module Text.Highlighting.Kate.Syntax.Html
(highlight, parseExpression, syntaxName, syntaxExtensions)
where
import Text.Highlighting.Kate.Types
import Text.Highlighting.Kate.Common
import qualified Text.Highlighting.Kate.Syntax.Alert
import qualified Text.Highlighting.Kate.Syntax.Css
import qualified Text.Highlighting.Kate.Syntax.Javascript
import Text.ParserCombinators.Parsec hiding (State)
import Data.Map (fromList)
import Control.Monad.State
import Data.Char (isSpace)
import Data.Maybe (fromMaybe)
syntaxName :: String
syntaxName = "HTML"
syntaxExtensions :: String
syntaxExtensions = "*.htm;*.html;*.shtml;*.shtm"
highlight :: String -> [SourceLine]
highlight input = evalState (mapM parseSourceLine $ lines input) startingState
parseSourceLine :: String -> State SyntaxState SourceLine
parseSourceLine = mkParseSourceLine parseExpressionInternal pEndLine
parseExpression :: KateParser Token
parseExpression = do
st <- getState
let oldLang = synStLanguage st
setState $ st { synStLanguage = "HTML" }
context <- currentContext <|> (pushContext "Start" >> currentContext)
result <- parseRules context
optional $ eof >> pEndLine
updateState $ \st -> st { synStLanguage = oldLang }
return result
startingState = SyntaxState {synStContexts = fromList [("HTML",["Start"])], synStLanguage = "HTML", synStLineNumber = 0, synStPrevChar = '\n', synStPrevNonspace = False, synStCaseSensitive = True, synStKeywordCaseSensitive = True, synStCaptures = []}
pEndLine = do
updateState $ \st -> st{ synStPrevNonspace = False }
context <- currentContext
case context of
"Start" -> return ()
"FindHTML" -> return ()
"FindEntityRefs" -> return ()
"FindPEntityRefs" -> return ()
"FindAttributes" -> return ()
"FindDTDRules" -> return ()
"Comment" -> return ()
"CDATA" -> return ()
"PI" -> return ()
"Doctype" -> return ()
"Doctype Internal Subset" -> return ()
"Doctype Markupdecl" -> return ()
"Doctype Markupdecl DQ" -> return ()
"Doctype Markupdecl SQ" -> return ()
"El Open" -> return ()
"El Close" -> return ()
"El Close 2" -> return ()
"El Close 3" -> return ()
"CSS" -> return ()
"CSS content" -> return ()
"JS" -> return ()
"JS content" -> return ()
"JS comment close" -> (popContext) >> pEndLine
"Value" -> return ()
"Value NQ" -> (popContext >> popContext) >> pEndLine
"Value DQ" -> return ()
"Value SQ" -> return ()
_ -> return ()
withAttribute attr txt = do
when (null txt) $ fail "Parser matched no text"
updateState $ \st -> st { synStPrevChar = last txt
, synStPrevNonspace = synStPrevNonspace st || not (all isSpace txt) }
return (attr, txt)
parseExpressionInternal = do
context <- currentContext
parseRules context <|> (pDefault >>= withAttribute (fromMaybe NormalTok $ lookup context defaultAttributes))
regex_'3c'21DOCTYPE'5cs'2b = compileRegex "<!DOCTYPE\\s+"
regex_'3c'5c'3f'5b'5cw'3a'2d'5d'2a = compileRegex "<\\?[\\w:-]*"
regex_'3cstyle'5cb = compileRegex "<style\\b"
regex_'3cscript'5cb = compileRegex "<script\\b"
regex_'3cpre'5cb = compileRegex "<pre\\b"
regex_'3cdiv'5cb = compileRegex "<div\\b"
regex_'3ctable'5cb = compileRegex "<table\\b"
regex_'3cul'5cb = compileRegex "<ul\\b"
regex_'3col'5cb = compileRegex "<ol\\b"
regex_'3cdl'5cb = compileRegex "<dl\\b"
regex_'3c'5bA'2dZa'2dz'5f'3a'5d'5b'5cw'2e'3a'5f'2d'5d'2a = compileRegex "<[A-Za-z_:][\\w.:_-]*"
regex_'3c'2fpre'5cb = compileRegex "</pre\\b"
regex_'3c'2fdiv'5cb = compileRegex "</div\\b"
regex_'3c'2ftable'5cb = compileRegex "</table\\b"
regex_'3c'2ful'5cb = compileRegex "</ul\\b"
regex_'3c'2fol'5cb = compileRegex "</ol\\b"
regex_'3c'2fdl'5cb = compileRegex "</dl\\b"
regex_'3c'2f'5bA'2dZa'2dz'5f'3a'5d'5b'5cw'2e'3a'5f'2d'5d'2a = compileRegex "</[A-Za-z_:][\\w.:_-]*"
regex_'26'28'23'5b0'2d9'5d'2b'7c'23'5bxX'5d'5b0'2d9A'2dFa'2df'5d'2b'7c'5bA'2dZa'2dz'5f'3a'5d'5b'5cw'2e'3a'5f'2d'5d'2a'29'3b = compileRegex "&(#[0-9]+|#[xX][0-9A-Fa-f]+|[A-Za-z_:][\\w.:_-]*);"
regex_'25'5bA'2dZa'2dz'5f'3a'5d'5b'5cw'2e'3a'5f'2d'5d'2a'3b = compileRegex "%[A-Za-z_:][\\w.:_-]*;"
regex_'5bA'2dZa'2dz'5f'3a'5d'5b'5cw'2e'3a'5f'2d'5d'2a = compileRegex "[A-Za-z_:][\\w.:_-]*"
regex_'5cs'2b'5bA'2dZa'2dz'5f'3a'5d'5b'5cw'2e'3a'5f'2d'5d'2a = compileRegex "\\s+[A-Za-z_:][\\w.:_-]*"
regex_'3c'21'28ELEMENT'7cENTITY'7cATTLIST'7cNOTATION'29'5cb = compileRegex "<!(ELEMENT|ENTITY|ATTLIST|NOTATION)\\b"
regex_'2d'28'2d'28'3f'21'2d'3e'29'29'2b = compileRegex "-(-(?!->))+"
regex_'5cS = compileRegex "\\S"
regex_'3c'2fstyle'5cb = compileRegex "</style\\b"
regex_'3c'2fscript'5cb = compileRegex "</script\\b"
regex_'2f'2f'28'3f'3d'2e'2a'3c'2fscript'5cb'29 = compileRegex "//(?=.*</script\\b)"
regex_'2f'28'3f'21'3e'29 = compileRegex "/(?!>)"
regex_'5b'5e'2f'3e'3c'22'27'5cs'5d = compileRegex "[^/><\"'\\s]"
defaultAttributes = [("Start",NormalTok),("FindHTML",NormalTok),("FindEntityRefs",NormalTok),("FindPEntityRefs",NormalTok),("FindAttributes",NormalTok),("FindDTDRules",NormalTok),("Comment",CommentTok),("CDATA",NormalTok),("PI",NormalTok),("Doctype",NormalTok),("Doctype Internal Subset",NormalTok),("Doctype Markupdecl",NormalTok),("Doctype Markupdecl DQ",StringTok),("Doctype Markupdecl SQ",StringTok),("El Open",NormalTok),("El Close",NormalTok),("El Close 2",NormalTok),("El Close 3",NormalTok),("CSS",NormalTok),("CSS content",NormalTok),("JS",NormalTok),("JS content",NormalTok),("JS comment close",CommentTok),("Value",NormalTok),("Value NQ",NormalTok),("Value DQ",StringTok),("Value SQ",StringTok)]
parseRules "Start" =
((parseRules "FindHTML"))
parseRules "FindHTML" =
(((pDetectSpaces >>= withAttribute NormalTok))
<|>
((pDetectIdentifier >>= withAttribute NormalTok))
<|>
((pString False "<!--" >>= withAttribute CommentTok) >>~ pushContext "Comment")
<|>
((pString False "<![CDATA[" >>= withAttribute BaseNTok) >>~ pushContext "CDATA")
<|>
((pRegExpr regex_'3c'21DOCTYPE'5cs'2b >>= withAttribute DataTypeTok) >>~ pushContext "Doctype")
<|>
((pRegExpr regex_'3c'5c'3f'5b'5cw'3a'2d'5d'2a >>= withAttribute KeywordTok) >>~ pushContext "PI")
<|>
((pRegExpr regex_'3cstyle'5cb >>= withAttribute KeywordTok) >>~ pushContext "CSS")
<|>
((pRegExpr regex_'3cscript'5cb >>= withAttribute KeywordTok) >>~ pushContext "JS")
<|>
((pRegExpr regex_'3cpre'5cb >>= withAttribute KeywordTok) >>~ pushContext "El Open")
<|>
((pRegExpr regex_'3cdiv'5cb >>= withAttribute KeywordTok) >>~ pushContext "El Open")
<|>
((pRegExpr regex_'3ctable'5cb >>= withAttribute KeywordTok) >>~ pushContext "El Open")
<|>
((pRegExpr regex_'3cul'5cb >>= withAttribute KeywordTok) >>~ pushContext "El Open")
<|>
((pRegExpr regex_'3col'5cb >>= withAttribute KeywordTok) >>~ pushContext "El Open")
<|>
((pRegExpr regex_'3cdl'5cb >>= withAttribute KeywordTok) >>~ pushContext "El Open")
<|>
((pRegExpr regex_'3c'5bA'2dZa'2dz'5f'3a'5d'5b'5cw'2e'3a'5f'2d'5d'2a >>= withAttribute KeywordTok) >>~ pushContext "El Open")
<|>
((pRegExpr regex_'3c'2fpre'5cb >>= withAttribute KeywordTok) >>~ pushContext "El Close")
<|>
((pRegExpr regex_'3c'2fdiv'5cb >>= withAttribute KeywordTok) >>~ pushContext "El Close")
<|>
((pRegExpr regex_'3c'2ftable'5cb >>= withAttribute KeywordTok) >>~ pushContext "El Close")
<|>
((pRegExpr regex_'3c'2ful'5cb >>= withAttribute KeywordTok) >>~ pushContext "El Close")
<|>
((pRegExpr regex_'3c'2fol'5cb >>= withAttribute KeywordTok) >>~ pushContext "El Close")
<|>
((pRegExpr regex_'3c'2fdl'5cb >>= withAttribute KeywordTok) >>~ pushContext "El Close")
<|>
((pRegExpr regex_'3c'2f'5bA'2dZa'2dz'5f'3a'5d'5b'5cw'2e'3a'5f'2d'5d'2a >>= withAttribute KeywordTok) >>~ pushContext "El Close")
<|>
((parseRules "FindDTDRules"))
<|>
((parseRules "FindEntityRefs")))
parseRules "FindEntityRefs" =
(((pRegExpr regex_'26'28'23'5b0'2d9'5d'2b'7c'23'5bxX'5d'5b0'2d9A'2dFa'2df'5d'2b'7c'5bA'2dZa'2dz'5f'3a'5d'5b'5cw'2e'3a'5f'2d'5d'2a'29'3b >>= withAttribute DecValTok))
<|>
((pAnyChar "&<" >>= withAttribute ErrorTok)))
parseRules "FindPEntityRefs" =
(((pRegExpr regex_'26'28'23'5b0'2d9'5d'2b'7c'23'5bxX'5d'5b0'2d9A'2dFa'2df'5d'2b'7c'5bA'2dZa'2dz'5f'3a'5d'5b'5cw'2e'3a'5f'2d'5d'2a'29'3b >>= withAttribute DecValTok))
<|>
((pRegExpr regex_'25'5bA'2dZa'2dz'5f'3a'5d'5b'5cw'2e'3a'5f'2d'5d'2a'3b >>= withAttribute DecValTok))
<|>
((pAnyChar "&%" >>= withAttribute ErrorTok)))
parseRules "FindAttributes" =
(((pColumn 0 >> pRegExpr regex_'5bA'2dZa'2dz'5f'3a'5d'5b'5cw'2e'3a'5f'2d'5d'2a >>= withAttribute OtherTok))
<|>
((pRegExpr regex_'5cs'2b'5bA'2dZa'2dz'5f'3a'5d'5b'5cw'2e'3a'5f'2d'5d'2a >>= withAttribute OtherTok))
<|>
((pDetectChar False '=' >>= withAttribute OtherTok) >>~ pushContext "Value"))
parseRules "FindDTDRules" =
((pRegExpr regex_'3c'21'28ELEMENT'7cENTITY'7cATTLIST'7cNOTATION'29'5cb >>= withAttribute DataTypeTok) >>~ pushContext "Doctype Markupdecl")
parseRules "Comment" =
(((pDetectSpaces >>= withAttribute CommentTok))
<|>
((Text.Highlighting.Kate.Syntax.Alert.parseExpression >>= ((withAttribute CommentTok) . snd)))
<|>
((pDetectIdentifier >>= withAttribute CommentTok))
<|>
((pString False "-->" >>= withAttribute CommentTok) >>~ (popContext))
<|>
((pRegExpr regex_'2d'28'2d'28'3f'21'2d'3e'29'29'2b >>= withAttribute ErrorTok)))
parseRules "CDATA" =
(((pDetectSpaces >>= withAttribute NormalTok))
<|>
((pDetectIdentifier >>= withAttribute NormalTok))
<|>
((pString False "]]>" >>= withAttribute BaseNTok) >>~ (popContext))
<|>
((pString False "]]>" >>= withAttribute DecValTok)))
parseRules "PI" =
((pDetect2Chars False '?' '>' >>= withAttribute KeywordTok) >>~ (popContext))
parseRules "Doctype" =
(((pDetectChar False '>' >>= withAttribute DataTypeTok) >>~ (popContext))
<|>
((pDetectChar False '[' >>= withAttribute DataTypeTok) >>~ pushContext "Doctype Internal Subset"))
parseRules "Doctype Internal Subset" =
(((pDetectChar False ']' >>= withAttribute DataTypeTok) >>~ (popContext))
<|>
((parseRules "FindDTDRules"))
<|>
((pString False "<!--" >>= withAttribute CommentTok) >>~ pushContext "Comment")
<|>
((pRegExpr regex_'3c'5c'3f'5b'5cw'3a'2d'5d'2a >>= withAttribute KeywordTok) >>~ pushContext "PI")
<|>
((parseRules "FindPEntityRefs")))
parseRules "Doctype Markupdecl" =
(((pDetectChar False '>' >>= withAttribute DataTypeTok) >>~ (popContext))
<|>
((pDetectChar False '"' >>= withAttribute StringTok) >>~ pushContext "Doctype Markupdecl DQ")
<|>
((pDetectChar False '\'' >>= withAttribute StringTok) >>~ pushContext "Doctype Markupdecl SQ"))
parseRules "Doctype Markupdecl DQ" =
(((pDetectChar False '"' >>= withAttribute StringTok) >>~ (popContext))
<|>
((parseRules "FindPEntityRefs")))
parseRules "Doctype Markupdecl SQ" =
(((pDetectChar False '\'' >>= withAttribute StringTok) >>~ (popContext))
<|>
((parseRules "FindPEntityRefs")))
parseRules "El Open" =
(((pDetect2Chars False '/' '>' >>= withAttribute KeywordTok) >>~ (popContext))
<|>
((pDetectChar False '>' >>= withAttribute KeywordTok) >>~ (popContext))
<|>
((parseRules "FindAttributes"))
<|>
((pRegExpr regex_'5cS >>= withAttribute ErrorTok)))
parseRules "El Close" =
(((pDetectChar False '>' >>= withAttribute KeywordTok) >>~ (popContext))
<|>
((pRegExpr regex_'5cS >>= withAttribute ErrorTok)))
parseRules "El Close 2" =
(((pDetectChar False '>' >>= withAttribute KeywordTok) >>~ (popContext >> popContext >> popContext))
<|>
((pRegExpr regex_'5cS >>= withAttribute ErrorTok)))
parseRules "El Close 3" =
(((pDetectChar False '>' >>= withAttribute KeywordTok) >>~ (popContext >> popContext >> popContext >> popContext))
<|>
((pRegExpr regex_'5cS >>= withAttribute ErrorTok)))
parseRules "CSS" =
(((pDetect2Chars False '/' '>' >>= withAttribute KeywordTok) >>~ (popContext))
<|>
((pDetectChar False '>' >>= withAttribute KeywordTok) >>~ pushContext "CSS content")
<|>
((parseRules "FindAttributes"))
<|>
((pRegExpr regex_'5cS >>= withAttribute ErrorTok)))
parseRules "CSS content" =
(((pRegExpr regex_'3c'2fstyle'5cb >>= withAttribute KeywordTok) >>~ pushContext "El Close 2")
<|>
((Text.Highlighting.Kate.Syntax.Css.parseExpression)))
parseRules "JS" =
(((pDetect2Chars False '/' '>' >>= withAttribute KeywordTok) >>~ (popContext))
<|>
((pDetectChar False '>' >>= withAttribute KeywordTok) >>~ pushContext "JS content")
<|>
((parseRules "FindAttributes"))
<|>
((pRegExpr regex_'5cS >>= withAttribute ErrorTok)))
parseRules "JS content" =
(((pRegExpr regex_'3c'2fscript'5cb >>= withAttribute KeywordTok) >>~ pushContext "El Close 2")
<|>
((pRegExpr regex_'2f'2f'28'3f'3d'2e'2a'3c'2fscript'5cb'29 >>= withAttribute CommentTok) >>~ pushContext "JS comment close")
<|>
((Text.Highlighting.Kate.Syntax.Javascript.parseExpression)))
parseRules "JS comment close" =
(((pRegExpr regex_'3c'2fscript'5cb >>= withAttribute KeywordTok) >>~ pushContext "El Close 3")
<|>
((Text.Highlighting.Kate.Syntax.Alert.parseExpression >>= ((withAttribute CommentTok) . snd))))
parseRules "Value" =
(((pDetectChar False '"' >>= withAttribute StringTok) >>~ pushContext "Value DQ")
<|>
((pDetectChar False '\'' >>= withAttribute StringTok) >>~ pushContext "Value SQ")
<|>
((pDetectSpaces >>= withAttribute NormalTok))
<|>
(pushContext "Value NQ" >> currentContext >>= parseRules))
parseRules "Value NQ" =
(((parseRules "FindEntityRefs"))
<|>
((pRegExpr regex_'2f'28'3f'21'3e'29 >>= withAttribute StringTok))
<|>
((pRegExpr regex_'5b'5e'2f'3e'3c'22'27'5cs'5d >>= withAttribute StringTok))
<|>
((popContext >> popContext) >> currentContext >>= parseRules))
parseRules "Value DQ" =
(((pDetectChar False '"' >>= withAttribute StringTok) >>~ (popContext >> popContext))
<|>
((parseRules "FindEntityRefs")))
parseRules "Value SQ" =
(((pDetectChar False '\'' >>= withAttribute StringTok) >>~ (popContext >> popContext))
<|>
((parseRules "FindEntityRefs")))
parseRules "" = parseRules "Start"
parseRules x = fail $ "Unknown context" ++ x