bdb7648 · typeable/pandoc

277-- | Version of 'ppTopElement' that specifies UTF-8 encoding.

278ppTopElement :: Element -> String

279ppTopElement = ("<?xml version=\"1.0\" encoding=\"UTF-8\"?>\n" ++) . unEntity . ppElement

280− -- unEntity removes decimal entities introduced by ppElement

280+ -- unEntity removes numeric entities introduced by ppElement

281 -- (kindlegen seems to choke on these).

282 where unEntity [] = ""

283− unEntity ('&':'#':d:xs) | isDigit d =

283+ unEntity ('&':'#':xs) =

284− let ds = takeWhile isDigit xs

284+ let (ds,ys) = break (==';') xs

285+ c = if (all isDigit ds)

285− c = read $ '\'' : '\\' : d : ds ++ "'"

286+ then Just $ read $ '\'' : '\\' : ds ++ "'"

286− in if c > '\127'

287+ else Nothing

287− then c : unEntity (drop (length ds + 2) xs)

288+ rest = drop 1 ys

288− else '&':'#':d:ds ++ unEntity (drop (length ds + 2) xs)

289+ in case c of

290+ Just x | x > '\127' -> x : unEntity rest

291+ _ -> ('&':'#':ds) ++ ";" ++ unEntity rest

289 unEntity (x:xs) = x : unEntity xs

292 unEntity (x:xs) = x : unEntity xs

290

293

291imageTypeOf :: FilePath -> Maybe String

294imageTypeOf :: FilePath -> Maybe String