diff --git a/app/zbiorbielawa.hs b/app/zbiorbielawa.hs new file mode 100644 index 0000000..ebc4f5c --- /dev/null +++ b/app/zbiorbielawa.hs @@ -0,0 +1,46 @@ + +{-# 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='roczniki']" -- pary adres-tytuł + >>> second (arr $ replace "\r\n " " ") -- czyścimy drugi element pary, czyli tytuł z niepotrzebnych białych znaków + >>> first (arr ((++"tr") . init)) -- modyfikujemy pierwszy element pary, czyli adres URL + >>> first (extractLinksWithText "//li/a[contains(@href,'.pdf')]") -- pobieramy stronę z adresu URL i wyciągamy linki z tej strony pasujące do wyrażenia XPathowego + -- ostatecznie wyjdą trójki ((adres URL, tytuł artykułu), tytuł rocznika) + +-- ... a tutaj te trójki przerabiamy do docelowej struktury ShadowItem +toShadowItem :: ((String, String), String) -> ShadowItem +toShadowItem ((url, articleTitle), yearlyTitle) = + (defaultShadowItem url title) { + originalDate = Just date, + itype = "periodical", + format = Just "pdf", + finalUrl = url + } + where title = "Almanach Muszyny " ++ yearlyTitle ++ " " ++ (replace "\r\n" "" (replace "\r\n " "" articleTitle)) + date = getDate url + +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://www.almanachmuszyny.pl/" + let shadowLibrary = ShadowLibrary {logoUrl=Nothing, + lname="Almanach Muszyny", + abbrev="AlmMusz", + lLevel=0, + webpage=start} + extractItemsStartingFromUrl shadowLibrary start (extractRecords >>> arr toShadowItem) diff --git a/shadow-library.cabal b/shadow-library.cabal index cd77195..30f0764 100644 --- a/shadow-library.cabal +++ b/shadow-library.cabal @@ -60,6 +60,19 @@ executable almanachmuszyny default-language: Haskell2010 +executable zbiorbelawa + hs-source-dirs: app + main-is: zbiorbelawa.hs + ghc-options: -threaded -rtsopts -with-rtsopts=-N + build-depends: base + , hxt + , hxt-xpath + , MissingH + , regex-posix + , shadow-library + default-language: Haskell2010 + + source-repository head type: git location: https://github.com/name/project