diff --git a/Text/Regex/Less.hs b/Text/Regex/Less.hs --- a/Text/Regex/Less.hs +++ b/Text/Regex/Less.hs @@ -1,70 +1,46 @@ {- copyright (c) sreservoir. license bsd three-clause. -} -module Text.Regex.Less ((=~),truth,yank,base,subst,backref,unjust) where +module Text.Regex.Less ((=~),(<<),truth,subs,derefs,bref) where -import qualified Data.ByteString.Char8 as B -import qualified Text.Regex.PCRE.Light as R -import Data.List +import qualified Text.Regex.PCRE as R +import qualified Data.Array as A import Text.Regex.Less.Quackers -type Result = (String,Maybe [String]) +type Result = (String,[R.MatchArray]) -- standard usage: -- "" =~ "" - -- "" =~ "" << RECOpts - -- "" =~ "" << REEOpts - -- "" =~ "" << RECOpts << REEOpts + -- "" =~ "" << RECtOpts + -- "" =~ "" << RERtOpts + -- "" =~ "" << RECtOpts << RERtOpts -- value suitable for use below. -infixl 6 =~ +infixl 4 =~ (=~) :: QLR a => String -> a -> Result -a =~ b = - case R.match (compile b) (B.pack a) (runopts b) of - Just c -> (a,Just (map B.unpack c)) - Nothing -> (a,Nothing) - --- success/failure of a regex. -truth :: Result -> Bool -truth (_,a) = - case a of - Just _ -> True - Nothing -> False +a =~ b = a << R.matchAll (compile b) a --- yanks out the matches. -yank :: Result -> [String] -yank (_,a) = unjust a +infixl 5 << +(<<) :: a -> b -> (a,b) +a << b = (a,b) --- returns the original string. -base :: Result -> String -base (a,_) = a +truth :: Result -> Bool +truth (_,a) = not (null a) --- substitutes the matched part. --- does backrefs with derefs. -subst :: Result -> String -> String -subst a b = subst' (base a) b (yank a) - where subst' [] _ _ = [] - subst' _ _ [] = [] - subst' ca@(c:cs) d es@(e:_) - | e `isPrefixOf` ca = derefs es d ++ unjust (stripPrefix e ca) - | otherwise = c : subst' cs d es +subs :: Result -> String -> String +subs a b = s ++ derefs a b ++ f + where o = fst a + (i,c) = head (snd a) A.! 0 + (s,r) = splitAt i o + f = drop c r --- dereferences backrefs with ^ . -derefs :: [String] -> String -> String +derefs :: Result -> String -> String derefs a ('^':'^':bs) = '^' : derefs a bs -derefs a ('^':bs) = +derefs a ('^':bs) = case reads bs of - [(c,d)] -> a !! c ++ derefs a d - _ -> error "there's nothing here." -derefs a (b:bs) = b : derefs a bs -derefs _ [] = [] - --- takes the backrefs of a match. - -- backref 0 is the whole match. -backref :: Result -> Int -> String -backref a b = yank a !! b + [(c,d)] -> bref a c ++ derefs a d + _ -> undefined +derefs a (b:bs) = b : derefs a bs +derefs _ [] = [] --- takes a Maybe(Just)'s value. --- fails with [] on Nothing. -unjust :: Maybe a -> a -unjust (Just a) = a -unjust Nothing = error "not unjustifiable." +bref :: Result -> Int -> String +bref a i = R.extract (head (snd a) A.! i) (fst a) diff --git a/Text/Regex/Less/Quackers.hs b/Text/Regex/Less/Quackers.hs --- a/Text/Regex/Less/Quackers.hs +++ b/Text/Regex/Less/Quackers.hs @@ -5,13 +5,11 @@ module Text.Regex.Less.Quackers (QLR(..)) where -import qualified Text.Regex.PCRE.Light as R -import qualified Data.ByteString.Char8 as B +import qualified Text.Regex.PCRE as R import Text.Regex.Less.REOpts -- the QLR (QuacksLikeRegex) class: -- compile - -- runopts -- instances of QLR: -- String -- (String,[RECtOpts]) @@ -24,39 +22,24 @@ -- QuacksLikeRegex: can =~ . class QLR a where compile :: a -> R.Regex - runopts :: a -> [R.PCREExecOption] --- standard string. instance QLR String where - compile a = R.compile (B.pack a) [] - runopts _ = [] + compile a = R.makeRegexOpts (reCtOpts []) (reRtOpts []) a --- re << compile -instance QLR (String,[RECtOpt]) where - compile (a,b) = R.compile (B.pack a) (reCtOpts b) - runopts (_,_) = [] +instance QLR (String,[RECtOpts]) where + compile (a,b) = R.makeRegexOpts (reCtOpts b) (reRtOpts []) a --- re << runtime -instance QLR (String,[RERtOpt]) where - compile (a,_) = R.compile (B.pack a) [] - runopts (_,b) = reRtOpts b +instance QLR (String,[RERtOpts]) where + compile (a,b) = R.makeRegexOpts (reCtOpts []) (reRtOpts b) a --- re << compile << runtime -instance QLR (String,([RECtOpt],[RERtOpt])) where - compile (a,(b,_)) = R.compile (B.pack a) (reCtOpts b) - runopts (_,(_,c)) = reRtOpts c +instance QLR ((String,[RECtOpts]),[RERtOpts]) where + compile ((a,b),c) = R.makeRegexOpts (reCtOpts b) (reRtOpts c) a --- re << runtime << compile -instance QLR (String,([RERtOpt],[RECtOpt])) where - compile (a,(_,b)) = R.compile (B.pack a) (reCtOpts b) - runopts (_,(c,_)) = reRtOpts c +instance QLR ((String,[RERtOpts]),[RECtOpts]) where + compile ((a,b),c) = R.makeRegexOpts (reCtOpts c) (reRtOpts b) a --- (re << compile) << runtime -instance QLR ((String,[RECtOpt]),[RERtOpt]) where - compile ((a,b),_) = R.compile (B.pack a) (reCtOpts b) - runopts ((_,_),c) = reRtOpts c +instance QLR (String,([RECtOpts],[RERtOpts])) where + compile (a,(b,c)) = R.makeRegexOpts (reCtOpts b) (reRtOpts c) a --- (re << runtime) << compile -instance QLR ((String,[RERtOpt]),[RECtOpt]) where - compile ((a,_),b) = R.compile (B.pack a) (reCtOpts b) - runopts ((_,c),_) = reRtOpts c +instance QLR (String,([RERtOpts],[RECtOpts])) where + compile (a,(b,c)) = R.makeRegexOpts (reCtOpts c) (reRtOpts b) a diff --git a/Text/Regex/Less/RECtOpts.hs b/Text/Regex/Less/RECtOpts.hs --- a/Text/Regex/Less/RECtOpts.hs +++ b/Text/Regex/Less/RECtOpts.hs @@ -1,42 +1,34 @@ {- copyright (c) sreservoir. license bsd three-clause. -} -module Text.Regex.Less.RECtOpts (reCtOpts,RECtOpt(..)) where +module Text.Regex.Less.RECtOpts (reCtOpts,RECtOpts(..)) where -import qualified Text.Regex.PCRE.Light as R +import qualified Text.Regex.PCRE as R -- compile options. -data RECtOpt = CtAnchored | CtAutoCallout | CtCaseless | CtDollarEndOnly - | CtDotAll | CtDupNames | CtExtended | CtFirstLine | CtMultiLine - | CtNoAutoCapture | CtReverseGreedy | CtUtf8 | CtNoUtf8Check - | CtNlAnyCrLf | CtNlAnyUnicode | CtNlCr | CtNlLf | CtNlCrLf - | CtRAnyCrLf | CtRAnyUnicode | CtExtra +data RECtOpts = CtAnchored | CtAutoCallout | CtBlank | CtCaseless + | CtDollarEndOnly | CtDotAll | CtExtended | CtExtra + | CtFirstLine | CtMultiLine | CtNoAutoCapture | CtNoUtf8Check + | CtUtf8 | CtUngreedy deriving (Eq,Show) -- translates lists of compile option constructors. -reCtOpts :: [RECtOpt] -> [R.PCREOption] -reCtOpts = map reCtOpt1 +reCtOpts :: [RECtOpts] -> R.CompOption +reCtOpts = sum . map reCtOpt1 -- translates a compile option constructor. -reCtOpt1 :: RECtOpt -> R.PCREOption -reCtOpt1 CtAnchored = R.anchored -reCtOpt1 CtAutoCallout = R.auto_callout -reCtOpt1 CtCaseless = R.caseless -reCtOpt1 CtDollarEndOnly = R.dollar_endonly -reCtOpt1 CtDotAll = R.dotall -reCtOpt1 CtDupNames = R.dupnames -reCtOpt1 CtExtended = R.extended -reCtOpt1 CtFirstLine = R.firstline -reCtOpt1 CtMultiLine = R.multiline -reCtOpt1 CtNoAutoCapture = R.no_auto_capture -reCtOpt1 CtReverseGreedy = R.ungreedy -reCtOpt1 CtUtf8 = R.utf8 -reCtOpt1 CtNoUtf8Check = R.no_utf8_check -reCtOpt1 CtNlCr = R.newline_cr -reCtOpt1 CtNlLf = R.newline_lf -reCtOpt1 CtNlCrLf = R.newline_crlf ---reCtOpt1 CtNAnyCrLf = R.newline_anycrlf ---reCtOpt1 CtNAnyUnicode = R.newline_any ---reCtOpt1 CtNAnyCrLf = R.bsr_anycrlf ---reCtOpt1 CtNAnyUnicode = R.bsr_unicode -reCtOpt1 _ = undefined +reCtOpt1 :: RECtOpts -> R.CompOption +reCtOpt1 CtAnchored = R.compAnchored +reCtOpt1 CtAutoCallout = R.compAutoCallout +reCtOpt1 CtBlank = R.compBlank +reCtOpt1 CtCaseless = R.compCaseless +reCtOpt1 CtDollarEndOnly = R.compDollarEndOnly +reCtOpt1 CtDotAll = R.compDotAll +reCtOpt1 CtExtended = R.compExtended +reCtOpt1 CtExtra = R.compExtra +reCtOpt1 CtFirstLine = R.compFirstLine +reCtOpt1 CtMultiLine = R.compMultiline +reCtOpt1 CtNoAutoCapture = R.compNoAutoCapture +reCtOpt1 CtNoUtf8Check = R.compNoUTF8Check +reCtOpt1 CtUtf8 = R.compUTF8 +reCtOpt1 CtUngreedy = R.compUngreedy diff --git a/Text/Regex/Less/RERtOpts.hs b/Text/Regex/Less/RERtOpts.hs --- a/Text/Regex/Less/RERtOpts.hs +++ b/Text/Regex/Less/RERtOpts.hs @@ -1,31 +1,25 @@ {- copyright (c) sreservoir. license bsd three-clause. -} -module Text.Regex.Less.RERtOpts (reRtOpts,RERtOpt(..)) where +module Text.Regex.Less.RERtOpts (reRtOpts,RERtOpts(..)) where -import qualified Text.Regex.PCRE.Light as R +import qualified Text.Regex.PCRE as R -- runtime options. -data RERtOpt = RtAnchored | RtNlAny | RtNlAnyCrLf | RtNlCr | RtNlLf - | RtNlCrLf | RtNotBol | RtNotRtOl | RtNotEmpty | RtNoUtf8Check - | RtPartial +data RERtOpts = RtAnchored | RtBlank | RtNoUtf8Check | RtNotBol | RtNotEol + | RtNotEmpty | RtPartial deriving (Eq,Show) -- translates lists of runtime option constructors. -reRtOpts :: [RERtOpt] -> [R.PCREExecOption] -reRtOpts = map reRtOpt1 +reRtOpts :: [RERtOpts] -> R.ExecOption +reRtOpts = sum . map reRtOpt1 -- translates a runtime option constructor. -reRtOpt1 :: RERtOpt -> R.PCREExecOption -reRtOpt1 RtAnchored = R.exec_anchored -reRtOpt1 RtNlCr = R.exec_newline_cr -reRtOpt1 RtNlLf = R.exec_newline_lf -reRtOpt1 RtNlCrLf = R.exec_newline_crlf -reRtOpt1 RtNotBol = R.exec_notbol -reRtOpt1 RtNotRtOl = R.exec_noteol -reRtOpt1 RtNotEmpty = R.exec_notempty -reRtOpt1 RtNoUtf8Check = R.exec_no_utf8_check -reRtOpt1 RtPartial = R.exec_partial ; ---reRtOpt1 RtNlAny = R.exec_newline_any ---reRtOpt1 RtNlAnyCrLf = R.exec_newline_anycrlf -reRtOpt1 _ = undefined +reRtOpt1 :: RERtOpts -> R.ExecOption +reRtOpt1 RtAnchored = R.execAnchored +reRtOpt1 RtBlank = R.execBlank +reRtOpt1 RtNoUtf8Check = R.execNoUTF8Check +reRtOpt1 RtNotBol = R.execNotBOL +reRtOpt1 RtNotEol = R.execNotEOL +reRtOpt1 RtNotEmpty = R.execNotEmpty +reRtOpt1 RtPartial = R.execPartial diff --git a/pcre-less.cabal b/pcre-less.cabal --- a/pcre-less.cabal +++ b/pcre-less.cabal @@ -1,8 +1,8 @@ name: pcre-less -version: 0.0.0 -synopsis: Nicer interface to pcre-light. +version: 0.2.0 +synopsis: Nicer interface to regex-pcre description: - You'll have to read the comments until I get around to writing docs. + Unfortunately, docs don't exist yet. homepage: ~ license: BSD3 -- The file containing the license text. @@ -21,7 +21,7 @@ Text.Regex.Less.RECtOpts Text.Regex.Less.RERtOpts Text.Regex.Less.Quackers - build-depends: base >= 4 && < 5,pcre-light >= 0.4,bytestring >= 0.9 + build-depends: base >= 4 && < 5,regex-pcre >= 0.94,array >= 0.3 extensions: FlexibleInstances,TypeSynonymInstances - ghc-options: -O2 + ghc-options: -Wall ghc-prof-options: -auto-all