2b364b3 · typeable/pandoc

git.iko.soy/typeable/pandoc

on master help sign in

Merge pull request #1344 from mpickering/master

2b364b3John MacFarlane12 years ago

src/Text/Pandoc/Readers/HTML.hs+4−13

compare with master

unifiedsplit

@@ -50,7 +50,6 @@ import Data.Char ( isDigit )

50import Control.Monad ( liftM, guard, when, mzero )

51import Control.Applicative ( (<$>), (<$), (<*) )

52import Data.Monoid

53−import Data.Sequence (ViewL(..), ViewR(..), viewr, viewl)

55isSpace :: Char -> Bool

54isSpace :: Char -> Bool

56isSpace ' ' = True

55isSpace ' ' = True

@@ -369,9 +368,9 @@ pQ = do

369 then InSingleQuote

368 then InSingleQuote

370 else InDoubleQuote

369 else InDoubleQuote

371 let constructor = case quoteType of

370 let constructor = case quoteType of

372− SingleQuote -> B.singleQuoted

371+ SingleQuote -> B.singleQuoted

373 DoubleQuote -> B.doubleQuoted

372 DoubleQuote -> B.doubleQuoted

374− withQuoteContext innerQuoteContext $

373+ withQuoteContext innerQuoteContext $

375 pInlinesInTags "q" constructor

374 pInlinesInTags "q" constructor

376

375

377pEmph :: TagParser Inlines

376pEmph :: TagParser Inlines

@@ -406,7 +405,7 @@ pLink = try $ do

406 let url = fromAttrib "href" tag

405 let url = fromAttrib "href" tag

407 let title = fromAttrib "title" tag

406 let title = fromAttrib "title" tag

408 lab <- trimInlines . mconcat <$> manyTill inline (pCloses "a")

407 lab <- trimInlines . mconcat <$> manyTill inline (pCloses "a")

409− return $ B.link (escapeURI url) title lab

408+ return $ B.link (escapeURI url) title lab

410

409

411pImage :: TagParser Inlines

410pImage :: TagParser Inlines

412pImage = do

411pImage = do

@@ -439,15 +438,7 @@ pRawHtmlInline = do

439

438

440pInlinesInTags :: String -> (Inlines -> Inlines)

439pInlinesInTags :: String -> (Inlines -> Inlines)

441 -> TagParser Inlines

440 -> TagParser Inlines

442−pInlinesInTags tagtype f = do

441+pInlinesInTags tagtype f = extractSpaces f <$> pInTags tagtype inline

443− contents <- B.unMany <$> pInTags tagtype inline

444− let left = case viewl contents of

445− (Space :< _) -> B.space

446− _ -> mempty

447− let right = case viewr contents of

448− (_ :> Space) -> B.space

449− _ -> mempty

450− return (left <> f (trimInlines . B.Many $ contents) <> right)

451

442

452pInTags :: (Monoid a) => String -> TagParser a

443pInTags :: (Monoid a) => String -> TagParser a

453 -> TagParser a

444 -> TagParser a

src/Text/Pandoc/Shared.hs+16

@@ -53,6 +53,7 @@ module Text.Pandoc.Shared (

53 -- * Pandoc block and inline list processing

54 orderedListMarkers,

55 normalizeSpaces,

56+ extractSpaces,

56 normalize,

57 normalize,

57 stringify,

58 stringify,

58 compactify,

59 compactify,

@@ -113,6 +114,7 @@ import qualified Data.ByteString as BS

113import qualified Data.ByteString.Char8 as B8

114import qualified Data.ByteString.Char8 as B8

114import Text.Pandoc.Compat.Monoid

115import Text.Pandoc.Compat.Monoid

115import Data.ByteString.Base64 (decodeLenient)

116import Data.ByteString.Base64 (decodeLenient)

117+import Data.Sequence (ViewR(..), ViewL(..), viewl, viewr)

116

118

117#ifdef EMBED_DATA_FILES

119#ifdef EMBED_DATA_FILES

118import Text.Pandoc.Data (dataFiles)

120import Text.Pandoc.Data (dataFiles)

@@ -331,6 +333,20 @@ isSpaceOrEmpty Space = True

331isSpaceOrEmpty (Str "") = True

333isSpaceOrEmpty (Str "") = True

332isSpaceOrEmpty _ = False

334isSpaceOrEmpty _ = False

333

335

336+-- | Extract the leading and trailing spaces from inside an inline element

337+-- and place them outside the element.

338+

339+extractSpaces :: (Inlines -> Inlines) -> Inlines -> Inlines

340+extractSpaces f is =

341+ let contents = B.unMany is

342+ left = case viewl contents of

343+ (Space :< _) -> B.space

344+ _ -> mempty

345+ right = case viewr contents of

346+ (_ :> Space) -> B.space

347+ _ -> mempty in

348+ (left <> f (B.trimInlines . B.Many $ contents) <> right)

349+

334-- | Normalize @Pandoc@ document, consolidating doubled 'Space's,

350-- | Normalize @Pandoc@ document, consolidating doubled 'Space's,

335-- combining adjacent 'Str's and 'Emph's, remove 'Null's and

351-- combining adjacent 'Str's and 'Emph's, remove 'Null's and

336-- empty elements, etc.

352-- empty elements, etc.