@@ -1,4 +1,7 @@
11{-# LANGUAGE OverloadedStrings #-}
2+{-# LANGUAGE PatternGuards #-}
3+{-# LANGUAGE FlexibleInstances #-}
4+{-# LANGUAGE MultiParamTypeClasses #-}
25{-# LANGUAGE ScopedTypeVariables #-}
36{-
47Copyright (C) 2006-2017 John MacFarlane <jgm@berkeley.edu>
@@ -28,20 +31,26 @@ Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
2831 Portability : portable
2932
3033Conversion of LaTeX to 'Pandoc' document.
34+
3135-}
3236module Text.Pandoc.Readers.LaTeX ( readLaTeX,
37+ applyMacros,
3338 rawLaTeXInline,
3439 rawLaTeXBlock,
35− inlineCommand,
40+ macro,
41+ inlineCommand
3642 ) where
3743
3844import Control.Applicative (many, optional, (<|>))
3945import Control.Monad
4046import Control.Monad.Except (throwError)
41−import Data.Char (chr, isAlphaNum, isLetter, ord)
42−import Data.Text (Text, unpack)
47+import Data.Char (chr, isAlphaNum, isLetter, ord, isDigit)
48+import Data.Default
49+import Data.Text (Text)
50+import qualified Data.Text as T
4351import Data.List (intercalate, isPrefixOf)
4452import qualified Data.Map as M
53+import qualified Data.Set as Set
4554import Data.Maybe (fromMaybe, maybeToList)
4655import Safe (minimumDef)
4756import System.FilePath (addExtension, replaceExtension, takeExtension)
@@ -52,10 +61,19 @@ import Text.Pandoc.Highlighting (fromListingsLanguage, languagesByExtension)
5261import Text.Pandoc.ImageSize (numUnit, showFl)
5362import Text.Pandoc.Logging
5463import Text.Pandoc.Options
55−import Text.Pandoc.Parsing hiding (many, mathDisplay, mathInline, optional,
56− space, (<|>))
64+import Text.Pandoc.Parsing hiding (many, optional, withRaw,
65+ mathInline, mathDisplay,
66+ space, (<|>), spaces, blankline)
5767import Text.Pandoc.Shared
68+import Text.Pandoc.Readers.LaTeX.Types (Macro(..), Tok(..),
69+ TokType(..), Line, Column)
5870import Text.Pandoc.Walk
71+import Text.Pandoc.Error (PandocError(PandocParsecError, PandocMacroLoop))
72+
73+-- for debugging:
74+-- import Text.Pandoc.Extensions (getDefaultExtensions)
75+-- import Text.Pandoc.Class (runIOorExplode, PandocIO)
76+-- import Debug.Trace (traceShowId)
5977
6078-- | Parse LaTeX from string and return 'Pandoc' document.
6179readLaTeX :: PandocMonad m
@@ -63,18 +81,18 @@ readLaTeX :: PandocMonad m
6381 -> Text -- ^ String to parse (assumes @'\n'@ line endings)
6482 -> m Pandoc
6583readLaTeX opts ltx = do
66− parsed <- readWithM parseLaTeX def{ stateOptions = opts }
67− (unpack (crFilter ltx))
84+ parsed <- runParserT parseLaTeX def{ sOptions = opts } "source"
85+ (tokenize (crFilter ltx))
6886 case parsed of
6987 Right result -> return result
70− Left e -> throwError e
88+ Left e -> throwError $ PandocParsecError (T.unpack ltx) e
7189
7290parseLaTeX :: PandocMonad m => LP m Pandoc
7391parseLaTeX = do
7492 bs <- blocks
7593 eof
7694 st <- getState
77− let meta = stateMeta st
95+ let meta = sMeta st
7896 let doc' = doc bs
7997 let headerLevel (Header n _ _) = [n]
8098 headerLevel _ = []
@@ -88,177 +106,476 @@ parseLaTeX = do
88106 else id) doc'
89107 return $ Pandoc meta bs'
90108
91−type LP m = ParserT String ParserState m
92−
93−anyControlSeq :: PandocMonad m => LP m String
94−anyControlSeq = do
95− char '\\'
96− next <- option '\n' anyChar
97− case next of
98− '\n' -> return ""
99− c | isLetter c -> (c:) <$> (many letter <* optional sp)
100− | otherwise -> return [c]
101−
102−controlSeq :: PandocMonad m => String -> LP m String
103−controlSeq name = try $ do
104− char '\\'
105− case name of
106− "" -> mzero
107− [c] | not (isLetter c) -> string [c]
108− cs -> string cs <* notFollowedBy letter <* optional sp
109− return name
110−
111−dimenarg :: PandocMonad m => LP m String
112−dimenarg = try $ do
113− ch <- option "" $ string "="
114− num <- many1 digit
115− dim <- oneOfStrings ["pt","pc","in","bp","cm","mm","dd","cc","sp"]
116− return $ ch ++ num ++ dim
109+-- testParser :: LP PandocIO a -> Text -> IO a
110+-- testParser p t = do
111+-- res <- runIOorExplode (runParserT p defaultLaTeXState{
112+-- sOptions = def{ readerExtensions =
113+-- enableExtension Ext_raw_tex $
114+-- getDefaultExtensions "latex" }} "source" (tokenize t))
115+-- case res of
116+-- Left e -> error (show e)
117+-- Right r -> return r
118+
119+data LaTeXState = LaTeXState{ sOptions :: ReaderOptions
120+ , sMeta :: Meta
121+ , sQuoteContext :: QuoteContext
122+ , sMacros :: M.Map Text Macro
123+ , sContainers :: [String]
124+ , sHeaders :: M.Map Inlines String
125+ , sLogMessages :: [LogMessage]
126+ , sIdentifiers :: Set.Set String
127+ , sVerbatimMode :: Bool
128+ , sCaption :: Maybe Inlines
129+ , sInListItem :: Bool
130+ , sInTableCell :: Bool
131+ }
132+ deriving Show
133+
134+defaultLaTeXState :: LaTeXState
135+defaultLaTeXState = LaTeXState{ sOptions = def
136+ , sMeta = nullMeta
137+ , sQuoteContext = NoQuote
138+ , sMacros = M.empty
139+ , sContainers = []
140+ , sHeaders = M.empty
141+ , sLogMessages = []
142+ , sIdentifiers = Set.empty
143+ , sVerbatimMode = False
144+ , sCaption = Nothing
145+ , sInListItem = False
146+ , sInTableCell = False
147+ }
148+
149+instance PandocMonad m => HasQuoteContext LaTeXState m where
150+ getQuoteContext = sQuoteContext <$> getState
151+ withQuoteContext context parser = do
152+ oldState <- getState
153+ let oldQuoteContext = sQuoteContext oldState
154+ setState oldState { sQuoteContext = context }
155+ result <- parser
156+ newState <- getState
157+ setState newState { sQuoteContext = oldQuoteContext }
158+ return result
159+
160+instance HasLogMessages LaTeXState where
161+ addLogMessage msg st = st{ sLogMessages = msg : sLogMessages st }
162+ getLogMessages st = reverse $ sLogMessages st
163+
164+instance HasIdentifierList LaTeXState where
165+ extractIdentifierList = sIdentifiers
166+ updateIdentifierList f st = st{ sIdentifiers = f $ sIdentifiers st }
167+
168+instance HasIncludeFiles LaTeXState where
169+ getIncludeFiles = sContainers
170+ addIncludeFile f s = s{ sContainers = f : sContainers s }
171+ dropLatestIncludeFile s = s { sContainers = drop 1 $ sContainers s }
172+
173+instance HasHeaderMap LaTeXState where
174+ extractHeaderMap = sHeaders
175+ updateHeaderMap f st = st{ sHeaders = f $ sHeaders st }
176+
177+instance HasMacros LaTeXState where
178+ extractMacros st = sMacros st
179+ updateMacros f st = st{ sMacros = f (sMacros st) }
180+
181+instance HasReaderOptions LaTeXState where
182+ extractReaderOptions = sOptions
183+
184+instance HasMeta LaTeXState where
185+ setMeta field val st =
186+ st{ sMeta = setMeta field val $ sMeta st }
187+ deleteMeta field st =
188+ st{ sMeta = deleteMeta field $ sMeta st }
189+
190+instance Default LaTeXState where
191+ def = defaultLaTeXState
192+
193+type LP m = ParserT [Tok] LaTeXState m
194+
195+withVerbatimMode :: PandocMonad m => LP m a -> LP m a
196+withVerbatimMode parser = do
197+ updateState $ \st -> st{ sVerbatimMode = True }
198+ result <- parser
199+ updateState $ \st -> st{ sVerbatimMode = False }
200+ return result
201+
202+rawLaTeXBlock :: (PandocMonad m, HasMacros s, HasReaderOptions s)
203+ => ParserT String s m String
204+rawLaTeXBlock = do
205+ lookAhead (try (char '\\' >> letter))
206+ inp <- getInput
207+ let toks = tokenize $ T.pack inp
208+ let rawblock = do
209+ (_, raw) <- try $
210+ withRaw (environment <|> macroDef <|> blockCommand)
211+ return raw
212+ pstate <- getState
213+ let lstate = def{ sOptions = extractReaderOptions pstate }
214+ res <- runParserT rawblock lstate "source" toks
215+ case res of
216+ Left _ -> mzero
217+ Right raw -> count (T.length (untokenize raw)) anyChar
218+
219+macro :: (PandocMonad m, HasMacros s, HasReaderOptions s)
220+ => ParserT String s m Blocks
221+macro = do
222+ guardEnabled Ext_latex_macros
223+ lookAhead (char '\\' *> oneOfStrings ["new", "renew", "provide"] *>
224+ oneOfStrings ["command", "environment"])
225+ inp <- getInput
226+ let toks = tokenize $ T.pack inp
227+ let rawblock = do
228+ (_, raw) <- withRaw $ try macroDef
229+ st <- getState
230+ return (raw, st)
231+ pstate <- getState
232+ let lstate = def{ sOptions = extractReaderOptions pstate
233+ , sMacros = extractMacros pstate }
234+ res <- runParserT rawblock lstate "source" toks
235+ case res of
236+ Left _ -> mzero
237+ Right (raw, st) -> do
238+ updateState (updateMacros (const $ sMacros st))
239+ mempty <$ count (T.length (untokenize raw)) anyChar
240+
241+applyMacros :: (PandocMonad m, HasMacros s, HasReaderOptions s)
242+ => String -> ParserT String s m String
243+applyMacros s = do
244+ (guardEnabled Ext_latex_macros >>
245+ do let retokenize = doMacros 0 *> (toksToString <$> getInput)
246+ pstate <- getState
247+ let lstate = def{ sOptions = extractReaderOptions pstate
248+ , sMacros = extractMacros pstate }
249+ res <- runParserT retokenize lstate "math" (tokenize (T.pack s))
250+ case res of
251+ Left e -> fail (show e)
252+ Right s' -> return s') <|> return s
253+
254+rawLaTeXInline :: (PandocMonad m, HasMacros s, HasReaderOptions s)
255+ => ParserT String s m String
256+rawLaTeXInline = do
257+ lookAhead (try (char '\\' >> letter) <|> char '$')
258+ inp <- getInput
259+ let toks = tokenize $ T.pack inp
260+ let rawinline = do
261+ (_, raw) <- try $ withRaw (inlineEnvironment <|> inlineCommand')
262+ st <- getState
263+ return (raw, st)
264+ pstate <- getState
265+ let lstate = def{ sOptions = extractReaderOptions pstate
266+ , sMacros = extractMacros pstate }
267+ res <- runParserT rawinline lstate "source" toks
268+ case res of
269+ Left _ -> mzero
270+ Right (raw, s) -> do
271+ updateState $ updateMacros (const $ sMacros s)
272+ count (T.length (untokenize raw)) anyChar
273+
274+inlineCommand :: PandocMonad m => ParserT String ParserState m Inlines
275+inlineCommand = do
276+ lookAhead (try (char '\\' >> letter) <|> char '$')
277+ inp <- getInput
278+ let toks = tokenize $ T.pack inp
279+ let rawinline = do
280+ (il, raw) <- try $ withRaw (inlineEnvironment <|> inlineCommand')
281+ st <- getState
282+ return (il, raw, st)
283+ pstate <- getState
284+ let lstate = def{ sOptions = extractReaderOptions pstate
285+ , sMacros = extractMacros pstate }
286+ res <- runParserT rawinline lstate "source" toks
287+ case res of
288+ Left _ -> mzero
289+ Right (il, raw, s) -> do
290+ updateState $ updateMacros (const $ sMacros s)
291+ count (T.length (untokenize raw)) anyChar
292+ return il
293+
294+tokenize :: Text -> [Tok]
295+tokenize = totoks (1, 1)
296+
297+totoks :: (Line, Column) -> Text -> [Tok]
298+totoks (lin,col) t =
299+ case T.uncons t of
300+ Nothing -> []
301+ Just (c, rest)
302+ | c == '\n' ->
303+ Tok (lin, col) Newline "\n"
304+ : totoks (lin + 1,1) rest
305+ | isSpaceOrTab c ->
306+ let (sps, rest') = T.span isSpaceOrTab t
307+ in Tok (lin, col) Spaces sps
308+ : totoks (lin, col + T.length sps) rest'
309+ | isAlphaNum c ->
310+ let (ws, rest') = T.span isAlphaNum t
311+ in Tok (lin, col) Word ws
312+ : totoks (lin, col + T.length ws) rest'
313+ | c == '%' ->
314+ let (cs, rest') = T.break (== '\n') rest
315+ in Tok (lin, col) Comment ("%" <> cs)
316+ : totoks (lin, col + 1 + T.length cs) rest'
317+ | c == '\\' ->
318+ case T.uncons rest of
319+ Nothing -> [Tok (lin, col) Symbol (T.singleton c)]
320+ Just (d, rest')
321+ | isLetter d ->
322+ let (ws, rest'') = T.span isLetter rest
323+ (ss, rest''') = T.span isSpaceOrTab rest''
324+ in Tok (lin, col) (CtrlSeq ws) ("\\" <> ws <> ss)
325+ : totoks (lin,
326+ col + 1 + T.length ws + T.length ss) rest'''
327+ | d == '\t' || d == '\n' ->
328+ Tok (lin, col) Symbol ("\\")
329+ : totoks (lin, col + 1) rest
330+ | otherwise ->
331+ Tok (lin, col) (CtrlSeq (T.singleton d)) (T.pack [c,d])
332+ : totoks (lin, col + 2) rest'
333+ | c == '#' ->
334+ let (t1, t2) = T.span (\d -> d >= '0' && d <= '9') rest
335+ in case safeRead (T.unpack t1) of
336+ Just i ->
337+ Tok (lin, col) (Arg i) ("#" <> t1)
338+ : totoks (lin, col + 1 + T.length t1) t2
339+ Nothing ->
340+ Tok (lin, col) Symbol ("#")
341+ : totoks (lin, col + 1) t2
342+ | c == '^' ->
343+ case T.uncons rest of
344+ Just ('^', rest') ->
345+ case T.uncons rest' of
346+ Just (d, rest'')
347+ | isLowerHex d ->
348+ case T.uncons rest'' of
349+ Just (e, rest''') | isLowerHex e ->
350+ Tok (lin, col) Esc2 (T.pack ['^','^',d,e])
351+ : totoks (lin, col + 4) rest'''
352+ _ ->
353+ Tok (lin, col) Esc1 (T.pack ['^','^',d])
354+ : totoks (lin, col + 3) rest''
355+ | d < '\128' ->
356+ Tok (lin, col) Esc1 (T.pack ['^','^',d])
357+ : totoks (lin, col + 3) rest''
358+ _ -> [Tok (lin, col) Symbol ("^"),
359+ Tok (lin, col + 1) Symbol ("^")]
360+ _ -> Tok (lin, col) Symbol ("^")
361+ : totoks (lin, col + 1) rest
362+ | otherwise ->
363+ Tok (lin, col) Symbol (T.singleton c) : totoks (lin, col + 1) rest
364+
365+ where isSpaceOrTab ' ' = True
366+ isSpaceOrTab '\t' = True
367+ isSpaceOrTab _ = False
368+
369+isLowerHex :: Char -> Bool
370+isLowerHex x = x >= '0' && x <= '9' || x >= 'a' && x <= 'f'
371+
372+untokenize :: [Tok] -> Text
373+untokenize = mconcat . map untoken
374+
375+untoken :: Tok -> Text
376+untoken (Tok _ _ t) = t
377+
378+satisfyTok :: PandocMonad m => (Tok -> Bool) -> LP m Tok
379+satisfyTok f =
380+ try $ do
381+ res <- tokenPrim (T.unpack . untoken) updatePos matcher
382+ doMacros 0 -- apply macros on remaining input stream
383+ return res
384+ where matcher t | f t = Just t
385+ | otherwise = Nothing
386+ updatePos :: SourcePos -> Tok -> [Tok] -> SourcePos
387+ updatePos spos _ (Tok (lin,col) _ _ : _) =
388+ setSourceColumn (setSourceLine spos lin) col
389+ updatePos spos _ [] = spos
390+
391+doMacros :: PandocMonad m => Int -> LP m ()
392+doMacros n = do
393+ verbatimMode <- sVerbatimMode <$> getState
394+ when (not verbatimMode) $ do
395+ inp <- getInput
396+ case inp of
397+ Tok spos (CtrlSeq "begin") _ : Tok _ Symbol "{" :
398+ Tok _ Word name : Tok _ Symbol "}" : ts
399+ -> handleMacros spos name ts
400+ Tok spos (CtrlSeq "end") _ : Tok _ Symbol "{" :
401+ Tok _ Word name : Tok _ Symbol "}" : ts
402+ -> handleMacros spos ("end" <> name) ts
403+ Tok spos (CtrlSeq name) _ : ts
404+ -> handleMacros spos name ts
405+ _ -> return ()
406+ where handleMacros spos name ts = do
407+ macros <- sMacros <$> getState
408+ case M.lookup name macros of
409+ Nothing -> return ()
410+ Just (Macro numargs optarg newtoks) -> do
411+ setInput ts
412+ let getarg = spaces >> braced
413+ args <- case optarg of
414+ Nothing -> count numargs getarg
415+ Just o ->
416+ (:) <$> option o bracketedToks
417+ <*> count (numargs - 1) getarg
418+ let addTok (Tok _ (Arg i) _) acc | i > 0
419+ , i <= numargs =
420+ map (setpos spos) (args !! (i - 1)) ++ acc
421+ addTok t acc = setpos spos t : acc
422+ ts' <- getInput
423+ setInput $ foldr addTok ts' newtoks
424+ if n > 20 -- detect macro expansion loops
425+ then throwError $ PandocMacroLoop (T.unpack name)
426+ else doMacros (n + 1)
427+
428+setpos :: (Line, Column) -> Tok -> Tok
429+setpos spos (Tok _ tt txt) = Tok spos tt txt
430+
431+anyControlSeq :: PandocMonad m => LP m Tok
432+anyControlSeq = satisfyTok isCtrlSeq
433+ where isCtrlSeq (Tok _ (CtrlSeq _) _) = True
434+ isCtrlSeq _ = False
435+
436+anySymbol :: PandocMonad m => LP m Tok
437+anySymbol = satisfyTok isSym
438+ where isSym (Tok _ Symbol _) = True
439+ isSym _ = False
440+
441+spaces :: PandocMonad m => LP m ()
442+spaces = skipMany (satisfyTok (tokTypeIn [Comment, Spaces, Newline]))
443+
444+spaces1 :: PandocMonad m => LP m ()
445+spaces1 = skipMany1 (satisfyTok (tokTypeIn [Comment, Spaces, Newline]))
446+
447+tokTypeIn :: [TokType] -> Tok -> Bool
448+tokTypeIn toktypes (Tok _ tt _) = tt `elem` toktypes
449+
450+controlSeq :: PandocMonad m => Text -> LP m Tok
451+controlSeq name = satisfyTok isNamed
452+ where isNamed (Tok _ (CtrlSeq n) _) = n == name
453+ isNamed _ = False
454+
455+symbol :: PandocMonad m => Char -> LP m Tok
456+symbol c = satisfyTok isc
457+ where isc (Tok _ Symbol d) = case T.uncons d of
458+ Just (c',_) -> c == c'
459+ _ -> False
460+ isc _ = False
461+
462+symbolIn :: PandocMonad m => [Char] -> LP m Tok
463+symbolIn cs = satisfyTok isInCs
464+ where isInCs (Tok _ Symbol d) = case T.uncons d of
465+ Just (c,_) -> c `elem` cs
466+ _ -> False
467+ isInCs _ = False
117468
118469sp :: PandocMonad m => LP m ()
119470sp = whitespace <|> endline
120471
121472whitespace :: PandocMonad m => LP m ()
122−whitespace = skipMany1 $ satisfy (\c -> c == ' ' || c == '\t')
473+whitespace = () <$ satisfyTok isSpaceTok
474+ where isSpaceTok (Tok _ Spaces _) = True
475+ isSpaceTok _ = False
123476
124−endline :: PandocMonad m => LP m ()
125−endline = try (newline >> lookAhead anyChar >> notFollowedBy blankline)
126−
127−isLowerHex :: Char -> Bool
128−isLowerHex x = x >= '0' && x <= '9' || x >= 'a' && x <= 'f'
477+newlineTok :: PandocMonad m => LP m ()
478+newlineTok = () <$ satisfyTok isNewlineTok
129479
130−tildeEscape :: PandocMonad m => LP m Char
131−tildeEscape = try $ do
132− string "^^"
133− c <- satisfy (\x -> x >= '\0' && x <= '\128')
134− d <- if isLowerHex c
135− then option "" $ count 1 (satisfy isLowerHex)
136− else return ""
137− if null d
138− then case ord c of
139− x | x >= 64 && x <= 127 -> return $ chr (x - 64)
140− | otherwise -> return $ chr (x + 64)
141− else return $ chr $ read ('0':'x':c:d)
480+isNewlineTok :: Tok -> Bool
481+isNewlineTok (Tok _ Newline _) = True
482+isNewlineTok _ = False
142483
143484comment :: PandocMonad m => LP m ()
144−comment = do
145− char '%'
146− skipMany (satisfy (/='\n'))
147− optional newline
148− return ()
485+comment = () <$ satisfyTok isCommentTok
486+ where isCommentTok (Tok _ Comment _) = True
487+ isCommentTok _ = False
488+
489+anyTok :: PandocMonad m => LP m Tok
490+anyTok = satisfyTok (const True)
149491
150−bgroup :: PandocMonad m => LP m ()
492+endline :: PandocMonad m => LP m ()
493+endline = try $ do
494+ newlineTok
495+ lookAhead anyTok
496+ notFollowedBy blankline
497+
498+blankline :: PandocMonad m => LP m ()
499+blankline = try $ skipMany whitespace *> newlineTok
500+
501+primEscape :: PandocMonad m => LP m Char
502+primEscape = do
503+ Tok _ toktype t <- satisfyTok (tokTypeIn [Esc1, Esc2])
504+ case toktype of
505+ Esc1 -> case T.uncons (T.drop 2 t) of
506+ Just (c, _)
507+ | c >= '\64' && c <= '\127' -> return (chr (ord c - 64))
508+ | otherwise -> return (chr (ord c + 64))
509+ Nothing -> fail "Empty content of Esc1"
510+ Esc2 -> case safeRead ('0':'x':T.unpack (T.drop 2 t)) of
511+ Just x -> return (chr x)
512+ Nothing -> fail $ "Could not read: " ++ T.unpack t
513+ _ -> fail "Expected an Esc1 or Esc2 token" -- should not happen
514+
515+bgroup :: PandocMonad m => LP m Tok
151516bgroup = try $ do
152− skipMany (spaceChar <|> try (newline <* notFollowedBy blankline))
153− () <$ char '{'
154− <|> () <$ controlSeq "bgroup"
155− <|> () <$ controlSeq "begingroup"
517+ skipMany sp
518+ symbol '{' <|> controlSeq "bgroup" <|> controlSeq "begingroup"
156519
157−egroup :: PandocMonad m => LP m ()
158−egroup = () <$ char '}'
159− <|> () <$ controlSeq "egroup"
160− <|> () <$ controlSeq "endgroup"
520+egroup :: PandocMonad m => LP m Tok
521+egroup = (symbol '}' <|> controlSeq "egroup" <|> controlSeq "endgroup")
161522
162−grouped :: PandocMonad m => Monoid a => LP m a -> LP m a
523+grouped :: (PandocMonad m, Monoid a) => LP m a -> LP m a
163524grouped parser = try $ do
164525 bgroup
165526 -- first we check for an inner 'grouped', because
166527 -- {{a,b}} should be parsed the same as {a,b}
167− try (grouped parser <* egroup)
168− <|> (mconcat <$> manyTill parser egroup)
169−
170−braced :: PandocMonad m => LP m String
171−braced = grouped chunk
172− where chunk =
173− many1 (satisfy (\c -> c /= '\\' && c /= '}' && c /= '{'))
174− <|> try (string "\\}")
175− <|> try (string "\\{")
176− <|> try (string "\\\\")
177− <|> ((\x -> "{" ++ x ++ "}") <$> braced)
178− <|> count 1 anyChar
528+ try (grouped parser <* egroup) <|> (mconcat <$> manyTill parser egroup)
529+
530+braced :: PandocMonad m => LP m [Tok]
531+braced = bgroup *> braced' 1
532+ where braced' (n :: Int) =
533+ handleEgroup n <|> handleBgroup n <|> handleOther n
534+ handleEgroup n = do
535+ t <- egroup
536+ if n == 1
537+ then return []
538+ else (t:) <$> braced' (n - 1)
539+ handleBgroup n = do
540+ t <- bgroup
541+ (t:) <$> braced' (n + 1)
542+ handleOther n = do
543+ t <- anyTok
544+ (t:) <$> braced' n
179545
180546bracketed :: PandocMonad m => Monoid a => LP m a -> LP m a
181−bracketed parser = try $ char '[' *> (mconcat <$> manyTill parser (char ']'))
182−
183−mathDisplay :: PandocMonad m => LP m String -> LP m Inlines
184−mathDisplay p = displayMath <$> (try p >>= applyMacros' . trim)
185−
186−mathInline :: PandocMonad m => LP m String -> LP m Inlines
187−mathInline p = math <$> (try p >>= applyMacros')
188−
189−mathChars :: PandocMonad m => LP m String
190−mathChars =
191− concat <$> many (escapedChar
192− <|> (snd <$> withRaw braced)
193− <|> many1 (satisfy isOrdChar))
194− where escapedChar = try $ do char '\\'
195− c <- anyChar
196− return ['\\',c]
197− isOrdChar '$' = False
198− isOrdChar '{' = False
199− isOrdChar '}' = False
200− isOrdChar '\\' = False
201− isOrdChar _ = True
202−
203−quoted' :: PandocMonad m => (Inlines -> Inlines) -> LP m String -> LP m () -> LP m Inlines
204−quoted' f starter ender = do
205− startchs <- starter
206− smart <- extensionEnabled Ext_smart <$> getOption readerExtensions
207− if smart
208− then do
209− ils <- many (notFollowedBy ender >> inline)
210− (ender >> return (f (mconcat ils))) <|>
211− (<> mconcat ils) <$>
212− lit (case startchs of
213− "``" -> "“"
214− "`" -> "‘"
215− _ -> startchs)
216− else lit startchs
547+bracketed parser = try $ do
548+ symbol '['
549+ mconcat <$> manyTill parser (symbol ']')
217550
218−doubleQuote :: PandocMonad m => LP m Inlines
219−doubleQuote = do
220− quoted' doubleQuoted (try $ string "``") (void $ try $ string "''")
221− <|> quoted' doubleQuoted (string "“") (void $ char '”')
222− -- the following is used by babel for localized quotes:
223− <|> quoted' doubleQuoted (try $ string "\"`") (void $ try $ string "\"'")
224− <|> quoted' doubleQuoted (string "\"") (void $ char '"')
551+dimenarg :: PandocMonad m => LP m Text
552+dimenarg = try $ do
553+ ch <- option False $ True <$ symbol '='
554+ Tok _ _ s <- satisfyTok isWordTok
555+ guard $ (T.take 2 (T.reverse s)) `elem`
556+ ["pt","pc","in","bp","cm","mm","dd","cc","sp"]
557+ let num = T.take (T.length s - 2) s
558+ guard $ T.length num > 0
559+ guard $ T.all isDigit num
560+ return $ T.pack ['=' | ch] <> s
225561
226−singleQuote :: PandocMonad m => LP m Inlines
227−singleQuote = do
228− smart <- extensionEnabled Ext_smart <$> getOption readerExtensions
229− if smart
230− then quoted' singleQuoted (string "`") (try $ char '\'' >> notFollowedBy letter)
231− <|> quoted' singleQuoted (string "‘") (try $ char '’' >> notFollowedBy letter)
232− else str <$> many1 (oneOf "`\'‘’")
562+-- inline elements:
233563
234−inline :: PandocMonad m => LP m Inlines
235−inline = (mempty <$ comment)
236− <|> (space <$ whitespace)
237− <|> (softbreak <$ endline)
238− <|> inlineText
239− <|> inlineCommand
240− <|> inlineEnvironment
241− <|> inlineGroup
242− <|> (char '-' *> option (str "-")
243− (char '-' *> option (str "–") (str "—" <$ char '-')))
244− <|> doubleQuote
245− <|> singleQuote
246− <|> (str "”" <$ try (string "''"))
247− <|> (str "”" <$ char '”')
248− <|> (str "’" <$ char '\'')
249− <|> (str "’" <$ char '’')
250− <|> (str "\160" <$ char '~')
251− <|> mathDisplay (string "$$" *> mathChars <* string "$$")
252− <|> mathInline (char '$' *> mathChars <* char '$')
253− <|> (guardEnabled Ext_literate_haskell *> char '|' *> doLHSverb)
254− <|> (str . (:[]) <$> tildeEscape)
255− <|> (do res <- oneOf "#&~^'`\"[]"
256− pos <- getPosition
257− report $ ParsingUnescaped [res] pos
258− return $ str [res])
564+word :: PandocMonad m => LP m Inlines
565+word = (str . T.unpack . untoken) <$> satisfyTok isWordTok
259566
260−inlines :: PandocMonad m => LP m Inlines
261−inlines = mconcat <$> many (notFollowedBy (char '}') *> inline)
567+regularSymbol :: PandocMonad m => LP m Inlines
568+regularSymbol = (str . T.unpack . untoken) <$> satisfyTok isRegularSymbol
569+ where isRegularSymbol (Tok _ Symbol t) = not $ T.any isSpecial t
570+ isRegularSymbol _ = False
571+ isSpecial c = c `Set.member` specialChars
572+
573+specialChars :: Set.Set Char
574+specialChars = Set.fromList "#$%&~_^\\{}"
575+
576+isWordTok :: Tok -> Bool
577+isWordTok (Tok _ Word _) = True
578+isWordTok _ = False
262579
263580inlineGroup :: PandocMonad m => LP m Inlines
264581inlineGroup = do
@@ -269,233 +586,564 @@ inlineGroup = do
269586 -- we need the span so we can detitlecase bibtex entries;
270587 -- we need to know when something is {C}apitalized
271588
272−block :: PandocMonad m => LP m Blocks
273−block = (mempty <$ comment)
274− <|> (mempty <$ ((spaceChar <|> newline) *> spaces))
275− <|> environment
276− <|> include
277− <|> macro
278− <|> blockCommand
279− <|> paragraph
280− <|> grouped block
281−
282−blocks :: PandocMonad m => LP m Blocks
283−blocks = mconcat <$> many block
589+doLHSverb :: PandocMonad m => LP m Inlines
590+doLHSverb =
591+ (codeWith ("",["haskell"],[]) . T.unpack . untokenize)
592+ <$> manyTill (satisfyTok (not . isNewlineTok)) (symbol '|')
284593
285−getRawCommand :: PandocMonad m => String -> LP m String
286−getRawCommand name' = do
287− rawargs <- withRaw (many (try (optional sp *> opt)) *>
288− option "" (try (optional sp *> dimenarg)) *>
289− many braced)
290− return $ '\\' : name' ++ snd rawargs
594+mkImage :: PandocMonad m => [(String, String)] -> String -> LP m Inlines
595+mkImage options src = do
596+ let replaceTextwidth (k,v) =
597+ case numUnit v of
598+ Just (num, "\\textwidth") -> (k, showFl (num * 100) ++ "%")
599+ _ -> (k, v)
600+ let kvs = map replaceTextwidth
601+ $ filter (\(k,_) -> k `elem` ["width", "height"]) options
602+ let attr = ("",[], kvs)
603+ let alt = str "image"
604+ case takeExtension src of
605+ "" -> do
606+ defaultExt <- getOption readerDefaultImageExtension
607+ return $ imageWith attr (addExtension src defaultExt) "" alt
608+ _ -> return $ imageWith attr src "" alt
291609
292−lookupListDefault :: (Ord k) => v -> [k] -> M.Map k v -> v
293−lookupListDefault d = (fromMaybe d .) . lookupList
294− where
295− lookupList l m = msum $ map (`M.lookup` m) l
610+-- converts e.g. \SI{1}[\$]{} to "$ 1" or \SI{1}{\euro} to "1 €"
611+dosiunitx :: PandocMonad m => LP m Inlines
612+dosiunitx = do
613+ skipopts
614+ value <- tok
615+ valueprefix <- option "" $ bracketed tok
616+ unit <- tok
617+ let emptyOr160 "" = ""
618+ emptyOr160 _ = "\160"
619+ return . mconcat $ [valueprefix,
620+ emptyOr160 valueprefix,
621+ value,
622+ emptyOr160 unit,
623+ unit]
296624
297−blockCommand :: PandocMonad m => LP m Blocks
298−blockCommand = try $ do
299− name <- anyControlSeq
300− guard $ name /= "begin" && name /= "end"
301− star <- option "" (string "*" <* optional sp)
302− let name' = name ++ star
303− let raw = do
304− rawcommand <- getRawCommand name'
305− transformed <- applyMacros' rawcommand
306− guard $ transformed /= rawcommand
307− notFollowedBy $ parseFromString' inlines transformed
308− parseFromString' blocks transformed
309− lookupListDefault raw [name',name] blockCommands
625+lit :: String -> LP m Inlines
626+lit = pure . str
310627
311−inBrackets :: Inlines -> Inlines
312−inBrackets x = str "[" <> x <> str "]"
628+removeDoubleQuotes :: Text -> Text
629+removeDoubleQuotes t =
630+ maybe t id $ T.stripPrefix "\"" t >>= T.stripSuffix "\""
313631
314−-- eat an optional argument and one or more arguments in braces
315−ignoreInlines :: PandocMonad m => String -> (String, LP m Inlines)
316−ignoreInlines name = (name, p)
317− where
318− p = do oa <- optargs
319− let rawCommand = '\\':name ++ oa
320− let doraw = guardRaw >> return (rawInline "latex" rawCommand)
321− doraw <|> ignore rawCommand
632+doubleQuote :: PandocMonad m => LP m Inlines
633+doubleQuote = do
634+ quoted' doubleQuoted (try $ count 2 $ symbol '`')
635+ (void $ try $ count 2 $ symbol '\'')
636+ <|> quoted' doubleQuoted ((:[]) <$> symbol '“') (void $ symbol '”')
637+ -- the following is used by babel for localized quotes:
638+ <|> quoted' doubleQuoted (try $ sequence [symbol '"', symbol '`'])
639+ (void $ try $ sequence [symbol '"', symbol '\''])
640+ <|> quoted' doubleQuoted ((:[]) <$> symbol '"')
641+ (void $ symbol '"')
322642
323−guardRaw :: PandocMonad m => LP m ()
324−guardRaw = getOption readerExtensions >>= guard . extensionEnabled Ext_raw_tex
643+singleQuote :: PandocMonad m => LP m Inlines
644+singleQuote = do
645+ quoted' singleQuoted ((:[]) <$> symbol '`')
646+ (try $ symbol '\'' >>
647+ notFollowedBy (satisfyTok startsWithLetter))
648+ <|> quoted' singleQuoted ((:[]) <$> symbol '‘')
649+ (try $ symbol '’' >>
650+ notFollowedBy (satisfyTok startsWithLetter))
651+ where startsWithLetter (Tok _ Word t) =
652+ case T.uncons t of
653+ Just (c, _) | isLetter c -> True
654+ _ -> False
655+ startsWithLetter _ = False
656+
657+quoted' :: PandocMonad m
658+ => (Inlines -> Inlines)
659+ -> LP m [Tok]
660+ -> LP m ()
661+ -> LP m Inlines
662+quoted' f starter ender = do
663+ startchs <- (T.unpack . untokenize) <$> starter
664+ smart <- extensionEnabled Ext_smart <$> getOption readerExtensions
665+ if smart
666+ then do
667+ ils <- many (notFollowedBy ender >> inline)
668+ (ender >> return (f (mconcat ils))) <|>
669+ (<> mconcat ils) <$>
670+ lit (case startchs of
671+ "``" -> "“"
672+ "`" -> "‘"
673+ cs -> cs)
674+ else lit startchs
325675
326−optargs :: PandocMonad m => LP m String
327−optargs = snd <$> withRaw (skipopts *> skipMany (try $ optional sp *> braced))
676+enquote :: PandocMonad m => LP m Inlines
677+enquote = do
678+ skipopts
679+ quoteContext <- sQuoteContext <$> getState
680+ if quoteContext == InDoubleQuote
681+ then singleQuoted <$> withQuoteContext InSingleQuote tok
682+ else doubleQuoted <$> withQuoteContext InDoubleQuote tok
328683
329−ignore :: (Monoid a, PandocMonad m) => String -> ParserT s u m a
330−ignore raw = do
331− pos <- getPosition
332− report $ SkippedContent raw pos
333− return mempty
684+doverb :: PandocMonad m => LP m Inlines
685+doverb = do
686+ Tok _ Symbol t <- anySymbol
687+ marker <- case T.uncons t of
688+ Just (c, ts) | T.null ts -> return c
689+ _ -> mzero
690+ withVerbatimMode $
691+ (code . T.unpack . untokenize) <$>
692+ manyTill (verbTok marker) (symbol marker)
693+
694+verbTok :: PandocMonad m => Char -> LP m Tok
695+verbTok stopchar = do
696+ t@(Tok (lin, col) toktype txt) <- satisfyTok (not . isNewlineTok)
697+ case T.findIndex (== stopchar) txt of
698+ Nothing -> return t
699+ Just i -> do
700+ let (t1, t2) = T.splitAt i txt
701+ inp <- getInput
702+ setInput $ Tok (lin, col + i) Symbol (T.singleton stopchar)
703+ : (totoks (lin, col + i + 1) (T.drop 1 t2)) ++ inp
704+ return $ Tok (lin, col) toktype t1
334705
335−ignoreBlocks :: PandocMonad m => String -> (String, LP m Blocks)
336−ignoreBlocks name = (name, p)
337− where
338− p = do oa <- optargs
339− let rawCommand = '\\':name ++ oa
340− let doraw = guardRaw >> return (rawBlock "latex" rawCommand)
341− doraw <|> ignore rawCommand
706+dolstinline :: PandocMonad m => LP m Inlines
707+dolstinline = do
708+ options <- option [] keyvals
709+ let classes = maybeToList $ lookup "language" options >>= fromListingsLanguage
710+ Tok _ Symbol t <- anySymbol
711+ marker <- case T.uncons t of
712+ Just (c, ts) | T.null ts -> return c
713+ _ -> mzero
714+ let stopchar = if marker == '{' then '}' else marker
715+ withVerbatimMode $
716+ (codeWith ("",classes,[]) . T.unpack . untokenize) <$>
717+ manyTill (verbTok stopchar) (symbol stopchar)
342718
343−blockCommands :: PandocMonad m => M.Map String (LP m Blocks)
344−blockCommands = M.fromList $
345− [ ("par", mempty <$ skipopts)
346− , ("parbox", braced >> grouped blocks)
347− , ("title", mempty <$ (skipopts *>
348− (grouped inline >>= addMeta "title")
349− <|> (grouped block >>= addMeta "title")))
350− , ("subtitle", mempty <$ (skipopts *> tok >>= addMeta "subtitle"))
351− , ("author", mempty <$ (skipopts *> authors))
352− -- -- in letter class, temp. store address & sig as title, author
353− , ("address", mempty <$ (skipopts *> tok >>= addMeta "address"))
354− , ("signature", mempty <$ (skipopts *> authors))
355− , ("date", mempty <$ (skipopts *> tok >>= addMeta "date"))
356− -- Koma-script metadata commands
357− , ("dedication", mempty <$ (skipopts *> tok >>= addMeta "dedication"))
358− -- sectioning
359− , ("part", section nullAttr (-1))
360− , ("part*", section nullAttr (-1))
361− , ("chapter", section nullAttr 0)
362− , ("chapter*", section ("",["unnumbered"],[]) 0)
363− , ("section", section nullAttr 1)
364− , ("section*", section ("",["unnumbered"],[]) 1)
365− , ("subsection", section nullAttr 2)
366− , ("subsection*", section ("",["unnumbered"],[]) 2)
367− , ("subsubsection", section nullAttr 3)
368− , ("subsubsection*", section ("",["unnumbered"],[]) 3)
369− , ("paragraph", section nullAttr 4)
370− , ("paragraph*", section ("",["unnumbered"],[]) 4)
371− , ("subparagraph", section nullAttr 5)
372− , ("subparagraph*", section ("",["unnumbered"],[]) 5)
373− -- beamer slides
374− , ("frametitle", section nullAttr 3)
375− , ("framesubtitle", section nullAttr 4)
376− -- letters
377− , ("opening", (para . trimInlines) <$> (skipopts *> tok))
378− , ("closing", skipopts *> closing)
379− --
380− , ("hrule", pure horizontalRule)
381− , ("strut", pure mempty)
382− , ("rule", skipopts *> tok *> tok *> pure horizontalRule)
383− , ("item", skipopts *> looseItem)
384− , ("documentclass", skipopts *> braced *> preamble)
385− , ("centerline", (para . trimInlines) <$> (skipopts *> tok))
386− , ("caption", skipopts *> setCaption)
387− , ("bibliography", mempty <$ (skipopts *> braced >>=
388− addMeta "bibliography" . splitBibs))
389− , ("addbibresource", mempty <$ (skipopts *> braced >>=
390− addMeta "bibliography" . splitBibs))
391− -- includes
392− , ("lstinputlisting", inputListing)
393− , ("graphicspath", graphicsPath)
394− -- hyperlink
395− , ("hypertarget", braced >> grouped block)
396− ] ++ map ignoreBlocks
397− -- these commands will be ignored unless --parse-raw is specified,
398− -- in which case they will appear as raw latex blocks
399− [ "newcommand", "renewcommand", "newenvironment", "renewenvironment"
400− -- newcommand, etc. should be parsed by macro, but we need this
401− -- here so these aren't parsed as inline commands to ignore
402− , "special", "pdfannot", "pdfstringdef"
403− , "bibliographystyle"
404− , "maketitle", "makeindex", "makeglossary"
405− , "addcontentsline", "addtocontents", "addtocounter"
406− -- \ignore{} is used conventionally in literate haskell for definitions
407− -- that are to be processed by the compiler but not printed.
408− , "ignore"
409− , "hyperdef"
410− , "markboth", "markright", "markleft"
411− , "hspace", "vspace"
412− , "newpage"
413− , "clearpage"
414− , "pagebreak"
415− ]
719+keyval :: PandocMonad m => LP m (String, String)
720+keyval = try $ do
721+ Tok _ Word key <- satisfyTok isWordTok
722+ let isSpecSym (Tok _ Symbol t) = t `elem` [".",":","-","|","\\"]
723+ isSpecSym _ = False
724+ val <- option [] $ do
725+ symbol '='
726+ braced <|> (many1 (satisfyTok isWordTok <|> satisfyTok isSpecSym
727+ <|> anyControlSeq))
728+ optional sp
729+ optional (symbol ',')
730+ optional sp
731+ return (T.unpack key, T.unpack . untokenize $ val)
416732
417−graphicsPath :: PandocMonad m => LP m Blocks
418−graphicsPath = do
419− ps <- bgroup *> (manyTill braced egroup)
420− getResourcePath >>= setResourcePath . (++ ps)
421− return mempty
733+keyvals :: PandocMonad m => LP m [(String, String)]
734+keyvals = try $ symbol '[' >> manyTill keyval (symbol ']')
422735
423−addMeta :: PandocMonad m => ToMetaValue a => String -> a -> LP m ()
424−addMeta field val = updateState $ \st ->
425− st{ stateMeta = addMetaField field val $ stateMeta st }
736+accent :: (Char -> String) -> Inlines -> LP m Inlines
737+accent f ils =
738+ case toList ils of
739+ (Str (x:xs) : ys) -> return $ fromList (Str (f x ++ xs) : ys)
740+ [] -> mzero
741+ _ -> return ils
426742
427−splitBibs :: String -> [Inlines]
428−splitBibs = map (str . flip replaceExtension "bib" . trim) . splitBy (==',')
743+grave :: Char -> String
744+grave 'A' = "À"
745+grave 'E' = "È"
746+grave 'I' = "Ì"
747+grave 'O' = "Ò"
748+grave 'U' = "Ù"
749+grave 'a' = "à"
750+grave 'e' = "è"
751+grave 'i' = "ì"
752+grave 'o' = "ò"
753+grave 'u' = "ù"
754+grave c = [c]
429755
430−setCaption :: PandocMonad m => LP m Blocks
431−setCaption = do
432− ils <- tok
433− mblabel <- option Nothing $
434− try $ spaces' >> controlSeq "label" >> (Just <$> tok)
435− let ils' = case mblabel of
436− Just lab -> ils <> spanWith
437− ("",[],[("data-label", stringify lab)]) mempty
438− Nothing -> ils
439− updateState $ \st -> st{ stateCaption = Just ils' }
440− return mempty
756+acute :: Char -> String
757+acute 'A' = "Á"
758+acute 'E' = "É"
759+acute 'I' = "Í"
760+acute 'O' = "Ó"
761+acute 'U' = "Ú"
762+acute 'Y' = "Ý"
763+acute 'a' = "á"
764+acute 'e' = "é"
765+acute 'i' = "í"
766+acute 'o' = "ó"
767+acute 'u' = "ú"
768+acute 'y' = "ý"
769+acute 'C' = "Ć"
770+acute 'c' = "ć"
771+acute 'L' = "Ĺ"
772+acute 'l' = "ĺ"
773+acute 'N' = "Ń"
774+acute 'n' = "ń"
775+acute 'R' = "Ŕ"
776+acute 'r' = "ŕ"
777+acute 'S' = "Ś"
778+acute 's' = "ś"
779+acute 'Z' = "Ź"
780+acute 'z' = "ź"
781+acute c = [c]
441782
442−resetCaption :: PandocMonad m => LP m ()
443−resetCaption = updateState $ \st -> st{ stateCaption = Nothing }
783+circ :: Char -> String
784+circ 'A' = "Â"
785+circ 'E' = "Ê"
786+circ 'I' = "Î"
787+circ 'O' = "Ô"
788+circ 'U' = "Û"
789+circ 'a' = "â"
790+circ 'e' = "ê"
791+circ 'i' = "î"
792+circ 'o' = "ô"
793+circ 'u' = "û"
794+circ 'C' = "Ĉ"
795+circ 'c' = "ĉ"
796+circ 'G' = "Ĝ"
797+circ 'g' = "ĝ"
798+circ 'H' = "Ĥ"
799+circ 'h' = "ĥ"
800+circ 'J' = "Ĵ"
801+circ 'j' = "ĵ"
802+circ 'S' = "Ŝ"
803+circ 's' = "ŝ"
804+circ 'W' = "Ŵ"
805+circ 'w' = "ŵ"
806+circ 'Y' = "Ŷ"
807+circ 'y' = "ŷ"
808+circ c = [c]
444809
445−authors :: PandocMonad m => LP m ()
446−authors = try $ do
447− bgroup
448− let oneAuthor = mconcat <$>
449− many1 (notFollowedBy' (controlSeq "and") >>
450− (inline <|> mempty <$ blockCommand))
451− -- skip e.g. \vspace{10pt}
452− auths <- sepBy oneAuthor (controlSeq "and")
453− egroup
454− addMeta "author" (map trimInlines auths)
810+tilde :: Char -> String
811+tilde 'A' = "Ã"
812+tilde 'a' = "ã"
813+tilde 'O' = "Õ"
814+tilde 'o' = "õ"
815+tilde 'I' = "Ĩ"
816+tilde 'i' = "ĩ"
817+tilde 'U' = "Ũ"
818+tilde 'u' = "ũ"
819+tilde 'N' = "Ñ"
820+tilde 'n' = "ñ"
821+tilde c = [c]
455822
456−section :: PandocMonad m => Attr -> Int -> LP m Blocks
457−section (ident, classes, kvs) lvl = do
458− skipopts
459− contents <- grouped inline
460− lab <- option ident $ try (spaces' >> controlSeq "label" >> spaces' >> braced)
461− attr' <- registerHeader (lab, classes, kvs) contents
462− return $ headerWith attr' lvl contents
823+umlaut :: Char -> String
824+umlaut 'A' = "Ä"
825+umlaut 'E' = "Ë"
826+umlaut 'I' = "Ï"
827+umlaut 'O' = "Ö"
828+umlaut 'U' = "Ü"
829+umlaut 'a' = "ä"
830+umlaut 'e' = "ë"
831+umlaut 'i' = "ï"
832+umlaut 'o' = "ö"
833+umlaut 'u' = "ü"
834+umlaut c = [c]
463835
464−inlineCommand :: PandocMonad m => LP m Inlines
465−inlineCommand = try $ do
466− (name, raw') <- withRaw anyControlSeq
467− guard $ name /= "begin" && name /= "end"
468− star <- option "" (string "*")
469− let name' = name ++ star
836+hungarumlaut :: Char -> String
837+hungarumlaut 'A' = "A̋"
838+hungarumlaut 'E' = "E̋"
839+hungarumlaut 'I' = "I̋"
840+hungarumlaut 'O' = "Ő"
841+hungarumlaut 'U' = "Ű"
842+hungarumlaut 'Y' = "ӳ"
843+hungarumlaut 'a' = "a̋"
844+hungarumlaut 'e' = "e̋"
845+hungarumlaut 'i' = "i̋"
846+hungarumlaut 'o' = "ő"
847+hungarumlaut 'u' = "ű"
848+hungarumlaut 'y' = "ӳ"
849+hungarumlaut c = [c]
850+
851+dot :: Char -> String
852+dot 'C' = "Ċ"
853+dot 'c' = "ċ"
854+dot 'E' = "Ė"
855+dot 'e' = "ė"
856+dot 'G' = "Ġ"
857+dot 'g' = "ġ"
858+dot 'I' = "İ"
859+dot 'Z' = "Ż"
860+dot 'z' = "ż"
861+dot c = [c]
862+
863+macron :: Char -> String
864+macron 'A' = "Ā"
865+macron 'E' = "Ē"
866+macron 'I' = "Ī"
867+macron 'O' = "Ō"
868+macron 'U' = "Ū"
869+macron 'a' = "ā"
870+macron 'e' = "ē"
871+macron 'i' = "ī"
872+macron 'o' = "ō"
873+macron 'u' = "ū"
874+macron c = [c]
875+
876+cedilla :: Char -> String
877+cedilla 'c' = "ç"
878+cedilla 'C' = "Ç"
879+cedilla 's' = "ş"
880+cedilla 'S' = "Ş"
881+cedilla 't' = "ţ"
882+cedilla 'T' = "Ţ"
883+cedilla 'e' = "ȩ"
884+cedilla 'E' = "Ȩ"
885+cedilla 'h' = "ḩ"
886+cedilla 'H' = "Ḩ"
887+cedilla 'o' = "o̧"
888+cedilla 'O' = "O̧"
889+cedilla c = [c]
890+
891+hacek :: Char -> String
892+hacek 'A' = "Ǎ"
893+hacek 'a' = "ǎ"
894+hacek 'C' = "Č"
895+hacek 'c' = "č"
896+hacek 'D' = "Ď"
897+hacek 'd' = "ď"
898+hacek 'E' = "Ě"
899+hacek 'e' = "ě"
900+hacek 'G' = "Ǧ"
901+hacek 'g' = "ǧ"
902+hacek 'H' = "Ȟ"
903+hacek 'h' = "ȟ"
904+hacek 'I' = "Ǐ"
905+hacek 'i' = "ǐ"
906+hacek 'j' = "ǰ"
907+hacek 'K' = "Ǩ"
908+hacek 'k' = "ǩ"
909+hacek 'L' = "Ľ"
910+hacek 'l' = "ľ"
911+hacek 'N' = "Ň"
912+hacek 'n' = "ň"
913+hacek 'O' = "Ǒ"
914+hacek 'o' = "ǒ"
915+hacek 'R' = "Ř"
916+hacek 'r' = "ř"
917+hacek 'S' = "Š"
918+hacek 's' = "š"
919+hacek 'T' = "Ť"
920+hacek 't' = "ť"
921+hacek 'U' = "Ǔ"
922+hacek 'u' = "ǔ"
923+hacek 'Z' = "Ž"
924+hacek 'z' = "ž"
925+hacek c = [c]
926+
927+breve :: Char -> String
928+breve 'A' = "Ă"
929+breve 'a' = "ă"
930+breve 'E' = "Ĕ"
931+breve 'e' = "ĕ"
932+breve 'G' = "Ğ"
933+breve 'g' = "ğ"
934+breve 'I' = "Ĭ"
935+breve 'i' = "ĭ"
936+breve 'O' = "Ŏ"
937+breve 'o' = "ŏ"
938+breve 'U' = "Ŭ"
939+breve 'u' = "ŭ"
940+breve c = [c]
941+
942+toksToString :: [Tok] -> String
943+toksToString = T.unpack . untokenize
944+
945+mathDisplay :: String -> Inlines
946+mathDisplay = displayMath . trim
947+
948+mathInline :: String -> Inlines
949+mathInline = math . trim
950+
951+dollarsMath :: PandocMonad m => LP m Inlines
952+dollarsMath = do
953+ symbol '$'
954+ display <- option False (True <$ symbol '$')
955+ contents <- trim . toksToString <$>
956+ many (notFollowedBy (symbol '$') >> anyTok)
957+ if display
958+ then do
959+ mathDisplay contents <$ try (symbol '$' >> symbol '$')
960+ <|> (guard (null contents) >> return (mathInline ""))
961+ else mathInline contents <$ (symbol '$')
962+
963+-- citations
964+
965+addPrefix :: [Inline] -> [Citation] -> [Citation]
966+addPrefix p (k:ks) = k {citationPrefix = p ++ citationPrefix k} : ks
967+addPrefix _ _ = []
968+
969+addSuffix :: [Inline] -> [Citation] -> [Citation]
970+addSuffix s ks@(_:_) =
971+ let k = last ks
972+ in init ks ++ [k {citationSuffix = citationSuffix k ++ s}]
973+addSuffix _ _ = []
974+
975+simpleCiteArgs :: PandocMonad m => LP m [Citation]
976+simpleCiteArgs = try $ do
977+ first <- optionMaybe $ toList <$> opt
978+ second <- optionMaybe $ toList <$> opt
979+ keys <- try $ bgroup *> (manyTill citationLabel egroup)
980+ let (pre, suf) = case (first , second ) of
981+ (Just s , Nothing) -> (mempty, s )
982+ (Just s , Just t ) -> (s , t )
983+ _ -> (mempty, mempty)
984+ conv k = Citation { citationId = k
985+ , citationPrefix = []
986+ , citationSuffix = []
987+ , citationMode = NormalCitation
988+ , citationHash = 0
989+ , citationNoteNum = 0
990+ }
991+ return $ addPrefix pre $ addSuffix suf $ map conv keys
992+
993+citationLabel :: PandocMonad m => LP m String
994+citationLabel = do
995+ optional sp
996+ toksToString <$>
997+ (many1 (satisfyTok isWordTok <|> symbolIn bibtexKeyChar)
998+ <* optional sp
999+ <* optional (symbol ',')
1000+ <* optional sp)
1001+ where bibtexKeyChar = ".:;?!`'()/*@_+=-[]" :: [Char]
1002+
1003+cites :: PandocMonad m => CitationMode -> Bool -> LP m [Citation]
1004+cites mode multi = try $ do
1005+ cits <- if multi
1006+ then many1 simpleCiteArgs
1007+ else count 1 simpleCiteArgs
1008+ let cs = concat cits
1009+ return $ case mode of
1010+ AuthorInText -> case cs of
1011+ (c:rest) -> c {citationMode = mode} : rest
1012+ [] -> []
1013+ _ -> map (\a -> a {citationMode = mode}) cs
1014+
1015+citation :: PandocMonad m => String -> CitationMode -> Bool -> LP m Inlines
1016+citation name mode multi = do
1017+ (c,raw) <- withRaw $ cites mode multi
1018+ return $ cite c (rawInline "latex" $ "\\" ++ name ++ (toksToString raw))
1019+
1020+handleCitationPart :: Inlines -> [Citation]
1021+handleCitationPart ils =
1022+ let isCite Cite{} = True
1023+ isCite _ = False
1024+ (pref, rest) = break isCite (toList ils)
1025+ in case rest of
1026+ (Cite cs _:suff) -> addPrefix pref $ addSuffix suff cs
1027+ _ -> []
1028+
1029+complexNatbibCitation :: PandocMonad m => CitationMode -> LP m Inlines
1030+complexNatbibCitation mode = try $ do
1031+ (cs, raw) <-
1032+ withRaw $ concat <$> do
1033+ bgroup
1034+ items <- mconcat <$>
1035+ many1 (notFollowedBy (symbol ';') >> inline)
1036+ `sepBy1` (symbol ';')
1037+ egroup
1038+ return $ map handleCitationPart items
1039+ case cs of
1040+ [] -> mzero
1041+ (c:cits) -> return $ cite (c{ citationMode = mode }:cits)
1042+ (rawInline "latex" $ "\\citetext" ++ toksToString raw)
1043+
1044+inNote :: Inlines -> Inlines
1045+inNote ils =
1046+ note $ para $ ils <> str "."
1047+
1048+inlineCommand' :: PandocMonad m => LP m Inlines
1049+inlineCommand' = try $ do
1050+ Tok _ (CtrlSeq name) cmd <- anyControlSeq
1051+ guard $ name /= "begin" && name /= "end"
1052+ (star, rawstar) <- withRaw $ option "" ("*" <$ symbol '*' <* optional sp)
1053+ let name' = name <> star
1054+ let names = ordNub [name', name] -- check non-starred as fallback
4701055 let raw = do
471− guard $ not (isBlockCommand name)
472− rawargs <- withRaw
473− (skipangles *> skipopts *> option "" dimenarg *> many braced)
474− let rawcommand = raw' ++ star ++ snd rawargs
475− transformed <- applyMacros' rawcommand
476− exts <- getOption readerExtensions
477− if transformed /= rawcommand
478− then parseFromString' inlines transformed
479− else if extensionEnabled Ext_raw_tex exts
480− then return $ rawInline "latex" rawcommand
481− else ignore rawcommand
482− (lookupListDefault raw [name',name] inlineCommands <*
483− optional (try (string "{}")))
484−
485−rawInlineOr :: PandocMonad m => String -> LP m Inlines -> LP m Inlines
486−rawInlineOr name' fallback = do
487− parseRaw <- extensionEnabled Ext_raw_tex <$> getOption readerExtensions
488− if parseRaw
489− then rawInline "latex" <$> getRawCommand name'
490− else fallback
1056+ guard $ isInlineCommand name || not (isBlockCommand name)
1057+ (_, rawargs) <- withRaw
1058+ (skipangles *> skipopts *> option "" dimenarg *> many braced)
1059+ let rawcommand = T.unpack $ cmd <> untokenize (rawstar ++ rawargs)
1060+ (guardEnabled Ext_raw_tex >> return (rawInline "latex" rawcommand))
1061+ <|> ignore rawcommand
1062+ lookupListDefault raw names inlineCommands
1063+
1064+tok :: PandocMonad m => LP m Inlines
1065+tok = grouped inline <|> inlineCommand' <|> singleChar
1066+ where singleChar = try $ do
1067+ Tok (lin,col) toktype t <- satisfyTok (tokTypeIn [Word, Symbol])
1068+ guard $ not $ toktype == Symbol &&
1069+ T.any (`Set.member` specialChars) t
1070+ if T.length t > 1
1071+ then do
1072+ let (t1, t2) = (T.take 1 t, T.drop 1 t)
1073+ inp <- getInput
1074+ setInput $ (Tok (lin, col + 1) toktype t2) : inp
1075+ return $ str (T.unpack t1)
1076+ else return $ str (T.unpack t)
1077+
1078+opt :: PandocMonad m => LP m Inlines
1079+opt = bracketed inline
1080+
1081+rawopt :: PandocMonad m => LP m Text
1082+rawopt = do
1083+ symbol '['
1084+ inner <- untokenize <$> manyTill anyTok (symbol ']')
1085+ optional sp
1086+ return $ "[" <> inner <> "]"
1087+
1088+skipopts :: PandocMonad m => LP m ()
1089+skipopts = skipMany rawopt
1090+
1091+-- opts in angle brackets are used in beamer
1092+rawangle :: PandocMonad m => LP m ()
1093+rawangle = try $ do
1094+ symbol '<'
1095+ () <$ manyTill anyTok (symbol '>')
1096+
1097+skipangles :: PandocMonad m => LP m ()
1098+skipangles = skipMany rawangle
1099+
1100+ignore :: (Monoid a, PandocMonad m) => String -> ParserT s u m a
1101+ignore raw = do
1102+ pos <- getPosition
1103+ report $ SkippedContent raw pos
1104+ return mempty
1105+
1106+withRaw :: PandocMonad m => LP m a -> LP m (a, [Tok])
1107+withRaw parser = do
1108+ inp <- getInput
1109+ result <- parser
1110+ nxt <- option (Tok (0,0) Word "") (lookAhead anyTok)
1111+ let raw = takeWhile (/= nxt) inp
1112+ return (result, raw)
1113+
1114+inBrackets :: Inlines -> Inlines
1115+inBrackets x = str "[" <> x <> str "]"
1116+
1117+unescapeURL :: String -> String
1118+unescapeURL ('\\':x:xs) | isEscapable x = x:unescapeURL xs
1119+ where isEscapable c = c `elem` ("#$%&~_^\\{}" :: String)
1120+unescapeURL (x:xs) = x:unescapeURL xs
1121+unescapeURL [] = ""
1122+
1123+mathEnvWith :: PandocMonad m
1124+ => (Inlines -> a) -> Maybe Text -> Text -> LP m a
1125+mathEnvWith f innerEnv name = f . mathDisplay . inner <$> mathEnv name
1126+ where inner x = case innerEnv of
1127+ Nothing -> x
1128+ Just y -> "\\begin{" ++ T.unpack y ++ "}\n" ++ x ++
1129+ "\\end{" ++ T.unpack y ++ "}"
4911130
492−isBlockCommand :: String -> Bool
493−isBlockCommand s = s `M.member` (blockCommands :: M.Map String (LP PandocPure Blocks))
1131+mathEnv :: PandocMonad m => Text -> LP m String
1132+mathEnv name = do
1133+ skipopts
1134+ optional blankline
1135+ res <- manyTill anyTok (end_ name)
1136+ return $ stripTrailingNewlines $ T.unpack $ untokenize res
4941137
1138+inlineEnvironment :: PandocMonad m => LP m Inlines
1139+inlineEnvironment = try $ do
1140+ controlSeq "begin"
1141+ name <- untokenize <$> braced
1142+ M.findWithDefault mzero name inlineEnvironments
4951143
496−inlineEnvironments :: PandocMonad m => M.Map String (LP m Inlines)
497−inlineEnvironments = M.fromList
498− [ ("displaymath", mathEnvWith id Nothing "displaymath")
1144+inlineEnvironments :: PandocMonad m => M.Map Text (LP m Inlines)
1145+inlineEnvironments = M.fromList [
1146+ ("displaymath", mathEnvWith id Nothing "displaymath")
4991147 , ("math", math <$> mathEnv "math")
5001148 , ("equation", mathEnvWith id Nothing "equation")
5011149 , ("equation*", mathEnvWith id Nothing "equation*")
@@ -511,7 +1159,7 @@ inlineEnvironments = M.fromList
5111159 , ("alignat*", mathEnvWith id (Just "aligned") "alignat*")
5121160 ]
5131161
514−inlineCommands :: PandocMonad m => M.Map String (LP m Inlines)
1162+inlineCommands :: PandocMonad m => M.Map Text (LP m Inlines)
5151163inlineCommands = M.fromList $
5161164 [ ("emph", extractSpaces emph <$> tok)
5171165 , ("textit", extractSpaces emph <$> tok)
@@ -540,9 +1188,9 @@ inlineCommands = M.fromList $
5401188 , ("textgreek", tok)
5411189 , ("sep", lit ",")
5421190 , ("cref", rawInlineOr "cref" (inBrackets <$> tok)) -- from cleveref.sty
543− , ("(", mathInline $ manyTill anyChar (try $ string "\\)"))
544− , ("[", mathDisplay $ manyTill anyChar (try $ string "\\]"))
545− , ("ensuremath", mathInline braced)
1191+ , ("(", mathInline . toksToString <$> manyTill anyTok (controlSeq ")"))
1192+ , ("[", mathDisplay . toksToString <$> manyTill anyTok (controlSeq "]"))
1193+ , ("ensuremath", mathInline . toksToString <$> braced)
5461194 , ("texorpdfstring", (\_ x -> x) <$> tok <*> tok)
5471195 , ("P", lit "¶")
5481196 , ("S", lit "§")
@@ -592,7 +1240,10 @@ inlineCommands = M.fromList $
5921240 , ("v", option (str "v") $ try $ tok >>= accent hacek)
5931241 , ("u", option (str "u") $ try $ tok >>= accent breve)
5941242 , ("i", lit "i")
595− , ("\\", linebreak <$ (optional (bracketed inline) *> spaces'))
1243+ , ("\\", linebreak <$ (do inTableCell <- sInTableCell <$> getState
1244+ guard $ not inTableCell
1245+ optional (bracketed inline)
1246+ spaces))
5961247 , (",", lit "\8198")
5971248 , ("@", pure mempty)
5981249 , (" ", lit "\160")
@@ -607,13 +1258,14 @@ inlineCommands = M.fromList $
6071258 , ("verb", doverb)
6081259 , ("lstinline", dolstinline)
6091260 , ("Verb", doverb)
610− , ("url", (unescapeURL <$> braced) >>= \url ->
611− pure (link url "" (str url)))
612− , ("href", (unescapeURL <$> braced <* optional sp) >>= \url ->
613− tok >>= \lab ->
614− pure (link url "" lab))
1261+ , ("url", ((unescapeURL . T.unpack . untokenize) <$> braced) >>= \url ->
1262+ pure (link url "" (str url)))
1263+ , ("href", (unescapeURL . toksToString <$>
1264+ braced <* optional sp) >>= \url ->
1265+ tok >>= \lab -> pure (link url "" lab))
6151266 , ("includegraphics", do options <- option [] keyvals
616− src <- unescapeURL . removeDoubleQuotes <$> braced
1267+ src <- unescapeURL . T.unpack .
1268+ removeDoubleQuotes . untokenize <$> braced
6171269 mkImage options src)
6181270 , ("enquote", enquote)
6191271 , ("cite", citation "cite" NormalCitation False)
@@ -686,362 +1338,456 @@ inlineCommands = M.fromList $
6861338 -- fontawesome
6871339 , ("faCheck", lit "\10003")
6881340 , ("faClose", lit "\10007")
689− ] ++ map ignoreInlines
690− -- these commands will be ignored unless --parse-raw is specified,
691− -- in which case they will appear as raw latex blocks:
1341+ ]
1342+
1343+ttfamily :: PandocMonad m => LP m Inlines
1344+ttfamily = (code . stringify . toList) <$> tok
1345+
1346+rawInlineOr :: PandocMonad m => Text -> LP m Inlines -> LP m Inlines
1347+rawInlineOr name' fallback = do
1348+ parseRaw <- extensionEnabled Ext_raw_tex <$> getOption readerExtensions
1349+ if parseRaw
1350+ then rawInline "latex" <$> getRawCommand name'
1351+ else fallback
1352+
1353+getRawCommand :: PandocMonad m => Text -> LP m String
1354+getRawCommand txt = do
1355+ (_, rawargs) <- withRaw
1356+ (many (try (optional sp *> opt)) *>
1357+ option "" (try (optional sp *> dimenarg)) *>
1358+ many braced)
1359+ return $ T.unpack (txt <> untokenize rawargs)
1360+
1361+isBlockCommand :: Text -> Bool
1362+isBlockCommand s =
1363+ s `M.member` (blockCommands :: M.Map Text (LP PandocPure Blocks))
1364+ || s `Set.member` treatAsBlock
1365+
1366+treatAsBlock :: Set.Set Text
1367+treatAsBlock = Set.fromList
1368+ [ "newcommand", "renewcommand"
1369+ , "newenvironment", "renewenvironment"
1370+ , "providecommand", "provideenvironment"
1371+ -- newcommand, etc. should be parsed by macroDef, but we need this
1372+ -- here so these aren't parsed as inline commands to ignore
1373+ , "special", "pdfannot", "pdfstringdef"
1374+ , "bibliographystyle"
1375+ , "maketitle", "makeindex", "makeglossary"
1376+ , "addcontentsline", "addtocontents", "addtocounter"
1377+ -- \ignore{} is used conventionally in literate haskell for definitions
1378+ -- that are to be processed by the compiler but not printed.
1379+ , "ignore"
1380+ , "hyperdef"
1381+ , "markboth", "markright", "markleft"
1382+ , "hspace", "vspace"
1383+ , "newpage"
1384+ , "clearpage"
1385+ , "pagebreak"
1386+ ]
1387+
1388+isInlineCommand :: Text -> Bool
1389+isInlineCommand s =
1390+ s `M.member` (inlineCommands :: M.Map Text (LP PandocPure Inlines))
1391+ || s `Set.member` treatAsInline
1392+
1393+treatAsInline :: Set.Set Text
1394+treatAsInline = Set.fromList
6921395 [ "index"
6931396 , "hspace"
6941397 , "vspace"
1398+ , "noindent"
6951399 , "newpage"
6961400 , "clearpage"
6971401 , "pagebreak"
6981402 ]
6991403
700−ttfamily :: PandocMonad m => LP m Inlines
701−ttfamily = (code . stringify . toList) <$> tok
1404+lookupListDefault :: (Show k, Ord k) => v -> [k] -> M.Map k v -> v
1405+lookupListDefault d = (fromMaybe d .) . lookupList
1406+ where lookupList l m = msum $ map (`M.lookup` m) l
7021407
703−mkImage :: PandocMonad m => [(String, String)] -> String -> LP m Inlines
704−mkImage options src = do
705− let replaceTextwidth (k,v) = case numUnit v of
706− Just (num, "\\textwidth") -> (k, showFl (num * 100) ++ "%")
707− _ -> (k, v)
708− let kvs = map replaceTextwidth $ filter (\(k,_) -> k `elem` ["width", "height"]) options
709− let attr = ("",[], kvs)
710− let alt = str "image"
711− case takeExtension src of
712− "" -> do
713− defaultExt <- getOption readerDefaultImageExtension
714− return $ imageWith attr (addExtension src defaultExt) "" alt
715− _ -> return $ imageWith attr src "" alt
1408+inline :: PandocMonad m => LP m Inlines
1409+inline = (mempty <$ comment)
1410+ <|> (space <$ whitespace)
1411+ <|> (softbreak <$ endline)
1412+ <|> word
1413+ <|> inlineCommand'
1414+ <|> inlineEnvironment
1415+ <|> inlineGroup
1416+ <|> (symbol '-' *>
1417+ option (str "-") (symbol '-' *>
1418+ option (str "–") (str "—" <$ symbol '-')))
1419+ <|> doubleQuote
1420+ <|> singleQuote
1421+ <|> (str "”" <$ try (symbol '\'' >> symbol '\''))
1422+ <|> (str "”" <$ symbol '”')
1423+ <|> (str "’" <$ symbol '\'')
1424+ <|> (str "’" <$ symbol '’')
1425+ <|> (str "\160" <$ symbol '~')
1426+ <|> dollarsMath
1427+ <|> (guardEnabled Ext_literate_haskell *> symbol '|' *> doLHSverb)
1428+ <|> (str . (:[]) <$> primEscape)
1429+ <|> regularSymbol
1430+ <|> (do res <- symbolIn "#^'`\"[]"
1431+ pos <- getPosition
1432+ let s = T.unpack (untoken res)
1433+ report $ ParsingUnescaped s pos
1434+ return $ str s)
7161435
717−inNote :: Inlines -> Inlines
718−inNote ils =
719− note $ para $ ils <> str "."
1436+inlines :: PandocMonad m => LP m Inlines
1437+inlines = mconcat <$> many inline
7201438
721−unescapeURL :: String -> String
722−unescapeURL ('\\':x:xs) | isEscapable x = x:unescapeURL xs
723− where isEscapable c = c `elem` ("#$%&~_^\\{}" :: String)
724−unescapeURL (x:xs) = x:unescapeURL xs
725−unescapeURL [] = ""
726−
727−enquote :: PandocMonad m => LP m Inlines
728−enquote = do
729− skipopts
730− context <- stateQuoteContext <$> getState
731− if context == InDoubleQuote
732− then singleQuoted <$> withQuoteContext InSingleQuote tok
733− else doubleQuoted <$> withQuoteContext InDoubleQuote tok
734−
735−doverb :: PandocMonad m => LP m Inlines
736−doverb = do
737− marker <- anyChar
738− code <$> manyTill (satisfy (/='\n')) (char marker)
739−
740−dolstinline :: PandocMonad m => LP m Inlines
741−dolstinline = do
742− options <- option [] keyvals
743− let classes = maybeToList $ lookup "language" options >>= fromListingsLanguage
744− marker <- char '{' <|> anyChar
745− codeWith ("",classes,[]) <$> manyTill (satisfy (/='\n')) (char '}' <|> char marker)
746−
747−doLHSverb :: PandocMonad m => LP m Inlines
748−doLHSverb = codeWith ("",["haskell"],[]) <$> manyTill (satisfy (/='\n')) (char '|')
749−
750−-- converts e.g. \SI{1}[\$]{} to "$ 1" or \SI{1}{\euro} to "1 €"
751−dosiunitx :: PandocMonad m => LP m Inlines
752−dosiunitx = do
753− skipopts
754− value <- tok
755− valueprefix <- option "" $ char '[' >> (mconcat <$> manyTill tok (char ']'))
756− unit <- tok
757− let emptyOr160 "" = ""
758− emptyOr160 _ = "\160"
759− return . mconcat $ [valueprefix,
760− emptyOr160 valueprefix,
761− value,
762− emptyOr160 unit,
763− unit]
764−
765−lit :: String -> LP m Inlines
766−lit = pure . str
767−
768−accent :: (Char -> String) -> Inlines -> LP m Inlines
769−accent f ils =
770− case toList ils of
771− (Str (x:xs) : ys) -> return $ fromList (Str (f x ++ xs) : ys)
772− [] -> mzero
773− _ -> return ils
774−
775−grave :: Char -> String
776−grave 'A' = "À"
777−grave 'E' = "È"
778−grave 'I' = "Ì"
779−grave 'O' = "Ò"
780−grave 'U' = "Ù"
781−grave 'a' = "à"
782−grave 'e' = "è"
783−grave 'i' = "ì"
784−grave 'o' = "ò"
785−grave 'u' = "ù"
786−grave c = [c]
787−
788−acute :: Char -> String
789−acute 'A' = "Á"
790−acute 'E' = "É"
791−acute 'I' = "Í"
792−acute 'O' = "Ó"
793−acute 'U' = "Ú"
794−acute 'Y' = "Ý"
795−acute 'a' = "á"
796−acute 'e' = "é"
797−acute 'i' = "í"
798−acute 'o' = "ó"
799−acute 'u' = "ú"
800−acute 'y' = "ý"
801−acute 'C' = "Ć"
802−acute 'c' = "ć"
803−acute 'L' = "Ĺ"
804−acute 'l' = "ĺ"
805−acute 'N' = "Ń"
806−acute 'n' = "ń"
807−acute 'R' = "Ŕ"
808−acute 'r' = "ŕ"
809−acute 'S' = "Ś"
810−acute 's' = "ś"
811−acute 'Z' = "Ź"
812−acute 'z' = "ź"
813−acute c = [c]
814−
815−circ :: Char -> String
816−circ 'A' = "Â"
817−circ 'E' = "Ê"
818−circ 'I' = "Î"
819−circ 'O' = "Ô"
820−circ 'U' = "Û"
821−circ 'a' = "â"
822−circ 'e' = "ê"
823−circ 'i' = "î"
824−circ 'o' = "ô"
825−circ 'u' = "û"
826−circ 'C' = "Ĉ"
827−circ 'c' = "ĉ"
828−circ 'G' = "Ĝ"
829−circ 'g' = "ĝ"
830−circ 'H' = "Ĥ"
831−circ 'h' = "ĥ"
832−circ 'J' = "Ĵ"
833−circ 'j' = "ĵ"
834−circ 'S' = "Ŝ"
835−circ 's' = "ŝ"
836−circ 'W' = "Ŵ"
837−circ 'w' = "ŵ"
838−circ 'Y' = "Ŷ"
839−circ 'y' = "ŷ"
840−circ c = [c]
1439+-- block elements:
8411440
842−tilde :: Char -> String
843−tilde 'A' = "Ã"
844−tilde 'a' = "ã"
845−tilde 'O' = "Õ"
846−tilde 'o' = "õ"
847−tilde 'I' = "Ĩ"
848−tilde 'i' = "ĩ"
849−tilde 'U' = "Ũ"
850−tilde 'u' = "ũ"
851−tilde 'N' = "Ñ"
852−tilde 'n' = "ñ"
853−tilde c = [c]
1441+begin_ :: PandocMonad m => Text -> LP m ()
1442+begin_ t = (try $ do
1443+ controlSeq "begin"
1444+ spaces
1445+ symbol '{'
1446+ spaces
1447+ Tok _ Word txt <- satisfyTok isWordTok
1448+ spaces
1449+ symbol '}'
1450+ guard (t == txt)) <?> ("\\begin{" ++ T.unpack t ++ "}")
8541451
855−umlaut :: Char -> String
856−umlaut 'A' = "Ä"
857−umlaut 'E' = "Ë"
858−umlaut 'I' = "Ï"
859−umlaut 'O' = "Ö"
860−umlaut 'U' = "Ü"
861−umlaut 'a' = "ä"
862−umlaut 'e' = "ë"
863−umlaut 'i' = "ï"
864−umlaut 'o' = "ö"
865−umlaut 'u' = "ü"
866−umlaut c = [c]
1452+end_ :: PandocMonad m => Text -> LP m ()
1453+end_ t = (try $ do
1454+ controlSeq "end"
1455+ spaces
1456+ symbol '{'
1457+ spaces
1458+ Tok _ Word txt <- satisfyTok isWordTok
1459+ spaces
1460+ symbol '}'
1461+ guard $ t == txt) <?> ("\\end{" ++ T.unpack t ++ "}")
8671462
868−hungarumlaut :: Char -> String
869−hungarumlaut 'A' = "A̋"
870−hungarumlaut 'E' = "E̋"
871−hungarumlaut 'I' = "I̋"
872−hungarumlaut 'O' = "Ő"
873−hungarumlaut 'U' = "Ű"
874−hungarumlaut 'Y' = "ӳ"
875−hungarumlaut 'a' = "a̋"
876−hungarumlaut 'e' = "e̋"
877−hungarumlaut 'i' = "i̋"
878−hungarumlaut 'o' = "ő"
879−hungarumlaut 'u' = "ű"
880−hungarumlaut 'y' = "ӳ"
881−hungarumlaut c = [c]
1463+preamble :: PandocMonad m => LP m Blocks
1464+preamble = mempty <$ many preambleBlock
1465+ where preambleBlock = spaces1
1466+ <|> void include
1467+ <|> void macroDef
1468+ <|> void blockCommand
1469+ <|> void braced
1470+ <|> (notFollowedBy (begin_ "document") >> void anyTok)
8821471
883−dot :: Char -> String
884−dot 'C' = "Ċ"
885−dot 'c' = "ċ"
886−dot 'E' = "Ė"
887−dot 'e' = "ė"
888−dot 'G' = "Ġ"
889−dot 'g' = "ġ"
890−dot 'I' = "İ"
891−dot 'Z' = "Ż"
892−dot 'z' = "ż"
893−dot c = [c]
1472+paragraph :: PandocMonad m => LP m Blocks
1473+paragraph = do
1474+ x <- trimInlines . mconcat <$> many1 inline
1475+ if x == mempty
1476+ then return mempty
1477+ else return $ para x
8941478
895−macron :: Char -> String
896−macron 'A' = "Ā"
897−macron 'E' = "Ē"
898−macron 'I' = "Ī"
899−macron 'O' = "Ō"
900−macron 'U' = "Ū"
901−macron 'a' = "ā"
902−macron 'e' = "ē"
903−macron 'i' = "ī"
904−macron 'o' = "ō"
905−macron 'u' = "ū"
906−macron c = [c]
1479+include :: PandocMonad m => LP m Blocks
1480+include = do
1481+ (Tok _ (CtrlSeq name) _) <-
1482+ controlSeq "include" <|> controlSeq "input" <|>
1483+ controlSeq "subfile" <|> controlSeq "usepackage"
1484+ skipMany $ bracketed inline -- skip options
1485+ fs <- (map trim . splitBy (==',') . T.unpack . untokenize) <$> braced
1486+ let fs' = if name == "usepackage"
1487+ then map (maybeAddExtension ".sty") fs
1488+ else map (maybeAddExtension ".tex") fs
1489+ dirs <- (splitBy (==':') . fromMaybe ".") <$> lookupEnv "TEXINPUTS"
1490+ mconcat <$> mapM (insertIncludedFile blocks (tokenize . T.pack) dirs) fs'
9071491
908−cedilla :: Char -> String
909−cedilla 'c' = "ç"
910−cedilla 'C' = "Ç"
911−cedilla 's' = "ş"
912−cedilla 'S' = "Ş"
913−cedilla 't' = "ţ"
914−cedilla 'T' = "Ţ"
915−cedilla 'e' = "ȩ"
916−cedilla 'E' = "Ȩ"
917−cedilla 'h' = "ḩ"
918−cedilla 'H' = "Ḩ"
919−cedilla 'o' = "o̧"
920−cedilla 'O' = "O̧"
921−cedilla c = [c]
1492+maybeAddExtension :: String -> FilePath -> FilePath
1493+maybeAddExtension ext fp =
1494+ if null (takeExtension fp)
1495+ then addExtension fp ext
1496+ else fp
9221497
923−hacek :: Char -> String
924−hacek 'A' = "Ǎ"
925−hacek 'a' = "ǎ"
926−hacek 'C' = "Č"
927−hacek 'c' = "č"
928−hacek 'D' = "Ď"
929−hacek 'd' = "ď"
930−hacek 'E' = "Ě"
931−hacek 'e' = "ě"
932−hacek 'G' = "Ǧ"
933−hacek 'g' = "ǧ"
934−hacek 'H' = "Ȟ"
935−hacek 'h' = "ȟ"
936−hacek 'I' = "Ǐ"
937−hacek 'i' = "ǐ"
938−hacek 'j' = "ǰ"
939−hacek 'K' = "Ǩ"
940−hacek 'k' = "ǩ"
941−hacek 'L' = "Ľ"
942−hacek 'l' = "ľ"
943−hacek 'N' = "Ň"
944−hacek 'n' = "ň"
945−hacek 'O' = "Ǒ"
946−hacek 'o' = "ǒ"
947−hacek 'R' = "Ř"
948−hacek 'r' = "ř"
949−hacek 'S' = "Š"
950−hacek 's' = "š"
951−hacek 'T' = "Ť"
952−hacek 't' = "ť"
953−hacek 'U' = "Ǔ"
954−hacek 'u' = "ǔ"
955−hacek 'Z' = "Ž"
956−hacek 'z' = "ž"
957−hacek c = [c]
1498+addMeta :: PandocMonad m => ToMetaValue a => String -> a -> LP m ()
1499+addMeta field val = updateState $ \st ->
1500+ st{ sMeta = addMetaField field val $ sMeta st }
9581501
959−breve :: Char -> String
960−breve 'A' = "Ă"
961−breve 'a' = "ă"
962−breve 'E' = "Ĕ"
963−breve 'e' = "ĕ"
964−breve 'G' = "Ğ"
965−breve 'g' = "ğ"
966−breve 'I' = "Ĭ"
967−breve 'i' = "ĭ"
968−breve 'O' = "Ŏ"
969−breve 'o' = "ŏ"
970−breve 'U' = "Ŭ"
971−breve 'u' = "ŭ"
972−breve c = [c]
1502+authors :: PandocMonad m => LP m ()
1503+authors = try $ do
1504+ bgroup
1505+ let oneAuthor = mconcat <$>
1506+ many1 (notFollowedBy' (controlSeq "and") >>
1507+ (inline <|> mempty <$ blockCommand))
1508+ -- skip e.g. \vspace{10pt}
1509+ auths <- sepBy oneAuthor (controlSeq "and")
1510+ egroup
1511+ addMeta "author" (map trimInlines auths)
9731512
974−tok :: PandocMonad m => LP m Inlines
975−tok = try $ grouped inline <|> inlineCommand <|> str <$> count 1 inlineChar
1513+macroDef :: PandocMonad m => LP m Blocks
1514+macroDef = do
1515+ guardEnabled Ext_latex_macros
1516+ mempty <$ ((commandDef <|> environmentDef) <* doMacros 0)
1517+ where commandDef = do
1518+ (name, macro') <- newcommand
1519+ updateState $ \s -> s{ sMacros = M.insert name macro' (sMacros s) }
1520+ environmentDef = do
1521+ (name, macro1, macro2) <- newenvironment
1522+ updateState $ \s -> s{ sMacros =
1523+ M.insert name macro1 (sMacros s) }
1524+ updateState $ \s -> s{ sMacros =
1525+ M.insert ("end" <> name) macro2 (sMacros s) }
1526+ -- @\newenvironment{envname}[n-args][default]{begin}{end}@
1527+ -- is equivalent to
1528+ -- @\newcommand{\envname}[n-args][default]{begin}@
1529+ -- @\newcommand{\endenvname}@
1530+
1531+newcommand :: PandocMonad m => LP m (Text, Macro)
1532+newcommand = do
1533+ pos <- getPosition
1534+ Tok _ (CtrlSeq mtype) _ <- controlSeq "newcommand" <|>
1535+ controlSeq "renewcommand" <|>
1536+ controlSeq "providecommand"
1537+ optional $ symbol '*'
1538+ Tok _ (CtrlSeq name) txt <- withVerbatimMode $ anyControlSeq <|>
1539+ (symbol '{' *> spaces *> anyControlSeq <* spaces <* symbol '}')
1540+ spaces
1541+ numargs <- option 0 $ try bracketedNum
1542+ spaces
1543+ optarg <- option Nothing $ Just <$> try bracketedToks
1544+ spaces
1545+ contents <- braced
1546+ when (mtype == "newcommand") $ do
1547+ macros <- sMacros <$> getState
1548+ case M.lookup name macros of
1549+ Just _ -> report $ MacroAlreadyDefined (T.unpack txt) pos
1550+ Nothing -> return ()
1551+ return (name, Macro numargs optarg contents)
1552+
1553+newenvironment :: PandocMonad m => LP m (Text, Macro, Macro)
1554+newenvironment = do
1555+ pos <- getPosition
1556+ Tok _ (CtrlSeq mtype) _ <- controlSeq "newenvironment" <|>
1557+ controlSeq "renewenvironment" <|>
1558+ controlSeq "provideenvironment"
1559+ optional $ symbol '*'
1560+ symbol '{'
1561+ spaces
1562+ Tok _ Word name <- satisfyTok isWordTok
1563+ spaces
1564+ symbol '}'
1565+ spaces
1566+ numargs <- option 0 $ try bracketedNum
1567+ spaces
1568+ optarg <- option Nothing $ Just <$> try bracketedToks
1569+ spaces
1570+ startcontents <- braced
1571+ spaces
1572+ endcontents <- braced
1573+ when (mtype == "newenvironment") $ do
1574+ macros <- sMacros <$> getState
1575+ case M.lookup name macros of
1576+ Just _ -> report $ MacroAlreadyDefined (T.unpack name) pos
1577+ Nothing -> return ()
1578+ return (name, Macro numargs optarg startcontents,
1579+ Macro 0 Nothing endcontents)
1580+
1581+bracketedToks :: PandocMonad m => LP m [Tok]
1582+bracketedToks = do
1583+ symbol '['
1584+ manyTill anyTok (symbol ']')
1585+
1586+bracketedNum :: PandocMonad m => LP m Int
1587+bracketedNum = do
1588+ ds <- untokenize <$> bracketedToks
1589+ case safeRead (T.unpack ds) of
1590+ Just i -> return i
1591+ _ -> return 0
9761592
977−opt :: PandocMonad m => LP m Inlines
978−opt = bracketed inline
1593+setCaption :: PandocMonad m => LP m Blocks
1594+setCaption = do
1595+ ils <- tok
1596+ mblabel <- option Nothing $
1597+ try $ spaces >> controlSeq "label" >> (Just <$> tok)
1598+ let ils' = case mblabel of
1599+ Just lab -> ils <> spanWith
1600+ ("",[],[("data-label", stringify lab)]) mempty
1601+ Nothing -> ils
1602+ updateState $ \st -> st{ sCaption = Just ils' }
1603+ return mempty
9791604
980−rawopt :: PandocMonad m => LP m String
981−rawopt = do
982− contents <- bracketed (many1 (noneOf "[]") <|> try (string "\\]") <|>
983− try (string "\\[") <|> rawopt)
984− optional sp
985− return $ "[" ++ contents ++ "]"
1605+looseItem :: PandocMonad m => LP m Blocks
1606+looseItem = do
1607+ inListItem <- sInListItem <$> getState
1608+ guard $ not inListItem
1609+ skipopts
1610+ return mempty
9861611
987−skipopts :: PandocMonad m => LP m ()
988−skipopts = skipMany rawopt
1612+resetCaption :: PandocMonad m => LP m ()
1613+resetCaption = updateState $ \st -> st{ sCaption = Nothing }
9891614
990−-- opts in angle brackets are used in beamer
991−rawangle :: PandocMonad m => LP m ()
992−rawangle = try $ do
993− char '<'
994− skipMany (noneOf ">")
995− char '>'
996− return ()
1615+section :: PandocMonad m => Attr -> Int -> LP m Blocks
1616+section (ident, classes, kvs) lvl = do
1617+ skipopts
1618+ contents <- grouped inline
1619+ lab <- option ident $
1620+ try (spaces >> controlSeq "label"
1621+ >> spaces >> toksToString <$> braced)
1622+ attr' <- registerHeader (lab, classes, kvs) contents
1623+ return $ headerWith attr' lvl contents
9971624
998−skipangles :: PandocMonad m => LP m ()
999−skipangles = skipMany rawangle
1625+blockCommand :: PandocMonad m => LP m Blocks
1626+blockCommand = try $ do
1627+ Tok _ (CtrlSeq name) txt <- anyControlSeq
1628+ guard $ name /= "begin" && name /= "end"
1629+ star <- option "" ("*" <$ symbol '*' <* optional sp)
1630+ let name' = name <> star
1631+ let names = ordNub [name', name]
1632+ let raw = do
1633+ guard $ isBlockCommand name || not (isInlineCommand name)
1634+ rawBlock "latex" <$> getRawCommand txt
1635+ lookupListDefault raw names blockCommands
10001636
1001−inlineText :: PandocMonad m => LP m Inlines
1002−inlineText = str <$> many1 inlineChar
1637+closing :: PandocMonad m => LP m Blocks
1638+closing = do
1639+ contents <- tok
1640+ st <- getState
1641+ let extractInlines (MetaBlocks [Plain ys]) = ys
1642+ extractInlines (MetaBlocks [Para ys ]) = ys
1643+ extractInlines _ = []
1644+ let sigs = case lookupMeta "author" (sMeta st) of
1645+ Just (MetaList xs) ->
1646+ para $ trimInlines $ fromList $
1647+ intercalate [LineBreak] $ map extractInlines xs
1648+ _ -> mempty
1649+ return $ para (trimInlines contents) <> sigs
10031650
1004−inlineChar :: PandocMonad m => LP m Char
1005−inlineChar = noneOf "\\$%&~#{}^'`\"‘’“”-[] \t\n"
1651+blockCommands :: PandocMonad m => M.Map Text (LP m Blocks)
1652+blockCommands = M.fromList $
1653+ [ ("par", mempty <$ skipopts)
1654+ , ("parbox", braced >> grouped blocks)
1655+ , ("title", mempty <$ (skipopts *>
1656+ (grouped inline >>= addMeta "title")
1657+ <|> (grouped block >>= addMeta "title")))
1658+ , ("subtitle", mempty <$ (skipopts *> tok >>= addMeta "subtitle"))
1659+ , ("author", mempty <$ (skipopts *> authors))
1660+ -- -- in letter class, temp. store address & sig as title, author
1661+ , ("address", mempty <$ (skipopts *> tok >>= addMeta "address"))
1662+ , ("signature", mempty <$ (skipopts *> authors))
1663+ , ("date", mempty <$ (skipopts *> tok >>= addMeta "date"))
1664+ -- Koma-script metadata commands
1665+ , ("dedication", mempty <$ (skipopts *> tok >>= addMeta "dedication"))
1666+ -- sectioning
1667+ , ("part", section nullAttr (-1))
1668+ , ("part*", section nullAttr (-1))
1669+ , ("chapter", section nullAttr 0)
1670+ , ("chapter*", section ("",["unnumbered"],[]) 0)
1671+ , ("section", section nullAttr 1)
1672+ , ("section*", section ("",["unnumbered"],[]) 1)
1673+ , ("subsection", section nullAttr 2)
1674+ , ("subsection*", section ("",["unnumbered"],[]) 2)
1675+ , ("subsubsection", section nullAttr 3)
1676+ , ("subsubsection*", section ("",["unnumbered"],[]) 3)
1677+ , ("paragraph", section nullAttr 4)
1678+ , ("paragraph*", section ("",["unnumbered"],[]) 4)
1679+ , ("subparagraph", section nullAttr 5)
1680+ , ("subparagraph*", section ("",["unnumbered"],[]) 5)
1681+ -- beamer slides
1682+ , ("frametitle", section nullAttr 3)
1683+ , ("framesubtitle", section nullAttr 4)
1684+ -- letters
1685+ , ("opening", (para . trimInlines) <$> (skipopts *> tok))
1686+ , ("closing", skipopts *> closing)
1687+ --
1688+ , ("hrule", pure horizontalRule)
1689+ , ("strut", pure mempty)
1690+ , ("rule", skipopts *> tok *> tok *> pure horizontalRule)
1691+ , ("item", looseItem)
1692+ , ("documentclass", skipopts *> braced *> preamble)
1693+ , ("centerline", (para . trimInlines) <$> (skipopts *> tok))
1694+ , ("caption", skipopts *> setCaption)
1695+ , ("bibliography", mempty <$ (skipopts *> braced >>=
1696+ addMeta "bibliography" . splitBibs . toksToString))
1697+ , ("addbibresource", mempty <$ (skipopts *> braced >>=
1698+ addMeta "bibliography" . splitBibs . toksToString))
1699+ -- includes
1700+ , ("lstinputlisting", inputListing)
1701+ , ("graphicspath", graphicsPath)
1702+ -- hyperlink
1703+ , ("hypertarget", try $ braced >> grouped block)
1704+ ]
1705+
1706+
1707+environments :: PandocMonad m => M.Map Text (LP m Blocks)
1708+environments = M.fromList
1709+ [ ("document", env "document" blocks)
1710+ , ("abstract", mempty <$ (env "abstract" blocks >>= addMeta "abstract"))
1711+ , ("letter", env "letter" letterContents)
1712+ , ("minipage", env "minipage" $
1713+ skipopts *> spaces *> optional braced *> spaces *> blocks)
1714+ , ("figure", env "figure" $ skipopts *> figure)
1715+ , ("subfigure", env "subfigure" $ skipopts *> tok *> figure)
1716+ , ("center", env "center" blocks)
1717+ , ("longtable", env "longtable" $
1718+ resetCaption *> simpTable "longtable" False >>= addTableCaption)
1719+ , ("table", env "table" $
1720+ resetCaption *> skipopts *> blocks >>= addTableCaption)
1721+ , ("tabular*", env "tabular" $ simpTable "tabular*" True)
1722+ , ("tabularx", env "tabularx" $ simpTable "tabularx" True)
1723+ , ("tabular", env "tabular" $ simpTable "tabular" False)
1724+ , ("quote", blockQuote <$> env "quote" blocks)
1725+ , ("quotation", blockQuote <$> env "quotation" blocks)
1726+ , ("verse", blockQuote <$> env "verse" blocks)
1727+ , ("itemize", bulletList <$> listenv "itemize" (many item))
1728+ , ("description", definitionList <$> listenv "description" (many descItem))
1729+ , ("enumerate", orderedList')
1730+ , ("alltt", alltt <$> env "alltt" blocks)
1731+ , ("code", guardEnabled Ext_literate_haskell *>
1732+ (codeBlockWith ("",["sourceCode","literate","haskell"],[]) <$>
1733+ verbEnv "code"))
1734+ , ("comment", mempty <$ verbEnv "comment")
1735+ , ("verbatim", codeBlock <$> verbEnv "verbatim")
1736+ , ("Verbatim", fancyverbEnv "Verbatim")
1737+ , ("BVerbatim", fancyverbEnv "BVerbatim")
1738+ , ("lstlisting", do attr <- parseListingsOptions <$> option [] keyvals
1739+ codeBlockWith attr <$> verbEnv "lstlisting")
1740+ , ("minted", minted)
1741+ , ("obeylines", obeylines)
1742+ , ("displaymath", mathEnvWith para Nothing "displaymath")
1743+ , ("equation", mathEnvWith para Nothing "equation")
1744+ , ("equation*", mathEnvWith para Nothing "equation*")
1745+ , ("gather", mathEnvWith para (Just "gathered") "gather")
1746+ , ("gather*", mathEnvWith para (Just "gathered") "gather*")
1747+ , ("multline", mathEnvWith para (Just "gathered") "multline")
1748+ , ("multline*", mathEnvWith para (Just "gathered") "multline*")
1749+ , ("eqnarray", mathEnvWith para (Just "aligned") "eqnarray")
1750+ , ("eqnarray*", mathEnvWith para (Just "aligned") "eqnarray*")
1751+ , ("align", mathEnvWith para (Just "aligned") "align")
1752+ , ("align*", mathEnvWith para (Just "aligned") "align*")
1753+ , ("alignat", mathEnvWith para (Just "aligned") "alignat")
1754+ , ("alignat*", mathEnvWith para (Just "aligned") "alignat*")
1755+ , ("tikzpicture", rawVerbEnv "tikzpicture")
1756+ ]
10061757
10071758environment :: PandocMonad m => LP m Blocks
10081759environment = do
10091760 controlSeq "begin"
1010− name <- braced
1761+ name <- untokenize <$> braced
10111762 M.findWithDefault mzero name environments
10121763 <|> rawEnv name
10131764
1014−inlineEnvironment :: PandocMonad m => LP m Inlines
1015−inlineEnvironment = try $ do
1016− controlSeq "begin"
1017− name <- braced
1018− M.findWithDefault mzero name inlineEnvironments
1765+env :: PandocMonad m => Text -> LP m a -> LP m a
1766+env name p = p <* end_ name
10191767
1020−rawEnv :: PandocMonad m => String -> LP m Blocks
1768+rawEnv :: PandocMonad m => Text -> LP m Blocks
10211769rawEnv name = do
10221770 exts <- getOption readerExtensions
10231771 let parseRaw = extensionEnabled Ext_raw_tex exts
10241772 rawOptions <- mconcat <$> many rawopt
1025− let beginCommand = "\\begin{" ++ name ++ "}" ++ rawOptions
1773+ let beginCommand = "\\begin{" <> name <> "}" <> rawOptions
10261774 pos1 <- getPosition
10271775 (bs, raw) <- withRaw $ env name blocks
1028− raw' <- applyMacros' $ beginCommand ++ raw
1029− if raw' /= beginCommand ++ raw
1030− then parseFromString' blocks raw'
1031− else if parseRaw
1032− then return $ rawBlock "latex" $ beginCommand ++ raw'
1033− else do
1034− unless parseRaw $ do
1035− report $ SkippedContent beginCommand pos1
1036− pos2 <- getPosition
1037− report $ SkippedContent ("\\end{" ++ name ++ "}") pos2
1038− return bs
1039−
1040−rawVerbEnv :: PandocMonad m => String -> LP m Blocks
1776+ if parseRaw
1777+ then return $ rawBlock "latex"
1778+ $ T.unpack $ beginCommand <> untokenize raw
1779+ else do
1780+ unless parseRaw $ do
1781+ report $ SkippedContent (T.unpack beginCommand) pos1
1782+ pos2 <- getPosition
1783+ report $ SkippedContent ("\\end{" ++ T.unpack name ++ "}") pos2
1784+ return bs
1785+
1786+rawVerbEnv :: PandocMonad m => Text -> LP m Blocks
10411787rawVerbEnv name = do
10421788 pos <- getPosition
10431789 (_, raw) <- withRaw $ verbEnv name
1044− let raw' = "\\begin{tikzpicture}" ++ raw
1790+ let raw' = "\\begin{tikzpicture}" ++ toksToString raw
10451791 exts <- getOption readerExtensions
10461792 let parseRaw = extensionEnabled Ext_raw_tex exts
10471793 if parseRaw
@@ -1050,36 +1796,106 @@ rawVerbEnv name = do
10501796 report $ SkippedContent raw' pos
10511797 return mempty
10521798
1053−----
1799+verbEnv :: PandocMonad m => Text -> LP m String
1800+verbEnv name = withVerbatimMode $ do
1801+ skipopts
1802+ optional blankline
1803+ res <- manyTill anyTok (end_ name)
1804+ return $ stripTrailingNewlines $ toksToString res
1805+
1806+fancyverbEnv :: PandocMonad m => Text -> LP m Blocks
1807+fancyverbEnv name = do
1808+ options <- option [] keyvals
1809+ let kvs = [ (if k == "firstnumber"
1810+ then "startFrom"
1811+ else k, v) | (k,v) <- options ]
1812+ let classes = [ "numberLines" |
1813+ lookup "numbers" options == Just "left" ]
1814+ let attr = ("",classes,kvs)
1815+ codeBlockWith attr <$> verbEnv name
1816+
1817+obeylines :: PandocMonad m => LP m Blocks
1818+obeylines = do
1819+ para . fromList . removeLeadingTrailingBreaks .
1820+ walk softBreakToHard . toList <$> env "obeylines" inlines
1821+ where softBreakToHard SoftBreak = LineBreak
1822+ softBreakToHard x = x
1823+ removeLeadingTrailingBreaks = reverse . dropWhile isLineBreak .
1824+ reverse . dropWhile isLineBreak
1825+ isLineBreak LineBreak = True
1826+ isLineBreak _ = False
1827+
1828+minted :: PandocMonad m => LP m Blocks
1829+minted = do
1830+ options <- option [] keyvals
1831+ lang <- toksToString <$> braced
1832+ let kvs = [ (if k == "firstnumber"
1833+ then "startFrom"
1834+ else k, v) | (k,v) <- options ]
1835+ let classes = [ lang | not (null lang) ] ++
1836+ [ "numberLines" |
1837+ lookup "linenos" options == Just "true" ]
1838+ let attr = ("",classes,kvs)
1839+ codeBlockWith attr <$> verbEnv "minted"
1840+
1841+letterContents :: PandocMonad m => LP m Blocks
1842+letterContents = do
1843+ bs <- blocks
1844+ st <- getState
1845+ -- add signature (author) and address (title)
1846+ let addr = case lookupMeta "address" (sMeta st) of
1847+ Just (MetaBlocks [Plain xs]) ->
1848+ para $ trimInlines $ fromList xs
1849+ _ -> mempty
1850+ return $ addr <> bs -- sig added by \closing
1851+
1852+figure :: PandocMonad m => LP m Blocks
1853+figure = try $ do
1854+ resetCaption
1855+ blocks >>= addImageCaption
1856+
1857+addImageCaption :: PandocMonad m => Blocks -> LP m Blocks
1858+addImageCaption = walkM go
1859+ where go (Image attr alt (src,tit))
1860+ | not ("fig:" `isPrefixOf` tit) = do
1861+ mbcapt <- sCaption <$> getState
1862+ return $ case mbcapt of
1863+ Just ils -> Image attr (toList ils) (src, "fig:" ++ tit)
1864+ Nothing -> Image attr alt (src,tit)
1865+ go x = return x
10541866
1055−maybeAddExtension :: String -> FilePath -> FilePath
1056−maybeAddExtension ext fp =
1057− if null (takeExtension fp)
1058− then addExtension fp ext
1059− else fp
1867+graphicsPath :: PandocMonad m => LP m Blocks
1868+graphicsPath = do
1869+ ps <- map toksToString <$> (bgroup *> manyTill braced egroup)
1870+ getResourcePath >>= setResourcePath . (++ ps)
1871+ return mempty
10601872
1061−include :: PandocMonad m => LP m Blocks
1062−include = do
1063− fs' <- try $ do
1064− char '\\'
1065− name <- try (string "include")
1066− <|> try (string "input")
1067− <|> try (string "subfile")
1068− <|> string "usepackage"
1069− -- skip options
1070− skipMany $ try $ char '[' *> manyTill anyChar (char ']')
1071− fs <- (map trim . splitBy (==',')) <$> braced
1072− return $ if name == "usepackage"
1073− then map (maybeAddExtension ".sty") fs
1074− else map (maybeAddExtension ".tex") fs
1075− dirs <- (splitBy (==':') . fromMaybe ".") <$> lookupEnv "TEXINPUTS"
1076− mconcat <$> mapM (insertIncludedFile blocks dirs) fs'
1873+splitBibs :: String -> [Inlines]
1874+splitBibs = map (str . flip replaceExtension "bib" . trim) . splitBy (==',')
1875+
1876+alltt :: Blocks -> Blocks
1877+alltt = walk strToCode
1878+ where strToCode (Str s) = Code nullAttr s
1879+ strToCode Space = RawInline (Format "latex") "\\ "
1880+ strToCode SoftBreak = LineBreak
1881+ strToCode x = x
1882+
1883+parseListingsOptions :: [(String, String)] -> Attr
1884+parseListingsOptions options =
1885+ let kvs = [ (if k == "firstnumber"
1886+ then "startFrom"
1887+ else k, v) | (k,v) <- options ]
1888+ classes = [ "numberLines" |
1889+ lookup "numbers" options == Just "left" ]
1890+ ++ maybeToList (lookup "language" options
1891+ >>= fromListingsLanguage)
1892+ in (fromMaybe "" (lookup "label" options), classes, kvs)
10771893
10781894inputListing :: PandocMonad m => LP m Blocks
10791895inputListing = do
10801896 pos <- getPosition
10811897 options <- option [] keyvals
1082− f <- filter (/='"') <$> braced
1898+ f <- filter (/='"') . toksToString <$> braced
10831899 dirs <- (splitBy (==':') . fromMaybe ".") <$> lookupEnv "TEXINPUTS"
10841900 mbCode <- readFileFromDirs dirs f
10851901 codeLines <- case mbCode of
@@ -1098,169 +1914,10 @@ inputListing = do
10981914 drop (firstline - 1) codeLines
10991915 return $ codeBlockWith (ident,ordNub (classes ++ language),kvs) codeContents
11001916
1101−parseListingsOptions :: [(String, String)] -> Attr
1102−parseListingsOptions options =
1103− let kvs = [ (if k == "firstnumber"
1104− then "startFrom"
1105− else k, v) | (k,v) <- options ]
1106− classes = [ "numberLines" |
1107− lookup "numbers" options == Just "left" ]
1108− ++ maybeToList (lookup "language" options
1109− >>= fromListingsLanguage)
1110− in (fromMaybe "" (lookup "label" options), classes, kvs)
1111−
1112−----
1113−
1114−keyval :: PandocMonad m => LP m (String, String)
1115−keyval = try $ do
1116− key <- many1 alphaNum
1117− val <- option "" $ char '=' >> braced <|> (many1 (alphaNum <|> oneOf ".:-|\\"))
1118− skipMany spaceChar
1119− optional (char ',')
1120− skipMany spaceChar
1121− return (key, val)
1122−
1123−
1124−keyvals :: PandocMonad m => LP m [(String, String)]
1125−keyvals = try $ char '[' *> manyTill keyval (char ']')
1126−
1127−alltt :: PandocMonad m => String -> LP m Blocks
1128−alltt t = walk strToCode <$> parseFromString' blocks
1129− (substitute " " "\\ " $ substitute "%" "\\%" $
1130− intercalate "\\\\\n" $ lines t)
1131− where strToCode (Str s) = Code nullAttr s
1132− strToCode x = x
1133−
1134−rawLaTeXBlock :: PandocMonad m => LP m String
1135−rawLaTeXBlock = snd <$> try (withRaw (environment <|> blockCommand))
1136−
1137−rawLaTeXInline :: PandocMonad m => LP m Inline
1138−rawLaTeXInline = do
1139− raw <- (snd <$> withRaw inlineCommand)
1140− <|> (snd <$> withRaw inlineEnvironment)
1141− <|> (snd <$> withRaw blockCommand)
1142− RawInline "latex" <$> applyMacros' raw
1143−
1144−addImageCaption :: PandocMonad m => Blocks -> LP m Blocks
1145−addImageCaption = walkM go
1146− where go (Image attr alt (src,tit))
1147− | not ("fig:" `isPrefixOf` tit) = do
1148− mbcapt <- stateCaption <$> getState
1149− return $ case mbcapt of
1150− Just ils -> Image attr (toList ils) (src, "fig:" ++ tit)
1151− Nothing -> Image attr alt (src,tit)
1152− go x = return x
1153−
1154−addTableCaption :: PandocMonad m => Blocks -> LP m Blocks
1155−addTableCaption = walkM go
1156− where go (Table c als ws hs rs) = do
1157− mbcapt <- stateCaption <$> getState
1158− return $ case mbcapt of
1159− Just ils -> Table (toList ils) als ws hs rs
1160− Nothing -> Table c als ws hs rs
1161− go x = return x
1162−
1163−environments :: PandocMonad m => M.Map String (LP m Blocks)
1164−environments = M.fromList
1165− [ ("document", env "document" blocks <* skipMany anyChar)
1166− , ("abstract", mempty <$ (env "abstract" blocks >>= addMeta "abstract"))
1167− , ("letter", env "letter" letterContents)
1168− , ("minipage", env "minipage" $
1169− skipopts *> spaces' *> optional braced *> spaces' *> blocks)
1170− , ("figure", env "figure" $ skipopts *> figure)
1171− , ("subfigure", env "subfigure" $ skipopts *> tok *> figure)
1172− , ("center", env "center" blocks)
1173− , ("longtable", env "longtable" $
1174− resetCaption *> simpTable "longtable" False >>= addTableCaption)
1175− , ("table", env "table" $
1176− resetCaption *> skipopts *> blocks >>= addTableCaption)
1177− , ("tabular*", env "tabular" $ simpTable "tabular*" True)
1178− , ("tabularx", env "tabularx" $ simpTable "tabularx" True)
1179− , ("tabular", env "tabular" $ simpTable "tabular" False)
1180− , ("quote", blockQuote <$> env "quote" blocks)
1181− , ("quotation", blockQuote <$> env "quotation" blocks)
1182− , ("verse", blockQuote <$> env "verse" blocks)
1183− , ("itemize", bulletList <$> listenv "itemize" (many item))
1184− , ("description", definitionList <$> listenv "description" (many descItem))
1185− , ("enumerate", orderedList')
1186− , ("alltt", alltt =<< verbEnv "alltt")
1187− , ("code", guardEnabled Ext_literate_haskell *>
1188− (codeBlockWith ("",["sourceCode","literate","haskell"],[]) <$>
1189− verbEnv "code"))
1190− , ("comment", mempty <$ verbEnv "comment")
1191− , ("verbatim", codeBlock <$> verbEnv "verbatim")
1192− , ("Verbatim", fancyverbEnv "Verbatim")
1193− , ("BVerbatim", fancyverbEnv "BVerbatim")
1194− , ("lstlisting", do attr <- parseListingsOptions <$> option [] keyvals
1195− codeBlockWith attr <$> verbEnv "lstlisting")
1196− , ("minted", do options <- option [] keyvals
1197− lang <- grouped (many1 $ satisfy (/='}'))
1198− let kvs = [ (if k == "firstnumber"
1199− then "startFrom"
1200− else k, v) | (k,v) <- options ]
1201− let classes = [ lang | not (null lang) ] ++
1202− [ "numberLines" |
1203− lookup "linenos" options == Just "true" ]
1204− let attr = ("",classes,kvs)
1205− codeBlockWith attr <$> verbEnv "minted")
1206− , ("obeylines", parseFromString
1207− (para . trimInlines . mconcat <$> many inline) =<<
1208− intercalate "\\\\\n" . lines <$> verbEnv "obeylines")
1209− , ("displaymath", mathEnvWith para Nothing "displaymath")
1210− , ("equation", mathEnvWith para Nothing "equation")
1211− , ("equation*", mathEnvWith para Nothing "equation*")
1212− , ("gather", mathEnvWith para (Just "gathered") "gather")
1213− , ("gather*", mathEnvWith para (Just "gathered") "gather*")
1214− , ("multline", mathEnvWith para (Just "gathered") "multline")
1215− , ("multline*", mathEnvWith para (Just "gathered") "multline*")
1216− , ("eqnarray", mathEnvWith para (Just "aligned") "eqnarray")
1217− , ("eqnarray*", mathEnvWith para (Just "aligned") "eqnarray*")
1218− , ("align", mathEnvWith para (Just "aligned") "align")
1219− , ("align*", mathEnvWith para (Just "aligned") "align*")
1220− , ("alignat", mathEnvWith para (Just "aligned") "alignat")
1221− , ("alignat*", mathEnvWith para (Just "aligned") "alignat*")
1222− , ("tikzpicture", rawVerbEnv "tikzpicture")
1223− ]
1224−
1225−figure :: PandocMonad m => LP m Blocks
1226−figure = try $ do
1227− resetCaption
1228− blocks >>= addImageCaption
1229−
1230−letterContents :: PandocMonad m => LP m Blocks
1231−letterContents = do
1232− bs <- blocks
1233− st <- getState
1234− -- add signature (author) and address (title)
1235− let addr = case lookupMeta "address" (stateMeta st) of
1236− Just (MetaBlocks [Plain xs]) ->
1237− para $ trimInlines $ fromList xs
1238− _ -> mempty
1239− return $ addr <> bs -- sig added by \closing
1240−
1241−closing :: PandocMonad m => LP m Blocks
1242−closing = do
1243− contents <- tok
1244− st <- getState
1245− let extractInlines (MetaBlocks [Plain ys]) = ys
1246− extractInlines (MetaBlocks [Para ys ]) = ys
1247− extractInlines _ = []
1248− let sigs = case lookupMeta "author" (stateMeta st) of
1249− Just (MetaList xs) ->
1250− para $ trimInlines $ fromList $
1251− intercalate [LineBreak] $ map extractInlines xs
1252− _ -> mempty
1253− return $ para (trimInlines contents) <> sigs
1917+-- lists
12541918
12551919item :: PandocMonad m => LP m Blocks
1256−item = blocks *> controlSeq "item" *> skipopts *> blocks
1257−
1258−looseItem :: PandocMonad m => LP m Blocks
1259−looseItem = do
1260− ctx <- stateParserContext `fmap` getState
1261− if ctx == ListItemState
1262− then mzero
1263− else return mempty
1920+item = void blocks *> controlSeq "item" *> skipopts *> blocks
12641921
12651922descItem :: PandocMonad m => LP m (Inlines, [Blocks])
12661923descItem = do
@@ -1271,302 +1928,210 @@ descItem = do
12711928 bs <- blocks
12721929 return (ils, [bs])
12731930
1274−env :: PandocMonad m => String -> LP m a -> LP m a
1275−env name p = p <*
1276− (try (controlSeq "end" *> braced >>= guard . (== name))
1277− <?> ("\\end{" ++ name ++ "}"))
1278−
1279−listenv :: PandocMonad m => String -> LP m a -> LP m a
1931+listenv :: PandocMonad m => Text -> LP m a -> LP m a
12801932listenv name p = try $ do
1281− oldCtx <- stateParserContext `fmap` getState
1282− updateState $ \st -> st{ stateParserContext = ListItemState }
1933+ oldInListItem <- sInListItem `fmap` getState
1934+ updateState $ \st -> st{ sInListItem = True }
12831935 res <- env name p
1284− updateState $ \st -> st{ stateParserContext = oldCtx }
1936+ updateState $ \st -> st{ sInListItem = oldInListItem }
12851937 return res
12861938
1287−mathEnvWith :: PandocMonad m
1288− => (Inlines -> a) -> Maybe String -> String -> LP m a
1289−mathEnvWith f innerEnv name = f <$> mathDisplay (inner <$> mathEnv name)
1290− where inner x = case innerEnv of
1291− Nothing -> x
1292− Just y -> "\\begin{" ++ y ++ "}\n" ++ x ++
1293− "\\end{" ++ y ++ "}"
1294−
1295−mathEnv :: PandocMonad m => String -> LP m String
1296−mathEnv name = do
1297− skipopts
1298− optional blankline
1299− let endEnv = try $ controlSeq "end" *> braced >>= guard . (== name)
1300− charMuncher = skipMany comment *>
1301− (many1 (noneOf "\\%") <|> try (string "\\%")
1302− <|> try (string "\\\\") <|> count 1 anyChar)
1303− res <- concat <$> manyTill charMuncher endEnv
1304− return $ stripTrailingNewlines res
1305−
1306−verbEnv :: PandocMonad m => String -> LP m String
1307−verbEnv name = do
1308− skipopts
1309− optional blankline
1310− let endEnv = try $ controlSeq "end" *> braced >>= guard . (== name)
1311− charMuncher = anyChar
1312− res <- manyTill charMuncher endEnv
1313− return $ stripTrailingNewlines res
1314−
1315−fancyverbEnv :: PandocMonad m => String -> LP m Blocks
1316−fancyverbEnv name = do
1317− options <- option [] keyvals
1318− let kvs = [ (if k == "firstnumber"
1319− then "startFrom"
1320− else k, v) | (k,v) <- options ]
1321− let classes = [ "numberLines" |
1322− lookup "numbers" options == Just "left" ]
1323− let attr = ("",classes,kvs)
1324− codeBlockWith attr <$> verbEnv name
1325−
13261939orderedList' :: PandocMonad m => LP m Blocks
13271940orderedList' = try $ do
1328− optional sp
1329− (_, style, delim) <- option (1, DefaultStyle, DefaultDelim) $
1330− try $ char '[' *> anyOrderedListMarker <* char ']'
13311941 spaces
1332− optional $ try $ controlSeq "setlength" *> grouped (controlSeq "itemindent") *> braced
1942+ let markerSpec = do
1943+ symbol '['
1944+ ts <- toksToString <$> manyTill anyTok (symbol ']')
1945+ case runParser anyOrderedListMarker def "option" ts of
1946+ Right r -> return r
1947+ Left _ -> do
1948+ pos <- getPosition
1949+ report $ SkippedContent ("[" ++ ts ++ "]") pos
1950+ return (1, DefaultStyle, DefaultDelim)
1951+ (_, style, delim) <- option (1, DefaultStyle, DefaultDelim) markerSpec
13331952 spaces
1334− start <- option 1 $ try $ do controlSeq "setcounter"
1335− grouped (string "enum" *> many1 (oneOf "iv"))
1953+ optional $ try $ controlSeq "setlength"
1954+ *> grouped (count 1 $ controlSeq "itemindent")
1955+ *> braced
1956+ spaces
1957+ start <- option 1 $ try $ do pos <- getPosition
1958+ controlSeq "setcounter"
1959+ ctr <- toksToString <$> braced
1960+ guard $ "enum" `isPrefixOf` ctr
1961+ guard $ all (`elem` ['i','v']) (drop 4 ctr)
13361962 optional sp
1337− num <- grouped (many1 digit)
1338− spaces
1339− return (read num + 1 :: Int)
1963+ num <- toksToString <$> braced
1964+ case safeRead num of
1965+ Just i -> return (i + 1 :: Int)
1966+ Nothing -> do
1967+ report $ SkippedContent
1968+ ("\\setcounter{" ++ ctr ++
1969+ "}{" ++ num ++ "}") pos
1970+ return 1
13401971 bs <- listenv "enumerate" (many item)
13411972 return $ orderedListWith (start, style, delim) bs
13421973
1343−paragraph :: PandocMonad m => LP m Blocks
1344−paragraph = do
1345− x <- trimInlines . mconcat <$> many1 inline
1346− if x == mempty
1347− then return mempty
1348− else return $ para x
1349−
1350−preamble :: PandocMonad m => LP m Blocks
1351−preamble = mempty <$> manyTill preambleBlock beginDoc
1352− where beginDoc = lookAhead $ try $ controlSeq "begin" *> string "{document}"
1353− preambleBlock = void comment
1354− <|> void sp
1355− <|> void blanklines
1356− <|> void include
1357− <|> void macro
1358− <|> void blockCommand
1359− <|> void anyControlSeq
1360− <|> void braced
1361− <|> void anyChar
1362−
1363−-------
1364−
1365−-- citations
1366−
1367−addPrefix :: [Inline] -> [Citation] -> [Citation]
1368−addPrefix p (k:ks) = k {citationPrefix = p ++ citationPrefix k} : ks
1369−addPrefix _ _ = []
1370−
1371−addSuffix :: [Inline] -> [Citation] -> [Citation]
1372−addSuffix s ks@(_:_) =
1373− let k = last ks
1374− in init ks ++ [k {citationSuffix = citationSuffix k ++ s}]
1375−addSuffix _ _ = []
1376−
1377−simpleCiteArgs :: PandocMonad m => LP m [Citation]
1378−simpleCiteArgs = try $ do
1379− first <- optionMaybe $ toList <$> opt
1380− second <- optionMaybe $ toList <$> opt
1381− keys <- try $ bgroup *> (manyTill citationLabel egroup)
1382− let (pre, suf) = case (first , second ) of
1383− (Just s , Nothing) -> (mempty, s )
1384− (Just s , Just t ) -> (s , t )
1385− _ -> (mempty, mempty)
1386− conv k = Citation { citationId = k
1387− , citationPrefix = []
1388− , citationSuffix = []
1389− , citationMode = NormalCitation
1390− , citationHash = 0
1391− , citationNoteNum = 0
1392− }
1393− return $ addPrefix pre $ addSuffix suf $ map conv keys
1394−
1395−citationLabel :: PandocMonad m => LP m String
1396−citationLabel = optional sp *>
1397− (many1 (satisfy isBibtexKeyChar)
1398− <* optional sp
1399− <* optional (char ',')
1400− <* optional sp)
1401− where isBibtexKeyChar c = isAlphaNum c || c `elem` (".:;?!`'()/*@_+=-[]" :: String)
1974+-- tables
14021975
1403−cites :: PandocMonad m => CitationMode -> Bool -> LP m [Citation]
1404−cites mode multi = try $ do
1405− cits <- if multi
1406− then many1 simpleCiteArgs
1407− else count 1 simpleCiteArgs
1408− let cs = concat cits
1409− return $ case mode of
1410− AuthorInText -> case cs of
1411− (c:rest) -> c {citationMode = mode} : rest
1412− [] -> []
1413− _ -> map (\a -> a {citationMode = mode}) cs
1976+hline :: PandocMonad m => LP m ()
1977+hline = try $ do
1978+ spaces
1979+ controlSeq "hline" <|>
1980+ -- booktabs rules:
1981+ controlSeq "toprule" <|>
1982+ controlSeq "bottomrule" <|>
1983+ controlSeq "midrule" <|>
1984+ controlSeq "endhead" <|>
1985+ controlSeq "endfirsthead"
1986+ spaces
1987+ optional $ bracketed inline
1988+ return ()
14141989
1415−citation :: PandocMonad m => String -> CitationMode -> Bool -> LP m Inlines
1416−citation name mode multi = do
1417− (c,raw) <- withRaw $ cites mode multi
1418− return $ cite c (rawInline "latex" $ "\\" ++ name ++ raw)
1990+lbreak :: PandocMonad m => LP m Tok
1991+lbreak = (controlSeq "\\" <|> controlSeq "tabularnewline") <* spaces
14191992
1420−complexNatbibCitation :: PandocMonad m => CitationMode -> LP m Inlines
1421−complexNatbibCitation mode = try $ do
1422− let ils = (toList . trimInlines . mconcat) <$>
1423− many (notFollowedBy (oneOf "\\};") >> inline)
1424− let parseOne = try $ do
1425− skipSpaces
1426− pref <- ils
1427− cit' <- inline -- expect a citation
1428− let citlist = toList cit'
1429− cits' <- case citlist of
1430− [Cite cs _] -> return cs
1431− _ -> mzero
1432− suff <- ils
1433− skipSpaces
1434− optional $ char ';'
1435− return $ addPrefix pref $ addSuffix suff cits'
1436− (c:cits, raw) <- withRaw $ grouped parseOne
1437− return $ cite (c{ citationMode = mode }:cits)
1438− (rawInline "latex" $ "\\citetext" ++ raw)
1993+amp :: PandocMonad m => LP m Tok
1994+amp = symbol '&'
14391995
1440−-- tables
1996+-- Split a Word into individual Symbols (for parseAligns)
1997+splitWordTok :: PandocMonad m => LP m ()
1998+splitWordTok = do
1999+ inp <- getInput
2000+ case inp of
2001+ (Tok spos Word t : rest) -> do
2002+ setInput $ map (Tok spos Symbol . T.singleton) (T.unpack t) ++ rest
2003+ _ -> return ()
14412004
1442−parseAligns :: PandocMonad m => LP m [(Alignment, Double, (String, String))]
2005+parseAligns :: PandocMonad m => LP m [(Alignment, Double, ([Tok], [Tok]))]
14432006parseAligns = try $ do
1444− bgroup
1445− let maybeBar = skipMany $ sp <|> () <$ char '|' <|> () <$ (char '@' >> braced)
1446− maybeBar
1447− let cAlign = AlignCenter <$ char 'c'
1448− let lAlign = AlignLeft <$ char 'l'
1449− let rAlign = AlignRight <$ char 'r'
1450− let parAlign = AlignLeft <$ char 'p'
1451− -- algins from tabularx
1452− let xAlign = AlignLeft <$ char 'X'
1453− let mAlign = AlignLeft <$ char 'm'
1454− let bAlign = AlignLeft <$ char 'b'
1455− let alignChar = cAlign <|> lAlign <|> rAlign <|> parAlign
1456− <|> xAlign <|> mAlign <|> bAlign
1457− let alignPrefix = char '>' >> braced
1458− let alignSuffix = char '<' >> braced
2007+ let maybeBar = skipMany $
2008+ sp <|> () <$ symbol '|' <|> () <$ (symbol '@' >> braced)
2009+ let cAlign = AlignCenter <$ symbol 'c'
2010+ let lAlign = AlignLeft <$ symbol 'l'
2011+ let rAlign = AlignRight <$ symbol 'r'
2012+ let parAlign = AlignLeft <$ symbol 'p'
2013+ -- aligns from tabularx
2014+ let xAlign = AlignLeft <$ symbol 'X'
2015+ let mAlign = AlignLeft <$ symbol 'm'
2016+ let bAlign = AlignLeft <$ symbol 'b'
2017+ let alignChar = splitWordTok *> ( cAlign <|> lAlign <|> rAlign <|> parAlign
2018+ <|> xAlign <|> mAlign <|> bAlign )
2019+ let alignPrefix = symbol '>' >> braced
2020+ let alignSuffix = symbol '<' >> braced
14592021 let colWidth = try $ do
1460− char '{'
1461− ds <- many1 (oneOf "0123456789.")
2022+ symbol '{'
2023+ ds <- trim . toksToString <$> manyTill anyTok (controlSeq "linewidth")
14622024 spaces
1463− string "\\linewidth"
1464− char '}'
2025+ symbol '}'
14652026 case safeRead ds of
14662027 Just w -> return w
14672028 Nothing -> return 0.0
1468− let alignSpec = do
2029+ let alignSpec = try $ do
14692030 spaces
1470− pref <- option "" alignPrefix
2031+ pref <- option [] alignPrefix
14712032 spaces
14722033 al <- alignChar
1473− width <- colWidth <|> option 0.0 (do s <- braced
2034+ width <- colWidth <|> option 0.0 (do s <- toksToString <$> braced
14742035 pos <- getPosition
14752036 report $ SkippedContent s pos
14762037 return 0.0)
14772038 spaces
1478− suff <- option "" alignSuffix
2039+ suff <- option [] alignSuffix
14792040 return (al, width, (pref, suff))
1480− aligns' <- sepEndBy alignSpec maybeBar
2041+ bgroup
2042+ spaces
2043+ maybeBar
2044+ aligns' <- many (alignSpec <* maybeBar)
14812045 spaces
14822046 egroup
14832047 spaces
1484− return $ aligns'
1485−
1486−hline :: PandocMonad m => LP m ()
1487−hline = try $ do
1488− spaces'
1489− controlSeq "hline" <|>
1490− -- booktabs rules:
1491− controlSeq "toprule" <|>
1492− controlSeq "bottomrule" <|>
1493− controlSeq "midrule" <|>
1494− controlSeq "endhead" <|>
1495− controlSeq "endfirsthead"
1496− spaces'
1497− optional $ bracketed (many1 (satisfy (/=']')))
1498− return ()
1499−
1500−lbreak :: PandocMonad m => LP m ()
1501−lbreak = () <$ try (spaces' *>
1502− (controlSeq "\\" <|> controlSeq "tabularnewline") <*
1503− spaces')
1504−
1505−amp :: PandocMonad m => LP m ()
1506−amp = () <$ try (spaces' *> char '&' <* spaces')
2048+ return aligns'
15072049
15082050parseTableRow :: PandocMonad m
1509− => String -- ^ table environment name
1510− -> [(String, String)] -- ^ pref/suffixes
2051+ => Text -- ^ table environment name
2052+ -> [([Tok], [Tok])] -- ^ pref/suffixes
15112053 -> LP m [Blocks]
1512−parseTableRow envname prefsufs = try $ do
2054+parseTableRow envname prefsufs = do
2055+ notFollowedBy (spaces *> end_ envname)
15132056 let cols = length prefsufs
1514− let tableCellRaw = concat <$> many
1515− (do notFollowedBy amp
1516− notFollowedBy lbreak
1517− notFollowedBy $ () <$ try (string ("\\end{" ++ envname ++ "}"))
1518− many1 (noneOf "&%\n\r\\")
1519− <|> try (string "\\&")
1520− <|> count 1 anyChar)
1521− let plainify bs = case toList bs of
1522− [Para ils] -> plain (fromList ils)
1523− _ -> bs
1524− rawcells <- sepBy1 tableCellRaw amp
1525− guard $ length rawcells == cols
1526− let rawcells' = zipWith (\c (p, s) -> p ++ trim c ++ s) rawcells prefsufs
1527− let tableCell = plainify <$> blocks
1528− cells' <- mapM (parseFromString' tableCell) rawcells'
1529− let numcells = length cells'
2057+ -- add prefixes and suffixes in token stream:
2058+ let celltoks (pref, suff) = do
2059+ prefpos <- getPosition
2060+ contents <- many (notFollowedBy
2061+ (() <$ amp <|> () <$ lbreak <|> end_ envname)
2062+ >> anyTok)
2063+ suffpos <- getPosition
2064+ option [] (count 1 amp)
2065+ return $ map (setpos (sourceLine prefpos, sourceColumn prefpos)) pref
2066+ ++ contents ++
2067+ map (setpos (sourceLine suffpos, sourceColumn suffpos)) suff
2068+ rawcells <- sequence (map celltoks prefsufs)
2069+ oldInput <- getInput
2070+ cells <- sequence $ map (\ts -> setInput ts >> parseTableCell) rawcells
2071+ setInput oldInput
2072+ spaces
2073+ let numcells = length cells
15302074 guard $ numcells <= cols && numcells >= 1
1531− guard $ cells' /= [mempty]
2075+ guard $ cells /= [mempty]
15322076 -- note: a & b in a three-column table leaves an empty 3rd cell:
1533− let cells'' = cells' ++ replicate (cols - numcells) mempty
1534− spaces'
1535− return cells''
2077+ return $ cells ++ replicate (cols - numcells) mempty
15362078
1537−spaces' :: PandocMonad m => LP m ()
1538−spaces' = spaces *> skipMany (comment *> spaces)
2079+parseTableCell :: PandocMonad m => LP m Blocks
2080+parseTableCell = do
2081+ let plainify bs = case toList bs of
2082+ [Para ils] -> plain (fromList ils)
2083+ _ -> bs
2084+ updateState $ \st -> st{ sInTableCell = True }
2085+ cells <- plainify <$> blocks
2086+ updateState $ \st -> st{ sInTableCell = False }
2087+ return cells
15392088
1540−simpTable :: PandocMonad m => String -> Bool -> LP m Blocks
2089+simpTable :: PandocMonad m => Text -> Bool -> LP m Blocks
15412090simpTable envname hasWidthParameter = try $ do
1542− when hasWidthParameter $ () <$ (spaces' >> tok)
2091+ when hasWidthParameter $ () <$ (spaces >> tok)
15432092 skipopts
15442093 colspecs <- parseAligns
15452094 let (aligns, widths, prefsufs) = unzip3 colspecs
15462095 let cols = length colspecs
15472096 optional $ controlSeq "caption" *> skipopts *> setCaption
15482097 optional lbreak
1549− spaces'
2098+ spaces
15502099 skipMany hline
1551− spaces'
2100+ spaces
15522101 header' <- option [] $ try (parseTableRow envname prefsufs <*
15532102 lbreak <* many1 hline)
1554− spaces'
2103+ spaces
15552104 rows <- sepEndBy (parseTableRow envname prefsufs)
15562105 (lbreak <* optional (skipMany hline))
1557− spaces'
2106+ spaces
15582107 optional $ controlSeq "caption" *> skipopts *> setCaption
15592108 optional lbreak
1560− spaces'
2109+ spaces
15612110 let header'' = if null header'
15622111 then replicate cols mempty
15632112 else header'
15642113 lookAhead $ controlSeq "end" -- make sure we're at end
15652114 return $ table mempty (zip aligns widths) header'' rows
15662115
1567−removeDoubleQuotes :: String -> String
1568−removeDoubleQuotes ('"':xs) =
1569− case reverse xs of
1570− '"':ys -> reverse ys
1571− _ -> '"':xs
1572−removeDoubleQuotes xs = xs
2116+addTableCaption :: PandocMonad m => Blocks -> LP m Blocks
2117+addTableCaption = walkM go
2118+ where go (Table c als ws hs rs) = do
2119+ mbcapt <- sCaption <$> getState
2120+ return $ case mbcapt of
2121+ Just ils -> Table (toList ils) als ws hs rs
2122+ Nothing -> Table c als ws hs rs
2123+ go x = return x
2124+
2125+
2126+block :: PandocMonad m => LP m Blocks
2127+block = (mempty <$ spaces1)
2128+ <|> environment
2129+ <|> include
2130+ <|> macroDef
2131+ <|> paragraph
2132+ <|> blockCommand
2133+ <|> grouped block
2134+
2135+blocks :: PandocMonad m => LP m Blocks
2136+blocks = mconcat <$> many block
2137+