module Processing (getAttrs,
                   extractLink,
                   extractLinks) where

import Text.HTML.TagSoup
import Text.Regex.Posix
import Data.Maybe

-- | Implementation

getAttrs :: [Tag String] -> String -> String -> [String]
getAttrs tags tagName attrName = mapMaybe link tags
  where
    firstAttr [] = Nothing
    firstAttr ((n, v):as) | n == attrName = Just v
                          | otherwise = firstAttr as
    link (TagOpen tag attrs) | tag == tagName = firstAttr attrs
    link _ = Nothing

extractLinks :: String -> String -> String -> String -> [String]
extractLinks r tag attr regexp =
  let tags = parseTags r :: [Tag String]
      links = getAttrs tags tag attr
      isPicture x = (x =~ regexp) --(x =~ "\\.jpg$") || (x =~ "\\.png$")
      isPictureUrl x = (isPicture x) --(&&) (isURI x) (True) --(isPicture x)
      urls = filter isPictureUrl links
  in urls

extractLink :: String -> String -> String -> String -> Maybe String
extractLink r tag attr regexp = passUrl (extractLinks r tag attr regexp) where
  passUrl [] = Nothing
  passUrl (x:_) = Just x