da37023 · typeable/pandoc

git.iko.soy/typeable/pandoc

on master help sign in

Put smart, strict in separate options field in state.

da37023John MacFarlane14 years ago

src/Tests/Readers/Markdown.hs+3−1

compare with master

unifiedsplit

@@ -13,7 +13,9 @@ markdown :: String -> Pandoc

13markdown = readMarkdown defaultParserState{ stateStandalone = True }

15markdownSmart :: String -> Pandoc

16−markdownSmart = readMarkdown defaultParserState{ stateSmart = True }

16+markdownSmart = readMarkdown defaultParserState{ stateOptions =

17+ let oldOpts = stateOptions defaultParserState in

18+ oldOpts { optionSmart = True } }

18infix 4 =:

20infix 4 =:

19(=:) :: ToString c

21(=:) :: ToString c

src/Text/Pandoc.hs+3

@@ -60,6 +60,8 @@ module Text.Pandoc

60 module Text.Pandoc.Definition

61 -- * Generics

62 , module Text.Pandoc.Generic

63+ -- * Options

64+ , module Text.Pandoc.Options

63 -- * Lists of readers and writers

65 -- * Lists of readers and writers

64 , readers

66 , readers

65 , writers

67 , writers

@@ -151,6 +153,7 @@ import Text.Pandoc.Writers.AsciiDoc

151import Text.Pandoc.Templates

153import Text.Pandoc.Templates

152import Text.Pandoc.Parsing

154import Text.Pandoc.Parsing

153import Text.Pandoc.Shared

155import Text.Pandoc.Shared

156+import Text.Pandoc.Options

154import Data.ByteString.Lazy (ByteString)

157import Data.ByteString.Lazy (ByteString)

155import Data.Version (showVersion)

158import Data.Version (showVersion)

156import Text.JSON.Generic

159import Text.JSON.Generic

src/Text/Pandoc/Options.hs+16−2

@@ -29,10 +29,13 @@ Data structures and functions for representing parser and writer

29options.

30-}

31module Text.Pandoc.Options ( Extension(..)

32+ , Options(..)

32 ) where

33 ) where

33import Data.Set (Set)

34import Data.Set (Set)

35+import qualified Data.Set as Set

36+import Data.Default

35−-- | Individually selectable markdown syntax extensions.

38+-- | Individually selectable syntax extensions.

36data Extension = Footnotes

39data Extension = Footnotes

37 | TeX_math

40 | TeX_math

38 | Delimited_code_blocks

41 | Delimited_code_blocks

@@ -45,6 +48,17 @@ data Extension = Footnotes

45 | Blank_before_blockquote

48 | Blank_before_blockquote

46 | Blank_before_header

49 | Blank_before_header

47 | Significant_bullets

50 | Significant_bullets

48− deriving (Show, Read, Enum, Eq, Bounded)

51+ deriving (Show, Read, Enum, Eq, Ord, Bounded)

53+data Options = Options{

54+ optionExtensions :: Set Extension

55+ , optionSmart :: Bool

56+ , optionStrict :: Bool -- FOR TRANSITION ONLY

57+ } deriving (Show, Read)

59+instance Default Options

60+ where def = Options{

61+ optionExtensions = Set.fromList [minBound..maxBound]

62+ , optionSmart = False

63+ , optionStrict = False

64+ }

src/Text/Pandoc/Parsing.hs+7−8

@@ -129,6 +129,7 @@ module Text.Pandoc.Parsing ( (>>~),

129where

130

131import Text.Pandoc.Definition

132+import Text.Pandoc.Options

132import Text.Pandoc.Generic

133import Text.Pandoc.Generic

133import qualified Text.Pandoc.UTF8 as UTF8 (putStrLn)

134import qualified Text.Pandoc.UTF8 as UTF8 (putStrLn)

134import Text.Parsec

135import Text.Parsec

@@ -391,7 +392,7 @@ nullBlock = anyChar >> return Null

391failIfStrict :: Parsec [a] ParserState ()

392failIfStrict :: Parsec [a] ParserState ()

392failIfStrict = do

393failIfStrict = do

393 state <- getState

394 state <- getState

394− if stateStrict state then fail "strict mode" else return ()

395+ if optionStrict (stateOptions state) then fail "strict mode" else return ()

395

396

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

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

397failUnlessLHS :: Parsec [tok] ParserState ()

398failUnlessLHS :: Parsec [tok] ParserState ()

@@ -688,7 +689,8 @@ testStringWith parser str = UTF8.putStrLn $ show $

688

689

689-- | Parsing options.

690-- | Parsing options.

690data ParserState = ParserState

691data ParserState = ParserState

692+ { stateOptions :: Options, -- ^ User options

691− { stateParseRaw :: Bool, -- ^ Parse raw HTML and LaTeX?

693+ stateParseRaw :: Bool, -- ^ Parse raw HTML and LaTeX?

692 stateParserContext :: ParserContext, -- ^ Inside list?

694 stateParserContext :: ParserContext, -- ^ Inside list?

693 stateQuoteContext :: QuoteContext, -- ^ Inside quoted environment?

695 stateQuoteContext :: QuoteContext, -- ^ Inside quoted environment?

694 stateMaxNestingLevel :: Int, -- ^ Max # of nested Strong/Emph

696 stateMaxNestingLevel :: Int, -- ^ Max # of nested Strong/Emph

@@ -701,8 +703,6 @@ data ParserState = ParserState

701 stateTitle :: [Inline], -- ^ Title of document

703 stateTitle :: [Inline], -- ^ Title of document

702 stateAuthors :: [[Inline]], -- ^ Authors of document

704 stateAuthors :: [[Inline]], -- ^ Authors of document

703 stateDate :: [Inline], -- ^ Date of document

705 stateDate :: [Inline], -- ^ Date of document

704− stateStrict :: Bool, -- ^ Use strict markdown syntax?

705− stateSmart :: Bool, -- ^ Use smart typography?

706 stateOldDashes :: Bool, -- ^ Use pandoc <= 1.8.2.1 behavior

707 -- in parsing dashes; -- is em-dash;

708 -- before numeral is en-dash

@@ -724,7 +724,8 @@ instance Default ParserState where

724

725defaultParserState :: ParserState

726defaultParserState =

727+ ParserState { stateOptions = def,

727− ParserState { stateParseRaw = False,

728+ stateParseRaw = False,

728 stateParserContext = NullState,

729 stateParserContext = NullState,

729 stateQuoteContext = NoQuote,

730 stateQuoteContext = NoQuote,

730 stateMaxNestingLevel = 6,

731 stateMaxNestingLevel = 6,

@@ -737,8 +738,6 @@ defaultParserState =

737 stateTitle = [],

738 stateTitle = [],

738 stateAuthors = [],

739 stateAuthors = [],

739 stateDate = [],

740 stateDate = [],

740− stateStrict = False,

741− stateSmart = False,

742 stateOldDashes = False,

741 stateOldDashes = False,

743 stateLiterateHaskell = False,

742 stateLiterateHaskell = False,

744 stateColumns = 80,

743 stateColumns = 80,

@@ -796,7 +795,7 @@ lookupKeySrc table key = case M.lookup key table of

796

795

797-- | Fail unless we're in "smart typography" mode.

796-- | Fail unless we're in "smart typography" mode.

798failUnlessSmart :: Parsec [tok] ParserState ()

797failUnlessSmart :: Parsec [tok] ParserState ()

799−failUnlessSmart = getState >>= guard . stateSmart

798+failUnlessSmart = getState >>= guard . optionSmart . stateOptions

800

799

801smartPunctuation :: Parsec [Char] ParserState Inline

800smartPunctuation :: Parsec [Char] ParserState Inline

802 -> Parsec [Char] ParserState Inline

801 -> Parsec [Char] ParserState Inline

src/Text/Pandoc/Readers/HTML.hs+3−2

@@ -41,6 +41,7 @@ import Text.HTML.TagSoup.Match

41import Text.Pandoc.Definition

42import Text.Pandoc.Builder (text, toList)

43import Text.Pandoc.Shared

44+import Text.Pandoc.Options

44import Text.Pandoc.Parsing

45import Text.Pandoc.Parsing

45import Data.Maybe ( fromMaybe, isJust )

46import Data.Maybe ( fromMaybe, isJust )

46import Data.List ( intercalate )

47import Data.List ( intercalate )

@@ -125,7 +126,7 @@ pOrderedList :: TagParser [Block]

125pOrderedList = try $ do

126pOrderedList = try $ do

126 TagOpen _ attribs <- pSatisfy (~== TagOpen "ol" [])

127 TagOpen _ attribs <- pSatisfy (~== TagOpen "ol" [])

127 st <- getState

128 st <- getState

128− let (start, style) = if stateStrict st

129+ let (start, style) = if optionStrict (stateOptions st)

129 then (1, DefaultStyle)

130 then (1, DefaultStyle)

130 else (sta', sty')

131 else (sta', sty')

131 where sta = fromMaybe "1" $

132 where sta = fromMaybe "1" $

@@ -280,7 +281,7 @@ pCodeBlock = try $ do

280 let attribsClasses = words $ fromMaybe "" $ lookup "class" attr

281 let attribsClasses = words $ fromMaybe "" $ lookup "class" attr

281 let attribsKV = filter (\(k,_) -> k /= "class" && k /= "id") attr

282 let attribsKV = filter (\(k,_) -> k /= "class" && k /= "id") attr

282 st <- getState

283 st <- getState

283− let attribs = if stateStrict st

284+ let attribs = if optionStrict (stateOptions st)

284 then ("",[],[])

285 then ("",[],[])

285 else (attribsId, attribsClasses, attribsKV)

286 else (attribsId, attribsClasses, attribsKV)

286 return [CodeBlock attribs result]

287 return [CodeBlock attribs result]

src/Text/Pandoc/Readers/Markdown.hs+11−10

@@ -37,6 +37,7 @@ import Data.Char ( isAlphaNum )

37import Data.Maybe

38import Text.Pandoc.Definition

39import Text.Pandoc.Generic

40+import Text.Pandoc.Options

40import Text.Pandoc.Shared

41import Text.Pandoc.Shared

41import Text.Pandoc.Parsing

42import Text.Pandoc.Parsing

42import Text.Pandoc.Readers.LaTeX ( rawLaTeXInline, rawLaTeXBlock )

43import Text.Pandoc.Readers.LaTeX ( rawLaTeXInline, rawLaTeXBlock )

@@ -181,7 +182,7 @@ parseMarkdown = do

181 -- docMinusKeys is the raw document with blanks where the keys/notes were...

182 -- docMinusKeys is the raw document with blanks where the keys/notes were...

182 st <- getState

183 st <- getState

183 let firstPassParser = referenceKey

184 let firstPassParser = referenceKey

184− <|> (if stateStrict st then mzero else noteBlock)

185+ <|> (if optionStrict (stateOptions st) then mzero else noteBlock)

185 <|> liftM snd (withRaw codeBlockDelimited)

186 <|> liftM snd (withRaw codeBlockDelimited)

186 <|> lineClump

187 <|> lineClump

187 docMinusKeys <- liftM concat $ manyTill firstPassParser eof

188 docMinusKeys <- liftM concat $ manyTill firstPassParser eof

@@ -292,7 +293,7 @@ parseBlocks = manyTill block eof

292block :: Parser [Char] ParserState Block

293block :: Parser [Char] ParserState Block

293block = do

294block = do

294 st <- getState

295 st <- getState

295− choice (if stateStrict st

296+ choice (if optionStrict (stateOptions st)

296 then [ header

297 then [ header

297 , codeBlockIndented

298 , codeBlockIndented

298 , blockQuote

299 , blockQuote

@@ -533,7 +534,7 @@ anyOrderedListStart = try $ do

533 skipNonindentSpaces

534 skipNonindentSpaces

534 notFollowedBy $ string "p." >> spaceChar >> digit -- page number

535 notFollowedBy $ string "p." >> spaceChar >> digit -- page number

535 state <- getState

536 state <- getState

536− if stateStrict state

537+ if optionStrict (stateOptions state)

537 then do many1 digit

538 then do many1 digit

538 char '.'

539 char '.'

539 spaceChar

540 spaceChar

@@ -694,7 +695,7 @@ para = try $ do

694 option (Plain result) $ try $ do

695 option (Plain result) $ try $ do

695 newline

696 newline

696 blanklines <|>

697 blanklines <|>

697− (getState >>= guard . stateStrict >>

698+ (getState >>= guard . optionStrict . stateOptions >>

698 lookAhead (blockQuote <|> header) >> return "")

699 lookAhead (blockQuote <|> header) >> return "")

699 return $ Para result

700 return $ Para result

700

701

@@ -1008,7 +1009,7 @@ escapedChar' :: Parser [Char] ParserState Char

1008escapedChar' = try $ do

1009escapedChar' = try $ do

1009 char '\\'

1010 char '\\'

1010 state <- getState

1011 state <- getState

1011− if stateStrict state

1012+ if optionStrict (stateOptions state)

1012 then oneOf "\\`*_{}[]()>#+-.!~"

1013 then oneOf "\\`*_{}[]()>#+-.!~"

1013 else satisfy (not . isAlphaNum)

1014 else satisfy (not . isAlphaNum)

1014

1015

@@ -1023,7 +1024,7 @@ escapedChar = do

1023ltSign :: Parser [Char] ParserState Inline

1024ltSign :: Parser [Char] ParserState Inline

1024ltSign = do

1025ltSign = do

1025 st <- getState

1026 st <- getState

1026− if stateStrict st

1027+ if optionStrict (stateOptions st)

1027 then char '<'

1028 then char '<'

1028 else notFollowedBy' rawHtmlBlocks >> char '<' -- unless it starts html

1029 else notFollowedBy' rawHtmlBlocks >> char '<' -- unless it starts html

1029 return $ Str ['<']

1030 return $ Str ['<']

@@ -1159,7 +1160,7 @@ nonEndline = satisfy (/='\n')

1159

1160

1160str :: Parser [Char] ParserState Inline

1161str :: Parser [Char] ParserState Inline

1161str = do

1162str = do

1162− smart <- stateSmart `fmap` getState

1163+ smart <- (optionSmart . stateOptions) `fmap` getState

1163 a <- alphaNum

1164 a <- alphaNum

1164 as <- many $ alphaNum

1165 as <- many $ alphaNum

1165 <|> (try $ char '_' >>~ lookAhead alphaNum)

1166 <|> (try $ char '_' >>~ lookAhead alphaNum)

@@ -1200,7 +1201,7 @@ endline = try $ do

1200 newline

1201 newline

1201 notFollowedBy blankline

1202 notFollowedBy blankline

1202 st <- getState

1203 st <- getState

1203− when (stateStrict st) $ do

1204+ when (optionStrict (stateOptions st)) $ do

1204 notFollowedBy emailBlockQuoteStart

1205 notFollowedBy emailBlockQuoteStart

1205 notFollowedBy (char '#') -- atx header

1206 notFollowedBy (char '#') -- atx header

1206 -- parse potential list-starts differently if in a list:

1207 -- parse potential list-starts differently if in a list:

@@ -1282,7 +1283,7 @@ autoLink = try $ do

1282 (orig, src) <- uri <|> emailAddress

1283 (orig, src) <- uri <|> emailAddress

1283 char '>'

1284 char '>'

1284 st <- getState

1285 st <- getState

1285− return $ if stateStrict st

1286+ return $ if optionStrict (stateOptions st)

1286 then Link [Str orig] (src, "")

1287 then Link [Str orig] (src, "")

1287 else Link [Code ("",["url"],[]) orig] (src, "")

1288 else Link [Code ("",["url"],[]) orig] (src, "")

1288

1289

@@ -1343,7 +1344,7 @@ inBrackets parser = do

1343rawHtmlInline :: Parser [Char] ParserState Inline

1344rawHtmlInline :: Parser [Char] ParserState Inline

1344rawHtmlInline = do

1345rawHtmlInline = do

1345 st <- getState

1346 st <- getState

1346− (_,result) <- if stateStrict st

1347+ (_,result) <- if optionStrict (stateOptions st)

1347 then htmlTag (not . isTextTag)

1348 then htmlTag (not . isTextTag)

1348 else htmlTag isInlineTag

1349 else htmlTag isInlineTag

1349 return $ RawInline "html" result

1350 return $ RawInline "html" result

src/Text/Pandoc/Readers/Textile.hs+5−1

@@ -57,6 +57,7 @@ module Text.Pandoc.Readers.Textile ( readTextile) where

58import Text.Pandoc.Definition

59import Text.Pandoc.Shared

60+import Text.Pandoc.Options

60import Text.Pandoc.Parsing

61import Text.Pandoc.Parsing

61import Text.Pandoc.Readers.HTML ( htmlTag, isInlineTag, isBlockTag )

62import Text.Pandoc.Readers.HTML ( htmlTag, isInlineTag, isBlockTag )

62import Text.Pandoc.Readers.LaTeX ( rawLaTeXInline, rawLaTeXBlock )

63import Text.Pandoc.Readers.LaTeX ( rawLaTeXInline, rawLaTeXBlock )

@@ -77,7 +78,10 @@ readTextile state s =

77parseTextile :: Parser [Char] ParserState Pandoc

78parseTextile :: Parser [Char] ParserState Pandoc

78parseTextile = do

79parseTextile = do

79 -- textile allows raw HTML and does smart punctuation by default

80 -- textile allows raw HTML and does smart punctuation by default

81+ oldOpts <- stateOptions `fmap` getState

80− updateState (\state -> state { stateParseRaw = True, stateSmart = True })

82+ updateState $ \state -> state { stateParseRaw = True

83+ , stateOptions = oldOpts{ optionSmart = True }

84+ }

81 many blankline

85 many blankline

82 startPos <- getPosition

86 startPos <- getPosition

83 -- go through once just to get list of reference keys and notes

87 -- go through once just to get list of reference keys and notes

src/pandoc.hs+4−3

@@ -942,11 +942,12 @@ main = do

942 lhsExtension sources,

943 stateStandalone = standalone',

944 stateCitations = map CSL.refId refs,

945− stateSmart = smart || (texLigatures &&

946− (laTeXOutput || writerName' == "context")),

947 stateOldDashes = oldDashes,

945 stateOldDashes = oldDashes,

948 stateColumns = columns,

946 stateColumns = columns,

949− stateStrict = strict,

947+ stateOptions = def{ optionStrict = strict

948+ , optionSmart = smart ||

949+ (texLigatures &&

950+ (laTeXOutput || writerName' == "context")) },

950 stateIndentedCodeClasses = codeBlockClasses,

951 stateIndentedCodeClasses = codeBlockClasses,

951 stateApplyMacros = not laTeXOutput

952 stateApplyMacros = not laTeXOutput

952 }

953 }