f6cf038 · typeable/pandoc

git.iko.soy/typeable/pandoc

on master help sign in

LaTeX reader efficiency improvements.

f6cf038John MacFarlane6 years ago

src/Text/Pandoc/Readers/LaTeX.hs+42−31

compare with master

unifiedsplit

@@ -191,12 +191,6 @@ inlineCommand = do

191word :: PandocMonad m => LP m Inlines

192word = str . untoken <$> satisfyTok isWordTok

193

194−regularSymbol :: PandocMonad m => LP m Inlines

195−regularSymbol = str . untoken <$> satisfyTok isRegularSymbol

196− where isRegularSymbol (Tok _ Symbol t) = not $ T.any isSpecial t

197− isRegularSymbol _ = False

198− isSpecial c = c `Set.member` specialChars

199−

200inlineGroup :: PandocMonad m => LP m Inlines

194inlineGroup :: PandocMonad m => LP m Inlines

201inlineGroup = do

195inlineGroup = do

202 ils <- grouped inline

196 ils <- grouped inline

@@ -961,31 +955,48 @@ lookupListDefault d = (fromMaybe d .) . lookupList

961 where lookupList l m = msum $ map (`M.lookup` m) l

955 where lookupList l m = msum $ map (`M.lookup` m) l

962

956

963inline :: PandocMonad m => LP m Inlines

957inline :: PandocMonad m => LP m Inlines

958+inline = do

959+ Tok pos toktype t <- lookAhead anyTok

960+ let symbolAsString = str . untoken <$> anySymbol

961+ let unescapedSymbolAsString =

962+ do s <- untoken <$> anySymbol

963+ report $ ParsingUnescaped s pos

964+ return $ str s

965+ case toktype of

964−inline = (mempty <$ comment)

966+ Comment -> mempty <$ comment

965− <|> (space <$ whitespace)

967+ Spaces -> space <$ whitespace

966− <|> (softbreak <$ endline)

968+ Newline -> softbreak <$ endline

967− <|> word

969+ Word -> word

968− <|> macroDef (rawInline "latex")

970+ Esc1 -> str . T.singleton <$> primEscape

969− <|> inlineCommand'

971+ Esc2 -> str . T.singleton <$> primEscape

970− <|> inlineEnvironment

972+ Symbol ->

971− <|> inlineGroup

973+ case t of

972− <|> (symbol '-' *>

974+ "-" -> symbol '-' *>

973− option (str "-") (symbol '-' *>

975+ option (str "-") (symbol '-' *>

974− option (str "–") (str "—" <$ symbol '-')))

976+ option (str "–") (str "—" <$ symbol '-'))

975− <|> doubleQuote

977+ "'" -> symbol '\'' *>

976− <|> singleQuote

977− <|> (str "”" <$ try (symbol '\'' >> symbol '\''))

978+ option (str "’") (str "”" <$ symbol '\'')

978− <|> (str "’" <$ symbol '\'')

979− <|> (str "\160" <$ symbol '~')

979+ "~" -> str "\160" <$ symbol '~'

980+ "`" -> doubleQuote <|> singleQuote <|> symbolAsString

981+ "\"" -> doubleQuote <|> singleQuote <|> symbolAsString

982+ "“" -> doubleQuote <|> symbolAsString

983+ "‘" -> singleQuote <|> symbolAsString

980− <|> dollarsMath

984+ "$" -> dollarsMath <|> unescapedSymbolAsString

981− <|> (guardEnabled Ext_literate_haskell *> symbol '|' *> doLHSverb)

985+ "|" -> (guardEnabled Ext_literate_haskell *>

982− <|> (str . T.singleton <$> primEscape)

986+ symbol '|' *> doLHSverb) <|> symbolAsString

983− <|> regularSymbol

987+ "{" -> inlineGroup

984− <|> (do res <- symbolIn "#^'`\"[]&"

988+ "#" -> unescapedSymbolAsString

985− pos <- getPosition

989+ "&" -> unescapedSymbolAsString

986− let s = untoken res

990+ "_" -> unescapedSymbolAsString

987− report $ ParsingUnescaped s pos

991+ "^" -> unescapedSymbolAsString

988− return $ str s)

992+ "\\" -> mzero

993+ "}" -> mzero

994+ _ -> symbolAsString

995+ CtrlSeq _ -> macroDef (rawInline "latex")

996+ <|> inlineCommand'

997+ <|> inlineEnvironment

998+ <|> inlineGroup

999+ _ -> mzero

989

1000

990inlines :: PandocMonad m => LP m Inlines

1001inlines :: PandocMonad m => LP m Inlines

991inlines = mconcat <$> many inline

1002inlines = mconcat <$> many inline