{-# 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, split, join)

import Text.Regex.Posix
import Text.Printf


extractRecords = extractLinksWithText "//section[@class='widget widget_zzm_gm_archive']//a"  -- pary adres-tytuł
	>>> first (arr $ replace "https" "http")
	>>> first (extractLinksWithText "//article//h2//a")
	>>> first (arr $  first (arr $ replace "https" "http"))
	>>> first (arr $  second (arr $ replace "ń" "n"))
	>>> first (arr $  second (arr $ replace "ś" "s"))
	>>> first (arr $  second (arr $ replace "ź" "z"))
	>>> first (arr $  second (arr $ replace "ł" "l"))
	>>> first (first (extractLinksWithText "//h1[@class='entry-title']//a[contains(@href,'.pdf')]"))
	>>> first (arr $  first (arr $ second(arr $ replace "ń" "n")))
	>>> first (arr $  first (arr $ second(arr $ replace "ś" "s")))
	>>> first (arr $  first (arr $ second(arr $ replace "ź" "z")))
	>>> first (arr $  first (arr $ second(arr $ replace "ł" "l")))
	>>> first (arr $  second (arr $ replace "Styczen" "01"))
 	>>> first (arr $  second (arr $ replace "styczen" "01"))
 	>>> first (arr $  second (arr $ replace "Luty" "02"))
 	>>> first (arr $  second (arr $ replace "luty" "02"))
 	>>> first (arr $  second (arr $ replace "Marzec" "03"))
 	>>> first (arr $  second (arr $ replace "marzec" "03"))
 	>>> first (arr $  second (arr $ replace "Kwiecien" "04"))
 	>>> first (arr $  second (arr $ replace "kwiecien" "04"))
 	>>> first (arr $  second (arr $ replace "Maj" "05"))
 	>>> first (arr $  second (arr $ replace "maj" "05"))
 	>>> first (arr $  second (arr $ replace "Czerwiec" "06"))
 	>>> first (arr $  second (arr $ replace "czerwiec" "06"))
 	>>> first (arr $  second (arr $ replace "Lipiec" "07"))
 	>>> first (arr $  second (arr $ replace "lipiec" "07"))
 	>>> first (arr $  second (arr $ replace "Sierpien" "08"))
 	>>> first (arr $  second (arr $ replace "sierpien" "08"))
 	>>> first (arr $  second (arr $ replace "sierpnia" "08"))
 	>>> first (arr $  second (arr $ replace "Wrzesien" "09"))
 	>>> first (arr $  second (arr $ replace "wrzesien" "09" ))
 	>>> first (arr $  second (arr $ replace "wrzesnia" "09" ))
 	>>> first (arr $  second (arr $ replace "pazdziernika" "10" ))
 	>>> first (arr $  second (arr $ replace "Pazdziernik" "10" ))
 	>>> first (arr $  second (arr $ replace "pazdziernik" "10" ))
 	>>> first (arr $  second (arr $ replace "listopada" "11" ))
 	>>> first (arr $  second (arr $ replace "Listopad" "11" ))
 	>>> first (arr $  second (arr $ replace "listopad" "11" ))
 	>>> first (arr $  second (arr $ replace "Grudzien" "12"))
 	>>> first (arr $  second (arr $ replace "grudzien" "12"))
 	>>> first (arr $  second (arr $ replace "grudnia" "12"))
 	>>> first (arr $  second (arr $ replace " " "-"))
 	

              
-- ... a tutaj te trójki przerabiamy do docelowej struktury ShadowItem


toShadowItem :: (((String, String), String), String) -> ShadowItem
toShadowItem (((url, articleTitle), articleDatePart), year) =
  (defaultShadowItem url title) {
    originalDate = getDate articleDatePart,
    itype = "periodical",
    format = Just "pdf",
    finalUrl = url
    }
  where title = fixTitle articleTitle

fixTitle :: String -> String
fixTitle title 
  | title =~ "Glos Maszynisty .*" = title
  | otherwise = "Glos Maszynisty " ++ title

getDate :: String -> String
getDate url = 
  case url =~~ ("[0-9]*-?[0-9]{2}-[0-9]{4}" :: String) of
  Just date -> concat([join "-" (reverse (split "-" date))]) :: String
  otherwise -> ""



main = do
    let start = "http://zzm.org.pl/glos-maszynisty/"
    let shadowLibrary = ShadowLibrary {logoUrl=Nothing,
                                       lname="Głos Maszynisty",
                                       abbrev="GloMasz",
                                       lLevel=0,
                                       webpage=start}
    extractItemsStartingFromUrl shadowLibrary start (extractRecords >>> arr toShadowItem)