GetAllPagePan2.hs

GetAllPagePan2.hs

#!/usr/bin/env runhaskell
{-# Language OverloadedStrings #-}
import qualified Text.Pandoc as P
import System.Process
import System.Exit
import qualified System.IO.UTF8 as U8
import Data.ByteString.Char8 ()

findLink' :: [P.Inline] -> [(String, String)]
findLink' inlines = concat $ fmap go inlines
  where
    go :: P.Inline -> [(String, String)]
    go (P.Link [P.Str title] (url, _)) = [(url, title)]
    go _ = []

findLink :: [P.Block] -> [(String, String)]
findLink blocks = concat $ fmap go blocks
  where
    go :: P.Block -> [(String, String)]
    go (P.Plain inlines) = findLink' inlines
    go (P.Para inlines) = findLink' inlines
    go (P.BulletList blockss) = concat $ fmap findLink blockss
    go _ = []

htmlToPandoc :: String -> P.Pandoc
htmlToPandoc = P.readHtml P.defaultParserState{ P.stateStandalone = True }

curlIt :: (String, String) -> IO ()
curlIt ss = putStrLn ("echo curl -d p=\"" ++ postp ++ 
                      "\" -d c=e http://www.sampou.org/cgi-bin/haskell.cgi -o \"" 
                      ++ outfile ++ "\"")
  where postp = dropWhile (/= '?') $ fst ss
        outfile = (snd ss) ++ ".html"

{--
curlIt :: (String, String) -> IO ExitCode
curlIt ss = system ("echo curl -d p=\"" ++ postp ++ 
                    "\" -d c=e http://www.sampou.org/cgi-bin/haskell.cgi -o \"" 
                    ++ outfile ++ "\"")
  where postp = tail . init $ dropWhile (/= '?') $ fst ss
        outfile = (tail . init $ snd ss) ++ ".html"
--}

main :: IO ()
main = do
  con <- getContents
  let P.Pandoc _ blocks = htmlToPandoc con
  print $ findLink blocks
  mapM_ curlIt $ findLink blocks