54 lines
1.8 KiB
Haskell
54 lines
1.8 KiB
Haskell
module ParseHtml (printDef, printDifferentDef) where
|
|
|
|
import Text.HTML.TagSoup
|
|
import Prelude
|
|
import WordDef
|
|
import qualified Data.Text.Lazy as T
|
|
|
|
|
|
printDifferentDef :: T.Text -> IO ()
|
|
printDifferentDef x = putStrLn (getDifferentDef x)
|
|
|
|
getDifferentDef :: T.Text -> String
|
|
getDifferentDef x = getDifferentDefMessages (length xs) ++ renderWordList xs
|
|
where xs = (parseDefTags . differentDefTags) x
|
|
|
|
getDifferentDefMessages :: Int -> String
|
|
getDifferentDefMessages 0 = "Aucune définition disponible."
|
|
getDifferentDefMessages 1 = "Une définition disponible : \n"
|
|
getDifferentDefMessages _ = "Plusieurs définitions disponibles : \n"
|
|
|
|
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 :: T.Text -> [Tag String]
|
|
differentDefTags =
|
|
filter (~== TagText "") . (takeWhile (~/= "<div id=contentbox>"))
|
|
. (dropWhile (~/= "<div id=vtoolbar>")) . parseTags . T.unpack
|
|
|
|
printDef :: T.Text -> IO ()
|
|
printDef = putStrLn . getDef
|
|
|
|
getDef :: T.Text -> String
|
|
getDef = (renderFullDef . getFullDef . parseFullDefTags)
|
|
|
|
getFullDef :: [Tag String] -> WordFullDef
|
|
getFullDef x = WordFullDef (WordDefHeader (WordName "lol") (WordType
|
|
"Interjection")) [WordDefSentence "Laughing Out Loud",
|
|
WordExampleSentence "ur wrong lol"]
|
|
|
|
parseFullDefTags :: T.Text -> [Tag String]
|
|
parseFullDefTags = filter (~== "<span class=tlf_cdefinition") . parseTags
|
|
. T.unpack
|