Compare commits

...

3 Commits

Author SHA1 Message Date
0114a102d4 title extraction 2022-06-15 22:27:32 +02:00
1f23d94e7c Robot for Teatr Lalek 2022-04-10 23:54:07 +02:00
953d7242e8 Robot for Teatr Lalek 2022-04-10 23:51:13 +02:00
2 changed files with 52 additions and 12 deletions

39
app/teatrLalek.hs Normal file
View File

@ -0,0 +1,39 @@
{-# 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 = extractLinks "//a[contains(@href, '.pdf')]"
toShadowItem :: String -> ShadowItem
toShadowItem url =
(defaultShadowItem url title) {
originalDate = Just date,
itype = "periodical",
format = Just "pdf",
finalUrl = 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
let start = "http://polunima.pl/teatr-lalek/"
let shadowLibrary = ShadowLibrary {logoUrl=Nothing,
lname="Teatr Lalek",
abbrev="Teatr",
lLevel=0,
webpage=start}
extractItemsStartingFromUrl shadowLibrary start (extractRecords >>> arr toShadowItem)

View File

@ -34,18 +34,19 @@ library
, tz
default-language: Haskell2010
-- executable maly-modelarz-exe
-- hs-source-dirs: app
-- main-is: malymodelarz.hs
-- ghc-options: -threaded -rtsopts -with-rtsopts=-N
-- build-depends: base
-- , hxt
-- , hxt-curl
-- , hxt-xpath
-- , MissingH
-- , regex-posix
-- , shadow-library
-- default-language: Haskell2010
executable teatrLalek
hs-source-dirs: app
main-is: teatrLalek.hs
ghc-options: -threaded -rtsopts -with-rtsopts=-N
build-depends: base
, hxt
, hxt-curl
, hxt-xpath
, MissingH
, regex-posix
, shadow-library
, regex-tdfa
default-language: Haskell2010
executable almanachmuszyny
hs-source-dirs: app