title extraction

This commit is contained in:
s470623 2022-06-15 22:27:32 +02:00
parent 1f23d94e7c
commit 0114a102d4

View File

@ -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