41663e9 · typeable/pandoc

git.iko.soy/typeable/pandoc

on master help sign in

More refactoring of LaTeX reader code.

41663e9John MacFarlane8 years ago

src/Text/Pandoc/Readers/LaTeX.hs+1−1

compare with master

unifiedsplit

@@ -1486,7 +1486,7 @@ authors = try $ do

1486

1487macroDef :: PandocMonad m => LP m Blocks

1488macroDef =

1489− mempty <$ ((commandDef <|> environmentDef) <* doMacros 0)

1489+ mempty <$ ((commandDef <|> environmentDef) <* doMacros)

1490 where commandDef = do

1491 (name, macro') <- newcommand <|> letmacro <|> defmacro

1492 guardDisabled Ext_latex_macros <|>

src/Text/Pandoc/Readers/LaTeX/Parsing.hs+36−33

@@ -110,6 +110,8 @@ import Text.Pandoc.Readers.LaTeX.Types (ExpansionPoint (..), Macro (..),

110import Text.Pandoc.Shared

111import Text.Parsec.Pos

112

113+-- import Debug.Trace (traceShowId)

114+

113newtype DottedNum = DottedNum [Int]

115newtype DottedNum = DottedNum [Int]

114 deriving (Show)

116 deriving (Show)

115

117

@@ -231,7 +233,7 @@ rawLaTeXParser retokenize parser valParser = do

231 Right toks' -> do

233 Right toks' -> do

232 res <- lift $ runParserT (do when retokenize $ do

234 res <- lift $ runParserT (do when retokenize $ do

233 -- retokenize, applying macros

235 -- retokenize, applying macros

234− doMacros 0

236+ doMacros

235 ts <- many (satisfyTok (const True))

237 ts <- many (satisfyTok (const True))

236 setInput ts

238 setInput ts

237 rawparser)

239 rawparser)

@@ -246,7 +248,7 @@ rawLaTeXParser retokenize parser valParser = do

246applyMacros :: (PandocMonad m, HasMacros s, HasReaderOptions s)

248applyMacros :: (PandocMonad m, HasMacros s, HasReaderOptions s)

247 => String -> ParserT String s m String

249 => String -> ParserT String s m String

248applyMacros s = (guardDisabled Ext_latex_macros >> return s) <|>

250applyMacros s = (guardDisabled Ext_latex_macros >> return s) <|>

249− do let retokenize = doMacros 0 *>

251+ do let retokenize = doMacros *>

250 (toksToString <$> many (satisfyTok (const True)))

252 (toksToString <$> many (satisfyTok (const True)))

251 pstate <- getState

253 pstate <- getState

252 let lstate = def{ sOptions = extractReaderOptions pstate

254 let lstate = def{ sOptions = extractReaderOptions pstate

@@ -371,7 +373,7 @@ satisfyTok :: PandocMonad m => (Tok -> Bool) -> LP m Tok

371satisfyTok f =

373satisfyTok f =

372 try $ do

374 try $ do

373 res <- tokenPrim (T.unpack . untoken) updatePos matcher

375 res <- tokenPrim (T.unpack . untoken) updatePos matcher

374− doMacros 0 -- apply macros on remaining input stream

376+ doMacros -- apply macros on remaining input stream

375 return res

377 return res

376 where matcher t | f t = Just t

378 where matcher t | f t = Just t

377 | otherwise = Nothing

379 | otherwise = Nothing

@@ -379,25 +381,29 @@ satisfyTok f =

379 updatePos _spos _ (Tok pos _ _ : _) = pos

381 updatePos _spos _ (Tok pos _ _ : _) = pos

380 updatePos spos _ [] = incSourceColumn spos 1

382 updatePos spos _ [] = incSourceColumn spos 1

381

383

382−doMacros :: PandocMonad m => Int -> LP m ()

384+doMacros :: PandocMonad m => LP m ()

383−doMacros n = do

385+doMacros = do

384 verbatimMode <- sVerbatimMode <$> getState

386 verbatimMode <- sVerbatimMode <$> getState

385 unless verbatimMode $ do

387 unless verbatimMode $ do

386− inp <- getInput

388+ mbNewInp <- getInput >>= doMacros' 1

389+ case mbNewInp of

390+ Nothing -> return ()

391+ Just inp -> setInput inp

392+

393+doMacros' :: PandocMonad m => Int -> [Tok] -> LP m (Maybe [Tok])

394+doMacros' n inp = do

387− case inp of

395+ case inp of

388− Tok spos (CtrlSeq "begin") _ : Tok _ Symbol "{" :

396+ Tok spos (CtrlSeq "begin") _ : Tok _ Symbol "{" :

389− Tok _ Word name : Tok _ Symbol "}" : ts

397+ Tok _ Word name : Tok _ Symbol "}" : ts

390− -> handleMacros spos name ts

398+ -> handleMacros n spos name ts

391− Tok spos (CtrlSeq "end") _ : Tok _ Symbol "{" :

399+ Tok spos (CtrlSeq "end") _ : Tok _ Symbol "{" :

392− Tok _ Word name : Tok _ Symbol "}" : ts

400+ Tok _ Word name : Tok _ Symbol "}" : ts

393− -> handleMacros spos ("end" <> name) ts

401+ -> handleMacros n spos ("end" <> name) ts

394− Tok _ (CtrlSeq "expandafter") _ : t : ts

402+ Tok _ (CtrlSeq "expandafter") _ : t : ts

395− -> do setInput ts

396− doMacros n

397− getInput >>= setInput . combineTok t

403+ -> (fmap (combineTok t)) <$> doMacros' n ts

398− Tok spos (CtrlSeq name) _ : ts

404+ Tok spos (CtrlSeq name) _ : ts

399− -> handleMacros spos name ts

405+ -> handleMacros n spos name ts

400− _ -> return ()

406+ _ -> return Nothing

401

407

402 where

408 where

403 combineTok (Tok spos (CtrlSeq name) x) (Tok _ Word w : ts)

409 combineTok (Tok spos (CtrlSeq name) x) (Tok _ Word w : ts)

@@ -419,9 +425,7 @@ doMacros n = do

419 getargs argmap rest

425 getargs argmap rest

420 getargs argmap (ArgNum i : Pattern toks : rest) =

426 getargs argmap (ArgNum i : Pattern toks : rest) =

421 try $ do

427 try $ do

422− x <- mconcat <$> manyTill

428+ x <- mconcat <$> manyTill bracedOrToken (matchPattern toks)

423− (braced <|> ((:[]) <$> anyTok))

424− (matchPattern toks)

425 getargs (M.insert i x argmap) rest

429 getargs (M.insert i x argmap) rest

426 getargs argmap (ArgNum i : rest) = do

430 getargs argmap (ArgNum i : rest) = do

427 x <- try $ spaces >> bracedOrToken

431 x <- try $ spaces >> bracedOrToken

@@ -439,10 +443,12 @@ doMacros n = do

439 Tok spos (CtrlSeq x) (txt <> " ") : acc

443 Tok spos (CtrlSeq x) (txt <> " ") : acc

440 addTok _ _ spos t acc = setpos spos t : acc

444 addTok _ _ spos t acc = setpos spos t : acc

441

445

442− handleMacros spos name ts = do

446+ handleMacros n' spos name ts = do

447+ when (n' > 20) -- detect macro expansion loops

448+ $ throwError $ PandocMacroLoop (T.unpack name)

443 macros <- sMacros <$> getState

449 macros <- sMacros <$> getState

444 case M.lookup name macros of

450 case M.lookup name macros of

445− Nothing -> return ()

451+ Nothing -> return Nothing

446 Just (Macro expansionPoint argspecs optarg newtoks) -> do

452 Just (Macro expansionPoint argspecs optarg newtoks) -> do

447 setInput ts

453 setInput ts

448 args <- case optarg of

454 args <- case optarg of

@@ -454,15 +460,12 @@ doMacros n = do

454 -- an argument (in which case we don't want to

460 -- an argument (in which case we don't want to

455 -- expand #1 etc.)

461 -- expand #1 etc.)

456 ts' <- getInput

462 ts' <- getInput

457− setInput $ foldr (addTok False args spos) ts' newtoks

463+ let result = foldr (addTok False args spos) ts' newtoks

458−

459 case expansionPoint of

464 case expansionPoint of

460− ExpandWhenUsed ->

465+ ExpandWhenUsed ->

461− if n > 20 -- detect macro expansion loops

462− then throwError $ PandocMacroLoop (T.unpack name)

463− else doMacros (n + 1)

466+ doMacros' (n' + 1) result >>=

467+ maybe (return (Just result)) (return . Just)

464− ExpandWhenDefined -> return ()

468+ ExpandWhenDefined -> return $ Just result

465−

466

469

467setpos :: SourcePos -> Tok -> Tok

470setpos :: SourcePos -> Tok -> Tok

468setpos spos (Tok _ tt txt) = Tok spos tt txt

471setpos spos (Tok _ tt txt) = Tok spos tt txt