conditional-bleh.hs

tokenise :: CleanHillEntry -> String -> String
tokenise hill str = unwords (render toks)
 where
  toks = case parse parseChunks "" str of
            Left e   -> error $ "bug (chunking should always work): " ++ show e
            Right cs -> cs
  --
  render [] = []
  render (Word x : ts)  = x : render ts
  render (Desig x : ts) = x : render ts
  render (Punct x : ts) = [x] : render ts
  render (Num x : ts) = num : render ts
    where
     num  = fromMaybe "_OTHER_NUMBER_" (possibilities ts)
     possibilities (Word w : _)
       | w `elem` [ "ft", "feet" ] && matchH chiHeightFeet x = Just "_HEIGHT_FEET_"
       | "m" `isPrefixOf` w =
         case () of _ | any (isWord "col")  toks && matchMH chiColHeight x -> Just "_COL_HEIGHT_"
                      | any (isWord "drop") toks && matchMH chiColDrop x   -> Just "_COL_DROP_"
                      | matchH chiHeightMetres x                           -> Just "_HEIGHT_METRES_"
                      | otherwise                                          -> Nothing
     possibilities _ = Nothing
  --
  matchMH f x =
   case f hill of
     MaybeJSON (Nothing) -> False
     MaybeJSON (Just h)  -> heightMatches h x
  matchH f = heightMatches (f hill)

-- ----------------------------------------------------------------------
-- parsers
-- ----------------------------------------------------------------------

-- should try to preserve the text as much as possible

data Chunk = Num String
           | Desig String -- ^ eg. A10, BN2
           | Word String
           | Punct Char
 deriving (Show, Eq, Ord)

parseChunks = (parseChunk `sepEndBy` optional space) <* eof
parseChunk  = nonStr <|> str
 where
  nonStr = desig <|> num <|> punct
  --
  desig = try $ do
             ds <- (:) <$> dchar <*> many1 dchar
             if any isDigit ds && any isUpper ds
                then return (Desig ds)
                else fail "designation (at least 1 digit and 1 upper)"
          where
           dchar = digit <|> satisfy isUpper
  --
  num = try $ do
           ds <- (:) <$> digit <*> many (digit <|> oneOf ".,")
           return (Num ds)
  punct = Punct <$> satisfy isPunctuation
  str   = Word  <$> many1 (notFollowedBy' nonStr >> satisfy (not . isSpace))

input.txt

Ben Nevis (Scottish Gaelic: Beinn Nibheis, Scottish Gaelic pronounciation: peˈɲivəʃ) is the highest mountain in the British Isles.
It is located at the western end of the Grampian Mountains in the Lochaber area of Scotland, close to the town of Fort William.
A route popular with experienced hillwalkers starts at Torlundy, a few miles north-east of Fort William on the A82 road, and follows the path alongside the Allt a' Mhuilinn.
In the late 1990s, Lochaber Mountain Rescue Team erected two posts on the summit plateau, in order to assist walkers attempting the descent in foggy conditions.

output.txt

Ben Nevis ( Scottish Gaelic : Beinn Nibheis , Scottish Gaelic pronounciation : peˈɲivəʃ ) is the highest mountain in the British Isles .
It is located at the western end of the Grampian Mountains in the Lochaber area of Scotland , close to the town of Fort William .
A route popular with experienced hillwalkers starts at Torlundy , a few miles north - east of Fort William on the A82 road , and follows the path alongside the Allt a ' Mhuilinn .
In the late _OTHER_NUMBER_ s , Lochaber Mountain Rescue Team erected two posts on the summit plateau , in order to assist walkers attempting the descent in foggy conditions .