42ba781 · typeable/pandoc

git.iko.soy/typeable/pandoc

on master help sign in

Added literate haskell support for markdown reader.

42ba781fiddlosopher18 years ago

Main.hs+24−9

compare with master

unifiedsplit

@@ -146,6 +146,7 @@ data Opt = Opt

146 , optReferenceLinks :: Bool -- ^ Use reference links in writing markdown, rst

147 , optWrapText :: Bool -- ^ Wrap text

148 , optSanitizeHTML :: Bool -- ^ Sanitize HTML

149+ , optLHSIn :: Bool -- ^ Treat input as literate haskell

149#ifdef _CITEPROC

150#ifdef _CITEPROC

150 , optModsFile :: String

151 , optModsFile :: String

151 , optCslFile :: String

152 , optCslFile :: String

@@ -179,6 +180,7 @@ defaultOpts = Opt

179 , optReferenceLinks = False

180 , optReferenceLinks = False

180 , optWrapText = True

181 , optWrapText = True

181 , optSanitizeHTML = False

182 , optSanitizeHTML = False

183+ , optLHSIn = False

182#ifdef _CITEPROC

184#ifdef _CITEPROC

183 , optModsFile = []

185 , optModsFile = []

184 , optCslFile = []

186 , optCslFile = []

@@ -288,6 +290,11 @@ options =

288 (\opt -> return opt { optSanitizeHTML = True }))

290 (\opt -> return opt { optSanitizeHTML = True }))

289 "" -- "Sanitize HTML"

291 "" -- "Sanitize HTML"

290

292

293+ , Option "" ["lhs-in"]

294+ (NoArg

295+ (\opt -> return opt { optLHSIn = True }))

296+ "" -- "Treat input as literate haskell"

297+

291 , Option "" ["toc", "table-of-contents"]

298 , Option "" ["toc", "table-of-contents"]

292 (NoArg

299 (NoArg

293 (\opt -> return opt { optTableOfContents = True }))

300 (\opt -> return opt { optTableOfContents = True }))

@@ -419,6 +426,11 @@ defaultReaderName (x:xs) =

419 ".native" -> "native"

426 ".native" -> "native"

420 _ -> defaultReaderName xs

427 _ -> defaultReaderName xs

421

428

429+-- Returns True if extension of first source is .lhs

430+lhsExtension :: [FilePath] -> Bool

431+lhsExtension (x:_) = takeExtension x == ".lhs"

432+lhsExtension _ = False

433+

422-- Determine default writer based on output file extension

434-- Determine default writer based on output file extension

423defaultWriterName :: FilePath -> String

435defaultWriterName :: FilePath -> String

424defaultWriterName "-" = "html" -- no output file

436defaultWriterName "-" = "html" -- no output file

@@ -498,6 +510,7 @@ main = do

498 , optReferenceLinks = referenceLinks

510 , optReferenceLinks = referenceLinks

499 , optWrapText = wrap

511 , optWrapText = wrap

500 , optSanitizeHTML = sanitize

512 , optSanitizeHTML = sanitize

513+ , optLHSIn = lhsIn

501#ifdef _CITEPROC

514#ifdef _CITEPROC

502 , optModsFile = modsFile

515 , optModsFile = modsFile

503 , optCslFile = cslFile

516 , optCslFile = cslFile

@@ -556,17 +569,19 @@ main = do

556#endif

569#endif

557

570

558 let startParserState =

571 let startParserState =

559− defaultParserState { stateParseRaw = parseRaw,

572+ defaultParserState { stateParseRaw = parseRaw,

560− stateTabStop = tabStop,

573+ stateTabStop = tabStop,

561− stateSanitizeHTML = sanitize,

574+ stateSanitizeHTML = sanitize,

575+ stateLiterateHaskell = lhsIn ||

576+ lhsExtension sources,

562− stateStandalone = standalone',

577+ stateStandalone = standalone',

563#ifdef _CITEPROC

578#ifdef _CITEPROC

564− stateCitations = map citeKey refs,

579+ stateCitations = map citeKey refs,

565#endif

580#endif

566− stateSmart = smart || writerName' `elem`

581+ stateSmart = smart || writerName' `elem`

567− ["latex", "context"],

582+ ["latex", "context"],

568− stateColumns = columns,

583+ stateColumns = columns,

569− stateStrict = strict }

584+ stateStrict = strict }

570 let csslink = if null css

585 let csslink = if null css

571 then ""

586 then ""

572 else concatMap

587 else concatMap

Text/Pandoc/Readers/Markdown.hs+27

@@ -45,6 +45,7 @@ import Text.Pandoc.Readers.HTML ( rawHtmlBlock, anyHtmlBlockTag,

45 htmlBlockElement, unsanitaryURI )

46import Text.Pandoc.CharacterReferences ( decodeCharacterReferences )

47import Text.ParserCombinators.Parsec

48+import Control.Monad (when)

49-- | Read markdown from an input string and return a Pandoc document.

50-- | Read markdown from an input string and return a Pandoc document.

50readMarkdown :: ParserState -> String -> Pandoc

51readMarkdown :: ParserState -> String -> Pandoc

@@ -102,6 +103,12 @@ failUnlessSmart = do

102 state <- getState

103 state <- getState

103 if stateSmart state then return () else fail "Smart typography feature"

104 if stateSmart state then return () else fail "Smart typography feature"

104

105

106+-- | Fail unless we're in literate haskell mode.

107+failUnlessLHS :: GenParser tok ParserState ()

108+failUnlessLHS = do

109+ state <- getState

110+ if stateLiterateHaskell state then return () else fail "Literate haskell feature"

111+

105-- | Parse a sequence of inline elements between square brackets,

112-- | Parse a sequence of inline elements between square brackets,

106-- including inlines between balanced pairs of square brackets.

113-- including inlines between balanced pairs of square brackets.

107inlinesInBalancedBrackets :: GenParser Char ParserState Inline

114inlinesInBalancedBrackets :: GenParser Char ParserState Inline

@@ -268,6 +275,7 @@ block = do

268 , header

275 , header

269 , table

276 , table

270 , codeBlockIndented

277 , codeBlockIndented

278+ , lhsCodeBlock

271 , blockQuote

279 , blockQuote

272 , hrule

280 , hrule

273 , bulletList

281 , bulletList

@@ -393,6 +401,25 @@ codeBlockIndented = do

393 optional blanklines

401 optional blanklines

394 return $ CodeBlock ("",[],[]) $ stripTrailingNewlines $ concat contents

402 return $ CodeBlock ("",[],[]) $ stripTrailingNewlines $ concat contents

395

403

404+lhsCodeBlock :: GenParser Char ParserState Block

405+lhsCodeBlock = do

406+ failUnlessLHS

407+ pos <- getPosition

408+ when (sourceColumn pos /= 1) $ fail "Not in first column"

409+ lns <- many1 birdTrackLine

410+ -- if (as is normal) there is always a space after >, drop it

411+ let lns' = if all (\ln -> null ln || take 1 ln == " ") lns

412+ then map (drop 1) lns

413+ else lns

414+ blanklines

415+ return $ CodeBlock ("",["haskell"],[]) $ intercalate "\n" lns'

416+

417+birdTrackLine :: GenParser Char st [Char]

418+birdTrackLine = do

419+ char '>'

420+ manyTill anyChar newline

421+

422+

396--

423--

397-- block quotes

424-- block quotes

398--

425--

Text/Pandoc/Shared.hs+2

@@ -635,6 +635,7 @@ data ParserState = ParserState

635 stateDate :: String, -- ^ Date of document

636 stateStrict :: Bool, -- ^ Use strict markdown syntax?

637 stateSmart :: Bool, -- ^ Use smart typography?

638+ stateLiterateHaskell :: Bool, -- ^ Treat input as literate haskell

638 stateColumns :: Int, -- ^ Number of columns in terminal

639 stateColumns :: Int, -- ^ Number of columns in terminal

639 stateHeaderTable :: [HeaderType] -- ^ Ordered list of header types used

640 stateHeaderTable :: [HeaderType] -- ^ Ordered list of header types used

640 }

641 }

@@ -658,6 +659,7 @@ defaultParserState =

658 stateDate = [],

659 stateDate = [],

659 stateStrict = False,

660 stateStrict = False,

660 stateSmart = False,

661 stateSmart = False,

662+ stateLiterateHaskell = False,

661 stateColumns = 80,

663 stateColumns = 80,

662 stateHeaderTable = [] }

664 stateHeaderTable = [] }

663

665