da37023 · typeable/pandoc

git.iko.soy/typeable/pandoc

Put smart, strict in separate options field in state.

This is the beginning of a larger transition that will make Options, not ParserState, the parameter of the read functions. (Options will also be used in writers, in place of WriterOptions.) Next step is to remove strict, replacing it with granular tests for different extensions.

commitda3702357dc9a310b460c82411fe5b8c870416d5 authorJohn MacFarlane <fiddlosopher@gmail.com> dateWed 25 Jul 2012, 10:45 (14 years ago) parent0d21b8d

8 files changed+52−27

compare with master

unifiedsplit

src/Tests/Readers/Markdown.hs+3−1 src/Text/Pandoc.hs+3 src/Text/Pandoc/Options.hs+16−2 src/Text/Pandoc/Parsing.hs+7−8 src/Text/Pandoc/Readers/HTML.hs+3−2 src/Text/Pandoc/Readers/Markdown.hs+11−10 src/Text/Pandoc/Readers/Textile.hs+5−1 src/pandoc.hs+4−3

src/Tests/Readers/Markdown.hs+3−1

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

1313markdown = readMarkdown defaultParserState{ stateStandalone = True }

1414

1515markdownSmart :: String -> Pandoc

16−markdownSmart = readMarkdown defaultParserState{ stateSmart = True }

16+markdownSmart = readMarkdown defaultParserState{ stateOptions =

17+ let oldOpts = stateOptions defaultParserState in

18+ oldOpts { optionSmart = True } }

1719

1820infix 4 =:

1921(=:) :: ToString c

src/Text/Pandoc.hs+3

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

6060 module Text.Pandoc.Definition

6161 -- * Generics

6262 , module Text.Pandoc.Generic

63+ -- * Options

64+ , module Text.Pandoc.Options

6365 -- * Lists of readers and writers

6466 , readers

6567 , writers

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

151153import Text.Pandoc.Templates

152154import Text.Pandoc.Parsing

153155import Text.Pandoc.Shared

156+import Text.Pandoc.Options

154157import Data.ByteString.Lazy (ByteString)

155158import Data.Version (showVersion)

156159import Text.JSON.Generic

src/Text/Pandoc/Options.hs+16−2

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

2929options.

3030-}

3131module Text.Pandoc.Options ( Extension(..)

32+ , Options(..)

3233 ) where

3334import Data.Set (Set)

35+import qualified Data.Set as Set

36+import Data.Default

3437

35−-- | Individually selectable markdown syntax extensions.

38+-- | Individually selectable syntax extensions.

3639data Extension = Footnotes

3740 | TeX_math

3841 | Delimited_code_blocks

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

4548 | Blank_before_blockquote

4649 | Blank_before_header

4750 | Significant_bullets

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

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

4952

53+data Options = Options{

54+ optionExtensions :: Set Extension

55+ , optionSmart :: Bool

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

57+ } deriving (Show, Read)

5058

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 ( (>>~),

129129where

130130

131131import Text.Pandoc.Definition

132+import Text.Pandoc.Options

132133import Text.Pandoc.Generic

133134import qualified Text.Pandoc.UTF8 as UTF8 (putStrLn)

134135import Text.Parsec

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

391392failIfStrict :: Parsec [a] ParserState ()

392393failIfStrict = do

393394 state <- getState

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

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

395396

396397-- | Fail unless we're in literate haskell mode.

397398failUnlessLHS :: Parsec [tok] ParserState ()

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

688689

689690-- | Parsing options.

690691data ParserState = ParserState

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

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

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

692694 stateParserContext :: ParserContext, -- ^ Inside list?

693695 stateQuoteContext :: QuoteContext, -- ^ Inside quoted environment?

694696 stateMaxNestingLevel :: Int, -- ^ Max # of nested Strong/Emph

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

701703 stateTitle :: [Inline], -- ^ Title of document

702704 stateAuthors :: [[Inline]], -- ^ Authors of document

703705 stateDate :: [Inline], -- ^ Date of document

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

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

706706 stateOldDashes :: Bool, -- ^ Use pandoc <= 1.8.2.1 behavior

707707 -- in parsing dashes; -- is em-dash;

708708 -- before numeral is en-dash

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

724724

725725defaultParserState :: ParserState

726726defaultParserState =

727− ParserState { stateParseRaw = False,

727+ ParserState { stateOptions = def,

728+ stateParseRaw = False,

728729 stateParserContext = NullState,

729730 stateQuoteContext = NoQuote,

730731 stateMaxNestingLevel = 6,

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

737738 stateTitle = [],

738739 stateAuthors = [],

739740 stateDate = [],

740− stateStrict = False,

741− stateSmart = False,

742741 stateOldDashes = False,

743742 stateLiterateHaskell = False,

744743 stateColumns = 80,

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

796795

797796-- | Fail unless we're in "smart typography" mode.

798797failUnlessSmart :: Parsec [tok] ParserState ()

799−failUnlessSmart = getState >>= guard . stateSmart

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

800799

801800smartPunctuation :: Parsec [Char] ParserState Inline

802801 -> Parsec [Char] ParserState Inline

src/Text/Pandoc/Readers/HTML.hs+3−2

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

4141import Text.Pandoc.Definition

4242import Text.Pandoc.Builder (text, toList)

4343import Text.Pandoc.Shared

44+import Text.Pandoc.Options

4445import Text.Pandoc.Parsing

4546import Data.Maybe ( fromMaybe, isJust )

4647import Data.List ( intercalate )

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

125126pOrderedList = try $ do

126127 TagOpen _ attribs <- pSatisfy (~== TagOpen "ol" [])

127128 st <- getState

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

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

129130 then (1, DefaultStyle)

130131 else (sta', sty')

131132 where sta = fromMaybe "1" $

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

280281 let attribsClasses = words $ fromMaybe "" $ lookup "class" attr

281282 let attribsKV = filter (\(k,_) -> k /= "class" && k /= "id") attr

282283 st <- getState

283− let attribs = if stateStrict st

284+ let attribs = if optionStrict (stateOptions st)

284285 then ("",[],[])

285286 else (attribsId, attribsClasses, attribsKV)

286287 return [CodeBlock attribs result]

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

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

3737import Data.Maybe

3838import Text.Pandoc.Definition

3939import Text.Pandoc.Generic

40+import Text.Pandoc.Options

4041import Text.Pandoc.Shared

4142import Text.Pandoc.Parsing

4243import Text.Pandoc.Readers.LaTeX ( rawLaTeXInline, rawLaTeXBlock )

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

181182 -- docMinusKeys is the raw document with blanks where the keys/notes were...

182183 st <- getState

183184 let firstPassParser = referenceKey

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

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

185186 <|> liftM snd (withRaw codeBlockDelimited)

186187 <|> lineClump

187188 docMinusKeys <- liftM concat $ manyTill firstPassParser eof

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

292293block :: Parser [Char] ParserState Block

293294block = do

294295 st <- getState

295− choice (if stateStrict st

296+ choice (if optionStrict (stateOptions st)

296297 then [ header

297298 , codeBlockIndented

298299 , blockQuote

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

533534 skipNonindentSpaces

534535 notFollowedBy $ string "p." >> spaceChar >> digit -- page number

535536 state <- getState

536− if stateStrict state

537+ if optionStrict (stateOptions state)

537538 then do many1 digit

538539 char '.'

539540 spaceChar

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

694695 option (Plain result) $ try $ do

695696 newline

696697 blanklines <|>

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

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

698699 lookAhead (blockQuote <|> header) >> return "")

699700 return $ Para result

700701

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

10081009escapedChar' = try $ do

10091010 char '\\'

10101011 state <- getState

1011− if stateStrict state

1012+ if optionStrict (stateOptions state)

10121013 then oneOf "\\`*_{}[]()>#+-.!~"

10131014 else satisfy (not . isAlphaNum)

10141015

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

10231024ltSign :: Parser [Char] ParserState Inline

10241025ltSign = do

10251026 st <- getState

1026− if stateStrict st

1027+ if optionStrict (stateOptions st)

10271028 then char '<'

10281029 else notFollowedBy' rawHtmlBlocks >> char '<' -- unless it starts html

10291030 return $ Str ['<']

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

11591160

11601161str :: Parser [Char] ParserState Inline

11611162str = do

1162− smart <- stateSmart `fmap` getState

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

11631164 a <- alphaNum

11641165 as <- many $ alphaNum

11651166 <|> (try $ char '_' >>~ lookAhead alphaNum)

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

12001201 newline

12011202 notFollowedBy blankline

12021203 st <- getState

1203− when (stateStrict st) $ do

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

12041205 notFollowedBy emailBlockQuoteStart

12051206 notFollowedBy (char '#') -- atx header

12061207 -- parse potential list-starts differently if in a list:

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

12821283 (orig, src) <- uri <|> emailAddress

12831284 char '>'

12841285 st <- getState

1285− return $ if stateStrict st

1286+ return $ if optionStrict (stateOptions st)

12861287 then Link [Str orig] (src, "")

12871288 else Link [Code ("",["url"],[]) orig] (src, "")

12881289

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

13431344rawHtmlInline :: Parser [Char] ParserState Inline

13441345rawHtmlInline = do

13451346 st <- getState

1346− (_,result) <- if stateStrict st

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

13471348 then htmlTag (not . isTextTag)

13481349 else htmlTag isInlineTag

13491350 return $ RawInline "html" result

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

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

5757

5858import Text.Pandoc.Definition

5959import Text.Pandoc.Shared

60+import Text.Pandoc.Options

6061import Text.Pandoc.Parsing

6162import Text.Pandoc.Readers.HTML ( htmlTag, isInlineTag, isBlockTag )

6263import Text.Pandoc.Readers.LaTeX ( rawLaTeXInline, rawLaTeXBlock )

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

7778parseTextile :: Parser [Char] ParserState Pandoc

7879parseTextile = do

7980 -- textile allows raw HTML and does smart punctuation by default

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

81+ oldOpts <- stateOptions `fmap` getState

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

83+ , stateOptions = oldOpts{ optionSmart = True }

84+ }

8185 many blankline

8286 startPos <- getPosition

8387 -- go through once just to get list of reference keys and notes

src/pandoc.hs+4−3

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

942942 lhsExtension sources,

943943 stateStandalone = standalone',

944944 stateCitations = map CSL.refId refs,

945− stateSmart = smart || (texLigatures &&

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

947945 stateOldDashes = oldDashes,

948946 stateColumns = columns,

949− stateStrict = strict,

947+ stateOptions = def{ optionStrict = strict

948+ , optionSmart = smart ||

949+ (texLigatures &&

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

950951 stateIndentedCodeClasses = codeBlockClasses,

951952 stateApplyMacros = not laTeXOutput

952953 }