archiwumharcerskie.hs changes

This commit is contained in:
Mariusz B 2021-04-11 12:24:25 +00:00
parent 0858766cc9
commit 65cf05c014

View File

@ -12,7 +12,7 @@ import Text.Regex.Posix
import Text.Printf
-- wyciaganie pdf-ow
extractRecords = extractLinksWithText "//nav//a[contains(@href,'title=Kategoria:Dokumenty')]"
extractRecords = extractLinksWithText "//nav//a[contains(@href,'title=Kategoria:')]"
>>> first (extractLinksWithText "//a[contains(@href,'title=Plik') and contains(@href,'pdf')]")
>>> first (first (extractLinksGeneralized "//div[@id='file']//iframe" "src"))