diff --git a/bin/main.hs b/bin/main.hs --- a/bin/main.hs +++ b/bin/main.hs @@ -114,6 +114,7 @@ "tango" -> return tango "haddock" -> return haddock "monochrome" -> return monochrome + "breeze-dark" -> return breezeDark _ -> err $ "Unknown style: " ++ s styleOf (_ : xs) = styleOf xs diff --git a/changelog.md b/changelog.md --- a/changelog.md +++ b/changelog.md @@ -8,3 +8,9 @@ * Fixed bug in LaTeX renderer (backslash in wrong position). +## 0.1.1 -- 2017-01-19 + +* Added breezeDark style, from breeze-dark.theme. +* Fixed performance bug in regex application (#1). This gives a + significant speedup, especially for inputs with long lines. + diff --git a/skylighting.cabal b/skylighting.cabal --- a/skylighting.cabal +++ b/skylighting.cabal @@ -1,5 +1,5 @@ name: skylighting -version: 0.1.0.1 +version: 0.1.1 synopsis: syntax highlighting library description: Skylighting is a syntax highlighting library with support for over one hundred languages. It derives diff --git a/src/Skylighting/Regex.hs b/src/Skylighting/Regex.hs --- a/src/Skylighting/Regex.hs +++ b/src/Skylighting/Regex.hs @@ -50,8 +50,7 @@ compileRegex caseSensitive regexpStr = let opts = compAnchored + compUTF8 + if caseSensitive then 0 else compCaseless - in case unsafePerformIO $ compile opts (execNotEmpty) - (BS.cons '.' regexpStr) of + in case unsafePerformIO $ compile opts (execNotEmpty) regexpStr of Left (off,msg) -> E.throw $ RegexException $ "Error compiling regex /" ++ toString regexpStr ++ "/ at offset " ++ show off ++ "\n" ++ msg diff --git a/src/Skylighting/Styles.hs b/src/Skylighting/Styles.hs --- a/src/Skylighting/Styles.hs +++ b/src/Skylighting/Styles.hs @@ -1,7 +1,8 @@ module Skylighting.Styles ( parseTheme - , pygments , kate + , breezeDark + , pygments , espresso , tango , haddock @@ -19,46 +20,6 @@ color :: Int -> Maybe Color color = toColor --- | Style based on pygments's default colors. -pygments :: Style -pygments = Style{ - backgroundColor = Nothing - , defaultColor = Nothing - , lineNumberColor = color 0xaaaaaa - , lineNumberBackgroundColor = Nothing - , tokenStyles = - [ (KeywordTok, defStyle{ tokenColor = color 0x007020, tokenBold = True }) - , (DataTypeTok, defStyle{ tokenColor = color 0x902000 }) - , (DecValTok, defStyle{ tokenColor = color 0x40a070 }) - , (BaseNTok, defStyle{ tokenColor = color 0x40a070 }) - , (FloatTok, defStyle{ tokenColor = color 0x40a070 }) - , (CharTok, defStyle{ tokenColor = color 0x4070a0 }) - , (StringTok, defStyle{ tokenColor = color 0x4070a0 }) - , (CommentTok, defStyle{ tokenColor = color 0x60a0b0, tokenItalic = True }) - , (OtherTok, defStyle{ tokenColor = color 0x007020 }) - , (AlertTok, defStyle{ tokenColor = color 0xff0000, tokenBold = True }) - , (FunctionTok, defStyle{ tokenColor = color 0x06287e }) - , (ErrorTok, defStyle{ tokenColor = color 0xff0000, tokenBold = True }) - , (WarningTok, defStyle{ tokenColor = color 0x60a0b0, tokenItalic = True, tokenBold = True }) - , (ConstantTok, defStyle{ tokenColor = color 0x880000 }) - , (SpecialCharTok, defStyle{ tokenColor = color 0x4070a0 }) - , (VerbatimStringTok, defStyle{ tokenColor = color 0x4070a0 }) - , (SpecialStringTok, defStyle{ tokenColor = color 0xBB6688 }) - , (ImportTok, defStyle) - , (VariableTok, defStyle{ tokenColor = color 0x19177C }) - , (ControlFlowTok, defStyle{ tokenColor = color 0x007020, tokenBold = True }) - , (OperatorTok, defStyle{ tokenColor = color 0x666666 }) - , (BuiltInTok, defStyle) - , (ExtensionTok, defStyle) - , (PreprocessorTok, defStyle{ tokenColor = color 0xBC7A00 }) - , (AttributeTok, defStyle{ tokenColor = color 0x7D9029 }) - , (DocumentationTok, defStyle{ tokenColor = color 0xBA2121, tokenItalic = True }) - , (AnnotationTok, defStyle{ tokenColor = color 0x60a0b0, tokenItalic = True, tokenBold = True }) - , (CommentVarTok, defStyle{ tokenColor = color 0x60a0b0, tokenItalic = True, tokenBold = True }) - , (InformationTok, defStyle{ tokenColor = color 0x60a0b0, tokenItalic = True, tokenBold = True }) - ] - } - -- | Style based on kate's default colors. kate :: Style kate = Style{ @@ -100,6 +61,89 @@ , lineNumberColor = Just (RGB 160 160 160) , lineNumberBackgroundColor = Just (RGB 255 255 255) } + +-- | Style from the breeze-dark KDE syntax highlighting theme. +breezeDark :: Style +breezeDark = Style + { tokenStyles = + [ ( KeywordTok, defStyle { tokenColor = Just (RGB 207 207 194) }) + , ( DataTypeTok, defStyle { tokenColor = Just (RGB 41 128 185) }) + , ( DecValTok, defStyle { tokenColor = Just (RGB 246 116 0) }) + , ( BaseNTok, defStyle { tokenColor = Just (RGB 246 116 0) }) + , ( FloatTok, defStyle { tokenColor = Just (RGB 246 116 0) }) + , ( ConstantTok, defStyle { tokenColor = Just (RGB 39 174 174) }) + , ( CharTok, defStyle { tokenColor = Just (RGB 61 174 233) }) + , ( SpecialCharTok, defStyle { tokenColor = Just (RGB 61 174 233) }) + , ( StringTok, defStyle { tokenColor = Just (RGB 244 79 79) }) + , ( VerbatimStringTok, defStyle { tokenColor = Just (RGB 218 68 83) }) + , ( SpecialStringTok, defStyle { tokenColor = Just (RGB 218 68 83) }) + , ( ImportTok, defStyle { tokenColor = Just (RGB 39 174 96) }) + , ( CommentTok, defStyle { tokenColor = Just (RGB 122 124 125) }) + , ( DocumentationTok, defStyle { tokenColor = Just (RGB 164 51 64) }) + , ( AnnotationTok, defStyle { tokenColor = Just (RGB 63 128 88) }) + , ( CommentVarTok, defStyle { tokenColor = Just (RGB 127 140 141) }) + , ( OtherTok, defStyle { tokenColor = Just (RGB 39 174 96) }) + , ( FunctionTok, defStyle { tokenColor = Just (RGB 142 68 173) }) + , ( VariableTok, defStyle { tokenColor = Just (RGB 39 174 174) }) + , ( ControlFlowTok, defStyle { tokenColor = Just (RGB 253 188 75) }) + , ( OperatorTok, defStyle { tokenColor = Just (RGB 207 207 194) }) + , ( BuiltInTok, defStyle { tokenColor = Just (RGB 127 140 141) }) + , ( ExtensionTok, defStyle { tokenColor = Just (RGB 0 153 255) }) + , ( PreprocessorTok, defStyle { tokenColor = Just (RGB 39 174 96) }) + , ( AttributeTok, defStyle { tokenColor = Just (RGB 41 128 185) }) + , ( RegionMarkerTok, defStyle { tokenColor = Just (RGB 41 128 185) }) + , ( InformationTok, defStyle { tokenColor = Just (RGB 196 91 0) }) + , ( WarningTok, defStyle { tokenColor = Just (RGB 218 68 83) }) + , ( AlertTok, defStyle { tokenColor = Just (RGB 149 218 76) }) + , ( ErrorTok, defStyle { tokenColor = Just (RGB 218 68 83) }) + , ( NormalTok, defStyle { tokenColor = Just (RGB 207 207 194) }) + ] + , defaultColor = Just (RGB 207 207 194) + , backgroundColor = Just (RGB 35 38 41) + , lineNumberColor = Just (RGB 122 124 125) + , lineNumberBackgroundColor = Just (RGB 35 38 41) + } + +-- | Style based on pygments's default colors. +pygments :: Style +pygments = Style{ + backgroundColor = Nothing + , defaultColor = Nothing + , lineNumberColor = color 0xaaaaaa + , lineNumberBackgroundColor = Nothing + , tokenStyles = + [ (KeywordTok, defStyle{ tokenColor = color 0x007020, tokenBold = True }) + , (DataTypeTok, defStyle{ tokenColor = color 0x902000 }) + , (DecValTok, defStyle{ tokenColor = color 0x40a070 }) + , (BaseNTok, defStyle{ tokenColor = color 0x40a070 }) + , (FloatTok, defStyle{ tokenColor = color 0x40a070 }) + , (CharTok, defStyle{ tokenColor = color 0x4070a0 }) + , (StringTok, defStyle{ tokenColor = color 0x4070a0 }) + , (CommentTok, defStyle{ tokenColor = color 0x60a0b0, tokenItalic = True }) + , (OtherTok, defStyle{ tokenColor = color 0x007020 }) + , (AlertTok, defStyle{ tokenColor = color 0xff0000, tokenBold = True }) + , (FunctionTok, defStyle{ tokenColor = color 0x06287e }) + , (ErrorTok, defStyle{ tokenColor = color 0xff0000, tokenBold = True }) + , (WarningTok, defStyle{ tokenColor = color 0x60a0b0, tokenItalic = True, tokenBold = True }) + , (ConstantTok, defStyle{ tokenColor = color 0x880000 }) + , (SpecialCharTok, defStyle{ tokenColor = color 0x4070a0 }) + , (VerbatimStringTok, defStyle{ tokenColor = color 0x4070a0 }) + , (SpecialStringTok, defStyle{ tokenColor = color 0xBB6688 }) + , (ImportTok, defStyle) + , (VariableTok, defStyle{ tokenColor = color 0x19177C }) + , (ControlFlowTok, defStyle{ tokenColor = color 0x007020, tokenBold = True }) + , (OperatorTok, defStyle{ tokenColor = color 0x666666 }) + , (BuiltInTok, defStyle) + , (ExtensionTok, defStyle) + , (PreprocessorTok, defStyle{ tokenColor = color 0xBC7A00 }) + , (AttributeTok, defStyle{ tokenColor = color 0x7D9029 }) + , (DocumentationTok, defStyle{ tokenColor = color 0xBA2121, tokenItalic = True }) + , (AnnotationTok, defStyle{ tokenColor = color 0x60a0b0, tokenItalic = True, tokenBold = True }) + , (CommentVarTok, defStyle{ tokenColor = color 0x60a0b0, tokenItalic = True, tokenBold = True }) + , (InformationTok, defStyle{ tokenColor = color 0x60a0b0, tokenItalic = True, tokenBold = True }) + ] + } + -- | Style based on pygments's tango colors. tango :: Style diff --git a/src/Skylighting/Tokenizer.hs b/src/Skylighting/Tokenizer.hs --- a/src/Skylighting/Tokenizer.hs +++ b/src/Skylighting/Tokenizer.hs @@ -403,14 +403,19 @@ $ reCompiled re inp <- gets input prev <- gets prevChar - -- we keep one preceding character, so initial \b can match: - let target = encodeUtf8 $ - if prev == '\n' - then Text.cons ' ' inp - else Text.cons prev inp + -- If regex starts with \b, determine if we're at a word + -- boundary and mzero if not (TODO - is this the correct + -- definition of a word boundary?) + when (BS.take 2 reStr == "\\b") $ + case Text.uncons inp of + Nothing -> return () + Just (c, _) + | isAlphaNum prev -> guard (not (isAlphaNum c)) + | otherwise -> guard (isAlphaNum c) + let target = encodeUtf8 inp case matchRegex regex target of Just (match:capts) -> do - match' <- decodeBS $ BS.drop 1 match -- drop the prevchar + match' <- decodeBS match capts' <- mapM decodeBS capts modify $ \st -> st{ captures = capts' } takeChars (Text.length match') diff --git a/test/expected/abc.python.native b/test/expected/abc.python.native --- a/test/expected/abc.python.native +++ b/test/expected/abc.python.native @@ -39,7 +39,8 @@ , ( NormalTok , ":i] " ) , ( OperatorTok , "+" ) , ( NormalTok , " b[i" ) - , ( DecValTok , "+1" ) + , ( OperatorTok , "+" ) + , ( DecValTok , "1" ) , ( NormalTok , ":])" ) ] , [ ( NormalTok , " " )