32785ad · streamly · git.iko.soy

git.iko.soy/streamly

on master help sign in

Fix doc and naming of wordWithQuotes/blockWithQuotes

32785adHarendra Kumar4 years ago

core/src/Streamly/Data/Parser.hs+2−2

compare with master

unifiedsplit

@@ -125,8 +125,8 @@ module Streamly.Data.Parser

125 -- , groupByRollingEither

126

127 -- ** Framing

128− , wordFramedBy

128+ -- , wordFramedBy

129− , wordQuotedBy

129+ , wordWithQuotes

130

131 -- * Alternative

132 , alt

core/src/Streamly/Internal/Data/Parser/ParserD.hs+52−41

@@ -145,8 +145,8 @@ module Streamly.Internal.Data.Parser.ParserD

145 , takeFramedByEsc_

146 , takeFramedByGeneric

147 , wordFramedBy

148− , wordQuotedBy

148+ , wordWithQuotes

149− , blockQuotedBy

149+ , blockWithQuotes

150

151 -- Matching strings

152 -- , prefixOf -- match any prefix of a given string

@@ -1504,7 +1504,7 @@ wordFramedBy :: Monad m =>

1504 (a -> Bool) -- ^ Matches escape elem?

1505 -> (a -> Bool) -- ^ Matches left quote?

1506 -> (a -> Bool) -- ^ matches right quote?

1507− -> (a -> Bool) -- ^ matches word seperator?

1507+ -> (a -> Bool) -- ^ matches word separator?

1508 -> Fold m a b

1509 -> Parser a m b

1510wordFramedBy isEsc isBegin isEnd isSep

@@ -1592,38 +1592,48 @@ data WordQuotedState s b a =

1592-- XXX If quote start and end chars are different then quotes can be nested

1593-- without escaping. However, is there a case for the use of nested quotes?

1594

1595−-- XXX Simplify like blockQuotedBy

1595+-- XXX Simplify like blockWithQuotes

1596

1597−-- | Like 'wordFramedBy' but the closing quote is determined by the opening

1597+-- | Like 'wordBy' but word separators within quotes are ignored. Quote

1598−-- quote. The first quote begin starts a quote that is closed by its

1598+-- character can be used within quotes if escaped using the supplied escape

1599−-- corresponding closing quote.

1599+-- char. A different quoting character can be used inside quotes without

1600+-- escaping. Starting quote determines the ending quote.

1600--

1601--

1601−-- 'wordFramedBy' and 'wordQuotedBy' both allow multiple quote characters based

1602+-- Example:

1602−-- on the predicates but 'wordQuotedBy' always fixes the quote at the first

1603−-- occurrence and then it is closed only by the corresponding closing quote.

1604−-- Therefore, other quoting characters can be embedded inside it as normal

1605−-- characters. On the other hand, 'wordFramedBy' would close the quote as soon

1606−-- as it encounters any of the closing quotes.

1607--

1603--

1604+-- >>> :{

1605+-- >>> lq x =

1606+-- >>> case x of

1607+-- >>> '"' -> Just x

1608+-- >>> '\'' -> Just x

1609+-- >>> _ -> Nothing

1610+-- >>> :}

1611+--

1608−-- >>> q = (`elem` ['"', '\''])

1612+-- >>> rq = (`elem` ['"', '\''])

1609−-- >>> p kQ = Parser.wordQuotedBy kQ (== '\\') q q id isSpace Fold.toList

1613+-- >>> p keep = Parser.wordWithQuotes keep (== '\\') lq rq isSpace Fold.toList

1610--

1614--

1611-- >>> Stream.parse (p False) $ Stream.fromList "a\"b'c\";'d\"e'f ghi"

1615-- >>> Stream.parse (p False) $ Stream.fromList "a\"b'c\";'d\"e'f ghi"

1612-- Right "ab'c;d\"ef"

1616-- Right "ab'c;d\"ef"

1613--

1617--

1618+-- Outer quotes and backslashes from the input string are removed by Haskell,

1619+-- therefore, the actual input string looks like:

1620+-- a"b'c";'d"e'f ghi

1621+-- From this the outer quotes are stripped by the parser, double quotes inside

1622+-- single quotes or vice-versa are kept.

1623+--

1614-- >>> Stream.parse (p True) $ Stream.fromList "a\"b'c\";'d\"e'f ghi"

1624-- >>> Stream.parse (p True) $ Stream.fromList "a\"b'c\";'d\"e'f ghi"

1615-- Right "a\"b'c\";'d\"e'f"

1625-- Right "a\"b'c\";'d\"e'f"

1616--

1626--

1617−{-# INLINE wordQuotedBy #-}

1627+{-# INLINE wordWithQuotes #-}

1618−wordQuotedBy :: (Monad m, Eq a) =>

1628+wordWithQuotes :: (Monad m, Eq a) =>

1619 Bool -- ^ keep the quotes in the output

1629 Bool -- ^ keep the quotes in the output

1620 -> (a -> Bool) -- ^ Matches an escape elem?

1630 -> (a -> Bool) -- ^ Matches an escape elem?

1621 -> (a -> Maybe a) -- ^ If left quote, return right quote, else nothing.

1631 -> (a -> Maybe a) -- ^ If left quote, return right quote, else nothing.

1622 -> (a -> Bool) -- ^ Matches a right quote?

1632 -> (a -> Bool) -- ^ Matches a right quote?

1623− -> (a -> Bool) -- ^ Matches a word seperator?

1633+ -> (a -> Bool) -- ^ Matches a word separator?

1624 -> Fold m a b

1634 -> Fold m a b

1625 -> Parser a m b

1635 -> Parser a m b

1626−wordQuotedBy keepQuotes isEsc toRight isEnd isSep

1636+wordWithQuotes keepQuotes isEsc toRight isEnd isSep

1627 (Fold fstep finitial fextract) =

1637 (Fold fstep finitial fextract) =

1628 Parser step initial extract

1638 Parser step initial extract

1629

1639

@@ -1635,7 +1645,7 @@ wordQuotedBy keepQuotes isEsc toRight isEnd isSep

1635 case res of

1645 case res of

1636 FL.Partial s -> IPartial (WordQuotedSkipPre s)

1646 FL.Partial s -> IPartial (WordQuotedSkipPre s)

1637 FL.Done _ ->

1647 FL.Done _ ->

1638− error "wordQuotedBy: fold done without input"

1648+ error "wordWithQuotes: fold done without input"

1639

1649

1640 {-# INLINE process #-}

1650 {-# INLINE process #-}

1641 process s a n ql qr = do

1651 process s a n ql qr = do

@@ -1664,7 +1674,7 @@ wordQuotedBy keepQuotes isEsc toRight isEnd isSep

1664 else return $ Continue 0 $ WordQuotedWord s 1 a qr

1674 else return $ Continue 0 $ WordQuotedWord s 1 a qr

1665 Nothing

1675 Nothing

1666 | isEnd a ->

1676 | isEnd a ->

1667− return $ Error "wordQuotedBy: missing frame start"

1677+ return $ Error "wordWithQuotes: missing frame start"

1668 | otherwise -> processUnquoted s a

1678 | otherwise -> processUnquoted s a

1669 step (WordUnquotedWord s) a

1679 step (WordUnquotedWord s) a

1670 | isEsc a = return $ Continue 0 $ WordUnquotedEsc s

1680 | isEsc a = return $ Continue 0 $ WordUnquotedEsc s

@@ -1679,7 +1689,7 @@ wordQuotedBy keepQuotes isEsc toRight isEnd isSep

1679 else return $ Continue 0 $ WordQuotedWord s 1 a qr

1689 else return $ Continue 0 $ WordQuotedWord s 1 a qr

1680 Nothing ->

1690 Nothing ->

1681 if isEnd a

1691 if isEnd a

1682− then return $ Error "wordQuotedBy: missing frame start"

1692+ then return $ Error "wordWithQuotes: missing frame start"

1683 else processUnquoted s a

1693 else processUnquoted s a

1684 step (WordQuotedWord s n ql qr) a

1694 step (WordQuotedWord s n ql qr) a

1685 | isEsc a = return $ Continue 0 $ WordQuotedEsc s n ql qr

1695 | isEsc a = return $ Continue 0 $ WordQuotedEsc s n ql qr

@@ -1713,11 +1723,11 @@ wordQuotedBy keepQuotes isEsc toRight isEnd isSep

1713 extract (WordQuotedWord s n _ _) =

1723 extract (WordQuotedWord s n _ _) =

1714 if n == 0

1724 if n == 0

1715 then fmap (Done 0) $ fextract s

1725 then fmap (Done 0) $ fextract s

1716− else err "wordQuotedBy: missing frame end"

1726+ else err "wordWithQuotes: missing frame end"

1717 extract WordQuotedEsc {} =

1727 extract WordQuotedEsc {} =

1718− err "wordQuotedBy: trailing escape"

1728+ err "wordWithQuotes: trailing escape"

1719 extract (WordUnquotedEsc _) =

1729 extract (WordUnquotedEsc _) =

1720− err "wordQuotedBy: trailing escape"

1730+ err "wordWithQuotes: trailing escape"

1721 extract (WordQuotedSkipPost b) = return (Done 0 b)

1731 extract (WordQuotedSkipPost b) = return (Done 0 b)

1722

1732

1723data BlockParseState s =

1733data BlockParseState s =

@@ -1733,26 +1743,27 @@ data BlockParseState s =

1733

1743

1734-- XXX Put a limit on nest level to keep the API safe.

1744-- XXX Put a limit on nest level to keep the API safe.

1735

1745

1736−-- | Parse a block using block enclosed by open, close identifiers, block

1746+-- | Parse a block enclosed within open, close brackets. Block contents may be

1737−-- identifiers inside quotes are ignored. Quoting characters can be escaped. A

1747+-- quoted, brackets inside quotes are ignored. Quoting characters can be used

1738−-- block can have a nested block inside it.

1748+-- within quotes if escaped. A block can have a nested block inside it.

1739--

1749--

1750+-- Quote begin and end chars are the same. Block brackets and quote chars must

1740−-- Block chars and quote chars must not overlap. Block start and end chars must

1751+-- not overlap. Block start and end brackets must be different for nesting

1741−-- not be the same for nesting without escaping.

1752+-- blocks within blocks.

1742--

1753--

1743−-- >>> p = Parser.blockQuotedBy '{' '}' '"' '\\' Fold.toList

1754+-- >>> p = Parser.blockWithQuotes '{' '}' '"' '\\' Fold.toList

1744-- >>> Stream.parse p $ Stream.fromList "{msg: \"hello world\"}"

1755-- >>> Stream.parse p $ Stream.fromList "{msg: \"hello world\"}"

1745-- Right "msg: \"hello world\""

1756-- Right "msg: \"hello world\""

1746--

1757--

1747−{-# INLINE blockQuotedBy #-}

1758+{-# INLINE blockWithQuotes #-}

1748−blockQuotedBy :: (Monad m, Eq a) =>

1759+blockWithQuotes :: (Monad m, Eq a) =>

1749− a -- ^ Block opening char

1760+ a -- ^ Block opening bracket

1750− -> a -- ^ Block closing char

1761+ -> a -- ^ Block closing bracket

1751 -> a -- ^ quote char

1762 -> a -- ^ quote char

1752 -> a -- ^ escape char

1763 -> a -- ^ escape char

1753 -> Fold m a b

1764 -> Fold m a b

1754 -> Parser a m b

1765 -> Parser a m b

1755−blockQuotedBy bopen bclose quote esc

1766+blockWithQuotes bopen bclose quote esc

1756 (Fold fstep finitial fextract) =

1767 (Fold fstep finitial fextract) =

1757 Parser step initial extract

1768 Parser step initial extract

1758

1769

@@ -1764,7 +1775,7 @@ blockQuotedBy bopen bclose quote esc

1764 case res of

1775 case res of

1765 FL.Partial s -> IPartial (BlockInit s)

1776 FL.Partial s -> IPartial (BlockInit s)

1766 FL.Done _ ->

1777 FL.Done _ ->

1767− error "wordQuotedBy: fold finished without input"

1778+ error "blockWithQuotes: fold finished without input"

1768

1779

1769 {-# INLINE process #-}

1780 {-# INLINE process #-}

1770 process s a nextState = do

1781 process s a nextState = do

@@ -1778,7 +1789,7 @@ blockQuotedBy bopen bclose quote esc

1778 return

1789 return

1779 $ if a == bopen

1790 $ if a == bopen

1780 then Continue 0 $ BlockUnquoted 1 s

1791 then Continue 0 $ BlockUnquoted 1 s

1781− else Error "blockQuotedBy: missing block start"

1792+ else Error "blockWithQuotes: missing block start"

1782 step (BlockUnquoted level s) a =

1793 step (BlockUnquoted level s) a =

1783 if a == bopen

1794 if a == bopen

1784 then process s a (BlockUnquoted (level + 1))

1795 then process s a (BlockUnquoted (level + 1))

@@ -1804,12 +1815,12 @@ blockQuotedBy bopen bclose quote esc

1804

1815

1805 extract (BlockInit s) = fmap (Done 0) $ fextract s

1816 extract (BlockInit s) = fmap (Done 0) $ fextract s

1806 extract (BlockUnquoted level _) =

1817 extract (BlockUnquoted level _) =

1807− err $ "blockQuotedBy: finished at block nest level " ++ show level

1818+ err $ "blockWithQuotes: finished at block nest level " ++ show level

1808 extract (BlockQuoted level _) =

1819 extract (BlockQuoted level _) =

1809− err $ "blockQuotedBy: finished, inside an unfinished quote, "

1820+ err $ "blockWithQuotes: finished, inside an unfinished quote, "

1810 ++ "at block nest level " ++ show level

1821 ++ "at block nest level " ++ show level

1811 extract (BlockQuotedEsc level _) =

1822 extract (BlockQuotedEsc level _) =

1812− err $ "blockQuotedBy: finished, inside an unfinished quote, "

1823+ err $ "blockWithQuotes: finished, inside an unfinished quote, "

1813 ++ "after an escape char, at block nest level " ++ show level

1824 ++ "after an escape char, at block nest level " ++ show level

1814

1825

1815{-# ANN type GroupByState Fuse #-}

1826{-# ANN type GroupByState Fuse #-}