twilight-library/app/jujitsu.hs

{-# LANGUAGE Arrows, NoMonomorphismRestriction #-}
import ShadowLibrary.Core

import Text.XML.HXT.Core
import Text.XML.HXT.XPath
-- import Text.XML.HXT.Curl
import Data.List
import Data.List.Utils (replace)

import Text.Regex.Posix
import Text.Printf


extractRecords = extractLinksWithText "//a[font]"  -- pary adres-tytuł

-- ... a tutaj te trójki przerabiamy do docelowej struktury ShadowItem
toShadowItem :: (String, String) -> ShadowItem
toShadowItem (url, title) =
  (defaultShadowItem url title) {
    originalDate = date,
    itype = "periodical",
    format = Just "pdf",
    finalUrl = url
    }
  where date = extractYear title

extractMonth :: String -> Maybe String
extractMonth n =
  case n =~~ ("/(styczeń|luty|marzec|kwiecien|maj|czerwiec|lipiec|sierpień|wrzesień|październik|listopad|grudzień)/" :: String) of
    Just month -> Just month
    otherwise -> Nothing

getDate url =
  case url =~~ "/(19[0-9][0-9]|20[0-9][0-9])/" :: Maybe [[String]] of
    Just [[_, year]] -> year
    otherwise -> error $ "unexpected url: " ++ url


main = do
    let start = "http://kobudo.pl/artykuly_jujutsu.html"
    let shadowLibrary = ShadowLibrary {logoUrl=Nothing,
                                       lname="Jujitsu",
                                       abbrev="Juji",
                                       lLevel=0,
                                       webpage="http://kobudo.pl"}

    extractItemsStartingFromUrl shadowLibrary start (extractRecords >>> arr toShadowItem)