From 2eb795f7973ff21670efa11d9d457c1dcb8a7896 Mon Sep 17 00:00:00 2001
From: = <=>
Date: Wed, 16 Mar 2022 11:33:07 +0100
Subject: [PATCH] Add jujistsu.hs

---
 app/jujitsu.hs       | 47 ++++++++++++++++++++++++++++++++++++++++++++
 shadow-library.cabal | 12 +++++++++++
 2 files changed, 59 insertions(+)
 create mode 100644 app/jujitsu.hs

diff --git a/app/jujitsu.hs b/app/jujitsu.hs
new file mode 100644
index 0000000..42bc88d
--- /dev/null
+++ b/app/jujitsu.hs
@@ -0,0 +1,47 @@
+
+{-# 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 "//a[img]"  -- pary adres-tytuł
+                --  >>> second (arr $ replace "\r\n            " " ") -- czyścimy drugi element pary, czyli tytuł z niepotrzebnych białych znaków
+                --  >>> first (arr ((++"tr") . init))  -- modyfikujemy pierwszy element pary, czyli adres URL
+                --  >>> first (extractLinksWithText "//li/a[contains(@href,'.pdf')]") -- pobieramy stronę z adresu URL i wyciągamy linki z tej strony pasujące do wyrażenia XPathowego
+                 -- ostatecznie wyjdą trójki ((adres URL, tytuł artykułu), tytuł rocznika)
+
+-- ... a tutaj te trójki przerabiamy do docelowej struktury ShadowItem
+toShadowItem :: ((String, String), String) -> ShadowItem
+toShadowItem ((url, articleTitle), yearlyTitle) =
+  (defaultShadowItem url title) {
+    originalDate = Just date,
+    itype = "periodical",
+    format = Just "pdf",
+    finalUrl = url
+    }
+  where title = "Jujitsu " ++ 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 = "http://kobudo.pl/artykuly_jujutsu.html"
+    let shadowLibrary = ShadowLibrary {logoUrl=Nothing,
+                                       lname="Jujitsu",
+                                       abbrev="Juji",
+                                       lLevel=0,
+                                       webpage="http://kobudo.pl"}
+
+    extractItemsStartingFromUrl shadowLibrary start (extractRecords)
diff --git a/shadow-library.cabal b/shadow-library.cabal
index cd77195..1791faa 100644
--- a/shadow-library.cabal
+++ b/shadow-library.cabal
@@ -59,6 +59,18 @@ executable almanachmuszyny
                       , shadow-library
    default-language:    Haskell2010
 
+executable jujitsu
+   hs-source-dirs:      app
+   main-is:             jujitsu.hs
+   ghc-options:         -threaded -rtsopts -with-rtsopts=-N
+   build-depends:       base
+                      , hxt
+                      , hxt-xpath
+                      , MissingH
+                      , regex-posix
+                      , shadow-library
+   default-language:    Haskell2010
+
 
 source-repository head
   type:     git