forked from filipg/twilight-library
Compare commits
No commits in common. "master" and "master" have entirely different histories.
@ -33,7 +33,7 @@ import Data.Tree.NTree.TypeDefs
|
|||||||
import Data.Maybe
|
import Data.Maybe
|
||||||
import Control.Monad.Trans
|
import Control.Monad.Trans
|
||||||
import Text.XML.HXT.XPath
|
import Text.XML.HXT.XPath
|
||||||
import Text.XML.HXT.Curl
|
-- import Text.XML.HXT.Curl
|
||||||
import Text.XML.HXT.HTTP
|
import Text.XML.HXT.HTTP
|
||||||
|
|
||||||
import Text.Regex.TDFA
|
import Text.Regex.TDFA
|
||||||
@ -64,8 +64,8 @@ downloadDocument = readFromDocument [withParseHTML yes,
|
|||||||
withEncodingErrors no,
|
withEncodingErrors no,
|
||||||
withPreserveComment yes,
|
withPreserveComment yes,
|
||||||
withStrictInput yes,
|
withStrictInput yes,
|
||||||
-- withHTTP []
|
withHTTP []
|
||||||
withCurl [("curl--user-agent","AMU Digital Libraries Indexing Agent")]
|
-- withCurl [("curl--user-agent","AMU Digital Libraries Indexing Agent")]
|
||||||
]
|
]
|
||||||
|
|
||||||
downloadDocumentWithEncoding enc = readFromDocument [withParseHTML yes,
|
downloadDocumentWithEncoding enc = readFromDocument [withParseHTML yes,
|
||||||
@ -73,13 +73,13 @@ downloadDocumentWithEncoding enc = readFromDocument [withParseHTML yes,
|
|||||||
withEncodingErrors no,
|
withEncodingErrors no,
|
||||||
withPreserveComment yes,
|
withPreserveComment yes,
|
||||||
withInputEncoding enc,
|
withInputEncoding enc,
|
||||||
-- withHTTP []]
|
withHTTP []]
|
||||||
withCurl []]
|
-- withCurl []]
|
||||||
|
|
||||||
downloadXmlDocument = readFromDocument [withWarnings no,
|
downloadXmlDocument = readFromDocument [withWarnings no,
|
||||||
withEncodingErrors no,
|
withEncodingErrors no,
|
||||||
-- withHTTP []]
|
withHTTP []]
|
||||||
withCurl [] ]
|
-- withCurl [] ]
|
||||||
|
|
||||||
|
|
||||||
data ShadowLibrary = ShadowLibrary { logoUrl :: Maybe String,
|
data ShadowLibrary = ShadowLibrary { logoUrl :: Maybe String,
|
||||||
|
|||||||
@ -1,47 +0,0 @@
|
|||||||
|
|
||||||
{-# 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 "//div[@class='span4']//h2[@itemprop='name']/a[contains(@href,'o-nas')]" -- pary adres-tytuł
|
|
||||||
>>> first (extractLinksWithText "//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)
|
|
||||||
|
|
||||||
|
|
||||||
toShadowItem :: ((String, String), String) -> ShadowItem
|
|
||||||
toShadowItem ((url, articleTitle), yearlyTitle) =
|
|
||||||
(defaultShadowItem url title) {
|
|
||||||
originalDate = Just date,
|
|
||||||
itype = "periodical",
|
|
||||||
format = Just "pdf",
|
|
||||||
finalUrl = url
|
|
||||||
}
|
|
||||||
where title = replace "\"" "'" articleTitle
|
|
||||||
date = getDate yearlyTitle
|
|
||||||
|
|
||||||
getDate yearlyTitle =
|
|
||||||
case yearlyTitle =~~ "(19[0-9][0-9]|20[0-9][0-9])" :: Maybe [[String]] of
|
|
||||||
Just [[_, year]] -> year
|
|
||||||
otherwise -> "unexpected yearlyTitle" ++ yearlyTitle
|
|
||||||
|
|
||||||
|
|
||||||
main = do
|
|
||||||
let start = "http://moksir.chelmek.pl/o-nas/echo-chelmka"
|
|
||||||
let shadowLibrary = ShadowLibrary {logoUrl=Nothing,
|
|
||||||
lname="Chelmek",
|
|
||||||
abbrev="Chelmek",
|
|
||||||
lLevel=0,
|
|
||||||
webpage=start}
|
|
||||||
extractItemsStartingFromUrl shadowLibrary start (extractRecords >>> arr toShadowItem)
|
|
||||||
|
|
||||||
|
|
||||||
|
|
||||||
@ -5,8 +5,8 @@ description: Please see README.md
|
|||||||
homepage: http://github.com/name/project
|
homepage: http://github.com/name/project
|
||||||
license: Proprietary
|
license: Proprietary
|
||||||
license-file: LICENSE
|
license-file: LICENSE
|
||||||
author: Maciej Ścigacz
|
author: Your name here
|
||||||
maintainer: macsci1@st.amu.edu.pl
|
maintainer: your.address@example.com
|
||||||
-- copyright:
|
-- copyright:
|
||||||
category: Web
|
category: Web
|
||||||
build-type: Simple
|
build-type: Simple
|
||||||
@ -20,7 +20,6 @@ library
|
|||||||
, HTTP
|
, HTTP
|
||||||
, hxt
|
, hxt
|
||||||
, hxt-http
|
, hxt-http
|
||||||
, hxt-curl
|
|
||||||
, hxt-xpath
|
, hxt-xpath
|
||||||
, MissingH
|
, MissingH
|
||||||
, monad-logger
|
, monad-logger
|
||||||
@ -55,25 +54,12 @@ executable almanachmuszyny
|
|||||||
build-depends: base
|
build-depends: base
|
||||||
, hxt
|
, hxt
|
||||||
, hxt-xpath
|
, hxt-xpath
|
||||||
, hxt-curl
|
|
||||||
, MissingH
|
, MissingH
|
||||||
, regex-posix
|
, regex-posix
|
||||||
, shadow-library
|
, shadow-library
|
||||||
default-language: Haskell2010
|
default-language: Haskell2010
|
||||||
|
|
||||||
executable chelmek
|
|
||||||
hs-source-dirs: app
|
|
||||||
main-is: chelmek.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
|
source-repository head
|
||||||
type: git
|
type: git
|
||||||
location: https://git.wmi.amu.edu.pl/s444476/twilight-library
|
location: https://github.com/name/project
|
||||||
|
|||||||
Loading…
Reference in New Issue
Block a user