title extraction

app/teatrLalek.hs

@@ -0,0 +1,39 @@
+
+{-# 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 = extractLinks "//a[contains(@href, '.pdf')]"
+
+
+toShadowItem :: String -> ShadowItem
+toShadowItem url =
+  (defaultShadowItem url title) {
+    originalDate = Just date,
+    itype = "periodical",
+    format = Just "pdf",
+    finalUrl = url
+    }
+  where
+    date = last $ getAllTextMatches $ url =~ "(19[0-9][0-9]|20[0-9][0-9])" :: String
+    titleToProcess = last $ getAllTextMatches $ url =~ "/[a-zA-Z ]+[-_]" :: String
+    title = titleToProcess =~ "[a-zA-Z ]+" :: String
+
+
+main = do
+    let start = "http://polunima.pl/teatr-lalek/"
+    let shadowLibrary = ShadowLibrary {logoUrl=Nothing,
+                                       lname="Teatr Lalek",
+                                       abbrev="Teatr",
+                                       lLevel=0,
+                                       webpage=start}
+    extractItemsStartingFromUrl shadowLibrary start (extractRecords >>> arr toShadowItem)

shadow-library.cabal

@@ -34,18 +34,19 @@ library
                      , tz
   default-language:    Haskell2010
 
--- executable maly-modelarz-exe
---    hs-source-dirs:      app
---    main-is:             malymodelarz.hs
---    ghc-options:         -threaded -rtsopts -with-rtsopts=-N
---    build-depends:       base
---                       , hxt
---                       , hxt-curl
---                       , hxt-xpath
---                       , MissingH
---                       , regex-posix
---                       , shadow-library
---    default-language:    Haskell2010
+executable teatrLalek
+    hs-source-dirs:      app
+    main-is:             teatrLalek.hs
+    ghc-options:         -threaded -rtsopts -with-rtsopts=-N
+    build-depends:       base
+                       , hxt
+                       , hxt-curl
+                       , hxt-xpath
+                       , MissingH
+                       , regex-posix
+                       , shadow-library
+                       , regex-tdfa
+    default-language:    Haskell2010
 
 executable almanachmuszyny
    hs-source-dirs:      app