) +import qualified Data.Map as Map + + +data Parkinson = Parkinson | Dil + + deriving (Enum, Show) + + +instance Encoding Parkinson where + + encode _ = recode encoder + + decode _ = recode decoder + + +recode :: (Ord a, Enum b, Enum a) => Map a b -> [a] -> [b] + +recode xry xs = [ Map.findWithDefault ((toEnum . fromEnum) x) x xry | x <- xs ] + + +recoder :: Ord a => [a] -> [b] -> Map a b + +recoder xs ys = Map.fromList (zip xs ys) + + +encoder :: Map UPoint Char + +encoder = recoder decoded encoded + + +decoder :: Map Char UPoint + +decoder = recoder encoded decoded + + +decoded :: [UPoint] + +decoded = map toEnum ( [] + + ++ [0x0623, 0x0624, 0x0625] + ++ [0x060C, 0x061B, 0x061F] + ++ [0x0621, 0x0622] ++ [0x0626 .. 0x063A] ++ [0x0641 .. 0x064A] + ++ [0x0660 .. 0x0669] + ++ [0x0671] + ++ [0x0651] + ++ [0x064B .. 0x0650] ++ [0x0670] ++ [0x0652] + ++ [0x0640] + + ) + + +encoded :: [Char] + +encoded = map id ( [] + + ++ "LWE" + ++ ",;?" + ++ "CM" ++ "YAbQtVjHxdvrzspSDTZcg" ++ "fqklmnhwey" + ++ ['0' .. '9'] + ++ "O" + ++ "~" + ++ "NUIaui" ++ "R" ++ "o" + ++ "_" + + ) diff --git a/Encode/ExPlus.hs b/Encode/ExPlus.hs --- a/Encode/ExPlus.hs +++ b/Encode/ExPlus.hs @@ -1,14 +1,10 @@ --- -------------------------------------------------------------------------- --- $Revision: 1006 $ $Date: 2011-04-21 09:39:00 +0200 (Thu, 21 Apr 2011) $ --- -------------------------------------------------------------------------- - -- | -- -- Module : Encode.ExPlus -- Copyright : Otakar Smrz 2005-2011 -- License : GPL -- --- Maintainer : otakar.smrz mff.cuni.cz +-- Maintainer : otakar-smrz users.sf.net -- Stability : provisional -- Portability : portable -- diff --git a/Encode/Extend.hs b/Encode/Extend.hs --- a/Encode/Extend.hs +++ b/Encode/Extend.hs @@ -1,14 +1,10 @@ --- -------------------------------------------------------------------------- --- $Revision: 1006 $ $Date: 2011-04-21 09:39:00 +0200 (Thu, 21 Apr 2011) $ --- -------------------------------------------------------------------------- - -- | -- -- Module : Encode.Extend -- Copyright : Otakar Smrz 2005-2011 -- License : GPL -- --- Maintainer : otakar.smrz mff.cuni.cz +-- Maintainer : otakar-smrz users.sf.net -- Stability : provisional -- Portability : portable -- diff --git a/Encode/Mapper.hs b/Encode/Mapper.hs --- a/Encode/Mapper.hs +++ b/Encode/Mapper.hs @@ -1,14 +1,10 @@ --- -------------------------------------------------------------------------- --- $Revision: 1006 $ $Date: 2011-04-21 09:39:00 +0200 (Thu, 21 Apr 2011) $ --- -------------------------------------------------------------------------- - -- | -- -- Module : Encode.Mapper -- Copyright : Otakar Smrz 2005-2011 -- License : GPL -- --- Maintainer : otakar.smrz mff.cuni.cz +-- Maintainer : otakar-smrz users.sf.net -- Stability : provisional -- Portability : portable -- diff --git a/Encode/Unicode.hs b/Encode/Unicode.hs --- a/Encode/Unicode.hs +++ b/Encode/Unicode.hs @@ -1,14 +1,10 @@ --- -------------------------------------------------------------------------- --- $Revision: 1006 $ $Date: 2011-04-21 09:39:00 +0200 (Thu, 21 Apr 2011) $ --- -------------------------------------------------------------------------- - -- | -- -- Module : Encode.Unicode -- Copyright : Otakar Smrz 2005-2011 -- License : GPL -- --- Maintainer : otakar.smrz mff.cuni.cz +-- Maintainer : otakar-smrz users.sf.net -- Stability : provisional -- Portability : portable -- diff --git a/Encode/Unicode/UTF8.hs b/Encode/Unicode/UTF8.hs --- a/Encode/Unicode/UTF8.hs +++ b/Encode/Unicode/UTF8.hs @@ -1,14 +1,10 @@ --- -------------------------------------------------------------------------- --- $Revision: 1006 $ $Date: 2011-04-21 09:39:00 +0200 (Thu, 21 Apr 2011) $ --- -------------------------------------------------------------------------- - -- | -- -- Module : Encode.Unicode.UTF8 -- Copyright : Otakar Smrz 2005-2011 -- License : GPL -- --- Maintainer : otakar.smrz mff.cuni.cz +-- Maintainer : otakar-smrz users.sf.net -- Stability : provisional -- Portability : portable -- diff --git a/Encode/Version.hs b/Encode/Version.hs --- a/Encode/Version.hs +++ b/Encode/Version.hs @@ -1,18 +1,14 @@ --- -------------------------------------------------------------------------- --- $Revision: 1006 $ $Date: 2011-04-21 09:39:00 +0200 (Thu, 21 Apr 2011) $ --- -------------------------------------------------------------------------- - -- | -- -- Module : Encode.Version --- Copyright : Otakar Smrz 2005-2011 +-- Copyright : Otakar Smrz 2005-2013 -- License : GPL -- --- Maintainer : otakar.smrz mff.cuni.cz +-- Maintainer : otakar-smrz users.sf.net -- Stability : provisional -- Portability : portable -- --- Exports the 'version' of the "Encode" library and provides support +-- Exports the 'version' of the "Encode" library and provides support -- for working with the SVN\/CVS revision keyword. The 'revised' method -- parses the '$Revision ... $' string supplied to it. Results have the -- type 'Version' of the "Data.Version" module, which is exported, too. @@ -26,21 +22,26 @@ -- * Functions - revised, version + showPretty, revised, version ) where import Data.Version +import Data.List + import Text.ParserCombinators.ReadP +showPretty :: Version -> String + +showPretty (Version x y) = unwords ((concat . intersperse "." . map show) x : y) + + revised :: String -> Version revised revision = fst . last . readP_to_S parseVersion $ words revision !! 1 -version = Version [1, 3, min build 5] [] - - where Version [build] [] = revised "$Revision: 1006 $" +version = Version [1, 3, 6] ["April 2013"] diff --git a/Exec/Decode/Main.hs b/Exec/Decode/Main.hs deleted file mode 100644 --- a/Exec/Decode/Main.hs +++ /dev/null @@ -1,155 +0,0 @@ --- -------------------------------------------------------------------------- --- $Revision: 1013 $ $Date: 2011-08-22 00:38:38 +0200 (Mon, 22 Aug 2011) $ --- -------------------------------------------------------------------------- - --- | --- --- Module : Exec.Decode.Main --- Copyright : Otakar Smrz 2005-2011 --- License : GPL --- --- Maintainer : otakar.smrz mff.cuni.cz --- Stability : provisional --- Portability : portable --- --- "Encode.Arabic" "Encode.Mapper" "Encode.Unicode" - - -module Main where - - -import Encode -import Encode.Arabic -import Encode.Unicode - -import System.Environment -import System.Console.GetOpt -import System.IO - -import Data.Char -import Data.List - -import Encode.Version - - -data Opts = DisplayUsage | PrintVersion | LineOriented | - - Prefix String | Suffix String - - deriving (Eq, Ord) - - -options :: [OptDescr Opts] - -options = [ Option ['h'] ["help"] (NoArg DisplayUsage) - "show usage information", - - Option ['l'] ["lines"] (NoArg LineOriented) - "use line-oriented mode", - - Option ['p'] ["prefix"] (ReqArg Prefix "text") - "prefix input with text", - - Option ['s'] ["suffix"] (ReqArg Suffix "text") - "suffix input with text", - - Option ['v'] ["version"] (NoArg PrintVersion) - "show program's version" ] - - -copyleft = unlines ["Encode Arabic (C) 2011-2005 Otakar Smrz", - "GNU General Public License http://www.gnu.org/licenses/"] - -synopsis = unlines [copyleft, - " http://sourceforge.net/projects/encode-arabic/", - " http://quest.ms.mff.cuni.cz/encode/", - " ", - "", - "decode [--OPTIONS] [ENCODING]"] - - -main = do args <- getArgs - - hSetBinaryMode stdin True - hSetBinaryMode stdout True - - hSetBuffering stdin LineBuffering - hSetBuffering stdout LineBuffering - - let (opts, nons, errs) = getOpt Permute options args - - if null errs then case (sort . nub) opts of - - DisplayUsage : _ -> tell (usageInfo synopsis options) - PrintVersion : _ -> tell (unlines [copyleft, - unwords ["Encode Arabic", - showVersion version, - "August 2011"]]) - LineOriented : _ -> execute True (tail opts) nons - _ -> execute False opts nons - - else warn (usageInfo synopsis options) - - -tell = hPutStr stdout - - -warn = hPutStr stderr - - -execute l opts nons = case opts of - - [Prefix p] -> interactPlus l e (p ++) - [Prefix p, Suffix s] - -> interactPlus l e ((p ++) . (++ s)) - [Suffix s] -> interactPlus l e (++ s) - _ -> interactPlus l e id - - where e = case nons of [] -> "" - _ -> head nons - - -interactPlus :: Bool -> String -> (String -> String) -> IO () - -interactPlus l e f = interact (if l then unlines . map encodes . lines - else encodes) - - where encodes = encode UTF . decodes . f - - decodes = case map toLower e of - - "arabtex" -> decode ArabTeX . encode UCS . decode UTF - "lagally" -> decode Lagally . encode UCS . decode UTF - "tex" -> decode TeX . encode UCS . decode UTF - - "zdmg" -> decode ZDMG . encode UCS . decode UTF - "arabtex_zdmg" -> decode ArabTeX_ZDMG . encode UCS . decode UTF - "arabtex-zdmg" -> decode ArabTeX_ZDMG . encode UCS . decode UTF - - "buckwalter" -> decode Buckwalter . encode UCS . decode UTF - "tim" -> decode Tim . encode UCS . decode UTF - - "winarabic" -> decode WinArabic - "cp1256" -> decode CP1256 - "windows_1256" -> decode Windows_1256 - "windows-1256" -> decode Windows_1256 - - "isoarabic" -> decode ISOArabic - "iso_8859_6" -> decode ISO_8859_6 - "iso-8859-6" -> decode ISO_8859_6 - - "macarabic" -> decode MacArabic - - "macfarsi" -> decode MacFarsi - - "dosarabic" -> decode DOSArabic - "cp864" -> decode CP864 - - "dosfarsi" -> decode DOSFarsi - "cp1006" -> decode CP1006 - - "asmo449" -> decode ASMO449 - - "isiri3342" -> decode ISIRI3342 - - _ -> decode UTF diff --git a/Exec/Encode/Main.hs b/Exec/Encode/Main.hs deleted file mode 100644 --- a/Exec/Encode/Main.hs +++ /dev/null @@ -1,155 +0,0 @@ --- -------------------------------------------------------------------------- --- $Revision: 1013 $ $Date: 2011-08-22 00:38:38 +0200 (Mon, 22 Aug 2011) $ --- -------------------------------------------------------------------------- - --- | --- --- Module : Exec.Encode.Main --- Copyright : Otakar Smrz 2005-2011 --- License : GPL --- --- Maintainer : otakar.smrz mff.cuni.cz --- Stability : provisional --- Portability : portable --- --- "Encode.Arabic" "Encode.Mapper" "Encode.Unicode" - - -module Main where - - -import Encode -import Encode.Arabic -import Encode.Unicode - -import System.Environment -import System.Console.GetOpt -import System.IO - -import Data.Char -import Data.List - -import Encode.Version - - -data Opts = DisplayUsage | PrintVersion | LineOriented | - - Prefix String | Suffix String - - deriving (Eq, Ord) - - -options :: [OptDescr Opts] - -options = [ Option ['h'] ["help"] (NoArg DisplayUsage) - "show usage information", - - Option ['l'] ["lines"] (NoArg LineOriented) - "use line-oriented mode", - - Option ['p'] ["prefix"] (ReqArg Prefix "text") - "prefix input with text", - - Option ['s'] ["suffix"] (ReqArg Suffix "text") - "suffix input with text", - - Option ['v'] ["version"] (NoArg PrintVersion) - "show program's version" ] - - -copyleft = unlines ["Encode Arabic (C) 2011-2005 Otakar Smrz", - "GNU General Public License http://www.gnu.org/licenses/"] - -synopsis = unlines [copyleft, - " http://sourceforge.net/projects/encode-arabic/", - " http://quest.ms.mff.cuni.cz/encode/", - " ", - "", - "encode [--OPTIONS] [ENCODING]"] - - -main = do args <- getArgs - - hSetBinaryMode stdin True - hSetBinaryMode stdout True - - hSetBuffering stdin LineBuffering - hSetBuffering stdout LineBuffering - - let (opts, nons, errs) = getOpt Permute options args - - if null errs then case (sort . nub) opts of - - DisplayUsage : _ -> tell (usageInfo synopsis options) - PrintVersion : _ -> tell (unlines [copyleft, - unwords ["Encode Arabic", - showVersion version, - "August 2011"]]) - LineOriented : _ -> execute True (tail opts) nons - _ -> execute False opts nons - - else warn (usageInfo synopsis options) - - -tell = hPutStr stdout - - -warn = hPutStr stderr - - -execute l opts nons = case opts of - - [Prefix p] -> interactPlus l e (p ++) - [Prefix p, Suffix s] - -> interactPlus l e ((p ++) . (++ s)) - [Suffix s] -> interactPlus l e (++ s) - _ -> interactPlus l e id - - where e = case nons of [] -> "" - _ -> head nons - - -interactPlus :: Bool -> String -> (String -> String) -> IO () - -interactPlus l e f = interact (if l then unlines . map decodes . lines - else decodes) - - where decodes = encodes . decode UTF . f - - encodes = case map toLower e of - - "arabtex" -> encode UTF . decode UCS . encode ArabTeX - "lagally" -> encode UTF . decode UCS . encode Lagally - "tex" -> encode UTF . decode UCS . encode TeX - - "zdmg" -> encode UTF . decode UCS . encode ZDMG - "arabtex_zdmg" -> encode UTF . decode UCS . encode ArabTeX_ZDMG - "arabtex-zdmg" -> encode UTF . decode UCS . encode ArabTeX_ZDMG - - "buckwalter" -> encode UTF . decode UCS . encode Buckwalter - "tim" -> encode UTF . decode UCS . encode Tim - - "winarabic" -> encode WinArabic - "cp1256" -> encode CP1256 - "windows_1256" -> encode Windows_1256 - "windows-1256" -> encode Windows_1256 - - "isoarabic" -> encode ISOArabic - "iso_8859_6" -> encode ISO_8859_6 - "iso-8859-6" -> encode ISO_8859_6 - - "macarabic" -> encode MacArabic - - "macfarsi" -> encode MacFarsi - - "dosarabic" -> encode DOSArabic - "cp864" -> encode CP864 - - "dosfarsi" -> encode DOSFarsi - "cp1006" -> encode CP1006 - - "asmo449" -> encode ASMO449 - - "isiri3342" -> encode ISIRI3342 - - _ -> encode UTF diff --git a/Main/Decode.hs b/Main/Decode.hs new file mode 100644 --- /dev/null +++ b/Main/Decode.hs @@ -0,0 +1,157 @@ +-- | +-- +-- Module : Main.Decode +-- Copyright : Otakar Smrz 2005-2013 +-- License : GPL +-- +-- Maintainer : otakar-smrz users.sf.net +-- Stability : provisional +-- Portability : portable +-- +-- "Encode.Arabic" "Encode.Mapper" "Encode.Unicode" + + +module Main where + + +import Encode +import Encode.Arabic +import Encode.Unicode + +import System.Environment +import System.Console.GetOpt +import System.IO + +import Data.Char +import Data.List + +import Encode.Version + + +data Opts = DisplayUsage | PrintVersion | LineOriented | + + Prefix String | Suffix String + + deriving (Eq, Ord) + + +options :: [OptDescr Opts] + +options = [ Option ['h'] ["help"] (NoArg DisplayUsage) + "show usage information", + + Option ['l'] ["lines"] (NoArg LineOriented) + "use line-oriented mode", + + Option ['p'] ["prefix"] (ReqArg Prefix "text") + "prefix input with text", + + Option ['s'] ["suffix"] (ReqArg Suffix "text") + "suffix input with text", + + Option ['v'] ["version"] (NoArg PrintVersion) + "show program's version" ] + + +copyleft = unlines ["Encode Arabic (C) 2013-2005 Otakar Smrz", + "GNU General Public License http://www.gnu.org/licenses/"] + +synopsis = unlines [copyleft, + " http://sourceforge.net/projects/encode-arabic/", + " http://encode-arabic.sourceforge.net/", + " ", + "", + "decode [--OPTIONS] [ENCODING]"] + + +main = do args <- getArgs + + hSetBinaryMode stdin True + hSetBinaryMode stdout True + + hSetBuffering stdin LineBuffering + hSetBuffering stdout LineBuffering + + let (opts, nons, errs) = getOpt Permute options args + + if null errs then case (sort . nub) opts of + + DisplayUsage : _ -> tell (usageInfo synopsis options) + PrintVersion : _ -> tell (unlines [copyleft, + unwords ["Encode Arabic", + showPretty version]]) + LineOriented : _ -> execute True (tail opts) nons + _ -> execute False opts nons + + else warn (usageInfo synopsis options) + + +tell = hPutStr stdout + + +warn = hPutStr stderr + + +execute l opts nons = case opts of + + [Prefix p] -> interactPlus l e (p ++) + [Prefix p, Suffix s] + -> interactPlus l e ((p ++) . (++ s)) + [Suffix s] -> interactPlus l e (++ s) + _ -> interactPlus l e id + + where e = case nons of [] -> "" + _ -> head nons + + +interactPlus :: Bool -> String -> (String -> String) -> IO () + +interactPlus l e f = interact (if l then unlines . map encodes . lines + else encodes) + + where encodes = encode UTF . decodes . f + + decodes = case map toLower e of + + "arabtex" -> decode ArabTeX . encode UCS . decode UTF + "lagally" -> decode Lagally . encode UCS . decode UTF + "tex" -> decode TeX . encode UCS . decode UTF + + "zdmg" -> decode ZDMG . encode UCS . decode UTF + "arabtex_zdmg" -> decode ArabTeX_ZDMG . encode UCS . decode UTF + "arabtex-zdmg" -> decode ArabTeX_ZDMG . encode UCS . decode UTF + + "buckwalter" -> decode Buckwalter . encode UCS . decode UTF + "tim" -> decode Tim . encode UCS . decode UTF + + "parkinson" -> decode Parkinson . encode UCS . decode UTF + "dil" -> decode Dil . encode UCS . decode UTF + + "habash" -> decode Habash . encode UCS . decode UTF + "soudi" -> decode Soudi . encode UCS . decode UTF + "hsb" -> decode HSB . encode UCS . decode UTF + + "winarabic" -> decode WinArabic + "cp1256" -> decode CP1256 + "windows_1256" -> decode Windows_1256 + "windows-1256" -> decode Windows_1256 + + "isoarabic" -> decode ISOArabic + "iso_8859_6" -> decode ISO_8859_6 + "iso-8859-6" -> decode ISO_8859_6 + + "macarabic" -> decode MacArabic + + "macfarsi" -> decode MacFarsi + + "dosarabic" -> decode DOSArabic + "cp864" -> decode CP864 + + "dosfarsi" -> decode DOSFarsi + "cp1006" -> decode CP1006 + + "asmo449" -> decode ASMO449 + + "isiri3342" -> decode ISIRI3342 + + _ -> decode UTF diff --git a/Main/Encode.hs b/Main/Encode.hs new file mode 100644 --- /dev/null +++ b/Main/Encode.hs @@ -0,0 +1,157 @@ +-- | +-- +-- Module : Main.Encode +-- Copyright : Otakar Smrz 2005-2013 +-- License : GPL +-- +-- Maintainer : otakar-smrz users.sf.net +-- Stability : provisional +-- Portability : portable +-- +-- "Encode.Arabic" "Encode.Mapper" "Encode.Unicode" + + +module Main where + + +import Encode +import Encode.Arabic +import Encode.Unicode + +import System.Environment +import System.Console.GetOpt +import System.IO + +import Data.Char +import Data.List + +import Encode.Version + + +data Opts = DisplayUsage | PrintVersion | LineOriented | + + Prefix String | Suffix String + + deriving (Eq, Ord) + + +options :: [OptDescr Opts] + +options = [ Option ['h'] ["help"] (NoArg DisplayUsage) + "show usage information", + + Option ['l'] ["lines"] (NoArg LineOriented) + "use line-oriented mode", + + Option ['p'] ["prefix"] (ReqArg Prefix "text") + "prefix input with text", + + Option ['s'] ["suffix"] (ReqArg Suffix "text") + "suffix input with text", + + Option ['v'] ["version"] (NoArg PrintVersion) + "show program's version" ] + + +copyleft = unlines ["Encode Arabic (C) 2013-2005 Otakar Smrz", + "GNU General Public License http://www.gnu.org/licenses/"] + +synopsis = unlines [copyleft, + " http://sourceforge.net/projects/encode-arabic/", + " http://encode-arabic.sourceforge.net/", + " ", + "", + "encode [--OPTIONS] [ENCODING]"] + + +main = do args <- getArgs + + hSetBinaryMode stdin True + hSetBinaryMode stdout True + + hSetBuffering stdin LineBuffering + hSetBuffering stdout LineBuffering + + let (opts, nons, errs) = getOpt Permute options args + + if null errs then case (sort . nub) opts of + + DisplayUsage : _ -> tell (usageInfo synopsis options) + PrintVersion : _ -> tell (unlines [copyleft, + unwords ["Encode Arabic", + showPretty version]]) + LineOriented : _ -> execute True (tail opts) nons + _ -> execute False opts nons + + else warn (usageInfo synopsis options) + + +tell = hPutStr stdout + + +warn = hPutStr stderr + + +execute l opts nons = case opts of + + [Prefix p] -> interactPlus l e (p ++) + [Prefix p, Suffix s] + -> interactPlus l e ((p ++) . (++ s)) + [Suffix s] -> interactPlus l e (++ s) + _ -> interactPlus l e id + + where e = case nons of [] -> "" + _ -> head nons + + +interactPlus :: Bool -> String -> (String -> String) -> IO () + +interactPlus l e f = interact (if l then unlines . map decodes . lines + else decodes) + + where decodes = encodes . decode UTF . f + + encodes = case map toLower e of + + "arabtex" -> encode UTF . decode UCS . encode ArabTeX + "lagally" -> encode UTF . decode UCS . encode Lagally + "tex" -> encode UTF . decode UCS . encode TeX + + "zdmg" -> encode UTF . decode UCS . encode ZDMG + "arabtex_zdmg" -> encode UTF . decode UCS . encode ArabTeX_ZDMG + "arabtex-zdmg" -> encode UTF . decode UCS . encode ArabTeX_ZDMG + + "buckwalter" -> encode UTF . decode UCS . encode Buckwalter + "tim" -> encode UTF . decode UCS . encode Tim + + "parkinson" -> encode UTF . decode UCS . encode Parkinson + "dil" -> encode UTF . decode UCS . encode Dil + + "habash" -> encode UTF . decode UCS . encode Habash + "soudi" -> encode UTF . decode UCS . encode Soudi + "hsb" -> encode UTF . decode UCS . encode HSB + + "winarabic" -> encode WinArabic + "cp1256" -> encode CP1256 + "windows_1256" -> encode Windows_1256 + "windows-1256" -> encode Windows_1256 + + "isoarabic" -> encode ISOArabic + "iso_8859_6" -> encode ISO_8859_6 + "iso-8859-6" -> encode ISO_8859_6 + + "macarabic" -> encode MacArabic + + "macfarsi" -> encode MacFarsi + + "dosarabic" -> encode DOSArabic + "cp864" -> encode CP864 + + "dosfarsi" -> encode DOSFarsi + "cp1006" -> encode CP1006 + + "asmo449" -> encode ASMO449 + + "isiri3342" -> encode ISIRI3342 + + _ -> encode UTF diff --git a/Parsek.hs b/Parsek.hs --- a/Parsek.hs +++ b/Parsek.hs @@ -1,14 +1,10 @@ --- -------------------------------------------------------------------------- --- $Revision: 531 $ $Date: 2008-04-11 00:55:45 +0200 (Fri, 11 Apr 2008) $ --- -------------------------------------------------------------------------- - -- | -- -- Module : Parsek -- Copyright : Koen Claessen 2003 -- License : GPL -- --- Maintainer : otakar.smrz mff.cuni.cz +-- Maintainer : otakar-smrz users.sf.net -- Stability : provisional -- Portability : portable -- @@ -16,11 +12,7 @@ -- functional pearl article /Parallel Parsing Processes/, Journal of Functional -- Programming, 14(6), 741–757, Cambridge University Press, 2004: -- --- --- --- --- --- +-- -- -- -- Copyright (C) 2003 Koen Claessen diff --git a/PureFP.hs b/PureFP.hs --- a/PureFP.hs +++ b/PureFP.hs @@ -1,14 +1,10 @@ --- -------------------------------------------------------------------------- --- $Revision: 1006 $ $Date: 2011-04-21 09:39:00 +0200 (Thu, 21 Apr 2011) $ --- -------------------------------------------------------------------------- - -- | -- -- Module : PureFP -- Copyright : Peter Ljunglof 2002 -- License : GPL -- --- Maintainer : otakar.smrz mff.cuni.cz +-- Maintainer : otakar-smrz users.sf.net -- Stability : provisional -- Portability : portable -- diff --git a/PureFP/OrdMap.hs b/PureFP/OrdMap.hs --- a/PureFP/OrdMap.hs +++ b/PureFP/OrdMap.hs @@ -1,14 +1,10 @@ --- -------------------------------------------------------------------------- --- $Revision: 262 $ $Date: 2007-04-12 12:19:50 +0200 (Thu, 12 Apr 2007) $ --- -------------------------------------------------------------------------- - -- | -- -- Module : PureFP.OrdMap -- Copyright : Peter Ljunglof 2002 -- License : GPL -- --- Maintainer : otakar.smrz mff.cuni.cz +-- Maintainer : otakar-smrz users.sf.net -- Stability : provisional -- Portability : portable -- diff --git a/PureFP/OrdSet.hs b/PureFP/OrdSet.hs --- a/PureFP/OrdSet.hs +++ b/PureFP/OrdSet.hs @@ -1,14 +1,10 @@ --- -------------------------------------------------------------------------- --- $Revision: 262 $ $Date: 2007-04-12 12:19:50 +0200 (Thu, 12 Apr 2007) $ --- -------------------------------------------------------------------------- - -- | -- -- Module : PureFP.OrdSet -- Copyright : Peter Ljunglof 2002 -- License : GPL -- --- Maintainer : otakar.smrz mff.cuni.cz +-- Maintainer : otakar-smrz users.sf.net -- Stability : provisional -- Portability : portable -- diff --git a/PureFP/Parsers.hs b/PureFP/Parsers.hs --- a/PureFP/Parsers.hs +++ b/PureFP/Parsers.hs @@ -1,14 +1,10 @@ --- -------------------------------------------------------------------------- --- $Revision: 1006 $ $Date: 2011-04-21 09:39:00 +0200 (Thu, 21 Apr 2011) $ --- -------------------------------------------------------------------------- - -- | -- -- Module : PureFP.Parsers -- Copyright : Peter Ljunglof 2002 -- License : GPL -- --- Maintainer : otakar.smrz mff.cuni.cz +-- Maintainer : otakar-smrz users.sf.net -- Stability : provisional -- Portability : portable -- diff --git a/PureFP/Parsers/AmbExTrie.hs b/PureFP/Parsers/AmbExTrie.hs --- a/PureFP/Parsers/AmbExTrie.hs +++ b/PureFP/Parsers/AmbExTrie.hs @@ -1,14 +1,10 @@ --- -------------------------------------------------------------------------- --- $Revision: 262 $ $Date: 2007-04-12 12:19:50 +0200 (Thu, 12 Apr 2007) $ --- -------------------------------------------------------------------------- - -- | -- -- Module : PureFP.Parsers.AmbExTrie -- Copyright : Peter Ljunglof 2002 -- License : GPL -- --- Maintainer : otakar.smrz mff.cuni.cz +-- Maintainer : otakar-smrz users.sf.net -- Stability : provisional -- Portability : portable -- diff --git a/PureFP/Parsers/AmbTrie.hs b/PureFP/Parsers/AmbTrie.hs --- a/PureFP/Parsers/AmbTrie.hs +++ b/PureFP/Parsers/AmbTrie.hs @@ -1,14 +1,10 @@ --- -------------------------------------------------------------------------- --- $Revision: 262 $ $Date: 2007-04-12 12:19:50 +0200 (Thu, 12 Apr 2007) $ --- -------------------------------------------------------------------------- - -- | -- -- Module : PureFP.Parsers.AmbTrie -- Copyright : Peter Ljunglof 2002 -- License : GPL -- --- Maintainer : otakar.smrz mff.cuni.cz +-- Maintainer : otakar-smrz users.sf.net -- Stability : provisional -- Portability : portable -- diff --git a/PureFP/Parsers/ExTrie.hs b/PureFP/Parsers/ExTrie.hs --- a/PureFP/Parsers/ExTrie.hs +++ b/PureFP/Parsers/ExTrie.hs @@ -1,14 +1,10 @@ --- -------------------------------------------------------------------------- --- $Revision: 262 $ $Date: 2007-04-12 12:19:50 +0200 (Thu, 12 Apr 2007) $ --- -------------------------------------------------------------------------- - -- | -- -- Module : PureFP.Parsers.ExTrie -- Copyright : Peter Ljunglof 2002 -- License : GPL -- --- Maintainer : otakar.smrz mff.cuni.cz +-- Maintainer : otakar-smrz users.sf.net -- Stability : provisional -- Portability : portable -- diff --git a/PureFP/Parsers/PairTrie.hs b/PureFP/Parsers/PairTrie.hs --- a/PureFP/Parsers/PairTrie.hs +++ b/PureFP/Parsers/PairTrie.hs @@ -1,14 +1,10 @@ --- -------------------------------------------------------------------------- --- $Revision: 262 $ $Date: 2007-04-12 12:19:50 +0200 (Thu, 12 Apr 2007) $ --- -------------------------------------------------------------------------- - -- | -- -- Module : PureFP.Parsers.PairTrie -- Copyright : Peter Ljunglof 2002 -- License : GPL -- --- Maintainer : otakar.smrz mff.cuni.cz +-- Maintainer : otakar-smrz users.sf.net -- Stability : provisional -- Portability : portable -- diff --git a/PureFP/Parsers/Parser.hs b/PureFP/Parsers/Parser.hs --- a/PureFP/Parsers/Parser.hs +++ b/PureFP/Parsers/Parser.hs @@ -1,14 +1,10 @@ --- -------------------------------------------------------------------------- --- $Revision: 262 $ $Date: 2007-04-12 12:19:50 +0200 (Thu, 12 Apr 2007) $ --- -------------------------------------------------------------------------- - -- | -- -- Module : PureFP.Parsers.Parser -- Copyright : Peter Ljunglof 2002 -- License : GPL -- --- Maintainer : otakar.smrz mff.cuni.cz +-- Maintainer : otakar-smrz users.sf.net -- Stability : provisional -- Portability : portable -- diff --git a/PureFP/Parsers/Standard.hs b/PureFP/Parsers/Standard.hs --- a/PureFP/Parsers/Standard.hs +++ b/PureFP/Parsers/Standard.hs @@ -1,14 +1,10 @@ --- -------------------------------------------------------------------------- --- $Revision: 287 $ $Date: 2007-04-25 23:25:57 +0200 (Wed, 25 Apr 2007) $ --- -------------------------------------------------------------------------- - -- | -- -- Module : PureFP.Parsers.Standard -- Copyright : Peter Ljunglof 2002 -- License : GPL -- --- Maintainer : otakar.smrz mff.cuni.cz +-- Maintainer : otakar-smrz users.sf.net -- Stability : provisional -- Portability : portable -- diff --git a/PureFP/Parsers/Stream.hs b/PureFP/Parsers/Stream.hs --- a/PureFP/Parsers/Stream.hs +++ b/PureFP/Parsers/Stream.hs @@ -1,14 +1,10 @@ --- -------------------------------------------------------------------------- --- $Revision: 264 $ $Date: 2007-04-13 18:24:56 +0200 (Fri, 13 Apr 2007) $ --- -------------------------------------------------------------------------- - -- | -- -- Module : PureFP.Parsers.Stream -- Copyright : Peter Ljunglof 2002 -- License : GPL -- --- Maintainer : otakar.smrz mff.cuni.cz +-- Maintainer : otakar-smrz users.sf.net -- Stability : provisional -- Portability : portable -- diff --git a/PureFP/Parsers/Trie.hs b/PureFP/Parsers/Trie.hs --- a/PureFP/Parsers/Trie.hs +++ b/PureFP/Parsers/Trie.hs @@ -1,14 +1,10 @@ --- -------------------------------------------------------------------------- --- $Revision: 262 $ $Date: 2007-04-12 12:19:50 +0200 (Thu, 12 Apr 2007) $ --- -------------------------------------------------------------------------- - -- | -- -- Module : PureFP.Parsers.Trie -- Copyright : Peter Ljunglof 2002 -- License : GPL -- --- Maintainer : otakar.smrz mff.cuni.cz +-- Maintainer : otakar-smrz users.sf.net -- Stability : provisional -- Portability : portable -- diff --git a/Setup.PL b/Setup.PL --- a/Setup.PL +++ b/Setup.PL @@ -4,9 +4,7 @@ # # Building Encode ############################################################################## -# $Id: Setup.PL 1006 2011-04-21 07:39:00Z smrz $ - -our $VERSION = join '.', 1, 3, 5, q $Revision: 1006 $ =~ /(\d+)/; +our $VERSION = '1.3.6'; use File::Spec;