feat(html): add paragraph_attributes extension

- Introduce `Ext_paragraph_attributes` to control preservation of
  attributes on paragraph tags.
- Enable the extension by default for all HTML-based formats.
- The HTML reader now wraps attributed paragraphs in a Div when
  the `native_divs` extension is enabled.
- The HTML writer unwraps the Div back to an attributed p tag.
- Add tests to verify attribute handling with and without the
  new extension.

Author: Valgard

src/Text/Pandoc/Extensions.hs

@@ -109,6 +109,7 @@ data Extension =
     | Ext_ntb                 -- ^ ConTeXt Natural Tables
     | Ext_old_dashes          -- ^ -- = em, - before number = en
     | Ext_pandoc_title_block  -- ^ Pandoc title block
+    | Ext_paragraph_attributes-- ^ Preserve attributes on paragraphs. By default, pandoc strips all attributes from paragraphs.
     | Ext_pipe_tables         -- ^ Pipe tables (as in PHP markdown extra)
     | Ext_raw_attribute           -- ^ Allow explicit raw blocks/inlines
     | Ext_raw_html            -- ^ Allow raw HTML
@@ -213,6 +214,7 @@ pandocExtensions = extensionsFromList
   [ Ext_footnotes
   , Ext_inline_notes
   , Ext_pandoc_title_block
+  , Ext_paragraph_attributes
   , Ext_yaml_metadata_block
   , Ext_table_captions
   , Ext_implicit_figures
@@ -435,7 +437,8 @@ getDefaultExtensions "html"            = extensionsFromList
                                           [Ext_auto_identifiers,
                                            Ext_native_divs,
                                            Ext_line_blocks,
-                                           Ext_native_spans]
+                                           Ext_native_spans,
+                                           Ext_paragraph_attributes]
 getDefaultExtensions "html4"           = getDefaultExtensions "html"
 getDefaultExtensions "html5"           = getDefaultExtensions "html"
 getDefaultExtensions "epub"            = extensionsFromList

src/Text/Pandoc/Readers/HTML.hs

@@ -59,7 +59,8 @@ import Text.Pandoc.Error
 import Text.Pandoc.Logging
 import Text.Pandoc.Options (
     Extension (Ext_epub_html_exts, Ext_empty_paragraphs, Ext_native_divs,
-               Ext_native_spans, Ext_raw_html, Ext_line_blocks, Ext_raw_tex),
+               Ext_paragraph_attributes, Ext_native_spans, Ext_raw_html,
+               Ext_line_blocks, Ext_raw_tex),
     ReaderOptions (readerExtensions, readerStripComments),
     extensionEnabled)
 import Text.Pandoc.Parsing hiding ((<|>))
@@ -626,14 +627,12 @@ pPlain = do
 -- Helper function for pPara when significant attributes are present
 pParaWithWrapper :: PandocMonad m => Attr -> TagParser m Blocks
 pParaWithWrapper (ident, classes, kvs) = do
-  guardEnabled Ext_native_divs -- Ensure native_divs is enabled for this behavior
   contents <- trimInlines <$> pInTags "p" inline
   (do guardDisabled Ext_empty_paragraphs
       guard (null contents)
       return mempty) <|> do
     let wrapperAttr = ("wrapper", "1")
-    let finalKVs = wrapperAttr : kvs
-    let finalAttrs = (ident, classes, finalKVs)
+    let finalAttrs = (ident, classes, wrapperAttr : kvs)
     return $ B.divWith finalAttrs (B.para contents)
 
 -- Helper function for pPara when no significant attributes are present
@@ -651,8 +650,10 @@ pPara = do
   let attr@(ident, classes, kvs) = toAttr attr'
   -- "Significant" attributes are any id, class, or key-value pair.
   let hasSignificantAttributes = not (T.null ident) || not (null classes) || not (null kvs)
-
-  if hasSignificantAttributes
+  exts <- getOption readerExtensions
+  if hasSignificantAttributes &&
+     (extensionEnabled Ext_paragraph_attributes exts ||
+      extensionEnabled Ext_native_divs exts)
     then pParaWithWrapper attr
     else pParaSimple
 

src/Text/Pandoc/Writers/HTML.hs

@@ -750,9 +750,11 @@ blockToHtmlInner opts (Para lst) = do
 blockToHtmlInner opts (LineBlock lns) = do
   htmlLines <- inlineListToHtml opts $ intercalate [LineBreak] lns
   return $ H.div ! A.class_ "line-block" $ htmlLines
-blockToHtmlInner opts (Div (ident, classes, kvs) [Para pans]) | Just "1" <- lookup "wrapper" kvs = do
+blockToHtmlInner opts (Div (ident, classes, kvs) [Para pans])
+  | Just "1" <- lookup "wrapper" kvs = do
   -- This is a paragraph that was wrapped in a Div by the reader
-  -- Unwrap it back to a <p> tag, transferring attributes from the Div
+  -- because of the paragraph_attributes extension.
+  -- Unwrap it back to a <p> tag, transferring attributes from the Div.
   let pKVs = filter (\(k,_) -> k /= "wrapper") kvs
   let pAttr = (ident, classes, pKVs)
   inner <- inlineListToHtml opts pans

test/Tests/Readers/HTML.hs

@@ -23,6 +23,7 @@ import Text.Pandoc.Shared (isHeaderBlock)
 import Text.Pandoc.Arbitrary ()
 import Text.Pandoc.Builder
 import Text.Pandoc.Walk (walk)
+import Test.Tasty.HUnit (testCase, assertEqual)
 
 html :: Text -> Pandoc
 html = purely $ readHtml def
@@ -55,6 +56,16 @@ roundTrip b = d'' == d'''
                   purely (writeHtml5String def
                             { writerWrapText = WrapPreserve })
 
+htmlHtmlTest :: TestName -> Extensions -> Extensions -> Text -> Text -> TestTree
+htmlHtmlTest name readerExts writerExts input expected =
+  testCase name $ do
+    let readerOpts = def { readerExtensions = readerExts }
+    let writerOpts = def { writerExtensions = writerExts, writerWrapText = WrapAuto }
+    actual <- runIOorExplode $ do
+      pandoc <- readHtml readerOpts input
+      writeHtml5String writerOpts pandoc
+    assertEqual name (T.strip expected) (T.strip actual)
+
 tests :: [TestTree]
 tests = [ testGroup "base tag"
           [ test html "simple" $
@@ -175,7 +186,21 @@ tests = [ testGroup "base tag"
             "<p id=\"baz\" align=\"invalid\">Invalid align with id.</p>" =?>
             doc (divWith ("baz", [], [("wrapper", "1"), ("align", "invalid")]) (para (text "Invalid align with id.")))
           ]
-        , askOption $ \(QuickCheckTests numtests) ->
+       , testGroup "paragraph-attributes-roundtrip"
+         [ htmlHtmlTest
+             "strip attributes when extension is disabled"
+             (readerExtensions def)
+             (writerExtensions def)
+             "<p id=\"foo\" class=\"bar\">Hello</p>"
+             "<p>Hello</p>"
+         , htmlHtmlTest
+             "keep attributes when extension is enabled"
+             (enableExtension Ext_native_divs $ enableExtension Ext_paragraph_attributes (readerExtensions def))
+             (enableExtension Ext_paragraph_attributes (writerExtensions def))
+             "<p id=\"foo\" class=\"bar\">Hello</p>"
+             "<p id=\"foo\" class=\"bar\">Hello</p>"
+         ]
+       , askOption $ \(QuickCheckTests numtests) ->
             testProperty "Round trip" $
               withMaxSuccess (if QuickCheckTests numtests == defaultValue
                                  then 25

test/command/paragraph_attributes_default.expected.html

@@ -0,0 +1 @@
+<p>Hallo Welt</p>

test/command/paragraph_attributes_default.html

@@ -0,0 +1 @@
+<p class="foo" id="bar">Hallo Welt</p>

test/command/paragraph_attributes_default.yaml

@@ -0,0 +1,6 @@
+description: 'Test paragraph attributes are removed by default (HTML to HTML)'
+input: paragraph_attributes_default.html
+output: paragraph_attributes_default.expected.html
+options:
+  - --from=html
+  - --to=html

test/command/paragraph_attributes_enabled.expected.html

@@ -0,0 +1 @@
+<p class="foo" id="bar">Hallo Welt</p>

test/command/paragraph_attributes_enabled.html

@@ -0,0 +1 @@
+<p class="foo" id="bar">Hallo Welt</p>

test/command/paragraph_attributes_enabled.yaml

@@ -0,0 +1,6 @@
+description: 'Test paragraph attributes are preserved with +paragraph_attributes (HTML to HTML)'
+input: paragraph_attributes_enabled.html
+output: paragraph_attributes_enabled.expected.html
+options:
+  - --from=html+paragraph_attributes
+  - --to=html+paragraph_attributes