regex-tre-0.91: Text/Regex/TRE/Sequence.hs

{-# OPTIONS_GHC -fglasgow-exts -fno-warn-orphans #-}
{-|
This exports instances of the high level API and the medium level
API of 'compile','execute', and 'regexec'.
-}
{- Copyright   :  (c) Chris Kuklewicz 2007 -}
module Text.Regex.TRE.Sequence(
  -- ** Types
  Regex,
  MatchOffset,
  MatchLength,
  CompOption(CompOption),
  ExecOption(ExecOption),
  ReturnCode,
  WrapError,
  -- ** Miscellaneous
  unusedOffset,
  getVersion,
  -- ** Medium level API functions
  compile,
  regexec,
  execute,
  -- ** CompOption flags
  compBlank,
  compExtended,   -- use extended regex syntax
  compIgnoreCase, -- ignore case when matching
  compNoSub,      -- no substring matching needed
  compNewline,    -- '.' doesn't match newline
  -- ** ExecOption flags
  execBlank,
  execNotBOL,     -- not at begining of line
  execNotEOL     -- not at end of line
  ) where

import Text.Regex.TRE.Wrap -- all
import Data.Array(Array,listArray)
import System.IO.Unsafe(unsafePerformIO)
import Text.Regex.Base.RegexLike(RegexMaker(..),RegexLike(..),RegexContext(..),MatchOffset,MatchLength,Extract(..))
import Text.Regex.Base.Impl(polymatch,polymatchM)
import Data.Sequence as S hiding (length)
import qualified Data.Sequence as S (length)
import Foreign.C.String
import Foreign.Marshal.Array
import Foreign.Marshal.Alloc
import Foreign.Storable

instance RegexContext Regex (Seq Char) (Seq Char) where
  match = polymatch
  matchM = polymatchM

unusedOffset :: Int
unusedOffset = fromIntegral unusedRegOffset

unwrap :: (Show e) => Either e v -> IO v
unwrap x = case x of Left err -> fail ("Text.Regex.TRE.Sequence died: "++show err)
                     Right v -> return v

instance RegexMaker Regex CompOption ExecOption (Seq Char) where
  makeRegexOpts c e pattern = unsafePerformIO $
    compile c e pattern >>= unwrap
  makeRegexOptsM c e pattern = either (fail.show) return $ unsafePerformIO $
    compile c e pattern

instance RegexLike Regex (Seq Char) where
  matchTest regex str = unsafePerformIO $
    withSeq str (wrapTest regex) >>= unwrap
  matchOnce regex str = unsafePerformIO $
    execute regex str >>= unwrap
  matchAll regex str = unsafePerformIO $ 
    withSeq str (wrapMatchAll regex) >>= unwrap
  matchCount regex str = unsafePerformIO $ 
    withSeq str (wrapCount regex) >>= unwrap

-- | Compiles a regular expression
compile :: CompOption -- ^ Flags (summed together)
        -> ExecOption -- ^ Flags (summed together)
        -> (Seq Char)     -- ^ The regular expression to compile
        -> IO (Either WrapError Regex) -- ^ Returns: an error string and offset or the compiled regular expression
compile c e pattern = withSeq pattern (wrapCompile c e)

-- | Matches a regular expression against a string
execute :: Regex      -- ^ Compiled regular expression
        -> (Seq Char)     -- ^ (Seq Char) to match against
        -> IO (Either WrapError (Maybe (Array Int (MatchOffset,MatchLength))))
                -- ^ Returns: 'Nothing' if the regex did not match the
                -- string, or:
                --   'Just' an array of (offset,length) pairs where index 0 is whole match, and the rest are the captured subexpressions.
execute regex str = do
  maybeStartEnd <- withSeq str (wrapMatch regex)
  case maybeStartEnd of
    Left err -> return (Left err)
    Right Nothing -> return (Right Nothing)
    Right (Just parts) -> 
      return . Right . Just . listArray (0,pred (length parts))
      . map (\(s,e)->(fromIntegral s, fromIntegral (e-s))) $ parts


-- | execute match and extract substrings rather than just offsets
regexec  :: Regex      -- ^ compiled regular expression
         -> (Seq Char)     -- ^ string to match
         -> IO (Either WrapError (Maybe ((Seq Char), (Seq Char),(Seq Char), [(Seq Char)])))
                      -- ^ Returns: Nothing if no match, else
                      --   (text before match, text after match, array of matches with 0 being the whole match)
regexec regex str = do
  let getSub (start',stop') | start == unusedOffset = S.empty
                            | otherwise = extract (start,stop-start) $ str
        where (start,stop) = (fromIntegral start', fromIntegral stop')
      matchedParts [] = (S.empty,S.empty,str,[]) -- no information
      matchedParts ((start',stop'):subStartStop) = 
        let matchedStartStop@(start,stop) = (fromIntegral start', fromIntegral stop')
        in (before start str
           ,getSub matchedStartStop
           ,after stop str
           ,map getSub subStartStop)
  maybeStartEnd <- withSeq str (wrapMatch regex)
  case maybeStartEnd of
    Left err -> return (Left err)
    Right Nothing -> return (Right Nothing)
    Right (Just parts) -> return . Right . Just . matchedParts $ parts

withSeq :: Seq Char -> (CStringLen -> IO a) -> IO a
withSeq s f =
  let -- Ensure null at end of s
      len = S.length s
      pokes p a = case viewl a of           -- bang pokes !p !a
                    EmptyL -> return ()
                    c :< a' -> poke p (castCharToCChar c) >> pokes (advancePtr p 1) a'
  in allocaBytes (S.length s) (\ptr -> pokes ptr s >> f (ptr,len))