module Text.Highlighting.Kate.Syntax.Ocaml
(highlight, parseExpression, syntaxName, syntaxExtensions)
where
import Text.Highlighting.Kate.Types
import Text.Highlighting.Kate.Common
import Text.ParserCombinators.Parsec hiding (State)
import Data.Map (fromList)
import Control.Monad.State
import Data.Char (isSpace)
import Data.Maybe (fromMaybe)
import qualified Data.Set as Set
syntaxName :: String
syntaxName = "Objective Caml"
syntaxExtensions :: String
syntaxExtensions = "*.ml;*.mli"
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 = "Objective Caml" }
context <- currentContext <|> (pushContext "Normal" >> currentContext)
result <- parseRules context
optional $ eof >> pEndLine
updateState $ \st -> st { synStLanguage = oldLang }
return result
startingState = SyntaxState {synStContexts = fromList [("Objective Caml",["Normal"])], synStLanguage = "Objective Caml", synStLineNumber = 0, synStPrevChar = '\n', synStPrevNonspace = False, synStCaseSensitive = True, synStKeywordCaseSensitive = True, synStCaptures = []}
pEndLine = do
updateState $ \st -> st{ synStPrevNonspace = False }
context <- currentContext
case context of
"Normal" -> return ()
"Multiline Comment" -> return ()
"String Constant" -> return ()
"Block" -> return ()
"Sig" -> return ()
"Struct" -> return ()
"Object" -> return ()
"ModuleEnv" -> return ()
"ModuleEnv2" -> return ()
"Camlp4 Quotation Constant" -> 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))
list_revised_syntax_keywords = Set.fromList $ words $ "declare value where"
list_keywords = Set.fromList $ words $ "and as assert asr class closed constraint downto else exception external false for fun function functor if in inherit initializer land lazy let lor lsl lsr lxor match method mod mutable new of or parser private rec then to true try type val virtual when while with"
list_core_types = Set.fromList $ words $ "exn lazy_t format unit int real char string ref array bool list option"
list_symbols = Set.fromList $ words $ "?"
list_end = Set.fromList $ words $ "end"
list_sig = Set.fromList $ words $ "sig"
list_struct = Set.fromList $ words $ "struct"
list_object = Set.fromList $ words $ "object"
list_begin = Set.fromList $ words $ "begin"
list_do = Set.fromList $ words $ "do"
list_done = Set.fromList $ words $ "done"
list_module'5fenvironment = Set.fromList $ words $ "module open include"
regex_'60'5cs'2a'5bA'2dZa'2dz'5c0300'2d'5c0326'5c0330'2d'5c0366'5c0370'2d'5c0377'5f'5d'5bA'2dZa'2dz'5c0300'2d'5c0326'5c0330'2d'5c0366'5c0370'2d'5c03770'2d9'5f'27'5d'2a = compileRegex "`\\s*[A-Za-z\\0300-\\0326\\0330-\\0366\\0370-\\0377_][A-Za-z\\0300-\\0326\\0330-\\0366\\0370-\\03770-9_']*"
regex_'5bA'2dZ'5d'5bA'2dZa'2dz'5c0300'2d'5c0326'5c0330'2d'5c0366'5c0370'2d'5c03770'2d9'5f'27'5d'2a'5cs'2a'5c'2e = compileRegex "[A-Z][A-Za-z\\0300-\\0326\\0330-\\0366\\0370-\\03770-9_']*\\s*\\."
regex_'5bA'2dZ'5d'5bA'2dZa'2dz'5c0300'2d'5c0326'5c0330'2d'5c0366'5c0370'2d'5c03770'2d9'5f'27'5d'2a = compileRegex "[A-Z][A-Za-z\\0300-\\0326\\0330-\\0366\\0370-\\03770-9_']*"
regex_'23'5bA'2dZa'2dz'5c0300'2d'5c0326'5c0330'2d'5c0366'5c0370'2d'5c0377'5f'5d'5bA'2dZa'2dz'5c0300'2d'5c0326'5c0330'2d'5c0366'5c0370'2d'5c03770'2d9'5f'27'5d'2a'2e'2a'24 = compileRegex "#[A-Za-z\\0300-\\0326\\0330-\\0366\\0370-\\0377_][A-Za-z\\0300-\\0326\\0330-\\0366\\0370-\\03770-9_']*.*$"
regex_'27'28'28'5c'5c'5bntbr'27'22'5c'5c'5d'7c'5c'5c'5b0'2d9'5d'7b3'7d'7c'5c'5cx'5b0'2d9A'2dFa'2df'5d'7b2'7d'29'7c'5b'5e'27'5d'29'27 = compileRegex "'((\\\\[ntbr'\"\\\\]|\\\\[0-9]{3}|\\\\x[0-9A-Fa-f]{2})|[^'])'"
regex_'3c'3a'5bA'2dZa'2dz'5c0300'2d'5c0326'5c0330'2d'5c0366'5c0370'2d'5c0377'5f'5d'5bA'2dZa'2dz'5c0300'2d'5c0326'5c0330'2d'5c0366'5c0370'2d'5c03770'2d9'5f'27'5d'2a'3c = compileRegex "<:[A-Za-z\\0300-\\0326\\0330-\\0366\\0370-\\0377_][A-Za-z\\0300-\\0326\\0330-\\0366\\0370-\\03770-9_']*<"
regex_'5bA'2dZa'2dz'5c0300'2d'5c0326'5c0330'2d'5c0366'5c0370'2d'5c0377'5f'5d'5bA'2dZa'2dz'5c0300'2d'5c0326'5c0330'2d'5c0366'5c0370'2d'5c03770'2d9'5f'27'5d'2a = compileRegex "[A-Za-z\\0300-\\0326\\0330-\\0366\\0370-\\0377_][A-Za-z\\0300-\\0326\\0330-\\0366\\0370-\\03770-9_']*"
regex_'2d'3f0'5bxX'5d'5b0'2d9A'2dFa'2df'5f'5d'2b = compileRegex "-?0[xX][0-9A-Fa-f_]+"
regex_'2d'3f0'5boO'5d'5b0'2d7'5f'5d'2b = compileRegex "-?0[oO][0-7_]+"
regex_'2d'3f0'5bbB'5d'5b01'5f'5d'2b = compileRegex "-?0[bB][01_]+"
regex_'2d'3f'5b0'2d9'5d'5b0'2d9'5f'5d'2a'28'28'5c'2e'28'5b0'2d9'5d'5b0'2d9'5f'5d'2a'29'3f'28'5beE'5d'5b'2d'2b'5d'3f'5b0'2d9'5d'5b0'2d9'5f'5d'2a'29'3f'29'7c'28'5beE'5d'5b'2d'2b'5d'3f'5b0'2d9'5d'5b0'2d9'5f'5d'2a'29'29 = compileRegex "-?[0-9][0-9_]*((\\.([0-9][0-9_]*)?([eE][-+]?[0-9][0-9_]*)?)|([eE][-+]?[0-9][0-9_]*))"
regex_'2d'3f'5b0'2d9'5d'5b0'2d9'5f'5d'2a = compileRegex "-?[0-9][0-9_]*"
regex_'28'5c'5c'5bntbr'27'22'5c'5c'5d'7c'5c'5c'5b0'2d9'5d'7b3'7d'7c'5c'5cx'5b0'2d9A'2dFa'2df'5d'7b2'7d'29 = compileRegex "(\\\\[ntbr'\"\\\\]|\\\\[0-9]{3}|\\\\x[0-9A-Fa-f]{2})"
regex_'5c'5c'24 = compileRegex "\\\\$"
regex_'5c'5c'28'5c'5c'7c'3e'3e'7c'3c'3c'29 = compileRegex "\\\\(\\\\|>>|<<)"
regex_'5c'5c'3c'3a'5bA'2dZa'2dz'5c0300'2d'5c0326'5c0330'2d'5c0366'5c0370'2d'5c0377'5f'5d'5bA'2dZa'2dz'5c0300'2d'5c0326'5c0330'2d'5c0366'5c0370'2d'5c03770'2d9'5f'27'5d'2a'3c = compileRegex "\\\\<:[A-Za-z\\0300-\\0326\\0330-\\0366\\0370-\\0377_][A-Za-z\\0300-\\0326\\0330-\\0366\\0370-\\03770-9_']*<"
defaultAttributes = [("Normal",NormalTok),("Multiline Comment",CommentTok),("String Constant",StringTok),("Block",NormalTok),("Sig",NormalTok),("Struct",NormalTok),("Object",NormalTok),("ModuleEnv",NormalTok),("ModuleEnv2",NormalTok),("Camlp4 Quotation Constant",StringTok)]
parseRules "Normal" =
(((pDetect2Chars False '(' '*' >>= withAttribute CommentTok) >>~ pushContext "Multiline Comment")
<|>
((pDetectChar False '(' >>= withAttribute NormalTok))
<|>
((pDetectChar False ')' >>= withAttribute NormalTok))
<|>
((pDetectChar False '{' >>= withAttribute NormalTok))
<|>
((pDetectChar False '}' >>= withAttribute NormalTok))
<|>
((pDetect2Chars False '[' '|' >>= withAttribute NormalTok))
<|>
((pDetect2Chars False '|' ']' >>= withAttribute NormalTok))
<|>
((pDetectChar False '[' >>= withAttribute NormalTok))
<|>
((pDetectChar False ']' >>= withAttribute NormalTok))
<|>
((pKeyword " \n\t.():!+,-<=>%&*/;?[]^{|}~\\" list_do >>= withAttribute KeywordTok))
<|>
((pKeyword " \n\t.():!+,-<=>%&*/;?[]^{|}~\\" list_done >>= withAttribute KeywordTok))
<|>
((pKeyword " \n\t.():!+,-<=>%&*/;?[]^{|}~\\" list_module'5fenvironment >>= withAttribute OtherTok) >>~ pushContext "ModuleEnv")
<|>
((pKeyword " \n\t.():!+,-<=>%&*/;?[]^{|}~\\" list_begin >>= withAttribute KeywordTok) >>~ pushContext "Block")
<|>
((pKeyword " \n\t.():!+,-<=>%&*/;?[]^{|}~\\" list_object >>= withAttribute KeywordTok) >>~ pushContext "Object")
<|>
((pKeyword " \n\t.():!+,-<=>%&*/;?[]^{|}~\\" list_sig >>= withAttribute KeywordTok) >>~ pushContext "Sig")
<|>
((pKeyword " \n\t.():!+,-<=>%&*/;?[]^{|}~\\" list_struct >>= withAttribute KeywordTok) >>~ pushContext "Struct")
<|>
((pRegExpr regex_'60'5cs'2a'5bA'2dZa'2dz'5c0300'2d'5c0326'5c0330'2d'5c0366'5c0370'2d'5c0377'5f'5d'5bA'2dZa'2dz'5c0300'2d'5c0326'5c0330'2d'5c0366'5c0370'2d'5c03770'2d9'5f'27'5d'2a >>= withAttribute DataTypeTok))
<|>
((lookAhead (pRegExpr regex_'5bA'2dZ'5d'5bA'2dZa'2dz'5c0300'2d'5c0326'5c0330'2d'5c0366'5c0370'2d'5c03770'2d9'5f'27'5d'2a'5cs'2a'5c'2e) >> pushContext "ModuleEnv2" >> currentContext >>= parseRules))
<|>
((pRegExpr regex_'5bA'2dZ'5d'5bA'2dZa'2dz'5c0300'2d'5c0326'5c0330'2d'5c0366'5c0370'2d'5c03770'2d9'5f'27'5d'2a >>= withAttribute DataTypeTok))
<|>
((pFirstNonSpace >> pRegExpr regex_'23'5bA'2dZa'2dz'5c0300'2d'5c0326'5c0330'2d'5c0366'5c0370'2d'5c0377'5f'5d'5bA'2dZa'2dz'5c0300'2d'5c0326'5c0330'2d'5c0366'5c0370'2d'5c03770'2d9'5f'27'5d'2a'2e'2a'24 >>= withAttribute OtherTok))
<|>
((pDetectChar False '"' >>= withAttribute StringTok) >>~ pushContext "String Constant")
<|>
((pRegExpr regex_'27'28'28'5c'5c'5bntbr'27'22'5c'5c'5d'7c'5c'5c'5b0'2d9'5d'7b3'7d'7c'5c'5cx'5b0'2d9A'2dFa'2df'5d'7b2'7d'29'7c'5b'5e'27'5d'29'27 >>= withAttribute CharTok))
<|>
((pDetect2Chars False '<' '<' >>= withAttribute StringTok) >>~ pushContext "Camlp4 Quotation Constant")
<|>
((pRegExpr regex_'3c'3a'5bA'2dZa'2dz'5c0300'2d'5c0326'5c0330'2d'5c0366'5c0370'2d'5c0377'5f'5d'5bA'2dZa'2dz'5c0300'2d'5c0326'5c0330'2d'5c0366'5c0370'2d'5c03770'2d9'5f'27'5d'2a'3c >>= withAttribute StringTok) >>~ pushContext "Camlp4 Quotation Constant")
<|>
((pKeyword " \n\t.():!+,-<=>%&*/;?[]^{|}~\\" list_keywords >>= withAttribute KeywordTok))
<|>
((pKeyword " \n\t.():!+,-<=>%&*/;?[]^{|}~\\" list_revised_syntax_keywords >>= withAttribute NormalTok))
<|>
((pKeyword " \n\t.():!+,-<=>%&*/;?[]^{|}~\\" list_core_types >>= withAttribute DataTypeTok))
<|>
((pRegExpr regex_'5bA'2dZa'2dz'5c0300'2d'5c0326'5c0330'2d'5c0366'5c0370'2d'5c0377'5f'5d'5bA'2dZa'2dz'5c0300'2d'5c0326'5c0330'2d'5c0366'5c0370'2d'5c03770'2d9'5f'27'5d'2a >>= withAttribute NormalTok))
<|>
((pRegExpr regex_'2d'3f0'5bxX'5d'5b0'2d9A'2dFa'2df'5f'5d'2b >>= withAttribute BaseNTok))
<|>
((pRegExpr regex_'2d'3f0'5boO'5d'5b0'2d7'5f'5d'2b >>= withAttribute BaseNTok))
<|>
((pRegExpr regex_'2d'3f0'5bbB'5d'5b01'5f'5d'2b >>= withAttribute BaseNTok))
<|>
((pRegExpr regex_'2d'3f'5b0'2d9'5d'5b0'2d9'5f'5d'2a'28'28'5c'2e'28'5b0'2d9'5d'5b0'2d9'5f'5d'2a'29'3f'28'5beE'5d'5b'2d'2b'5d'3f'5b0'2d9'5d'5b0'2d9'5f'5d'2a'29'3f'29'7c'28'5beE'5d'5b'2d'2b'5d'3f'5b0'2d9'5d'5b0'2d9'5f'5d'2a'29'29 >>= withAttribute FloatTok))
<|>
((pRegExpr regex_'2d'3f'5b0'2d9'5d'5b0'2d9'5f'5d'2a >>= withAttribute DecValTok)))
parseRules "Multiline Comment" =
(((pDetect2Chars False '*' ')' >>= withAttribute CommentTok) >>~ (popContext))
<|>
((pDetect2Chars False '(' '*' >>= withAttribute CommentTok) >>~ pushContext "Multiline Comment"))
parseRules "String Constant" =
(((pDetectChar False '"' >>= withAttribute StringTok) >>~ (popContext))
<|>
((pRegExpr regex_'28'5c'5c'5bntbr'27'22'5c'5c'5d'7c'5c'5c'5b0'2d9'5d'7b3'7d'7c'5c'5cx'5b0'2d9A'2dFa'2df'5d'7b2'7d'29 >>= withAttribute CharTok))
<|>
((pRegExpr regex_'5c'5c'24 >>= withAttribute CharTok)))
parseRules "Block" =
(((pKeyword " \n\t.():!+,-<=>%&*/;?[]^{|}~\\" list_end >>= withAttribute KeywordTok) >>~ (popContext))
<|>
((parseRules "Normal")))
parseRules "Sig" =
(((pKeyword " \n\t.():!+,-<=>%&*/;?[]^{|}~\\" list_end >>= withAttribute KeywordTok) >>~ (popContext))
<|>
((parseRules "Normal")))
parseRules "Struct" =
(((pKeyword " \n\t.():!+,-<=>%&*/;?[]^{|}~\\" list_end >>= withAttribute KeywordTok) >>~ (popContext))
<|>
((parseRules "Normal")))
parseRules "Object" =
(((pKeyword " \n\t.():!+,-<=>%&*/;?[]^{|}~\\" list_end >>= withAttribute KeywordTok) >>~ (popContext))
<|>
((parseRules "Normal")))
parseRules "ModuleEnv" =
(((pDetectSpaces >>= withAttribute NormalTok))
<|>
((lookAhead (pRegExpr regex_'5bA'2dZ'5d'5bA'2dZa'2dz'5c0300'2d'5c0326'5c0330'2d'5c0366'5c0370'2d'5c03770'2d9'5f'27'5d'2a'5cs'2a'5c'2e) >> pushContext "ModuleEnv2" >> currentContext >>= parseRules))
<|>
((pRegExpr regex_'5bA'2dZ'5d'5bA'2dZa'2dz'5c0300'2d'5c0326'5c0330'2d'5c0366'5c0370'2d'5c03770'2d9'5f'27'5d'2a >>= withAttribute NormalTok) >>~ (popContext))
<|>
((popContext) >> currentContext >>= parseRules))
parseRules "ModuleEnv2" =
(((pRegExpr regex_'5bA'2dZ'5d'5bA'2dZa'2dz'5c0300'2d'5c0326'5c0330'2d'5c0366'5c0370'2d'5c03770'2d9'5f'27'5d'2a >>= withAttribute NormalTok))
<|>
((pDetectSpaces >>= withAttribute NormalTok))
<|>
((pString False "." >>= withAttribute KeywordTok) >>~ (popContext)))
parseRules "Camlp4 Quotation Constant" =
(((pDetect2Chars False '>' '>' >>= withAttribute StringTok) >>~ (popContext))
<|>
((pDetect2Chars False '<' '<' >>= withAttribute StringTok) >>~ pushContext "Camlp4 Quotation Constant")
<|>
((pRegExpr regex_'3c'3a'5bA'2dZa'2dz'5c0300'2d'5c0326'5c0330'2d'5c0366'5c0370'2d'5c0377'5f'5d'5bA'2dZa'2dz'5c0300'2d'5c0326'5c0330'2d'5c0366'5c0370'2d'5c03770'2d9'5f'27'5d'2a'3c >>= withAttribute StringTok) >>~ pushContext "Camlp4 Quotation Constant")
<|>
((pRegExpr regex_'5c'5c'28'5c'5c'7c'3e'3e'7c'3c'3c'29 >>= withAttribute CharTok))
<|>
((pRegExpr regex_'5c'5c'3c'3a'5bA'2dZa'2dz'5c0300'2d'5c0326'5c0330'2d'5c0366'5c0370'2d'5c0377'5f'5d'5bA'2dZa'2dz'5c0300'2d'5c0326'5c0330'2d'5c0366'5c0370'2d'5c03770'2d9'5f'27'5d'2a'3c >>= withAttribute CharTok)))
parseRules "" = parseRules "Normal"
parseRules x = fail $ "Unknown context" ++ x