fix typos

This commit is contained in:
Norbert Litkowski 2021-04-07 04:07:35 +02:00
parent 5590d3d8ce
commit 45e471a312

View File

@ -17,7 +17,7 @@ getLinkAndText xpathCondition = proc doc -> do
name <- getElemName -< xpathTrees name <- getElemName -< xpathTrees
txt <- (listA (deep isText >>> getText) >>> arr (intercalate " ")) -< xpathTrees txt <- (listA (deep isText >>> getText) >>> arr (intercalate " ")) -< xpathTrees
href <- (getXPathTrees "//a" >>> getAttrValue "href") -< xpathTrees href <- (getXPathTrees "//a" >>> getAttrValue "href") -< xpathTrees
returnA -< href, txt returnA -< (href, txt)
extractNestedLinksWithText xpathCondition = proc url -> do extractNestedLinksWithText xpathCondition = proc url -> do
@ -29,8 +29,8 @@ extractNestedLinksWithText xpathCondition = proc url -> do
extractRecords = proc x -> do extractRecords = proc x -> do
(a, b) <- extractLinksWithText "//aside[@class='widget widget_maxmegamenu']//a[@class='mega-menu-link']" -< x -- pary adres-tytuł podstrony (a, b) <- extractLinksWithText "//aside[@class='widget widget_maxmegamenu']//a[@class='mega-menu-link']" -< x -- pary adres-tytuł podstrony
(a', b') <- (extractLinksWithText "//aside[@class='widget widget_maxmegamenu']//a[@class='mega-menu-link']") -< a -- pobieramy podstronę i kolejne podstrony z menu (a', b') <- extractLinksWithText "//aside[@class='widget widget_maxmegamenu']//a[@class='mega-menu-link']" -< a -- pobieramy podstronę i kolejne podstrony z menu
a'' <- (extractNestedLinksWithText "//big[a[contains(@href,'.pdf')][img]]") -< a' -- pobieramy stronę z adresu URL i wyciągamy linki z tej strony pasujące do wyrażenia XPathowego a'' <- extractNestedLinksWithText "(//big[a[contains(@href,'.pdf')][img]])[1]" -< a' -- pobieramy stronę z adresu URL i wyciągamy linki z tej strony pasujące do wyrażenia XPathowego
returnA -< ((a'', b'), b) returnA -< ((a'', b'), b)
-- ostatecznie wyjdą krotki (((adres URL, tytuł nr-u), tytuł podstrony 2), tytuł podstrony 1) -- ostatecznie wyjdą krotki (((adres URL, tytuł nr-u), tytuł podstrony 2), tytuł podstrony 1)
@ -48,7 +48,7 @@ toShadowItem (((url, releaseTitle), collectionTitle), categoryTitle) =
getDate yearlyTitle = getDate yearlyTitle =
case yearlyTitle =~~ "/.+(19[0-9][0-9]|20[0-9][0-9])/" :: Maybe [[String]] of case yearlyTitle =~~ "/(19[0-9][0-9]|20[0-9][0-9])/" :: Maybe [[String]] of
Just [[_, year]] -> year Just [[_, year]] -> year
-- otherwise -> error $ "unexpected yearlyTitle: " ++ yearlyTitle -- otherwise -> error $ "unexpected yearlyTitle: " ++ yearlyTitle
otherwise -> yearlyTitle otherwise -> yearlyTitle