pandocTools/extractByTag.hs

   1 -- extractByTag.hs
   2 --
   3 -- Extract any text that has been markup up with a given tag.
   4
   5 import Text.HTML.TagSoup
   6 import System.IO
   7 import System.Environment (getArgs)
   8 import Data.List.HT (breakAfter)
   9
  10 checkArgs :: [String] -> Bool
  11 checkArgs [] = True
  12 checkArgs x = any (elem '-') x
  13
  14 helpstring = "Bad Arguments!\n"
  15
  16 main :: IO ()
  17 main = do
  18         args <- getArgs
  19         if (checkArgs args)
  20                 then hPutStr stderr helpstring
  21                 else interact (unlines . extractByTag args)
  22
  23 extractByTag :: [String] -> String -> [String]
  24 extractByTag tags s = map renderTags (filterTags tags (parseTags s))
  25
  26 filterTags :: [String] -> [Tag String] -> [[Tag String]]
  27 filterTags _ [] = []
  28 filterTags tags (t:ts)
  29         | isStartTag tags t =
  30                 let (good, rest) = breakAfter (isStopTag (tagStr t)) (t:ts)
  31                 in good:(filterTags tags rest)
  32         | otherwise = filterTags tags ts
  33
  34 isStartTag :: [String] -> Tag String -> Bool
  35 isStartTag tags (TagOpen str _) = elem str tags
  36 isStartTag _ _ = False
  37
  38 tagStr :: Tag String -> String
  39 tagStr (TagOpen s _) = s
  40 tagStr (TagClose s)  = s
  41 tagStr _             = []
  42
  43 isStopTag :: String -> Tag String -> Bool
  44 isStopTag tag (TagClose str) = (tag == str)
  45 isStopTag _ _ = False
  46