44 lines
1.5 KiB
Haskell
44 lines
1.5 KiB
Haskell
|
|
{-# 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[@class='plik']" -- pary adres-tytuł
|
|
>>> second (arr $ replace "\r\n " " ") -- czyścimy drugi element pary, czyli tytuł z niepotrzebnych białych znaków
|
|
-- ostatecznie wyjdą dwójki ((adres URL, tytuł artykułu)
|
|
|
|
-- ... a tutaj te trójki przerabiamy do docelowej struktury ShadowItem
|
|
toShadowItem :: (String, String) -> ShadowItem
|
|
toShadowItem (url, articleTitle) =
|
|
(defaultShadowItem url title) {
|
|
originalDate = Just date,
|
|
itype = "periodical",
|
|
format = Just "pdf",
|
|
finalUrl = url
|
|
}
|
|
where title = articleTitle
|
|
date = getDate articleTitle
|
|
|
|
getDate articleTitle =
|
|
case articleTitle =~~ "(19[0-9][0-9]|20[0-9][0-9])" :: Maybe [[String]] of
|
|
Just [[_, year]] -> year
|
|
otherwise -> error $ "unexpected articleTitle: " ++ articleTitle
|
|
|
|
|
|
main = do
|
|
let start = "https://www.pipc.org.pl/publikacje/biuletyn-bezpieczna-chemia"
|
|
let shadowLibrary = ShadowLibrary {logoUrl=Nothing,
|
|
lname="Biuletyn Bezpieczna Chemia - Publikacje - PIPC",
|
|
lLevel=0,
|
|
webpage=start}
|
|
extractItemsStartingFromUrl shadowLibrary start (extractRecords >>> arr toShadowItem)
|