9e65867 · typeable/pandoc

git.iko.soy/typeable/pandoc

on master help sign in

Better definition of stringToURI.

9e65867John MacFarlane17 years ago

src/Text/Pandoc/Shared.hs+6−14

compare with master

unifiedsplit

@@ -115,10 +115,11 @@ import Text.ParserCombinators.Parsec

115import Text.PrettyPrint.HughesPJ ( Doc, fsep, ($$), (<>), empty, isEmpty, text, nest )

116import qualified Text.PrettyPrint.HughesPJ as PP

117import Text.Pandoc.CharacterReferences ( characterReference )

118−import Data.Char ( toLower, toUpper, ord, chr, isLower, isUpper, isAlpha,

118+import Data.Char ( toLower, toUpper, ord, isLower, isUpper, isAlpha,

119 isPunctuation )

120import Data.List ( find, isPrefixOf, intercalate )

121−import Network.URI ( parseURI, URI (..), isAllowedInURI )

121+import Network.URI ( parseURI, URI (..), isAllowedInURI, escapeURIString )

122+import Codec.Binary.UTF8.String ( encodeString )

122import System.Directory

123import System.Directory

123import System.FilePath ( (</>) )

124import System.FilePath ( (</>) )

124-- Note: ghc >= 6.12 (base >=4.2) supports unicode through iconv

125-- Note: ghc >= 6.12 (base >=4.2) supports unicode through iconv

@@ -131,10 +132,6 @@ import System.IO.UTF8

131import Data.Generics

132import Data.Generics

132import qualified Control.Monad.State as S

133import qualified Control.Monad.State as S

133import Control.Monad (join)

134import Control.Monad (join)

134−import Data.ByteString (unpack)

135−import Data.Word (Word8)

136−import Data.ByteString.UTF8 (fromString)

137−import Text.Printf (printf)

138import Paths_pandoc (getDataFileName)

135import Paths_pandoc (getDataFileName)

139

136

140--

137--

@@ -234,15 +231,10 @@ toRomanNumeral x =

234 _ | x >= 1 -> "I" ++ toRomanNumeral (x - 1)

231 _ | x >= 1 -> "I" ++ toRomanNumeral (x - 1)

235 _ -> ""

232 _ -> ""

236

233

237−-- | Escape unicode characters in a URI. This means converting

234+-- | Escape unicode characters in a URI. Characters that are

238−-- them to UTF-8, then URI-encoding the octets. We leave everything

235+-- already valid in a URI, including % and ?, are left alone.

239−-- else the same, assuming that the user has already escaped

240−-- special characters like & and %.

241stringToURI :: String -> String

236stringToURI :: String -> String

242−stringToURI = concatMap encodeOctet . unpack . fromString

237+stringToURI = escapeURIString isAllowedInURI . encodeString

243− where encodeOctet :: Word8 -> String

244− encodeOctet x | x > 127 = printf "%%%2x" x

245− encodeOctet x = [chr (fromIntegral x)]

246

238

247-- | Wrap inlines to line length.

239-- | Wrap inlines to line length.

248wrapped :: Monad m => ([Inline] -> m Doc) -> [Inline] -> m Doc

240wrapped :: Monad m => ([Inline] -> m Doc) -> [Inline] -> m Doc