Moved renderTags' from HTML reader & SelfContained to Shared.

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

4 files changed, 33 insertions, 28 deletions

diff --git a/src/Text/Pandoc/Readers/HTML.hs b/src/Text/Pandoc/Readers/HTML.hs
index 33846286d..e5c310ffc 100644
--- a/src/Text/Pandoc/Readers/HTML.hs
+++ b/src/Text/Pandoc/Readers/HTML.hs
@@ -45,7 +45,7 @@ import Text.Pandoc.Options
 import Text.Pandoc.Parsing
 import Data.Maybe ( fromMaybe, isJust )
 import Data.List ( intercalate )
-import Data.Char ( isDigit, toLower )
+import Data.Char ( isDigit )
 import Control.Monad ( liftM, guard, when, mzero )
 
 isSpace :: Char -> Bool
@@ -95,18 +95,6 @@ block = choice
             , pRawHtmlBlock
             ]
 
--- repeated in SelfContained -- consolidate eventually
-renderTags' :: [Tag String] -> String
-renderTags' = renderTagsOptions
-               renderOptions{ optMinimize = \x ->
-                                    let y = map toLower x
-                                    in  y == "hr" || y == "br" ||
-                                        y == "img" || y == "meta" ||
-                                        y == "link"
-                            , optRawTag = \x ->
-                                    let y = map toLower x
-                                    in  y == "script" || y == "style" }
-
 pList :: TagParser [Block]
 pList = pBulletList <|> pOrderedList <|> pDefinitionList
 
diff --git a/src/Text/Pandoc/Readers/Markdown.hs b/src/Text/Pandoc/Readers/Markdown.hs
index fe9be439e..b0925ac68 100644
--- a/src/Text/Pandoc/Readers/Markdown.hs
+++ b/src/Text/Pandoc/Readers/Markdown.hs
@@ -776,11 +776,10 @@ rawHtmlBlocks = do
                                                 if "markdown" `notElem`
                                                    map fst as
                                                    then mzero
-                                                   else return $ substitute
-                                                        " markdown=\"1\"" "" raw
+                                                   else return $
+                                                     stripMarkdownAttribute raw
                                           | otherwise -> return raw
                                         _ -> return raw )
-                          -- TODO remove markdown="1" attribute from raw tags
                           sps <- do sp1 <- many spaceChar
                                     sp2 <- option "" (blankline >> return "\n")
                                     sp3 <- many spaceChar
@@ -793,6 +792,13 @@ rawHtmlBlocks = do
   let combined = concat htmlBlocks
   return $ if last combined == '\n' then init combined else combined
 
+-- remove markdown="1" attribute
+stripMarkdownAttribute :: String -> String
+stripMarkdownAttribute s = renderTags' $ map filterAttrib $ parseTags s
+  where filterAttrib (TagOpen t as) = TagOpen t
+                                        [(k,v) | (k,v) <- as, k /= "markdown"]
+        filterAttrib              x = x
+
 --
 -- Tables
 --
diff --git a/src/Text/Pandoc/SelfContained.hs b/src/Text/Pandoc/SelfContained.hs
index 675b8366e..7a21f6f3a 100644
--- a/src/Text/Pandoc/SelfContained.hs
+++ b/src/Text/Pandoc/SelfContained.hs
@@ -42,7 +42,7 @@ import System.FilePath (takeExtension, dropExtension, takeDirectory, (</>))
 import Data.Char (toLower, isAscii, isAlphaNum)
 import Codec.Compression.GZip as Gzip
 import qualified Data.ByteString.Lazy as L
-import Text.Pandoc.Shared (findDataFile)
+import Text.Pandoc.Shared (findDataFile, renderTags')
 import Text.Pandoc.MIME (getMimeType)
 import System.Directory (doesFileExist)
 
@@ -163,14 +163,3 @@ makeSelfContained userdata inp = do
   out' <- mapM (convertTag userdata) tags
   return $ renderTags' out'
 
--- repeated from HTML reader:
-renderTags' :: [Tag String] -> String
-renderTags' = renderTagsOptions
-               renderOptions{ optMinimize = \x ->
-                                    let y = map toLower x
-                                    in  y == "hr" || y == "br" ||
-                                        y == "img" || y == "meta" ||
-                                        y == "link"
-                            , optRawTag = \x ->
-                                    let y = map toLower x
-                                    in  y == "script" || y == "style" }
diff --git a/src/Text/Pandoc/Shared.hs b/src/Text/Pandoc/Shared.hs
index ad28b7c23..d86f9a390 100644
--- a/src/Text/Pandoc/Shared.hs
+++ b/src/Text/Pandoc/Shared.hs
@@ -59,6 +59,8 @@ module Text.Pandoc.Shared (
                      uniqueIdent,
                      isHeaderBlock,
                      headerShift,
+                     -- * TagSoup HTML handling
+                     renderTags',
                      -- * File handling
                      inDirectory,
                      findDataFile,
@@ -89,6 +91,8 @@ import Text.Pandoc.Pretty (charWidth)
 import System.Locale (defaultTimeLocale)
 import Data.Time
 import System.IO (stderr)
+import Text.HTML.TagSoup (renderTagsOptions, RenderOptions(..), Tag(..),
+         renderOptions)
 
 --
 -- List processing
@@ -449,6 +453,22 @@ headerShift n = bottomUp shift
         shift x                    = x
 
 --
+-- TagSoup HTML handling
+--
+
+-- | Render HTML tags.
+renderTags' :: [Tag String] -> String
+renderTags' = renderTagsOptions
+               renderOptions{ optMinimize = \x ->
+                                    let y = map toLower x
+                                    in  y == "hr" || y == "br" ||
+                                        y == "img" || y == "meta" ||
+                                        y == "link"
+                            , optRawTag = \x ->
+                                    let y = map toLower x
+                                    in  y == "script" || y == "style" }
+
+--
 -- File handling
 --
 
@@ -501,3 +521,5 @@ safeRead s = case reads s of
                   (d,x):_
                     | all isSpace x -> return d
                   _                 -> fail $ "Could not read `" ++ s ++ "'"
+
+