From 0114a102d4424f00128d6a0d8257188cf936195c Mon Sep 17 00:00:00 2001
From: s470623 <cezgal@st.amu.edu.pl>
Date: Wed, 15 Jun 2022 22:27:32 +0200
Subject: [PATCH] title extraction

---
 app/teatrLalek.hs | 4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)

diff --git a/app/teatrLalek.hs b/app/teatrLalek.hs
index 575cc52..4347b23 100644
--- a/app/teatrLalek.hs
+++ b/app/teatrLalek.hs
@@ -17,7 +17,7 @@ extractRecords = extractLinks "//a[contains(@href, '.pdf')]"
 
 toShadowItem :: String -> ShadowItem
 toShadowItem url =
-  (defaultShadowItem url []) {
+  (defaultShadowItem url title) {
     originalDate = Just date,
     itype = "periodical",
     format = Just "pdf",
@@ -25,6 +25,8 @@ toShadowItem 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