diff --git a/regex-do.cabal b/regex-do.cabal --- a/regex-do.cabal +++ b/regex-do.cabal @@ -1,7 +1,7 @@ name: regex-do -version: 1.1 -synopsis: PCRE regex funs -description: Convenience functions to search, replace, format String | ByteString with PCRE regex +version: 1.3 +synopsis: PCRE wrapper +description: search, replace, format String | ByteString with PCRE regex Utf8-safe author: Imants Cekusins maintainer: Imants Cekusins category: Regex, Search, String diff --git a/src/Regexdo/Pcre/Match.hs b/src/Regexdo/Pcre/Match.hs --- a/src/Regexdo/Pcre/Match.hs +++ b/src/Regexdo/Pcre/Match.hs @@ -49,15 +49,15 @@ -} instance Match_cl String String --- | this instance accepts regex 'String' +-- | accepts regex 'String' instance Match_cl String ByteString --- | this instance accepts regex 'ByteString' +-- | accepts regex 'ByteString' instance Match_cl ByteString ByteString --- | this instance accepts regex 'ByteString' +-- | accepts regex 'ByteString' instance Match_cl ByteString String --- | this instance accepts 'Regex' made with 'makeRegexOpts' +-- | accepts 'Regex' made with 'makeRegexOpts' instance Match_cl Regex String --- | this instance accepts 'Regex' made with 'makeRegexOpts' +-- | accepts 'Regex' made with 'makeRegexOpts' instance Match_cl Regex ByteString diff --git a/src/Regexdo/Pcre/Replace.hs b/src/Regexdo/Pcre/Replace.hs --- a/src/Regexdo/Pcre/Replace.hs +++ b/src/Regexdo/Pcre/Replace.hs @@ -3,10 +3,11 @@ Replace_cl(..), replaceMatch, defaultReplacer, + getGroup, Mr ) where -import Data.Array +import Data.Array as A import Data.ByteString import Prelude as P import qualified Data.ByteString as B @@ -26,6 +27,7 @@ prefix::PosLen -> a -> a suffix::PosLen -> a -> a concat'::[a] -> a + len'::a -> Int toByteString'::a -> ByteString toA::ByteString -> a @@ -58,33 +60,34 @@ custom replacer fn returns replacement value >>> replacer::GroupReplacer String - replacer = defaultReplacer 3 tweak1 + replacer = defaultReplacer 1 tweak1 where tweak1 str1 = case str1 of - "101" -> "[A]" - "3" -> "[Be]" - + "101" -> "[сто один]" + "3" -> "[three]" + otherwise -> trace str1 "?" 'Once' vs 'All' options - >>> replaceGroup [Once] (Needle "(\\w)(=)(\\d{1,3})", replacer) $ Haystack "a=101 b=3 12" + >>> replaceGroup [Once,Utf8] (Needle "\\w=(\\d{1,3})", replacer) $ Haystack "a=101 b=3 12" - "a=[A] b=3 12" + "a=[сто один] b=3 12" - >>> replaceGroup [All] (Needle "(\\w)(=)(\\d{1,3})", replacer) $ Haystack "a=101 b=3 12" + >>> replaceGroup [All,Utf8] (Needle "\\w=(\\d{1,3})", replacer) $ Haystack "a=101 b=3 12" - "a=[A] b=[Be] 12" + "a=[сто один] b=[three] 12" == static replace for simple (no group) needle - >>> replace [Once,Utf8] (Needle "менее", Replacement "более") (Haystack "менее менее") + >>> replace [Once,Utf8] (Needle "менее", Replacement "более") $ Haystack "менее менее" "более менее" - >>> replace [Once,Utf8] (Needle "^a\\s", Replacement "A") (Haystack "a bc хол.гор.") + >>> replace [Once,Utf8] (Needle "^a\\s", Replacement "A") $ Haystack "a bc хол.гор." "Abc хол.гор." -} + instance Replace_cl String instance Replace_cl' String where @@ -95,6 +98,7 @@ concat' = P.concat toByteString' = toByteString toA = toString + len' = P.length instance Replace_cl B.ByteString @@ -107,81 +111,109 @@ concat' = B.concat toByteString' = id toA = id + len' = B.length --- | use in your custom 'GroupReplacer' passed to 'replaceGroup' --- --- see example replacer above or use 'defaultReplacer' --- -replaceMatch::Replace_cl' a => - (R.MatchOffset, R.MatchLength) -> - (a, a) -- ^ (new val, acc passed to 'GroupReplacer') - -> a -- ^ new acc -replaceMatch pl1 (repl1, bs_str1) = - concat' [prefix1,repl1,suffix1] - where prefix1 = prefix pl1 bs_str1 - suffix1 = suffix pl1 bs_str1 - -- static ronce::Mr a => (Needle Regex, Replacement a) -> Haystack a -> a -ronce (pat1, Replacement repl1) body1@(Haystack bs_str1) = +ronce (pat1, Replacement repl1) h1@(Haystack h0) = let pl2 = do - let m1 = match pat1 body1 + let m1 = match pat1 h1 poslen m1 in case pl2 of - Nothing -> bs_str1 - Just pl_arr -> firstGroup pl_arr (repl1,bs_str1) + Nothing -> h0 + Just lpl1 -> firstGroup lpl1 (repl1, h0) rall::Mr a => (Needle Regex, Replacement a) -> Haystack a -> a -rall (pat1, Replacement repl1) body1@(Haystack bs_str1) = - let pl_arr_arr1 = do - let m1 = matchAll pat1 body1 - poslen m1 - folderFn pl_arr acc1 = firstGroup pl_arr (repl1,acc1) - in P.foldr folderFn bs_str1 pl_arr_arr1 +rall (pat1, Replacement repl1) h1@(Haystack h0) = + let lpl1 = do + let m1 = matchAll pat1 h1 + poslen m1::[[PosLen]] + foldFn1 lpl1 acc1 = firstGroup lpl1 (repl1,acc1) + in P.foldr foldFn1 h0 lpl1 -firstGroup::Replace_cl' a => +firstGroup::(Replace_cl' a) => [PosLen] -> (a,a) -> a -firstGroup (pl1:_) r1@(repl1,bs_str1) = replaceMatch pl1 r1 +firstGroup (pl0:_) r1@(new0,a0) = acc_haystack $ replaceMatch pl0 (new0, acc1) + where acc1 = ReplaceAcc { + acc_haystack = a0, + position_adj = 0 + } -- dynamic -{- | replace with a tweak to specified (by idx) group match +{- | you can write a custom replacer. This is only one common use case. - see 'defaultReplacer' source for hints: how to write custom replacer - -} + Replaces specified (by idx) group match with tweaked value. -} defaultReplacer::(Replace_cl' a, R.Extract a) => - Int -- ^ idx of match within a group + Int -- ^ group idx -> (a -> a) -- ^ (group match -> replacement) tweak -> GroupReplacer a -defaultReplacer idx0 tweak0 (ma0::MatchArray) acc0 = - if idx0 >= P.length ma0 then acc0 -- safety catch - else fn1 val1 - where poslen1 = ma0 ! idx0 :: (R.MatchOffset, R.MatchLength) - val1 = extract poslen1 acc0 - fn1 str1 = replaceMatch poslen1 (str2,acc0) +defaultReplacer idx0 tweak0 (ma0::MatchArray) acc0 = maybe acc0 fn1 mval1 + where pl1 = ma0 A.! idx0 :: (R.MatchOffset, R.MatchLength) + mval1 = getGroup acc0 ma0 idx0 + fn1 str1 = replaceMatch pl1 (str2, acc0) where str2 = tweak0 str1 +{- | get group content safely + + call from your custom 'GroupReplacer' passed to 'replaceGroup' + -} +getGroup::R.Extract a => + ReplaceAcc a -> MatchArray -> Int -> Maybe a +getGroup acc0 ma0 idx0 = if idx0 >= P.length ma0 then Nothing -- safety catch + else Just val1 + where pl1 = ma0 A.! idx0 :: (R.MatchOffset, R.MatchLength) + pl2 = adjustPoslen pl1 acc0 + val1 = extract pl2 $ acc_haystack acc0 + + +adjustPoslen::PosLen -> ReplaceAcc a -> PosLen +adjustPoslen (p0,l0) acc0 = (p0 + position_adj acc0, l0) + + ronceGroup::Match_cl Regex a => (Needle Regex, GroupReplacer a) -> Haystack a -> a -ronceGroup (pat1, repl1) body1@(Haystack bs_str1) = - let m1 = match pat1 body1 +ronceGroup (pat1, repl1) h1@(Haystack h0) = + let m1 = match pat1 h1::Maybe MatchArray in case m1 of - Nothing -> bs_str1 - Just marr1 -> repl1 marr1 bs_str1 + Nothing -> h0 + Just ma1 -> let a1 = ReplaceAcc { + acc_haystack = h0, + position_adj = 0 + } + in acc_haystack $ repl1 ma1 a1 rallGroup::Match_cl Regex a => (Needle Regex, GroupReplacer a) -> Haystack a -> a -rallGroup (pat1, repl1) body1@(Haystack bs_str1) = - let marrList1 = matchAll pat1 body1 - in P.foldr repl1 bs_str1 marrList1 +rallGroup (pat1, repl1) h1@(Haystack h0) = + let ma1 = matchAll pat1 h1::[MatchArray] + acc1 = ReplaceAcc { acc_haystack = h0, position_adj = 0 } + in acc_haystack $ P.foldl (flip repl1) acc1 ma1 + +{- | call from your custom 'GroupReplacer' passed to 'replaceGroup' + + see example replacer above -} +replaceMatch::Replace_cl' a => + PosLen -- ^ replaced + -> (a, ReplaceAcc a) -- ^ (new val, acc passed to 'GroupReplacer') + -> ReplaceAcc a -- ^ new acc +replaceMatch pl0@(_,l0) (new0, acc0) = ReplaceAcc { + acc_haystack = acc1, + position_adj = position_adj acc0 + l1 - l0 + } + where pl1 = adjustPoslen pl0 acc0 + prefix1 = prefix pl1 $ acc_haystack acc0 + suffix1 = suffix pl1 $ acc_haystack acc0 + acc1 = concat' [prefix1, new0, suffix1] + l1 = len' new0 addOpt::Match_opt a => diff --git a/src/Regexdo/Search.hs b/src/Regexdo/Search.hs --- a/src/Regexdo/Search.hs +++ b/src/Regexdo/Search.hs @@ -1,8 +1,8 @@ -{- | from stringsearch package +{- | wraps functions from stringsearch package this module uses newtypes for args plus function names are tweaked - regex is treated as ordinary String. Do not use regex. + regex is treated as ordinary String -} module Regexdo.Search (break, diff --git a/src/Regexdo/TypeDo.hs b/src/Regexdo/TypeDo.hs --- a/src/Regexdo/TypeDo.hs +++ b/src/Regexdo/TypeDo.hs @@ -3,7 +3,13 @@ import Text.Regex.Base.RegexLike -- pcre -type GroupReplacer acc = (MatchArray -> acc -> acc) -- MatchArray -> acc -> acc +type GroupReplacer a = (MatchArray -> ReplaceAcc a -> ReplaceAcc a) -- MatchArray -> acc -> acc + + +data ReplaceAcc a = ReplaceAcc { + acc_haystack::a, -- ^ wip Haystack + position_adj::Int + } -- stringsearch, Pcre.Replace diff --git a/test/TestRegex/Replace.hs b/test/TestRegex/Replace.hs --- a/test/TestRegex/Replace.hs +++ b/test/TestRegex/Replace.hs @@ -7,6 +7,7 @@ import Regexdo.Pcre.Replace import Regexdo.Convert import Data.ByteString as B +import Debug.Trace main::IO() @@ -21,11 +22,11 @@ doc = hspec $ do describe "Pcre.Replace doc" $ do it "replaceGroup 1" $ do - replaceGroup [Once] (Needle "(\\w)(=)(\\d{1,3})", replacer) (Haystack "a=101 b=3 12") - `shouldBe` "a=[A] b=3 12" + replaceGroup [Once,Utf8] (Needle "\\w=(\\d{1,3})", replacer) (Haystack "a=101 b=3 12") + `shouldBe` "a=[сто один] b=3 12" it "replaceGroup 2" $ do - replaceGroup [All] (Needle "(\\w)(=)(\\d{1,3})", replacer) (Haystack "a=101 b=3 12") - `shouldBe` "a=[A] b=[Be] 12" + replaceGroup [All,Utf8] (Needle "\\w=(\\d{1,3})", replacer) (Haystack "a=101 b=3 12") + `shouldBe` "a=[сто один] b=[three] 12" it "replace 3" $ do replace [Once,Utf8] (Needle "менее", Replacement "более") (Haystack "менее менее") `shouldBe` "более менее" @@ -70,17 +71,18 @@ groupReplace = hspec $ do describe "Pcre.Replace group" $ do it "Once" $ do - runFn1 [Once] `shouldBe` "a=[A] b=3 12" + runFn1 [Once,Utf8] `shouldBe` "a=[сто один] b=3 12" it "All" $ do - runFn1 [All] `shouldBe` "a=[A] b=[Be] 12" + runFn1 [All,Utf8] `shouldBe` "a=[сто один] b=[three] 12" where runFn1 opts1 = - let rx1 = Needle "(\\w)(=)(\\d{1,3})" + let rx1 = Needle "\\w=(\\d{1,3})" body1 = Haystack "a=101 b=3 12" in replaceGroup opts1 (rx1,replacer) body1 replacer::GroupReplacer String -replacer = defaultReplacer 3 tweak1 +replacer = defaultReplacer 1 tweak1 where tweak1 str1 = case str1 of - "101" -> "[A]" - "3" -> "[Be]" + "101" -> "[сто один]" + "3" -> "[three]" + otherwise -> trace str1 "?" diff --git a/test/TestRegex/TestTrim.hs b/test/TestRegex/TestTrim.hs --- a/test/TestRegex/TestTrim.hs +++ b/test/TestRegex/TestTrim.hs @@ -1,3 +1,4 @@ +{-# LANGUAGE OverloadedStrings #-} module TestRegex.TestTrim where import Test.Hspec @@ -11,8 +12,7 @@ main = hspec $ do describe "Habase.Bin.Bistro" $ do it "trim" $ trace (show trimmed1) trimmed1 `shouldBe` (toByteString trimmed) - it "trim 2" $ - trim sTrim `shouldBe` "aiy \n pdsodfg987" - where sTrim = " aiy \n pdsodfg987 " + it "trim 2" $ trim sTrim `shouldBe` "aiy \n pdsodfg987" + where sTrim = " \t aiy \n pdsodfg987 \t\t "::String trimmed1 = trim $ toByteString sTrim trimmed = (T.unpack . T.strip . T.pack) sTrim