a6f2b96 · typeable/pandoc

git.iko.soy/typeable/pandoc

Moved renderTags' from HTML reader & SelfContained to Shared.

Improved removal of markdown="1" attribute in Markdow reader.

commita6f2b960844caa81d9c8dd4d18f94c3de50bdb49 authorJohn MacFarlane <jgm@berkeley.edu> dateWed 15 Aug 2012, 09:42 (14 years ago) parent3745706

4 files changed+33−28

compare with master

unifiedsplit

src/Text/Pandoc/Readers/HTML.hs+1−13 src/Text/Pandoc/Readers/Markdown.hs+9−3 src/Text/Pandoc/SelfContained.hs+1−12 src/Text/Pandoc/Shared.hs+22

src/Text/Pandoc/Readers/HTML.hs+1−13

@@ -45,7 +45,7 @@ import Text.Pandoc.Options

4545import Text.Pandoc.Parsing

4646import Data.Maybe ( fromMaybe, isJust )

4747import Data.List ( intercalate )

48−import Data.Char ( isDigit, toLower )

48+import Data.Char ( isDigit )

4949import Control.Monad ( liftM, guard, when, mzero )

5050

5151isSpace :: Char -> Bool

@@ -95,18 +95,6 @@ block = choice

9595 , pRawHtmlBlock

9696 ]

9797

98−-- repeated in SelfContained -- consolidate eventually

99−renderTags' :: [Tag String] -> String

100−renderTags' = renderTagsOptions

101− renderOptions{ optMinimize = \x ->

102− let y = map toLower x

103− in y == "hr" || y == "br" ||

104− y == "img" || y == "meta" ||

105− y == "link"

106− , optRawTag = \x ->

107− let y = map toLower x

108− in y == "script" || y == "style" }

109−

11098pList :: TagParser [Block]

11199pList = pBulletList <|> pOrderedList <|> pDefinitionList

112100

src/Text/Pandoc/Readers/Markdown.hs+9−3

@@ -776,11 +776,10 @@ rawHtmlBlocks = do

776776 if "markdown" `notElem`

777777 map fst as

778778 then mzero

779− else return $ substitute

780− " markdown=\"1\"" "" raw

779+ else return $

780+ stripMarkdownAttribute raw

781781 | otherwise -> return raw

782782 _ -> return raw )

783− -- TODO remove markdown="1" attribute from raw tags

784783 sps <- do sp1 <- many spaceChar

785784 sp2 <- option "" (blankline >> return "\n")

786785 sp3 <- many spaceChar

@@ -793,6 +792,13 @@ rawHtmlBlocks = do

793792 let combined = concat htmlBlocks

794793 return $ if last combined == '\n' then init combined else combined

795794

795+-- remove markdown="1" attribute

796+stripMarkdownAttribute :: String -> String

797+stripMarkdownAttribute s = renderTags' $ map filterAttrib $ parseTags s

798+ where filterAttrib (TagOpen t as) = TagOpen t

799+ [(k,v) | (k,v) <- as, k /= "markdown"]

800+ filterAttrib x = x

801+

796802--

797803-- Tables

798804--

src/Text/Pandoc/SelfContained.hs+1−12

@@ -42,7 +42,7 @@ import System.FilePath (takeExtension, dropExtension, takeDirectory, (</>))

4242import Data.Char (toLower, isAscii, isAlphaNum)

4343import Codec.Compression.GZip as Gzip

4444import qualified Data.ByteString.Lazy as L

45−import Text.Pandoc.Shared (findDataFile)

45+import Text.Pandoc.Shared (findDataFile, renderTags')

4646import Text.Pandoc.MIME (getMimeType)

4747import System.Directory (doesFileExist)

4848

@@ -163,14 +163,3 @@ makeSelfContained userdata inp = do

163163 out' <- mapM (convertTag userdata) tags

164164 return $ renderTags' out'

165165

166−-- repeated from HTML reader:

167−renderTags' :: [Tag String] -> String

168−renderTags' = renderTagsOptions

169− renderOptions{ optMinimize = \x ->

170− let y = map toLower x

171− in y == "hr" || y == "br" ||

172− y == "img" || y == "meta" ||

173− y == "link"

174− , optRawTag = \x ->

175− let y = map toLower x

176− in y == "script" || y == "style" }

src/Text/Pandoc/Shared.hs+22

@@ -59,6 +59,8 @@ module Text.Pandoc.Shared (

5959 uniqueIdent,

6060 isHeaderBlock,

6161 headerShift,

62+ -- * TagSoup HTML handling

63+ renderTags',

6264 -- * File handling

6365 inDirectory,

6466 findDataFile,

@@ -89,6 +91,8 @@ import Text.Pandoc.Pretty (charWidth)

8991import System.Locale (defaultTimeLocale)

9092import Data.Time

9193import System.IO (stderr)

94+import Text.HTML.TagSoup (renderTagsOptions, RenderOptions(..), Tag(..),

95+ renderOptions)

9296

9397--

9498-- List processing

@@ -448,6 +452,22 @@ headerShift n = bottomUp shift

448452 shift (Header level inner) = Header (level + n) inner

449453 shift x = x

450454

455+--

456+-- TagSoup HTML handling

457+--

458+

459+-- | Render HTML tags.

460+renderTags' :: [Tag String] -> String

461+renderTags' = renderTagsOptions

462+ renderOptions{ optMinimize = \x ->

463+ let y = map toLower x

464+ in y == "hr" || y == "br" ||

465+ y == "img" || y == "meta" ||

466+ y == "link"

467+ , optRawTag = \x ->

468+ let y = map toLower x

469+ in y == "script" || y == "style" }

470+

451471--

452472-- File handling

453473--

@@ -501,3 +521,5 @@ safeRead s = case reads s of

501521 (d,x):_

502522 | all isSpace x -> return d

503523 _ -> fail $ "Could not read `" ++ s ++ "'"

524+

525+