module Text.Highlighting.Kate.Syntax.Xml
(highlight, parseExpression, syntaxName, syntaxExtensions)
where
import Text.Highlighting.Kate.Types
import Text.Highlighting.Kate.Common
import qualified Text.Highlighting.Kate.Syntax.Alert
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 = "XML"
syntaxExtensions :: String
syntaxExtensions = "*.docbook;*.xml;*.rc;*.daml;*.rdf;*.rss;*.xspf;*.xsd;*.svg;*.ui;*.kcfg;*.qrc;*.wsdl"
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 = "XML" }
context <- currentContext <|> (pushContext "Start" >> currentContext)
result <- parseRules context
optional $ eof >> pEndLine
updateState $ \st -> st { synStLanguage = oldLang }
return result
startingState = SyntaxState {synStContexts = fromList [("XML",["Start"])], synStLanguage = "XML", 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 ()
"FindXML" -> return ()
"FindEntityRefs" -> return ()
"FindPEntityRefs" -> return ()
"Comment" -> return ()
"CDATA" -> return ()
"PI" -> return ()
"Doctype" -> return ()
"Doctype Internal Subset" -> return ()
"Doctype Markupdecl" -> return ()
"Doctype Markupdecl DQ" -> return ()
"Doctype Markupdecl SQ" -> return ()
"Element" -> return ()
"El Content" -> return ()
"El End" -> return ()
"Attribute" -> return ()
"Value" -> return ()
"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'5f'2d'5d'2a = compileRegex "<\\?[\\w:_-]*"
regex_'3c'28'3f'21'5b0'2d9'5d'29'5b'5cw'5f'3a'5d'5b'5cw'2e'3a'5f'2d'5d'2a = compileRegex "<(?![0-9])[\\w_:][\\w.:_-]*"
regex_'26'28'23'5b0'2d9'5d'2b'7c'23'5bxX'5d'5b0'2d9A'2dFa'2df'5d'2b'7c'28'3f'21'5b0'2d9'5d'29'5b'5cw'5f'3a'5d'5b'5cw'2e'3a'5f'2d'5d'2a'29'3b = compileRegex "&(#[0-9]+|#[xX][0-9A-Fa-f]+|(?![0-9])[\\w_:][\\w.:_-]*);"
regex_'25'28'3f'21'5b0'2d9'5d'29'5b'5cw'5f'3a'5d'5b'5cw'2e'3a'5f'2d'5d'2a'3b = compileRegex "%(?![0-9])[\\w_:][\\w.:_-]*;"
regex_'2d'28'2d'28'3f'21'2d'3e'29'29'2b = compileRegex "-(-(?!->))+"
regex_'3c'21'28ELEMENT'7cENTITY'7cATTLIST'7cNOTATION'29'5cb = compileRegex "<!(ELEMENT|ENTITY|ATTLIST|NOTATION)\\b"
regex_'28'3f'21'5b0'2d9'5d'29'5b'5cw'5f'3a'5d'5b'5cw'2e'3a'5f'2d'5d'2a = compileRegex "(?![0-9])[\\w_:][\\w.:_-]*"
regex_'5cs'2b'28'3f'21'5b0'2d9'5d'29'5b'5cw'5f'3a'5d'5b'5cw'2e'3a'5f'2d'5d'2a = compileRegex "\\s+(?![0-9])[\\w_:][\\w.:_-]*"
regex_'5cS = compileRegex "\\S"
regex_'3c'2f'28'3f'21'5b0'2d9'5d'29'5b'5cw'5f'3a'5d'5b'5cw'2e'3a'5f'2d'5d'2a = compileRegex "</(?![0-9])[\\w_:][\\w.:_-]*"
defaultAttributes = [("Start",NormalTok),("FindXML",NormalTok),("FindEntityRefs",NormalTok),("FindPEntityRefs",NormalTok),("Comment",CommentTok),("CDATA",NormalTok),("PI",NormalTok),("Doctype",NormalTok),("Doctype Internal Subset",NormalTok),("Doctype Markupdecl",NormalTok),("Doctype Markupdecl DQ",StringTok),("Doctype Markupdecl SQ",StringTok),("Element",NormalTok),("El Content",NormalTok),("El End",NormalTok),("Attribute",NormalTok),("Value",NormalTok),("Value DQ",StringTok),("Value SQ",StringTok)]
parseRules "Start" =
((parseRules "FindXML"))
parseRules "FindXML" =
(((pDetectSpaces >>= 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'5f'2d'5d'2a >>= withAttribute KeywordTok) >>~ pushContext "PI")
<|>
((pRegExpr regex_'3c'28'3f'21'5b0'2d9'5d'29'5b'5cw'5f'3a'5d'5b'5cw'2e'3a'5f'2d'5d'2a >>= withAttribute KeywordTok) >>~ pushContext "Element")
<|>
((parseRules "FindEntityRefs"))
<|>
((pDetectIdentifier >>= withAttribute NormalTok)))
parseRules "FindEntityRefs" =
(((pRegExpr regex_'26'28'23'5b0'2d9'5d'2b'7c'23'5bxX'5d'5b0'2d9A'2dFa'2df'5d'2b'7c'28'3f'21'5b0'2d9'5d'29'5b'5cw'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'28'3f'21'5b0'2d9'5d'29'5b'5cw'5f'3a'5d'5b'5cw'2e'3a'5f'2d'5d'2a'29'3b >>= withAttribute DecValTok))
<|>
((pRegExpr regex_'25'28'3f'21'5b0'2d9'5d'29'5b'5cw'5f'3a'5d'5b'5cw'2e'3a'5f'2d'5d'2a'3b >>= withAttribute DecValTok))
<|>
((pAnyChar "&%" >>= withAttribute ErrorTok)))
parseRules "Comment" =
(((pDetectSpaces >>= withAttribute CommentTok))
<|>
((pString False "-->" >>= withAttribute CommentTok) >>~ (popContext))
<|>
((pRegExpr regex_'2d'28'2d'28'3f'21'2d'3e'29'29'2b >>= withAttribute ErrorTok))
<|>
((Text.Highlighting.Kate.Syntax.Alert.parseExpression >>= ((withAttribute CommentTok) . snd)))
<|>
((pDetectIdentifier >>= withAttribute CommentTok)))
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))
<|>
((pRegExpr regex_'3c'21'28ELEMENT'7cENTITY'7cATTLIST'7cNOTATION'29'5cb >>= withAttribute DataTypeTok) >>~ pushContext "Doctype Markupdecl")
<|>
((pString False "<!--" >>= withAttribute CommentTok) >>~ pushContext "Comment")
<|>
((pRegExpr regex_'3c'5c'3f'5b'5cw'3a'5f'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 "Element" =
(((pDetect2Chars False '/' '>' >>= withAttribute KeywordTok) >>~ (popContext))
<|>
((pDetectChar False '>' >>= withAttribute KeywordTok) >>~ pushContext "El Content")
<|>
((pColumn 0 >> pRegExpr regex_'28'3f'21'5b0'2d9'5d'29'5b'5cw'5f'3a'5d'5b'5cw'2e'3a'5f'2d'5d'2a >>= withAttribute OtherTok) >>~ pushContext "Attribute")
<|>
((pRegExpr regex_'5cs'2b'28'3f'21'5b0'2d9'5d'29'5b'5cw'5f'3a'5d'5b'5cw'2e'3a'5f'2d'5d'2a >>= withAttribute OtherTok) >>~ pushContext "Attribute")
<|>
((pRegExpr regex_'5cS >>= withAttribute ErrorTok)))
parseRules "El Content" =
(((pRegExpr regex_'3c'2f'28'3f'21'5b0'2d9'5d'29'5b'5cw'5f'3a'5d'5b'5cw'2e'3a'5f'2d'5d'2a >>= withAttribute KeywordTok) >>~ pushContext "El End")
<|>
((parseRules "FindXML")))
parseRules "El End" =
(((pDetectChar False '>' >>= withAttribute KeywordTok) >>~ (popContext >> popContext >> popContext))
<|>
((pRegExpr regex_'5cS >>= withAttribute ErrorTok)))
parseRules "Attribute" =
(((pDetectChar False '=' >>= withAttribute OtherTok) >>~ pushContext "Value")
<|>
((pRegExpr regex_'5cS >>= withAttribute ErrorTok)))
parseRules "Value" =
(((pDetectChar False '"' >>= withAttribute StringTok) >>~ pushContext "Value DQ")
<|>
((pDetectChar False '\'' >>= withAttribute StringTok) >>~ pushContext "Value SQ")
<|>
((pRegExpr regex_'5cS >>= withAttribute ErrorTok)))
parseRules "Value DQ" =
(((pDetectChar False '"' >>= withAttribute StringTok) >>~ (popContext >> popContext >> popContext))
<|>
((parseRules "FindEntityRefs")))
parseRules "Value SQ" =
(((pDetectChar False '\'' >>= withAttribute StringTok) >>~ (popContext >> popContext >> popContext))
<|>
((parseRules "FindEntityRefs")))
parseRules "" = parseRules "Start"
parseRules x = fail $ "Unknown context" ++ x