PTD (work in progress)

This commit is contained in:
Anna Nowak 2021-04-05 18:48:35 +02:00
parent 3b8318d752
commit a7bf508ff4
3 changed files with 61 additions and 0 deletions

43
app/ptd.hs Normal file
View File

@ -0,0 +1,43 @@
{-# LANGUAGE Arrows, NoMonomorphismRestriction #-}
import ShadowLibrary.Core
import Text.XML.HXT.Core
import Text.XML.HXT.XPath
import Data.List
import Data.List.Utils (replace)
import Text.Regex.Posix
import Text.Printf
extractRecords = extractLinksWithText "//div[@class='entry-content']/p[count(a)=1]/a"
>>> second (arr $ replace "\r\n " " ")
>>> first (extractLinksWithText "//li/a[contains(@href,'.pdf')]")
-- ... 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 = "Almanach Muszyny " ++ 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 baseUrl = "https://www.ptd.pl/?page_id=7"
let shadowLibrary = ShadowLibrary {logoUrl=Nothing,
lname="Polskie Towarzystwo Dendrologiczne",
abbrev="ptd",
lLevel=0,
webpage=baseUrl}
extractItemsStartingFromUrl shadowLibrary baseUrl (extractRecords >>> arr toShadowItem)

5
readme.md Normal file
View File

@ -0,0 +1,5 @@
# Dodatkowe paczki zainstalowane na potrzebę uruchomienia programu:
```
apt-get install libcurl4-gnutls-dev
```

View File

@ -62,6 +62,19 @@ executable almanachmuszyny
default-language: Haskell2010
executable ptd
hs-source-dirs: app
main-is: ptd.hs
ghc-options: -threaded -rtsopts -with-rtsopts=-N
build-depends: base
, hxt
, hxt-xpath
, hxt-curl
, MissingH
, regex-posix
, shadow-library
default-language: Haskell2010
source-repository head
type: git
location: https://github.com/name/project