direct download links

.vscode/tasks.json

@@ -0,0 +1,50 @@
+
+{
+  // Automatically created by phoityne-vscode extension.
+
+  "version": "2.0.0",
+  "presentation": {
+    "reveal": "always",
+    "panel": "new"
+  },
+  "tasks": [
+    {
+      // F7
+      "group": {
+        "kind": "build",
+        "isDefault": true
+      },
+      "label": "haskell build",
+      "type": "shell",
+      //"command": "cabal configure && cabal build"
+      "command": "stack build"
+    },
+    {
+      // F6
+      "group": "build",
+      "type": "shell",
+      "label": "haskell clean & build",
+      //"command": "cabal clean && cabal configure && cabal build"
+      "command": "stack clean && stack build"
+      //"command": "stack clean ; stack build"  // for powershell
+    },
+    {
+      // F8
+      "group": {
+        "kind": "test",
+        "isDefault": true
+      },
+      "type": "shell",
+      "label": "haskell test",
+      //"command": "cabal test"
+      "command": "stack test"
+    },
+    {
+      // F6
+      "isBackground": true,
+      "type": "shell",
+      "label": "haskell watch",
+      "command": "stack build --test --no-run-tests --file-watch"
+    }
+  ]
+}

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 []
---                                     withCurl [("curl--user-agent","AMU Digital Libraries Indexing Agent")]
+                                    --  withHTTP,
+                                     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 []]		
---                                                     withCurl []]
+                                                    --  withHTTP,
+                                                     withCurl [("curl--user-agent","AMU Digital Libraries Indexing Agent")]]
 
 downloadXmlDocument = readFromDocument [withWarnings no,
                                         withEncodingErrors no,
-					withHTTP []]
---                                        withCurl [] ]
+                                        -- withHTTP,
+                                        withCurl [("curl--user-agent","AMU Digital Libraries Indexing Agent")]]
 
 
 data ShadowLibrary = ShadowLibrary { logoUrl :: Maybe String,

app/otwarteczasopismauam.hs

@@ -0,0 +1,50 @@
+
+{-# 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.TDFA
+-- import Text.Regex.TDFA.Text ()
+import Text.Regex.Posix
+import Text.Printf
+
+extractRecords = extractLinksWithText "//h3[@class='media-heading']/a"
+                 >>> first (arr (++"/issue/archive"))
+                 >>> first (extractLinksWithText "//*[@class='media-heading']/a"
+                            >>> first (extractLinksWithText "//*[@class='media-heading']/a" 
+                                        >>>  first (extractLinks "//div[@class='download']/a"
+                                                    >>>  extractLinks "//a[@class='download']")))
+
+toShadowItem :: (((String, String), String), String) -> ShadowItem
+toShadowItem (((url, articleTitle), yearlyTitle), journalTitle) =
+  (defaultShadowItem url title) {
+    originalDate = Just date,
+    itype = "periodical",
+    format = Just "pdf",
+    finalUrl = url
+    }
+  where title = replace "\n" "" (replace "\t" "" (journalTitle ++ " " ++  yearlyTitle ++ " " ++ articleTitle))
+        date = getDate yearlyTitle
+
+-- getDate yearlyTitle =
+--   case yearlyTitle =~~ "(19[0-9][0-9]|20[0-9][0-9])" :: Maybe [[String]] of
+--     Just [[_, year]] -> year
+--     otherwise -> error $ "unexpected text: " ++ yearlyTitle
+
+getDate :: String -> String
+getDate yearlyTitle = date where
+  date = yearlyTitle Text.Regex.TDFA.=~ "(19[0-9][0-9]|20[0-9][0-9])" :: String
+
+
+main = do
+    let start = "https://pressto.amu.edu.pl"
+    let shadowLibrary = ShadowLibrary {logoUrl=Nothing,
+                                       lname="Otwarte Czasopisma Naukowe UAM",
+                                       abbrev="OtwCzasNaukUAM",
+                                       lLevel=0,
+                                       webpage=start}
+    extractItemsStartingFromUrl shadowLibrary start (extractRecords >>> arr toShadowItem)

shadow-library.cabal

@@ -19,6 +19,7 @@ library
   build-depends:       base >= 4.7 && < 5
                      , HTTP
                      , hxt
+                     , hxt-curl
                      , hxt-http
                      , hxt-xpath
                      , MissingH
@@ -27,6 +28,7 @@ library
                      , network-uri
                      , regex-pcre
                      , regex-tdfa
+                     , regex-tdfa-text
                      , resourcet
                      , text
                      , time
@@ -59,6 +61,20 @@ executable almanachmuszyny
                       , shadow-library
    default-language:    Haskell2010
 
+executable otwarteczasopismauam
+   hs-source-dirs:      app
+   main-is:             otwarteczasopismauam.hs
+   ghc-options:         -threaded -rtsopts -with-rtsopts=-N
+   build-depends:       base
+                      , hxt
+                      , hxt-xpath
+                      , hxt-curl
+                      , MissingH
+                      , regex-posix
+                      , regex-tdfa
+                      , regex-tdfa-text
+                      , shadow-library
+   default-language:    Haskell2010
 
 source-repository head
   type:     git

stack.yaml.lock

@@ -0,0 +1,19 @@
+# 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
+snapshots:
+- completed:
+    size: 507596
+    url: https://raw.githubusercontent.com/commercialhaskell/stackage-snapshots/master/lts/11/9.yaml
+    sha256: 42f472dbf06482da1b3319241f3e3b3593a45bd7d4f537d2789f21386b9b2ad3
+  original: lts-11.9