clean output

app/ZborBielawa.hs

@@ -0,0 +1,67 @@
+
+{-# LANGUAGE Arrows, NoMonomorphismRestriction, TemplateHaskell, QuasiQuotes #-}
+import ShadowLibrary.Core
+
+import Text.XML.HXT.Core
+import Text.XML.HXT.XPath
+
+import Data.List
+import Data.List.Utils (replace)
+
+import Text.Printf
+
+import Control.Lens.Regex.Text
+import Control.Lens ((^?))
+import qualified Data.Text as T
+
+getLinkAndText xpathCondition = proc doc -> do
+  xpathTrees <- getXPathTrees xpathCondition -< doc
+  name <- getElemName -< xpathTrees
+  txt <- (getXPathTrees "//a/../text()" >>> listA (deep isText >>> getText)  >>> arr (intercalate " ")) -< xpathTrees
+  href <- (getXPathTrees "//a" >>> getAttrValue "href") -< xpathTrees
+  returnA -< (href, txt)
+
+
+extractNestedLinksWithText xpathCondition = proc url -> do
+  doc <- downloadDocument -< url
+  (link, text) <- getLinkAndText xpathCondition -< doc
+  uriFixed <- expandURIFixed -< (link, url)
+  returnA -< (uriFixed, text)
+
+
+extractRecords = proc startUrl -> do 
+  (catUrl, catText) <- extractLinksWithText "//aside[@class='widget widget_maxmegamenu']//a[@class='mega-menu-link']" -< startUrl  -- pary adres-tytuł podstrony
+  (collUrl, collText) <- (extractLinksWithText "//aside[@class='widget widget_maxmegamenu']//a[@class='mega-menu-link']") -< catUrl -- pobieramy podstronę kategorii i kolejne podstrony z menu
+  (relUrl, relText) <- (extractNestedLinksWithText "//big[a[contains(@href,'.pdf')]]") -< collUrl -- pobieramy stronę z wydaniami z danej kolekcji i linki do PDFów
+  returnA -< (relUrl, relText, collText, catText) -- ostatecznie wyjdą krotki (adres URL PDFa wydania, tytuł wydania, tytuł zbioru, tytuł kategorii)
+
+cleanReleaseTitle :: String -> String
+cleanReleaseTitle = proc text -> do
+  endCleaned <- replace " |\160" "" -< text
+  nbspCleaned <- replace "\160" " " -< endCleaned
+  res <- replace "\8211 " "" -< nbspCleaned
+  returnA -< res
+
+-- ... a tutaj te krotki przerabiamy do docelowej struktury ShadowItem
+toShadowItem :: (String, String, String, String) -> ShadowItem
+toShadowItem (url, releaseTitle, collectionTitle, categoryTitle) =
+  (defaultShadowItem url title) {
+    originalDate = T.unpack <$> date,
+    itype = "periodical",
+    format = Just "pdf",
+    finalUrl = url
+  }
+  where title = categoryTitle ++ (" " ++ arr cleanReleaseTitle releaseTitle)
+        date = getDate $ T.pack $ releaseTitle
+
+
+getDate txt = txt ^? [regex|19[0-9][0-9]|20[0-9][0-9]|] . match
+
+main = do
+    let start = "http://zborbielawa.pl/archiwum/"
+    let shadowLibrary = ShadowLibrary {logoUrl=Nothing,
+                                       lname="Zbór Bielawa",
+                                       abbrev="ZboBiel",
+                                       lLevel=0,
+                                       webpage=start}
+    extractItemsStartingFromUrl shadowLibrary start (extractRecords >>> arr toShadowItem)

shadow-library.cabal

@@ -60,6 +60,22 @@ executable almanachmuszyny
    default-language:    Haskell2010
 
 
+executable zborbielawa
+   hs-source-dirs:      app
+   main-is:             ZborBielawa.hs
+   ghc-options:         -threaded -rtsopts -with-rtsopts=-N
+   build-depends:       base
+                      , hxt
+                      , hxt-xpath
+                      , MissingH
+                      , regex-posix
+                      , shadow-library
+                      , lens-regex-pcre
+                      , lens
+                      , text
+   default-language:    Haskell2010
+
+
 source-repository head
   type:     git
   location: https://github.com/name/project

stack.yaml

@@ -1,5 +1,9 @@
 flags: {}
 packages:
 - '.'
-extra-deps: [hxt-xpath-9.1.2.2]
+extra-deps: [
+  hxt-xpath-9.1.2.2,
+  lens-regex-pcre-1.1.0.0,
+  pcre-light-0.4.1.0
+]
 resolver: lts-11.9

stack.yaml.lock

@@ -0,0 +1,33 @@
+# This file was autogenerated by Stack.
+# You should not edit this file by hand.
+# For more information, please see the documentation at:
+#   https://docs.haskellstack.org/en/stable/lock_files
+
+packages:
+- completed:
+    hackage: hxt-xpath-9.1.2.2@sha256:9cd590ae93a04573db8f90fa4094625ebd97dded45da7667c577ce6b38a42900,1999
+    pantry-tree:
+      size: 2225
+      sha256: aee2f75974e868ff429b8ff349a29667536c60397098f5dfedc968d1951511bb
+  original:
+    hackage: hxt-xpath-9.1.2.2
+- completed:
+    hackage: lens-regex-pcre-1.1.0.0@sha256:a6053fefae59f6b53b4741e5a75c5fae350af80dbf62e8673b565e1c3f34f8b9,2209
+    pantry-tree:
+      size: 630
+      sha256: 84a02d84433d92c5e2a9a7bfd662244be607b4004c7d4d6fd1ba53adbc655022
+  original:
+    hackage: lens-regex-pcre-1.1.0.0
+- completed:
+    hackage: pcre-light-0.4.1.0@sha256:4902ef06b037fd79448c7a63a65c1a4a1bf0bbc15ea7e67e71d52ec995e18460,1868
+    pantry-tree:
+      size: 467
+      sha256: 5e5f8d948ba577d4d838895ccb3533c6ef9876ef60884dae8342a77934cc2ec2
+  original:
+    hackage: pcre-light-0.4.1.0
+snapshots:
+- completed:
+    size: 507596
+    url: https://raw.githubusercontent.com/commercialhaskell/stackage-snapshots/master/lts/11/9.yaml
+    sha256: 42f472dbf06482da1b3319241f3e3b3593a45bd7d4f537d2789f21386b9b2ad3
+  original: lts-11.9