-- Hoogle documentation, generated by Haddock
-- See Hoogle, http://www.haskell.org/hoogle/


-- | Extracts text of main article from HTML document
--   
--   Give <tt>readability</tt> an HTML document and it will detect and
--   extract text of the article while removing everything unnecessary like
--   menus, advertisements or sidebars. It is more or less reimplementation
--   of <a>python-readability</a>.
@package readability
@version 0.0.1.0

module Readability.Helper
innerText :: Cursor -> Text
content' :: Cursor -> [Text]
elin :: Element -> [Name] -> Bool
getElement :: Node -> Maybe Element
guarded :: Alternative f => (a -> Bool) -> a -> f a

module Readability.Types

-- | Result of processing HTML through <i>readability</i>.
data Article
Article :: Document -> Article

-- | Body of article of original HTML.
[summary] :: Article -> Document
data Settings
Settings :: (Name -> Bool) -> Settings

-- | Remove HTML attributes for which the function returns true
[reRemoveAttributes] :: Settings -> Name -> Bool
newtype Scores
Scores :: Map Node (Cursor, Double) -> Scores
[scoreMap] :: Scores -> Map Node (Cursor, Double)
alterScores :: (Maybe (Cursor, Double) -> Maybe (Cursor, Double)) -> Node -> Scores -> Scores
emptyScores :: Scores
lookupCursor :: Node -> Scores -> Maybe Cursor
lookupScore :: Node -> Scores -> Maybe Double
mapScores :: (Cursor -> Double -> (Cursor, Double)) -> Scores -> Scores
maxScore :: Scores -> (Cursor, Double)
nullScores :: Scores -> Bool
instance GHC.Show.Show Readability.Types.Scores
instance GHC.Show.Show Readability.Types.Article

module Readability.Metrics
classWeight :: Element -> Double
contentScore :: Cursor -> Maybe Double
linkDensity :: Cursor -> Double
scoreAncestor :: Cursor -> Double
scoreParagraph :: Scores -> Cursor -> Scores
scoreParagraphs :: [Cursor] -> Maybe Scores
textualSibling :: Cursor -> Bool

module Readability.Clean
cleanElement :: Bool -> Element -> Element
paradivs :: Node -> Node

-- | Remove elements that do not contribute to the article.
--   
--   Removes:
--   
--   <ul>
--   <li>headings that are part of ads, menus, widgets etc.</li>
--   <li>forms, textareas</li>
--   <li>iframes</li>
--   <li>block elements (div, section, table etc.) with enough textual
--   content</li>
--   </ul>
--   
--   Preserves:
--   
--   <ul>
--   <li>textual content</li>
--   </ul>
sanitizeNode :: Settings -> Scores -> Node -> Maybe Node

module Readability.Internal
summary :: Settings -> Document -> Maybe Document
rootSummary :: Settings -> Bool -> Element -> Maybe Element

module Readability

-- | Result of processing HTML through <i>readability</i>.
data Article
Article :: Document -> Article

-- | Body of article of original HTML.
[summary] :: Article -> Document

-- | Extracts article from HTML in ByteString.
fromByteString :: ByteString -> Maybe Article

-- | Extracts article from HTML represented as HTML document.
fromDocument :: Document -> Maybe Article

-- | Extracts article from HTML in given file.
fromFile :: FilePath -> IO (Maybe Article)

-- | Extracts article from HTML in given text.
fromText :: Text -> Maybe Article