52df18f · typeable/pandoc

git.iko.soy/typeable/pandoc

Groff tokenizer: introduce TableRow type, handle .T&.

Closes #5020.

commit52df18f476b7eb7935c2c01f566fa1adee4a8621 authorJohn MacFarlane <jgm@berkeley.edu> dateFri 26 Oct 2018, 21:22 (8 years ago) parent7f70aaa

2 files changed+41−23

compare with master

unifiedsplit

src/Text/Pandoc/Readers/Groff.hs+29−12 src/Text/Pandoc/Readers/Man.hs+12−11

src/Text/Pandoc/Readers/Groff.hs+29−12

@@ -38,7 +38,8 @@ module Text.Pandoc.Readers.Groff

3838 , LinePart(..)

3939 , Arg

4040 , TableOption

41− , TableFormat(..)

41+ , CellFormat(..)

42+ , TableRow

4243 , GroffToken(..)

4344 , GroffTokens(..)

4445 , linePartsToString

@@ -47,6 +48,7 @@ module Text.Pandoc.Readers.Groff

4748where

4849

4950import Prelude

51+import Safe (lastDef)

5052import Control.Monad (void, mzero, guard, when)

5153import Control.Monad.Except (throwError)

5254import Text.Pandoc.Class

@@ -92,20 +94,21 @@ type Arg = [LinePart]

9294

9395type TableOption = (String, String)

9496

95−data TableFormat =

96− TableFormat

97+data CellFormat =

98+ CellFormat

9799 { columnType :: Char

98100 , pipePrefix :: Bool

99101 , pipeSuffix :: Bool

100102 , columnSuffixes :: [String]

101103 } deriving (Show, Eq, Ord)

102104

105+type TableRow = ([CellFormat], [GroffTokens])

103106

104107-- TODO parse tables (see man tbl)

105108data GroffToken = MLine [LinePart]

106109 | MEmptyLine

107110 | MMacro MacroKind [Arg] SourcePos

108− | MTable [TableOption] [[TableFormat]] [[GroffTokens]] SourcePos

111+ | MTable [TableOption] [TableRow] SourcePos

109112 deriving Show

110113

111114newtype GroffTokens = GroffTokens { unGroffTokens :: Seq.Seq GroffToken }

@@ -346,13 +349,27 @@ lexTable pos = do

346349 spaces

347350 skipMany lexComment

348351 spaces

352+ rows <- lexTableRows

353+ morerows <- many $ try $ do

354+ string ".T&"

355+ skipMany spacetab

356+ newline

357+ lexTableRows

358+ string ".TE"

359+ skipMany spacetab

360+ eofline

361+ return $ singleTok $ MTable opts (rows ++ concat morerows) pos

362+

363+lexTableRows :: PandocMonad m => GroffLexer m [TableRow]

364+lexTableRows = do

349365 aligns <- tableFormatSpec

350366 spaces

351367 skipMany lexComment

352368 spaces

353− rows <- manyTill tableRow (try (string ".TE" >> skipMany spacetab >> eofline))

354− return $ singleTok $ MTable opts aligns rows pos

355−

369+ rows <- many (notFollowedBy (try (string ".TE") <|> try (string ".T&")) >>

370+ tableRow)

371+ return $ zip aligns rows

372+

356373tableCell :: PandocMonad m => GroffLexer m GroffTokens

357374tableCell = (enclosedCell <|> simpleCell) >>= lexGroff . T.pack

358375 where

@@ -387,18 +404,18 @@ tableOption = do

387404 skipMany spacetab

388405 return (k,v)

389406

390−tableFormatSpec :: PandocMonad m => GroffLexer m [[TableFormat]]

407+tableFormatSpec :: PandocMonad m => GroffLexer m [[CellFormat]]

391408tableFormatSpec = do

392409 speclines <- tableFormatSpecLine `sepBy1` (newline <|> char ',')

393410 skipMany spacetab

394411 char '.'

395− return speclines

412+ return $ speclines ++ repeat (lastDef [] speclines) -- last line is default

396413

397−tableFormatSpecLine :: PandocMonad m => GroffLexer m [TableFormat]

414+tableFormatSpecLine :: PandocMonad m => GroffLexer m [CellFormat]

398415tableFormatSpecLine =

399416 many1 $ try $ skipMany spacetab >> tableColFormat

400417

401−tableColFormat :: PandocMonad m => GroffLexer m TableFormat

418+tableColFormat :: PandocMonad m => GroffLexer m CellFormat

402419tableColFormat = do

403420 pipePrefix' <- option False

404421 $ True <$ (try $ string "|" <* notFollowedBy spacetab)

@@ -416,7 +433,7 @@ tableColFormat = do

416433 else return ""

417434 return $ x : num

418435 pipeSuffix' <- option False $ True <$ string "|"

419− return $ TableFormat

436+ return $ CellFormat

420437 { columnType = c

421438 , pipePrefix = pipePrefix'

422439 , pipeSuffix = pipeSuffix'

src/Text/Pandoc/Readers/Man.hs+12−11

@@ -34,7 +34,6 @@ Conversion of man to 'Pandoc' document.

3434module Text.Pandoc.Readers.Man (readMan) where

3535

3636import Prelude

37−import Safe (lastMay)

3837import Data.Char (toLower)

3938import Data.Default (Default)

4039import Control.Monad (liftM, mzero, guard)

@@ -114,22 +113,22 @@ parseTable :: PandocMonad m => ManParser m Blocks

114113parseTable = do

115114 let isMTable (MTable{}) = True

116115 isMTable _ = False

117− MTable _opts aligns rows pos <- msatisfy isMTable

118− case lastMay aligns of

119− Just as -> try (do

116+ MTable _opts rows pos <- msatisfy isMTable

117+ case rows of

118+ ((as,_):_) -> try (do

120119 let as' = map (columnTypeToAlignment . columnType) as

121120 guard $ all isJust as'

122121 let alignments = catMaybes as'

123122 let (headerRow', bodyRows') =

124123 case rows of

125− (h:[x]:bs)

124+ (h:x:bs)

126125 | isHrule x -> (h, bs)

127− _ -> ([], rows)

128− headerRow <- mapM parseTableCell headerRow'

129− bodyRows <- mapM (mapM parseTableCell) bodyRows'

126+ _ -> (([],[]), rows)

127+ headerRow <- mapM parseTableCell $ snd headerRow'

128+ bodyRows <- mapM (mapM parseTableCell . snd) bodyRows'

130129 return $ B.table mempty (zip alignments (repeat 0.0))

131130 headerRow bodyRows) <|> fallback pos

132− Nothing -> fallback pos

131+ [] -> fallback pos

133132

134133 where

135134

@@ -146,11 +145,13 @@ parseTable = do

146145 Left e -> throwError e

147146 Right x -> return x

148147

149− isHrule :: GroffTokens -> Bool

150− isHrule (GroffTokens ss) =

148+ isHrule :: TableRow -> Bool

149+ isHrule ([cellfmt], _) = columnType cellfmt `elem` ['_','-','=']

150+ isHrule (_, [GroffTokens ss]) =

151151 case Foldable.toList ss of

152152 [MLine [RoffStr [c]]] -> c `elem` ['_','-','=']

153153 _ -> False

154+ isHrule _ = False

154155

155156 fallback pos = do

156157 report $ SkippedContent "TABLE" pos