master..f283d41 · typeable/hexml

.ghci+1−1

since splitsnapshots

unifiedsplit

@@ -1,4 +1,4 @@

1−:set -Wunused-binds -Wunused-imports -Worphans

1+:set -fwarn-unused-binds -fwarn-unused-imports -fwarn-orphans

2:set -isrc

3:load Main

4

.github/workflows/ci.ymldeleted−33

@@ -1,33 +0,0 @@

1−name: ci

2−on:

3− push:

4− pull_request:

5− schedule:

6− - cron: '0 3 * * 6' # 3am Saturday

7−jobs:

8− test:

9− runs-on: ${{ matrix.os }}

10−

11− strategy:

12− fail-fast: false

13− matrix:

14− os: [ubuntu-latest]

15− ghc: ['9.6', '9.4', '9.2', '9.0', '8.10', '8.8']

16− include:

17− - os: windows-latest

18− - os: macOS-latest

19−

20− steps:

21− - run: git config --global core.autocrlf false

22− - uses: actions/checkout@v2

23− - uses: haskell/actions/setup@v2

24− id: setup-haskell

25− with:

26− ghc-version: ${{ matrix.ghc }}

27− - run: cabal v2-freeze --enable-tests

28− - uses: actions/cache@v2

29− with:

30− path: ${{ steps.setup-haskell.outputs.cabal-store }}

31− key: ${{ runner.os }}-${{ matrix.ghc }}-${{ hashFiles('cabal.project.freeze') }}

32− restore-keys: ${{ runner.os }}-${{ matrix.ghc }}-

33− - uses: ndmitchell/neil@master

.travis.ymlnew file+14

@@ -0,0 +1,14 @@

1+sudo: required

2+env:

3+- GHCVER=7.6.3

4+- GHCVER=7.8.4

5+- GHCVER=7.10.3

6+- GHCVER=8.0.2

7+- GHCVER=8.2.2

8+- GHCVER=8.4.3

9+- GHCVER=8.6.1

10+- GHCVER=head

11+

12+script:

13+- curl -sSL https://raw.github.com/ndmitchell/neil/master/travis.sh | sh

14+- sh afl.sh

LICENSE+1−1

@@ -1,4 +1,4 @@

1−Copyright Neil Mitchell 2016-2023.

1+Copyright Neil Mitchell 2016-2018.

2

3Licensed under either of:

4

README.md+2−4

@@ -1,4 +1,4 @@

1−# Hexml [![Hackage version](https://img.shields.io/hackage/v/hexml.svg?label=Hackage)](https://hackage.haskell.org/package/hexml) [![Stackage version](https://www.stackage.org/package/hexml/badge/nightly?label=Stackage)](https://www.stackage.org/package/hexml) [![Build status](https://img.shields.io/github/actions/workflow/status/ndmitchell/hexml/ci.yml?branch=master)](https://github.com/ndmitchell/hexml/actions)

1+# Hexml [![Hackage version](https://img.shields.io/hackage/v/hexml.svg?label=Hackage)](https://hackage.haskell.org/package/hexml) [![Stackage version](https://www.stackage.org/package/hexml/badge/nightly?label=Stackage)](https://www.stackage.org/package/hexml) [![Linux Build Status](https://img.shields.io/travis/ndmitchell/hexml/master.svg?label=Linux%20build)](https://travis-ci.org/ndmitchell/hexml) [![Windows Build Status](https://img.shields.io/appveyor/ci/ndmitchell/hexml/master.svg?label=Windows%20build)](https://ci.appveyor.com/project/ndmitchell/hexml)

2

3An XML DOM-style parser, that only parses a subset of XML, but is designed to be fast. In particular:

4

@@ -8,10 +8,8 @@ An XML DOM-style parser, that only parses a subset of XML, but is designed to be

8

9The name "hexml" is a combination of "Hex" (a curse) and "XML". The "X" should not be capitalised because the parser is more curse and less XML.

10

11−Hexml may be suitable if you want to quickly parse XML, from known sources, and a full XML parser has been shown to be a bottleneck. As an alternative to hexml, which supports things like entities but is still pretty fast, see [Pugixml](http://pugixml.org/) (with a [Haskell binding](https://hackage.haskell.org/package/pugixml) - but be aware the Haskell binding [can segfault](https://github.com/philopon/pugixml-hs/issues/5)).

11+Hexml may be suitable if you want to quickly parse XML, from known sources, and a full XML parser has been shown to be a bottleneck. As an alternative to hexml, which supports things like entities but is still pretty fast, see [Pugixml](http://pugixml.org/) (with a [Haskell binding](https://hackage.haskell.org/package/pugixml)).

12

13Hexml is tested with [AFL](http://lcamtuf.coredump.cx/afl/).

14

15If you want lenses for Hexml, see [hexml-lens](http://hackage.haskell.org/package/hexml-lens).

16−

17−The optimisation work around Hexml spawned [Xeno](http://hackage.haskell.org/package/xeno), a Haskell-only alternative to Hexml. There is a [talk](https://ndmitchell.com/#hexml_12_oct_2017) covering the performance tricks of Hexml and Xeno.

appveyor.ymlnew file+5

@@ -0,0 +1,5 @@

1+build: off

2+cache: "c:\\sr -> appveyor.yml"

3+

4+test_script:

5+- ps: Invoke-Expression (Invoke-WebRequest 'https://raw.githubusercontent.com/ndmitchell/neil/master/appveyor.ps1')

cbits/hexml.c+2−2

@@ -40,7 +40,7 @@ typedef struct

40 // have a cursor at the end, which is stack scoped, so children write out, then I do

41 // when you commit, you copy over from end to front

42

43− // nodes

43+ // nodes

44 int size;

45 int used_front; // front entries, stored for good

46 int used_back; // back entries, stack based, copied into front

@@ -539,7 +539,7 @@ document* hexml_document_parse(const char* s, int slen)

539 d->nodes.nodes[0].outer = start_length(0, slen);

540 d->nodes.nodes[0].inner = start_length(0, slen);

541 d->nodes.nodes[0].attrs = start_length(0, 0);

542−

542+

543 // Introduce an intermediate result, otherwise behaviour is undefined

544 // because there is no guaranteed ordering between LHS and RHS evaluation

545 str content = parse_content(d);

gen/Lib.hs+3−3

@@ -10,7 +10,7 @@ module Lib(

10import Control.Monad.Extra

11import Control.Monad.Trans.Writer

12import Data.Maybe

13−import Data.List.Extra

13+import Data.List

14import Data.Bits

15import Data.Monoid

16import Numeric

@@ -67,7 +67,7 @@ choices :: [Parser o ()] -> Parser o ()

67choices = foldr1 choice

68

69optional :: Parser o () -> Parser o ()

70−optional x = choice x $ pure ()

70+optional x = choice x $ return ()

71

72

73---------------------------------------------------------------------

@@ -106,7 +106,7 @@ charSetC sets =

106 ,\x -> fromMaybe ("table[(int) *p] & " ++ showBit (bit $ fromJust $ elemIndex x hard)) $ simple x)

107 where

108 showBit x = "0x" ++ showHex (x :: Int) ""

109− disp c = showBit $ foldr (.|.) 0 [bit i | (i, h) <- zipFrom 0 hard, fromCharSet h c]

109+ disp c = showBit $ foldr (.|.) 0 [bit i | (i, h) <- zip [0..] hard, fromCharSet h c]

110

111 hard = nub $ filter (isNothing . simple) sets

112

gen/Main.hs+2−2

@@ -28,13 +28,13 @@ main = do

28

29writeFileChanged :: FilePath -> String -> IO ()

30writeFileChanged file new = do

31− old <- ifM (doesFileExist file) (readFile' file) (pure "")

31+ old <- ifM (doesFileExist file) (readFile' file) (return "")

32 when (old /= new) $

33 writeFile file new

34

35buildRule :: [FilePath] -> [FilePath] -> IO () -> IO ()

36buildRule from to act = do

37− let modTime x = ifM (doesFileExist x) (Just <$> getModificationTime x) (pure Nothing)

37+ let modTime x = ifM (doesFileExist x) (Just <$> getModificationTime x) (return Nothing)

38 from <- mapM modTime from

39 to <- mapM modTime to

40 when (any isNothing (from ++ to) || maximum from >= minimum to)

hexml.cabal+2−2

@@ -8,13 +8,13 @@ license-file: LICENSE

8category: XML

9author: Neil Mitchell <ndmitchell@gmail.com>

10maintainer: Neil Mitchell <ndmitchell@gmail.com>

11−copyright: Neil Mitchell 2016-2023

11+copyright: Neil Mitchell 2016-2018

12synopsis: XML subset DOM parser

13description:

14 An XML DOM-style parser, that only parses a subset of XML, but is designed to be fast.

15homepage: https://github.com/ndmitchell/hexml#readme

16bug-reports: https://github.com/ndmitchell/hexml/issues

17−tested-with: GHC==9.6, GHC==9.4, GHC==9.2, GHC==9.0, GHC==8.10, GHC==8.8

17+tested-with: GHC==8.6.1, GHC==8.4.4, GHC==8.2.2, GHC==8.0.2, GHC==7.10.3, GHC==7.8.4, GHC==7.6.3

18extra-doc-files:

19 README.md

20 CHANGES.txt

src/Main.hs+4−4

@@ -44,12 +44,12 @@ main = do

44 attributes (head $ children doc) === [Attribute "id" "1", Attribute "extra" "2"]

45 map (`attributeBy` "id") (childrenBy doc "test") === map (fmap (Attribute "id")) [Just "1", Just "2", Just "4", Nothing]

46

47− Right _ <- pure $ parse $ "<test " <> BS.unwords [BS.pack $ "x" ++ show i ++ "='value'" | i <- [1..10000]] <> " />"

47+ Right _ <- return $ parse $ "<test " <> BS.unwords [BS.pack $ "x" ++ show i ++ "='value'" | i <- [1..10000]] <> " />"

48− Right _ <- pure $ parse $ BS.unlines $ replicate 10000 "<test x='value' />"

48+ Right _ <- return $ parse $ BS.unlines $ replicate 10000 "<test x='value' />"

49

50 let attrs = ["usd:jpy","test","extra","more","stuff","jpy:usd","xxx","xxxx"]

51− Right doc <- pure $ parse $ "<test " <> BS.unwords [x <> "='" <> x <> "'" | x <- attrs] <> ">middle</test>"

51+ Right doc <- return $ parse $ "<test " <> BS.unwords [x <> "='" <> x <> "'" | x <- attrs] <> ">middle</test>"

52− [c] <- pure $ childrenBy doc "test"

52+ [c] <- return $ childrenBy doc "test"

53 forM_ attrs $ \a -> attributeBy c a === Just (Attribute a a)

54 forM_ ["missing","gone","nothing"] $ \a -> attributeBy c a === Nothing

55 putStrLn "\nSuccess"

src/Text/XML/Hexml.hs+9−9

@@ -90,11 +90,11 @@ parse src = do

90 if err /= nullPtr then do

91 bs <- BS.packCString =<< hexml_document_error doc

92 hexml_document_free doc

93− pure $ Left bs

93+ return $ Left bs

94 else do

95 node <- hexml_document_node doc

96 doc <- newForeignPtr hexml_document_free_funptr doc

97− pure $ Right $ Node src0 doc node

97+ return $ Right $ Node src0 doc node

98

99-- | Given a node, rerender it to something with an equivalent parse tree.

100-- Mostly useful for debugging - if you want the real source document use 'outer' instead.

@@ -103,7 +103,7 @@ render (Node src doc n) = unsafePerformIO $ withForeignPtr doc $ \d -> do

103 i <- hexml_node_render d n nullPtr 0

104 res <- BS.create (fromIntegral i) $ \ptr -> void $ hexml_node_render d n (castPtr ptr) i

105 touchBS src

106− pure res

106+ return res

107

108applyStr :: BS.ByteString -> Str -> BS.ByteString

109applyStr bs Str{..} = BS.take (fromIntegral strLength) $ BS.drop (fromIntegral strStart) bs

@@ -118,7 +118,7 @@ attrPeek :: BS.ByteString -> ForeignPtr CDocument -> Ptr CAttr -> Attribute

118attrPeek src doc a = unsafePerformIO $ withForeignPtr doc $ \_ -> do

119 name <- applyStr src <$> peekElemOff (castPtr a) 0

120 val <- applyStr src <$> peekElemOff (castPtr a) 1

121− pure $ Attribute name val

121+ return $ Attribute name val

122

123-- | Get the name of a node, e.g. @\<test /\>@ produces @\"test\"@.

124name :: Node -> BS.ByteString

@@ -142,7 +142,7 @@ outer = nodeBS 2

142contents :: Node -> [Either BS.ByteString Node]

143contents n@(Node src _ _) = f (strStart inner) outers

144 where

145− f i [] = string i (strEnd inner)

145+ f i [] = string i (strEnd inner) ++ []

146 f i ((x, n):xs) = string i (strStart x) ++ Right n : f (strEnd x) xs

147

148 string start end | start == end = []

@@ -156,7 +156,7 @@ children (Node src doc n) = unsafePerformIO $ withForeignPtr doc $ \d ->

156 alloca $ \count -> do

157 res <- hexml_node_children d n count

158 count <- fromIntegral <$> peek count

159− pure [Node src doc $ plusPtr res $ i*szNode | i <- [0..count-1]]

159+ return [Node src doc $ plusPtr res $ i*szNode | i <- [0..count-1]]

160

161-- | Get the attributes of this node.

162attributes :: Node -> [Attribute]

@@ -164,7 +164,7 @@ attributes (Node src doc n) = unsafePerformIO $ withForeignPtr doc $ \d ->

164 alloca $ \count -> do

165 res <- hexml_node_attributes d n count

166 count <- fromIntegral <$> peek count

167− pure [attrPeek src doc $ plusPtr res $ i*szAttr | i <- [0..count-1]]

167+ return [attrPeek src doc $ plusPtr res $ i*szAttr | i <- [0..count-1]]

168

169-- | Get the direct children of this node which have a specific name.

170-- A more efficient version of:

@@ -177,7 +177,7 @@ childrenBy (Node src doc n) str = go nullPtr

177 BS.unsafeUseAsCStringLen str $ \(bs, len) -> do

178 r <- hexml_node_child d n old bs $ fromIntegral len

179 touchBS src

180− pure $ if r == nullPtr then [] else Node src doc r : go r

180+ return $ if r == nullPtr then [] else Node src doc r : go r

181

182-- | Get the first attribute of this node which has a specific name, if there is one.

183-- A more efficient version of:

@@ -188,7 +188,7 @@ attributeBy (Node src doc n) str = unsafePerformIO $ withForeignPtr doc $ \d ->

188 BS.unsafeUseAsCStringLen str $ \(bs, len) -> do

189 r <- hexml_node_attribute d n bs $ fromIntegral len

190 touchBS src

191− pure $ if r == nullPtr then Nothing else Just $ attrPeek src doc r

191+ return $ if r == nullPtr then Nothing else Just $ attrPeek src doc r

192

193-- | Find the starting location of a node, the @<@ character.

194-- The first character will be reported as @(line 1,column 1)@, because thats