{-# LANGUAGE OverloadedStrings #-}
{-# LANGUAGE DeriveGeneric, DeriveAnyClass #-}

-- | A parser to transform 'Text' into a stream of 'BBToken's
module Text.BBCode.Lexer
       ( BBToken(..)
       , token
       , escapedText
       , escapedText'
       ) where

import Data.Attoparsec.Text

import Data.Text (Text)
import qualified Data.Text as T (singleton)

import Control.Applicative

-- | Our lexicographical unit
data BBToken = BBOpen Text -- ^ Tag open
             | BBClose Text -- ^ Tag close
             | BBStr Text -- ^ Content of a tag
             deriving (Eq, Show)

token :: Parser BBToken
-- ^ Tokenizer
token = BBClose <$> (string "[/" *> escapedText' [']'] <* string "]")
        <|> BBOpen <$> (string "[" *> escapedText' [']'] <* string "]")
        <|> BBStr <$> escapedText ['[']

escapedText :: [Char] -> Parser Text
-- ^ @escapedText cs@ consumes 'Text' up to (not including) the first occurence of a character from @cs@ that is not escaped using @\\@
--
-- Always consumes at least one character
--
-- @\\@ needs to be escaped (prefixed with @\\@) iff it precedes a character from @cs@
escapedText [] = takeText
escapedText cs = recurse $ choice [ takeWhile1 (not . special)
                                  , escapeSeq
                                  , escapeChar'
                                  ]
  where
    escapeChar = '\\'
    special = inClass $ escapeChar : cs
    escapeChar' = string $ T.singleton escapeChar
    escapeSeq = escapeChar' >> (T.singleton <$> satisfy special) -- s/\\([:cs])/\1/
    recurse p = mappend <$> p <*> escapedText' cs

escapedText' :: [Char] -> Parser Text
-- ^ @'option' "" $ 'escapedText' cs@
escapedText' cs = option "" $ escapedText cs