53 lines
1.7 KiB
Haskell
53 lines
1.7 KiB
Haskell
module ParseHtml (printDifferentDef, differentDefTags, parseDefTags) where
|
|
|
|
import Text.HTML.TagSoup
|
|
|
|
data WordDefHeader = WordDefHeader WordName WordType
|
|
data WordName = WordName String
|
|
data WordType = WordType String
|
|
|
|
printDifferentDef :: String -> IO ()
|
|
printDifferentDef x = putStrLn (getDifferentDef x)
|
|
|
|
|
|
getDifferentDef :: String -> String
|
|
getDifferentDef x = getDifferentDefMessages (length xs) ++ renderWordList xs
|
|
where xs = (parseDefTags . differentDefTags) x
|
|
|
|
renderWordList :: [WordDefHeader] -> String
|
|
renderWordList [] = ""
|
|
renderWordList (x:xs) = (renderString x) ++ "\n" ++ (renderWordList xs)
|
|
|
|
getDifferentDefMessages :: Int -> String
|
|
getDifferentDefMessages 0 = "Aucune définition disponible."
|
|
getDifferentDefMessages 1 = "Une définition disponible : \n"
|
|
getDifferentDefMessages _ = "Plusieurs définitions disponibles : \n"
|
|
|
|
renderString :: WordDefHeader -> String
|
|
renderString (WordDefHeader n t) = (getNameString n) ++ (getTypeString t)
|
|
|
|
getNameString :: WordName -> String
|
|
getNameString (WordName s) = s
|
|
|
|
getTypeString :: WordType -> String
|
|
getTypeString (WordType s) = s
|
|
|
|
parseDefTags :: [Tag String] -> [WordDefHeader]
|
|
parseDefTags [] = []
|
|
parseDefTags xs =
|
|
(WordDefHeader (WordName (renderTags (take 1 (fst st))))
|
|
(WordType (renderTags [(fst st) !! 1])))
|
|
: (parseDefTags (snd st))
|
|
where st = subTags xs
|
|
|
|
subTags :: [Tag String] -> ([Tag String],[Tag String])
|
|
subTags [] = ([],[])
|
|
subTags xs = if (length xs) `mod` 3 /= 0
|
|
then splitAt 2 xs
|
|
else ((take 1 xs) ++ (take 1 (drop 2 xs)) , drop 3 xs)
|
|
|
|
differentDefTags :: String -> [Tag String]
|
|
differentDefTags =
|
|
filter (~== TagText "") . (takeWhile (~/= "<div id=contentbox>"))
|
|
. (dropWhile (~/= "<div id=vtoolbar>")) . parseTags
|