module Text.Highlighting.Kate.Syntax.R
(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 Control.Monad.State
import Data.Char (isSpace)
import qualified Data.Set as Set
syntaxName :: String
syntaxName = "R Script"
syntaxExtensions :: String
syntaxExtensions = "*.R;*.r;*.S;*.s;*.q"
highlight :: String -> [SourceLine]
highlight input = evalState (mapM parseSourceLine $ lines input) startingState
parseSourceLine :: String -> State SyntaxState SourceLine
parseSourceLine = mkParseSourceLine (parseExpression Nothing)
parseExpression :: Maybe (String,String)
-> KateParser Token
parseExpression mbcontext = do
(lang,cont) <- maybe currentContext return mbcontext
result <- parseRules (lang,cont)
optional $ do eof
updateState $ \st -> st{ synStPrevChar = '\n' }
pEndLine
return result
startingState = SyntaxState {synStContexts = [("R Script","level0")], synStLineNumber = 0, synStPrevChar = '\n', synStPrevNonspace = False, synStContinuation = False, synStCaseSensitive = True, synStKeywordCaseSensitive = True, synStCaptures = []}
pEndLine = do
updateState $ \st -> st{ synStPrevNonspace = False }
context <- currentContext
contexts <- synStContexts `fmap` getState
st <- getState
if length contexts >= 2
then case context of
_ | synStContinuation st -> updateState $ \st -> st{ synStContinuation = False }
("R Script","level0") -> return ()
("R Script","ctx0") -> return ()
("R Script","parenthesis") -> return ()
("R Script","string") -> return ()
("R Script","string2") -> return ()
("R Script","backquotedsymbol") -> return ()
("R Script","operator_rhs") -> return ()
("R Script","Headline") -> (popContext) >> pEndLine
("R Script","Comment") -> (popContext) >> pEndLine
("R Script","CommonRules") -> return ()
_ -> return ()
else 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)
list_controls = Set.fromList $ words $ "for in next break while repeat if else switch function"
list_words = Set.fromList $ words $ "TRUE FALSE NULL NA NA_integer_ NA_real_ NA_complex_ NA_character_ Inf NaN"
regex_'5ba'2dzA'2dZ'5f'5c'2e'5d'5b0'2d9a'2dzA'2dZ'5f'5c'2e'5d'2a'5b'5cs'5d'2a'3d'28'3f'3d'28'5b'5e'3d'5d'7c'24'29'29 = compileRegex True "[a-zA-Z_\\.][0-9a-zA-Z_\\.]*[\\s]*=(?=([^=]|$))"
regex_'28'5c'2a'7c'2f'7c'3c'7c'3e'7c'5c'21'3d'7c'3d'7c'5c'7c'7c'26'7c'3a'7c'5c'5e'7c'40'7c'5c'24'7c'7e'29 = compileRegex True "(\\*|/|<|>|\\!=|=|\\||&|:|\\^|@|\\$|~)"
regex_'5ba'2dzA'2dZ'5f'5d'2b'5ba'2dzA'2dZ'5f'5c'2e0'2d9'5d'2a'28'3f'3d'5b'5cs'5d'2a'5b'28'5d'29 = compileRegex True "[a-zA-Z_]+[a-zA-Z_\\.0-9]*(?=[\\s]*[(])"
regex_'5c'2e'5ba'2dzA'2dZ'5f'5c'2e'5d'2b'5ba'2dzA'2dZ'5f'5c'2e0'2d9'5d'2a'28'3f'3d'5b'5cs'5d'2a'5b'28'5d'29 = compileRegex True "\\.[a-zA-Z_\\.]+[a-zA-Z_\\.0-9]*(?=[\\s]*[(])"
regex_'5c'28 = compileRegex True "\\("
regex_'5b'3c'5d'7b1'2c2'7d'5c'2d = compileRegex True "[<]{1,2}\\-"
regex_'5c'2d'5b'3e'5d'7b1'2c2'7d = compileRegex True "\\-[>]{1,2}"
regex_'3d'28'3f'21'3d'29 = compileRegex True "=(?!=)"
regex_'28'5c'2b'7c'5c'2d'7c'5c'2a'7b1'2c2'7d'7c'2f'7c'3c'3d'3f'7c'3e'3d'3f'7c'3d'7b1'2c2'7d'7c'5c'21'3d'3f'7c'5c'7c'7b1'2c2'7d'7c'26'7b1'2c2'7d'7c'3a'7b1'2c3'7d'7c'5c'5e'7c'40'7c'5c'24'7c'7e'29 = compileRegex True "(\\+|\\-|\\*{1,2}|/|<=?|>=?|={1,2}|\\!=?|\\|{1,2}|&{1,2}|:{1,3}|\\^|@|\\$|~)"
regex_'25'5b'5e'25'5d'2a'25 = compileRegex True "%[^%]*%"
parseRules ("R Script","level0") =
(((parseRules ("R Script","CommonRules")))
<|>
((pDetectChar False '}' >>= withAttribute ErrorTok))
<|>
((pDetectChar False ')' >>= withAttribute ErrorTok))
<|>
(currentContext >>= \x -> guard (x == ("R Script","level0")) >> pDefault >>= withAttribute NormalTok))
parseRules ("R Script","ctx0") =
(((parseRules ("R Script","CommonRules")))
<|>
((pDetectChar False '}' >>= withAttribute NormalTok) >>~ (popContext))
<|>
((pDetectChar False ')' >>= withAttribute ErrorTok))
<|>
(currentContext >>= \x -> guard (x == ("R Script","ctx0")) >> pDefault >>= withAttribute NormalTok))
parseRules ("R Script","parenthesis") =
(((pLineContinue >>= withAttribute NormalTok))
<|>
((pDetectChar False ')' >>= withAttribute NormalTok) >>~ (popContext))
<|>
((pRegExpr regex_'5ba'2dzA'2dZ'5f'5c'2e'5d'5b0'2d9a'2dzA'2dZ'5f'5c'2e'5d'2a'5b'5cs'5d'2a'3d'28'3f'3d'28'5b'5e'3d'5d'7c'24'29'29 >>= withAttribute DataTypeTok))
<|>
((parseRules ("R Script","CommonRules")))
<|>
((pDetectChar False '}' >>= withAttribute ErrorTok))
<|>
(currentContext >>= \x -> guard (x == ("R Script","parenthesis")) >> pDefault >>= withAttribute NormalTok))
parseRules ("R Script","string") =
(((pDetectChar False '"' >>= withAttribute StringTok) >>~ (popContext))
<|>
((pHlCStringChar >>= withAttribute CharTok))
<|>
(currentContext >>= \x -> guard (x == ("R Script","string")) >> pDefault >>= withAttribute StringTok))
parseRules ("R Script","string2") =
(((pDetectChar False '\'' >>= withAttribute StringTok) >>~ (popContext))
<|>
((pHlCStringChar >>= withAttribute CharTok))
<|>
(currentContext >>= \x -> guard (x == ("R Script","string2")) >> pDefault >>= withAttribute StringTok))
parseRules ("R Script","backquotedsymbol") =
(((pDetectChar False '`' >>= withAttribute StringTok) >>~ (popContext))
<|>
((pHlCStringChar >>= withAttribute CharTok))
<|>
(currentContext >>= \x -> guard (x == ("R Script","backquotedsymbol")) >> pDefault >>= withAttribute DataTypeTok))
parseRules ("R Script","operator_rhs") =
(((pString False "##" >>= withAttribute NormalTok) >>~ pushContext ("R Script","Headline"))
<|>
((pDetectChar False '#' >>= withAttribute CommentTok) >>~ pushContext ("R Script","Comment"))
<|>
((pDetectChar False ' ' >>= withAttribute StringTok))
<|>
((pRegExpr regex_'28'5c'2a'7c'2f'7c'3c'7c'3e'7c'5c'21'3d'7c'3d'7c'5c'7c'7c'26'7c'3a'7c'5c'5e'7c'40'7c'5c'24'7c'7e'29 >>= withAttribute ErrorTok))
<|>
((popContext) >> currentContext >>= parseRules))
parseRules ("R Script","Headline") =
(((Text.Highlighting.Kate.Syntax.Alert.parseExpression (Just ("Alerts",""))))
<|>
(currentContext >>= \x -> guard (x == ("R Script","Headline")) >> pDefault >>= withAttribute NormalTok))
parseRules ("R Script","Comment") =
(((Text.Highlighting.Kate.Syntax.Alert.parseExpression (Just ("Alerts","")) >>= ((withAttribute CommentTok) . snd)))
<|>
(currentContext >>= \x -> guard (x == ("R Script","Comment")) >> pDefault >>= withAttribute CommentTok))
parseRules ("R Script","CommonRules") =
(((pDetectChar False '"' >>= withAttribute StringTok) >>~ pushContext ("R Script","string"))
<|>
((pDetectChar False '\'' >>= withAttribute StringTok) >>~ pushContext ("R Script","string2"))
<|>
((pDetectChar False '`' >>= withAttribute StringTok) >>~ pushContext ("R Script","backquotedsymbol"))
<|>
((pKeyword " \n\t():!+,-<=>%&*/;?[]^{|}~\\$" list_controls >>= withAttribute NormalTok))
<|>
((pKeyword " \n\t():!+,-<=>%&*/;?[]^{|}~\\$" list_words >>= withAttribute OtherTok))
<|>
((pFloat >>= withAttribute FloatTok))
<|>
((pInt >>= withAttribute DecValTok))
<|>
((pRegExpr regex_'5ba'2dzA'2dZ'5f'5d'2b'5ba'2dzA'2dZ'5f'5c'2e0'2d9'5d'2a'28'3f'3d'5b'5cs'5d'2a'5b'28'5d'29 >>= withAttribute KeywordTok))
<|>
((pRegExpr regex_'5c'2e'5ba'2dzA'2dZ'5f'5c'2e'5d'2b'5ba'2dzA'2dZ'5f'5c'2e0'2d9'5d'2a'28'3f'3d'5b'5cs'5d'2a'5b'28'5d'29 >>= withAttribute KeywordTok))
<|>
((pRegExpr regex_'5c'28 >>= withAttribute NormalTok) >>~ pushContext ("R Script","parenthesis"))
<|>
((pString False "##" >>= withAttribute NormalTok) >>~ pushContext ("R Script","Headline"))
<|>
((pDetectChar False '#' >>= withAttribute CommentTok) >>~ pushContext ("R Script","Comment"))
<|>
((pRegExpr regex_'5b'3c'5d'7b1'2c2'7d'5c'2d >>= withAttribute NormalTok) >>~ pushContext ("R Script","operator_rhs"))
<|>
((pRegExpr regex_'5c'2d'5b'3e'5d'7b1'2c2'7d >>= withAttribute NormalTok) >>~ pushContext ("R Script","operator_rhs"))
<|>
((pRegExpr regex_'3d'28'3f'21'3d'29 >>= withAttribute NormalTok) >>~ pushContext ("R Script","operator_rhs"))
<|>
((pRegExpr regex_'28'5c'2b'7c'5c'2d'7c'5c'2a'7b1'2c2'7d'7c'2f'7c'3c'3d'3f'7c'3e'3d'3f'7c'3d'7b1'2c2'7d'7c'5c'21'3d'3f'7c'5c'7c'7b1'2c2'7d'7c'26'7b1'2c2'7d'7c'3a'7b1'2c3'7d'7c'5c'5e'7c'40'7c'5c'24'7c'7e'29 >>= withAttribute NormalTok) >>~ pushContext ("R Script","operator_rhs"))
<|>
((pRegExpr regex_'25'5b'5e'25'5d'2a'25 >>= withAttribute NormalTok) >>~ pushContext ("R Script","operator_rhs"))
<|>
((pDetectChar False '{' >>= withAttribute NormalTok) >>~ pushContext ("R Script","ctx0"))
<|>
((pDetectChar False '[' >>= withAttribute NormalTok))
<|>
((pDetectChar False ']' >>= withAttribute NormalTok))
<|>
(currentContext >>= \x -> guard (x == ("R Script","CommonRules")) >> pDefault >>= withAttribute NormalTok))
parseRules ("Alerts", _) = Text.Highlighting.Kate.Syntax.Alert.parseExpression Nothing
parseRules x = parseRules ("R Script","level0") <|> fail ("Unknown context" ++ show x)