BBCode lexer

1 files changed, 74 insertions, 0 deletions

diff --git a/bbcode/src/Text/BBCode/Lexer.hs b/bbcode/src/Text/BBCode/Lexer.hs
new file mode 100644
index 0000000..d2aa2bc
--- /dev/null
+++ b/bbcode/src/Text/BBCode/Lexer.hs
@@ -0,0 +1,74 @@
+{-# 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, head, last, tail, null)
+
+import Control.Applicative
+
+import Test.QuickCheck (Arbitrary(..), CoArbitrary, genericShrink)
+import Test.QuickCheck.Gen (oneof, suchThat)
+import Test.QuickCheck.Instances
+import GHC.Generics (Generic)
+
+-- | Our lexicographical unit
+data BBToken = BBOpen Text -- ^ Tag open
+             | BBClose Text -- ^ Tag close
+             | BBStr Text -- ^ Content of a tag
+             deriving (Generic, Eq, Show, CoArbitrary)
+
+-- | This instance does not produce:
+--
+--  * opening and closing tags whose 'Text' ends in @\\@
+--  * empty 'BBStr's
+instance Arbitrary BBToken where
+  shrink = genericShrink
+  arbitrary = oneof [ BBOpen <$> tagText
+                    , BBClose <$> tagText
+                    , BBStr <$> nonEmpty
+                    ]
+    where
+      tagText = arbitrary `suchThat` (not . lastIsEscape)
+      lastIsEscape t
+        | T.null t = False
+        | T.last t == '\\' = True
+        | otherwise = False
+      nonEmpty = (arbitrary `suchThat` (not . T.null))
+
+token :: Parser BBToken
+-- ^ Tokenizer
+token = BBClose <$> (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