{-# LANGUAGE RankNTypes, TypeFamilies, CPP #-} {-| This module provides @pipes@ utilities for \"text streams\", which are streams of 'Text' chunks. The individual chunks are uniformly @strict@, but can interact lazy 'Text's and 'IO.Handle's. To stream to or from 'IO.Handle's, use 'fromHandle' or 'toHandle'. For example, the following program copies a document from one file to another: > import Pipes > import qualified Data.Text.Pipes as Text > import System.IO > > main = > withFile "inFile.txt" ReadMode $ \hIn -> > withFile "outFile.txt" WriteMode $ \hOut -> > runEffect $ Text.fromHandle hIn >-> Text.toHandle hOut To stream from files, the following is perhaps more Prelude-like (note that it uses Pipes.Safe): > import Pipes > import qualified Data.Text.Pipes as Text > import Pipes.Safe > > main = runSafeT $ runEffect $ Text.readFile "inFile.txt" >-> Text.writeFile "outFile.txt" You can stream to and from 'stdin' and 'stdout' using the predefined 'stdin' and 'stdout' proxies, as with the following \"echo\" program: > main = runEffect $ Text.stdin >-> Text.stdout You can also translate pure lazy 'TL.Text's to and from proxies: > main = runEffect $ Text.fromLazy (TL.pack "Hello, world!\n") >-> Text.stdout In addition, this module provides many functions equivalent to lazy 'Text' functions so that you can transform or fold text streams. For example, to stream only the first three lines of 'stdin' to 'stdout' you might write: > import Pipes > import qualified Pipes.Text as Text > import qualified Pipes.Parse as Parse > > main = runEffect $ takeLines 3 Text.stdin >-> Text.stdout > where > takeLines n = Text.unlines . Parse.takeFree n . Text.lines The above program will never bring more than one chunk of text (~ 32 KB) into memory, no matter how long the lines are. Note that functions in this library are designed to operate on streams that are insensitive to text boundaries. This means that they may freely split text into smaller texts and /discard empty texts/. However, they will /never concatenate texts/ in order to provide strict upper bounds on memory usage. -} module Pipes.Text ( -- * Producers fromLazy, stdin, fromHandle, readFile, stdinLn, -- * Consumers stdout, stdoutLn, toHandle, writeFile, -- * Pipes map, concatMap, take, drop, takeWhile, dropWhile, filter, scan, encodeUtf8, pack, unpack, stripStart, -- * Folds toLazy, toLazyM, fold, head, last, null, length, any, all, maximum, minimum, find, index, -- elemIndex, -- findIndex, count, -- * Splitters splitAt, chunksOf, span, break, splitWith, split, groupBy, group, lines, words, #if MIN_VERSION_text(0,11,4) decodeUtf8, #endif -- * Transformations intersperse, -- * Joiners intercalate, unlines, unwords, -- * Character Parsers -- $parse nextChar, drawChar, unDrawChar, peekChar, isEndOfChars, -- * Re-exports -- $reexports module Data.Text, module Pipes.Parse ) where import Control.Exception (throwIO, try) import Control.Monad (liftM, unless) import Control.Monad.Trans.State.Strict (StateT) import qualified Data.Text as T import qualified Data.Text.IO as T import qualified Data.Text.Encoding as TE import Data.Text (Text) import qualified Data.Text.Lazy as TL import qualified Data.Text.Lazy.IO as TL import Data.Text.Lazy.Internal (foldrChunks, defaultChunkSize) import Data.ByteString.Unsafe (unsafeTake, unsafeDrop) import Data.ByteString (ByteString) import qualified Data.ByteString as B import Data.Char (ord) import Data.Functor.Identity (Identity) import qualified Data.List as List import Foreign.C.Error (Errno(Errno), ePIPE) import qualified GHC.IO.Exception as G import Pipes import qualified Pipes.ByteString.Parse as PBP import Data.Text.Pipes.Parse ( nextChar, drawChar, unDrawChar, peekChar, isEndOfChars ) import Pipes.Core (respond, Server') import qualified Pipes.Parse as PP import Pipes.Parse (input, concat, FreeT) import qualified Pipes.Safe.Prelude as Safe import qualified Pipes.Safe as Safe import Pipes.Safe (MonadSafe(..), Base(..)) import qualified Pipes.Prelude as P import qualified System.IO as IO import Data.Char (isSpace) import Prelude hiding ( all, any, break, concat, concatMap, drop, dropWhile, elem, filter, head, last, lines, length, map, maximum, minimum, notElem, null, readFile, span, splitAt, take, takeWhile, unlines, unwords, words, writeFile ) -- | Convert a lazy 'TL.Text' into a 'Producer' of strict 'Text's fromLazy :: (Monad m) => TL.Text -> Producer' Text m () fromLazy = foldrChunks (\e a -> yield e >> a) (return ()) {-# INLINABLE fromLazy #-} -- | Stream bytes from 'stdin' stdin :: MonadIO m => Producer' Text m () stdin = fromHandle IO.stdin {-# INLINABLE stdin #-} {-| Convert a 'IO.Handle' into a text stream using a text size determined by the good sense of the text library. -} fromHandle :: MonadIO m => IO.Handle -> Producer' Text m () fromHandle h = go where go = do txt <- liftIO (T.hGetChunk h) unless (T.null txt) $ do yield txt go {-# INLINABLE fromHandle#-} {-| Stream text from a file using Pipes.Safe >>> runSafeT $ runEffect $ Text.readFile "hello.hs" >-> Text.map toUpper >-> hoist lift Text.stdout MAIN = PUTSTRLN "HELLO WORLD" -} readFile :: (MonadSafe m, Base m ~ IO) => FilePath -> Producer' Text m () readFile file = Safe.withFile file IO.ReadMode fromHandle {-# INLINABLE readFile #-} {-| Stream lines of text from stdin (for testing in ghci etc.) >>> let safely = runSafeT . runEffect >>> safely $ for Text.stdinLn (lift . lift . print . T.length) hello 5 world 5 -} stdinLn :: MonadIO m => Producer' Text m () stdinLn = go where go = do eof <- liftIO (IO.hIsEOF IO.stdin) unless eof $ do txt <- liftIO (T.hGetLine IO.stdin) yield txt go {-| Stream text to 'stdout' Unlike 'toHandle', 'stdout' gracefully terminates on a broken output pipe. Note: For best performance, use @(for source (liftIO . putStr))@ instead of @(source >-> stdout)@ in suitable cases. -} stdout :: MonadIO m => Consumer' Text m () stdout = go where go = do txt <- await x <- liftIO $ try (T.putStr txt) case x of Left (G.IOError { G.ioe_type = G.ResourceVanished , G.ioe_errno = Just ioe }) | Errno ioe == ePIPE -> return () Left e -> liftIO (throwIO e) Right () -> go {-# INLINABLE stdout #-} stdoutLn :: (MonadIO m) => Consumer' Text m () stdoutLn = go where go = do str <- await x <- liftIO $ try (T.putStrLn str) case x of Left (G.IOError { G.ioe_type = G.ResourceVanished , G.ioe_errno = Just ioe }) | Errno ioe == ePIPE -> return () Left e -> liftIO (throwIO e) Right () -> go {-# INLINABLE stdoutLn #-} {-| Convert a text stream into a 'Handle' Note: again, for best performance, where possible use @(for source (liftIO . hPutStr handle))@ instead of @(source >-> toHandle handle)@. -} toHandle :: MonadIO m => IO.Handle -> Consumer' Text m r toHandle h = for cat (liftIO . T.hPutStr h) {-# INLINABLE toHandle #-} -- | Stream text into a file. Uses @pipes-safe@. writeFile :: (MonadSafe m, Base m ~ IO) => FilePath -> Consumer' Text m () writeFile file = Safe.withFile file IO.WriteMode toHandle -- | Apply a transformation to each 'Char' in the stream map :: (Monad m) => (Char -> Char) -> Pipe Text Text m r map f = P.map (T.map f) {-# INLINABLE map #-} -- | Map a function over the characters of a text stream and concatenate the results concatMap :: (Monad m) => (Char -> Text) -> Pipe Text Text m r concatMap f = P.map (T.concatMap f) {-# INLINABLE concatMap #-} -- | Transform a Pipe of 'Text' into a Pipe of 'ByteString's using UTF-8 -- encoding encodeUtf8 :: Monad m => Pipe Text ByteString m r encodeUtf8 = P.map TE.encodeUtf8 {-# INLINEABLE encodeUtf8 #-} --| Transform a Pipe of 'String's into one of 'Text' chunks pack :: Monad m => Pipe String Text m r pack = P.map T.pack {-# INLINEABLE pack #-} --| Transforma a Pipes of 'Text' chunks into one of 'String's unpack :: Monad m => Pipe Text String m r unpack = P.map T.unpack {-# INLINEABLE unpack #-} --| @toCaseFold@, @toLower@, @toUpper@ and @stripStart@ are standard 'Text' utility, -- here acting on a 'Text' pipe, rather as they would on a lazy text toCaseFold :: Monad m => Pipe Text Text m () toCaseFold = P.map T.toCaseFold {-# INLINEABLE toCaseFold #-} --| lowercase incoming 'Text' toLower :: Monad m => Pipe Text Text m () toLower = P.map T.toLower {-# INLINEABLE toLower #-} --| uppercase incoming 'Text' toUpper :: Monad m => Pipe Text Text m () toUpper = P.map T.toUpper {-# INLINEABLE toUpper #-} --| Remove leading white space from an incoming succession of 'Text's stripStart :: Monad m => Pipe Text Text m r stripStart = do chunk <- await let text = T.stripStart chunk if T.null text then stripStart else cat {-# INLINEABLE stripStart #-} -- | @(take n)@ only allows @n@ individual characters to pass; -- contrast @Pipes.Prelude.take@ which would let @n@ chunks pass. take :: (Monad m, Integral a) => a -> Pipe Text Text m () take n0 = go n0 where go n | n <= 0 = return () | otherwise = do txt <- await let len = fromIntegral (T.length txt) if (len > n) then yield (T.take (fromIntegral n) txt) else do yield txt go (n - len) {-# INLINABLE take #-} -- | @(drop n)@ drops the first @n@ characters drop :: (Monad m, Integral a) => a -> Pipe Text Text m r drop n0 = go n0 where go n | n <= 0 = cat | otherwise = do txt <- await let len = fromIntegral (T.length txt) if (len >= n) then do yield (T.drop (fromIntegral n) txt) cat else go (n - len) {-# INLINABLE drop #-} -- | Take characters until they fail the predicate takeWhile :: (Monad m) => (Char -> Bool) -> Pipe Text Text m () takeWhile predicate = go where go = do txt <- await let (prefix, suffix) = T.span predicate txt if (T.null suffix) then do yield txt go else yield prefix {-# INLINABLE takeWhile #-} -- | Drop characters until they fail the predicate dropWhile :: (Monad m) => (Char -> Bool) -> Pipe Text Text m r dropWhile predicate = go where go = do txt <- await case T.findIndex (not . predicate) txt of Nothing -> go Just i -> do yield (T.drop i txt) cat {-# INLINABLE dropWhile #-} -- | Only allows 'Char's to pass if they satisfy the predicate filter :: (Monad m) => (Char -> Bool) -> Pipe Text Text m r filter predicate = P.map (T.filter predicate) {-# INLINABLE filter #-} -- | Strict left scan over the characters scan :: (Monad m) => (Char -> Char -> Char) -> Char -> Pipe Text Text m r scan step begin = go begin where go c = do txt <- await let txt' = T.scanl step c txt c' = T.last txt' yield txt' go c' {-# INLINABLE scan #-} {-| Fold a pure 'Producer' of strict 'Text's into a lazy 'TL.Text' -} toLazy :: Producer Text Identity () -> TL.Text toLazy = TL.fromChunks . P.toList {-# INLINABLE toLazy #-} {-| Fold an effectful 'Producer' of strict 'Text's into a lazy 'TL.Text' Note: 'toLazyM' is not an idiomatic use of @pipes@, but I provide it for simple testing purposes. Idiomatic @pipes@ style consumes the chunks immediately as they are generated instead of loading them all into memory. -} toLazyM :: (Monad m) => Producer Text m () -> m TL.Text toLazyM = liftM TL.fromChunks . P.toListM {-# INLINABLE toLazyM #-} -- | Reduce the text stream using a strict left fold over characters fold :: Monad m => (x -> Char -> x) -> x -> (x -> r) -> Producer Text m () -> m r fold step begin done = P.fold (T.foldl' step) begin done {-# INLINABLE fold #-} -- | Retrieve the first 'Char' head :: (Monad m) => Producer Text m () -> m (Maybe Char) head = go where go p = do x <- nextChar p case x of Left _ -> return Nothing Right (c, _) -> return (Just c) {-# INLINABLE head #-} -- | Retrieve the last 'Char' last :: (Monad m) => Producer Text m () -> m (Maybe Char) last = go Nothing where go r p = do x <- next p case x of Left () -> return r Right (txt, p') -> if (T.null txt) then go r p' else go (Just $ T.last txt) p' {-# INLINABLE last #-} -- | Determine if the stream is empty null :: (Monad m) => Producer Text m () -> m Bool null = P.all T.null {-# INLINABLE null #-} -- | Count the number of bytes length :: (Monad m, Num n) => Producer Text m () -> m n length = P.fold (\n txt -> n + fromIntegral (T.length txt)) 0 id {-# INLINABLE length #-} -- | Fold that returns whether 'M.Any' received 'Char's satisfy the predicate any :: (Monad m) => (Char -> Bool) -> Producer Text m () -> m Bool any predicate = P.any (T.any predicate) {-# INLINABLE any #-} -- | Fold that returns whether 'M.All' received 'Char's satisfy the predicate all :: (Monad m) => (Char -> Bool) -> Producer Text m () -> m Bool all predicate = P.all (T.all predicate) {-# INLINABLE all #-} -- | Return the maximum 'Char' within a byte stream maximum :: (Monad m) => Producer Text m () -> m (Maybe Char) maximum = P.fold step Nothing id where step mc txt = if (T.null txt) then mc else Just $ case mc of Nothing -> T.maximum txt Just c -> max c (T.maximum txt) {-# INLINABLE maximum #-} -- | Return the minimum 'Char' within a byte stream minimum :: (Monad m) => Producer Text m () -> m (Maybe Char) minimum = P.fold step Nothing id where step mc txt = if (T.null txt) then mc else case mc of Nothing -> Just (T.minimum txt) Just c -> Just (min c (T.minimum txt)) {-# INLINABLE minimum #-} -- | Find the first element in the stream that matches the predicate find :: (Monad m) => (Char -> Bool) -> Producer Text m () -> m (Maybe Char) find predicate p = head (p >-> filter predicate) {-# INLINABLE find #-} -- | Index into a byte stream index :: (Monad m, Integral a) => a-> Producer Text m () -> m (Maybe Char) index n p = head (p >-> drop n) {-# INLINABLE index #-} -- | Find the index of an element that matches the given 'Char' -- elemIndex -- :: (Monad m, Num n) => Char -> Producer Text m () -> m (Maybe n) -- elemIndex w8 = findIndex (w8 ==) -- {-# INLINABLE elemIndex #-} -- | Store the first index of an element that satisfies the predicate -- findIndex -- :: (Monad m, Num n) -- => (Char -> Bool) -> Producer Text m () -> m (Maybe n) -- findIndex predicate p = P.head (p >-> findIndices predicate) -- {-# INLINABLE findIndex #-} -- -- | Store a tally of how many segments match the given 'Text' count :: (Monad m, Num n) => Text -> Producer Text m () -> m n count c p = P.fold (+) 0 id (p >-> P.map (fromIntegral . T.count c)) {-# INLINABLE count #-} #if MIN_VERSION_text(0,11,4) -- | Transform a Pipe of 'ByteString's expected to be UTF-8 encoded -- into a Pipe of Text decodeUtf8 :: Monad m => Producer ByteString m r -> Producer Text m (Producer ByteString m r) decodeUtf8 = go TE.streamDecodeUtf8 where go dec p = do x <- lift (next p) case x of Left r -> return (return r) Right (chunk, p') -> do let TE.Some text l dec' = dec chunk if B.null l then do yield text go dec' p' else return $ do yield l p' {-# INLINEABLE decodeUtf8 #-} #endif -- | Splits a 'Producer' after the given number of characters splitAt :: (Monad m, Integral n) => n -> Producer Text m r -> Producer' Text m (Producer Text m r) splitAt = go where go 0 p = return p go n p = do x <- lift (next p) case x of Left r -> return (return r) Right (txt, p') -> do let len = fromIntegral (T.length txt) if (len <= n) then do yield txt go (n - len) p' else do let (prefix, suffix) = T.splitAt (fromIntegral n) txt yield prefix return (yield suffix >> p') {-# INLINABLE splitAt #-} -- | Split a text stream into 'FreeT'-delimited text streams of fixed size chunksOf :: (Monad m, Integral n) => n -> Producer Text m r -> FreeT (Producer Text m) m r chunksOf n p0 = PP.FreeT (go p0) where go p = do x <- next p return $ case x of Left r -> PP.Pure r Right (txt, p') -> PP.Free $ do p'' <- splitAt n (yield txt >> p') return $ PP.FreeT (go p'') {-# INLINABLE chunksOf #-} {-| Split a text stream in two, where the first text stream is the longest consecutive group of text that satisfy the predicate -} span :: (Monad m) => (Char -> Bool) -> Producer Text m r -> Producer' Text m (Producer Text m r) span predicate = go where go p = do x <- lift (next p) case x of Left r -> return (return r) Right (txt, p') -> do let (prefix, suffix) = T.span predicate txt if (T.null suffix) then do yield txt go p' else do yield prefix return (yield suffix >> p') {-# INLINABLE span #-} {-| Split a byte stream in two, where the first byte stream is the longest consecutive group of bytes that don't satisfy the predicate -} break :: (Monad m) => (Char -> Bool) -> Producer Text m r -> Producer Text m (Producer Text m r) break predicate = span (not . predicate) {-# INLINABLE break #-} {-| Split a byte stream into sub-streams delimited by bytes that satisfy the predicate -} splitWith :: (Monad m) => (Char -> Bool) -> Producer Text m r -> PP.FreeT (Producer Text m) m r splitWith predicate p0 = PP.FreeT (go0 p0) where go0 p = do x <- next p case x of Left r -> return (PP.Pure r) Right (txt, p') -> if (T.null txt) then go0 p' else return $ PP.Free $ do p'' <- span (not . predicate) (yield txt >> p') return $ PP.FreeT (go1 p'') go1 p = do x <- nextChar p return $ case x of Left r -> PP.Pure r Right (_, p') -> PP.Free $ do p'' <- span (not . predicate) p' return $ PP.FreeT (go1 p'') {-# INLINABLE splitWith #-} -- | Split a text stream using the given 'Char' as the delimiter split :: (Monad m) => Char -> Producer Text m r -> FreeT (Producer Text m) m r split c = splitWith (c ==) {-# INLINABLE split #-} {-| Group a text stream into 'FreeT'-delimited byte streams using the supplied equality predicate -} groupBy :: (Monad m) => (Char -> Char -> Bool) -> Producer Text m r -> FreeT (Producer Text m) m r groupBy equal p0 = PP.FreeT (go p0) where go p = do x <- next p case x of Left r -> return (PP.Pure r) Right (txt, p') -> case (T.uncons txt) of Nothing -> go p' Just (c, _) -> do return $ PP.Free $ do p'' <- span (equal c) (yield txt >> p') return $ PP.FreeT (go p'') {-# INLINABLE groupBy #-} -- | Group a byte stream into 'FreeT'-delimited byte streams of identical bytes group :: (Monad m) => Producer Text m r -> FreeT (Producer Text m) m r group = groupBy (==) {-# INLINABLE group #-} {-| Split a byte stream into 'FreeT'-delimited lines Note: This function is purely for demonstration purposes since it assumes a particular encoding. You should prefer the 'Data.Text.Text' equivalent of this function from the upcoming @pipes-text@ library. -} lines :: (Monad m) => Producer Text m r -> FreeT (Producer Text m) m r lines p0 = PP.FreeT (go0 p0) where go0 p = do x <- next p case x of Left r -> return (PP.Pure r) Right (txt, p') -> if (T.null txt) then go0 p' else return $ PP.Free $ go1 (yield txt >> p') go1 p = do p' <- break ('\n' ==) p return $ PP.FreeT (go2 p') go2 p = do x <- nextChar p return $ case x of Left r -> PP.Pure r Right (_, p') -> PP.Free (go1 p') {-# INLINABLE lines #-} -- | Split a text stream into 'FreeT'-delimited words words :: (Monad m) => Producer Text m r -> FreeT (Producer Text m) m r words p0 = removeEmpty (splitWith isSpace p0) where removeEmpty f = PP.FreeT $ do x <- PP.runFreeT f case x of PP.Pure r -> return (PP.Pure r) PP.Free p -> do y <- next p case y of Left f' -> PP.runFreeT (removeEmpty f') Right (bs, p') -> return $ PP.Free $ do yield bs f' <- p' return (removeEmpty f') {-# INLINABLE words #-} -- | Intersperse a 'Char' in between the bytes of the byte stream intersperse :: (Monad m) => Char -> Producer Text m r -> Producer Text m r intersperse c = go0 where go0 p = do x <- lift (next p) case x of Left r -> return r Right (txt, p') -> do yield (T.intersperse c txt) go1 p' go1 p = do x <- lift (next p) case x of Left r -> return r Right (txt, p') -> do yield (T.singleton c) yield (T.intersperse c txt) go1 p' {-# INLINABLE intersperse #-} {-| 'intercalate' concatenates the 'FreeT'-delimited text streams after interspersing a text stream in between them -} intercalate :: (Monad m) => Producer Text m () -> FreeT (Producer Text m) m r -> Producer Text m r intercalate p0 = go0 where go0 f = do x <- lift (PP.runFreeT f) case x of PP.Pure r -> return r PP.Free p -> do f' <- p go1 f' go1 f = do x <- lift (PP.runFreeT f) case x of PP.Pure r -> return r PP.Free p -> do p0 f' <- p go1 f' {-# INLINABLE intercalate #-} {-| Join 'FreeT'-delimited lines into a byte stream -} unlines :: (Monad m) => FreeT (Producer Text m) m r -> Producer Text m r unlines = go where go f = do x <- lift (PP.runFreeT f) case x of PP.Pure r -> return r PP.Free p -> do f' <- p yield $ T.singleton '\n' go f' {-# INLINABLE unlines #-} {-| Join 'FreeT'-delimited words into a text stream -} unwords :: (Monad m) => FreeT (Producer Text m) m r -> Producer Text m r unwords = intercalate (yield $ T.pack " ") {-# INLINABLE unwords #-} {- $parse The following parsing utilities are single-character analogs of the ones found @pipes-parse@. -} {- $reexports @Pipes.Text.Parse@ re-exports 'nextChar', 'drawChar', 'unDrawChar', 'peekChar', and 'isEndOfChars'. @Data.Text@ re-exports the 'Text' type. @Pipes.Parse@ re-exports 'input', 'concat', and 'FreeT' (the type). -}