diff options
author | Joris Guyonvarch | 2015-04-11 11:50:48 +0200 |
---|---|---|
committer | Joris Guyonvarch | 2015-04-11 12:03:22 +0200 |
commit | 5977e1454d7738ddb086d37b20337e350e380790 (patch) | |
tree | e03261144e3d534434242c1dd037c2a4e4db5a9f /src/Parser/Utils.hs | |
download | ad-listener-5977e1454d7738ddb086d37b20337e350e380790.tar.gz ad-listener-5977e1454d7738ddb086d37b20337e350e380790.tar.bz2 ad-listener-5977e1454d7738ddb086d37b20337e350e380790.zip |
Fetch first page ads of a given leboncoin url, fetch also the description page of each item.
Diffstat (limited to 'src/Parser/Utils.hs')
-rw-r--r-- | src/Parser/Utils.hs | 30 |
1 files changed, 30 insertions, 0 deletions
diff --git a/src/Parser/Utils.hs b/src/Parser/Utils.hs new file mode 100644 index 0000000..4864e00 --- /dev/null +++ b/src/Parser/Utils.hs @@ -0,0 +1,30 @@ +module Parser.Utils + ( getTagAttribute + , getTagText + ) where + +import Data.List (find, findIndex) +import Data.Maybe (listToMaybe) + +import Text.HTML.TagSoup + +getTagAttribute :: String -> String -> [Tag String] -> Maybe String +getTagAttribute selector attribute item = + find (~== selector) item >>= maybeTagAttribute attribute + +getTagText :: String -> [Tag String] -> Maybe String +getTagText selector item = + case findIndex (~== selector) item of + Just index -> fmap trim $ safeGetAt (index + 1) item >>= maybeTagText + Nothing -> Nothing + +maybeTagAttribute :: String -> Tag String -> Maybe String +maybeTagAttribute name (TagOpen _ xs) = + fmap snd . find (\(x, _) -> x == name) $ xs +maybeTagAttribute attribute _ = Nothing + +trim :: String -> String +trim = unwords . words + +safeGetAt :: Int -> [a] -> Maybe a +safeGetAt index = listToMaybe . drop index |