version without file size

ShadowLibrary/Core.hs

@@ -33,7 +33,7 @@ import Data.Tree.NTree.TypeDefs
 import Data.Maybe
 import Control.Monad.Trans
 import Text.XML.HXT.XPath
--- import Text.XML.HXT.Curl
+import Text.XML.HXT.Curl
 import Text.XML.HXT.HTTP
 
 import Text.Regex.TDFA
@@ -64,8 +64,8 @@ downloadDocument = readFromDocument [withParseHTML yes,
                                      withEncodingErrors no,
                                      withPreserveComment yes,
                                      withStrictInput yes,
-                                     withHTTP []
+--                                     withHTTP []
---                                     withCurl [("curl--user-agent","AMU Digital Libraries Indexing Agent")]
+                                     withCurl [("curl--user-agent","AMU Digital Libraries Indexing Agent")]
                                     ]
 
 downloadDocumentWithEncoding enc = readFromDocument [withParseHTML yes,
@@ -73,13 +73,13 @@ downloadDocumentWithEncoding enc = readFromDocument [withParseHTML yes,
                                                      withEncodingErrors no,
                                                      withPreserveComment yes,
                                                      withInputEncoding enc,
-					             withHTTP []]		
+--					             withHTTP []]
---                                                     withCurl []]
+                                                     withCurl []]
 
 downloadXmlDocument = readFromDocument [withWarnings no,
                                         withEncodingErrors no,
-					withHTTP []]
+--					withHTTP []]
---                                        withCurl [] ]
+                                        withCurl [] ]
 
 
 data ShadowLibrary = ShadowLibrary { logoUrl :: Maybe String,

app/miastobierun.hs

@@ -20,15 +20,22 @@ toShadowItem (url, articleTitle) =
     originalDate = Just date,
     itype = "periodical",
     format = Just "pdf",
-    finalUrl = url
+    finalUrl = url,
+    description = Just desc
     }
-  where title = articleTitle
+  where title = "Miasto Bierun: " ++ articleTitle
-        date = getDate url
+        date = getDate articleTitle
+        desc = getArticleNr articleTitle
 
-getDate url =
+getDate title =
-  case url =~~ "/(19[0-9][0-9]|20[0-9][0-9])/" :: Maybe [[String]] of
+  case title =~~ "(19[0-9][0-9]|20[0-9][0-9])" :: Maybe [[String]] of
     Just [[_, year]] -> year
-    otherwise -> error $ "unexpected url: " ++ url
+    otherwise -> "No date for: " ++ title
+
+getArticleNr title =
+  case title =~~ "([0-9][0-9]/)" :: Maybe [[String]] of
+    Just [[_, nr]] -> "Article nr: " ++ (replace "/" "" nr)
+    otherwise -> "No article nr for: " ++ title
 
 
 main = do

shadow-library.cabal

@@ -20,6 +20,7 @@ library
                      , HTTP
                      , hxt
                      , hxt-http
+                     , hxt-curl
                      , hxt-xpath
                      , MissingH
                      , monad-logger
@@ -54,6 +55,7 @@ executable almanachmuszyny
    build-depends:       base
                       , hxt
                       , hxt-xpath
+                      , hxt-curl
                       , MissingH
                       , regex-posix
                       , shadow-library