geojournals skeleton added

app/geojournals.hs

@@ -0,0 +1,43 @@
+
+{-# 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 "//p//a[contains(@href,'gov.pl'), text()='View Journal']"
+
+
+toShadowItem :: ((String, String), String) -> ShadowItem
+toShadowItem ((url, articleTitle), yearlyTitle) =
+  (defaultShadowItem url title) {
+    originalDate = Just date,
+    itype = "periodical",
+    format = Just "pdf",
+    finalUrl = url
+    }
+  where title = "Geo Journal " ++ 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 = "https://geojournals.pgi.gov.pl"
+    let shadowLibrary = ShadowLibrary {logoUrl=Nothing,
+                                       lname="Polish Geological Journals",,
+                                       abbrev="geojour",
+                                       lLevel=0,
+                                       webpage=start}
+    extractItemsStartingFromUrl shadowLibrary start (extractRecords >>> arr toShadowItem)

shadow-library.cabal

@@ -34,18 +34,18 @@ library
                      , tz
   default-language:    Haskell2010
 
--- executable maly-modelarz-exe
+executable geojournals
---    hs-source-dirs:      app
+   hs-source-dirs:      app
---    main-is:             malymodelarz.hs
+   main-is:             geojournals.hs
---    ghc-options:         -threaded -rtsopts -with-rtsopts=-N
+   ghc-options:         -threaded -rtsopts -with-rtsopts=-N
---    build-depends:       base
+   build-depends:       base
---                       , hxt
+                      , hxt
---                       , hxt-curl
+                      , hxt-curl
---                       , hxt-xpath
+                      , hxt-xpath
---                       , MissingH
+                      , MissingH
---                       , regex-posix
+                      , regex-posix
---                       , shadow-library
+                      , shadow-library
---    default-language:    Haskell2010
+   default-language:    Haskell2010
 
 executable almanachmuszyny
    hs-source-dirs:      app