From 5cedcecd6ae31e2485dcab2ddd74c74a4779545d Mon Sep 17 00:00:00 2001 From: Joris Date: Tue, 3 Sep 2019 21:01:53 +0200 Subject: Make LBC to work Use request headers to simulate a normal browser --- src/lib/haskell/Parser/LeboncoinParser.hs | 17 +++++++++++------ 1 file changed, 11 insertions(+), 6 deletions(-) (limited to 'src/lib/haskell/Parser/LeboncoinParser.hs') diff --git a/src/lib/haskell/Parser/LeboncoinParser.hs b/src/lib/haskell/Parser/LeboncoinParser.hs index 77213cb..99d8116 100644 --- a/src/lib/haskell/Parser/LeboncoinParser.hs +++ b/src/lib/haskell/Parser/LeboncoinParser.hs @@ -11,14 +11,19 @@ import Model.Ad (Ad (Ad)) import Parser.Utils parse :: Text -> [Ad] -parse page = - catMaybes . fmap parseAd $ partitions (~== (T.unpack "")) tags - where tags = getTagsBetween "
  • " "
    " (parseTags page) +parse = + catMaybes + . fmap parseAd + . partitions (~== (T.unpack "
  • ")) + . parseTags parseAd :: [Tag Text] -> Maybe Ad parseAd tags = do - name <- getTagTextAfter "

    " tags - location <- getTagAttribute "" "content" tags - let price = getTagTextAfter "

    " tags + name <- getTagTextAfter "" tags + location <- getTagTextAfter "

    " tags + let price = + case getTagsBetween "" "" tags of + [] -> Nothing + xs -> Just $ innerText xs url <- getTagAttribute "" "href" tags return (Ad name location price (T.concat ["https:", url])) -- cgit v1.2.3