Better handling of one-liner Block HTML

Fixes #1074

docs/changelog.md

@@ -17,10 +17,11 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 ### Fixed
 
 * Backslash Unescape IDs set via `attr_list` on `toc` (#1493).
-* `md_in_html` should now process content inside "markdown" blocks a similar way
+* `md_in_html` will process content inside "markdown" blocks a similar way
  as they are parsed outside of "markdown" blocks giving a more consistent
  expectation to external extensions (#1503).
-* `md_in_html` should not handle tags within inline code blocks better (#1075).
+* `md_in_html` handle tags within inline code blocks better (#1075).
+* `md_in_html` fix handling of one-liner block HTML handling (#1074)
 
 ## [3.7] -- 2024-08-16
 

markdown/extensions/md_in_html.py

@@ -61,6 +61,7 @@ def reset(self):
  self.mdstack: list[str] = [] # When markdown=1, stack contains a list of tags
  self.treebuilder = etree.TreeBuilder()
  self.mdstate: list[Literal['block', 'span', 'off', None]] = []
+ self.mdstarted: list[bool] = []
  super().reset()
 
  def close(self):
@@ -126,8 +127,24 @@ def handle_starttag(self, tag, attrs):
  self.handle_endtag('p')
  self.mdstate.append(state)
  self.mdstack.append(tag)
+ self.mdstarted.append(True)
  attrs['markdown'] = state
  self.treebuilder.start(tag, attrs)
+
+ elif not self.inraw and tag in self.block_level_tags and self.mdstarted and self.mdstarted[-1]:
+ # Nested one-liner block tags `<tag><tag>...`
+ attrs = {key: value if value is not None else key for key, value in attrs}
+ state = self.get_state(tag, attrs)
+ if 'p' in self.mdstack and tag in self.block_level_tags:
+ # Close unclosed 'p' tag
+ self.handle_endtag('p')
+ self.mdstate.append(state)
+ self.mdstack.append(tag)
+ self.mdstarted.append(True)
+ attrs['markdown'] = state
+ self.treebuilder.start(tag, attrs)
+ return
+
  else:
  # Span level tag
  if self.inraw:
@@ -151,6 +168,7 @@ def handle_endtag(self, tag):
  while self.mdstack:
  item = self.mdstack.pop()
  self.mdstate.pop()
+ self.mdstarted.pop()
  self.treebuilder.end(item)
  if item == tag:
  break
@@ -247,6 +265,8 @@ def handle_data(self, data):
  if self.inraw or not self.mdstack:
  super().handle_data(data)
  else:
+ for i in range(len(self.mdstarted)):
+ self.mdstarted[i] = False
  self.treebuilder.data(data)
 
  def handle_empty_tag(self, data, is_block):

tests/test_syntax/extensions/test_md_in_html.py

@@ -1274,6 +1274,107 @@ def test_md1_code_void_tag_multiline(self):
  extensions=['md_in_html']
  )
 
+ def test_md1_oneliner_block(self):
+ self.assertMarkdownRenders(
+ self.dedent(
+ '<div class="outer" markdown="block"><div class="inner" markdown="block">*foo*</div></div>'
+ ),
+ '<div class="outer">\n'
+ '<div class="inner">\n'
+ '<p><em>foo</em></p>\n'
+ '</div>\n'
+ '</div>',
+ extensions=['md_in_html']
+ )
+
+ def test_md1_oneliner_block_mixed(self):
+ self.assertMarkdownRenders(
+ self.dedent(
+ """
+ <div class="a" markdown="block"><div class="b" markdown="block">
+
+ <div class="c" markdown="block"><div class="d" markdown="block">
+ *foo*
+ </div></div>
+
+ </div></div>
+ """
+ ),
+ '<div class="a">\n'
+ '<div class="b">\n'
+ '<div class="c">\n'
+ '<div class="d">\n'
+ '<p><em>foo</em></p>\n'
+ '</div>\n'
+ '</div>\n'
+ '</div>\n'
+ '</div>',
+ extensions=['md_in_html']
+ )
+
+ def test_md1_oneliner_block_start(self):
+ self.assertMarkdownRenders(
+ self.dedent(
+ """
+ <div class="outer" markdown="block"><div class="inner" markdown="block">
+ *foo*
+ </div></div>
+ """
+ ),
+ '<div class="outer">\n'
+ '<div class="inner">\n'
+ '<p><em>foo</em></p>\n'
+ '</div>\n'
+ '</div>',
+ extensions=['md_in_html']
+ )
+
+ def test_md1_oneliner_block_span(self):
+ self.assertMarkdownRenders(
+ self.dedent(
+ '<div class="outer" markdown="block"><div class="inner" markdown="span">*foo*</div></div>'
+ ),
+ '<div class="outer">\n'
+ '<div class="inner"><em>foo</em></div>\n'
+ '</div>',
+ extensions=['md_in_html']
+ )
+
+ def test_md1_oneliner_block_span_start(self):
+ self.assertMarkdownRenders(
+ self.dedent(
+ """
+ <div class="outer" markdown="block"><div class="inner" markdown="span">
+ *foo*
+ </div></div>
+ """
+ ),
+ '<div class="outer">\n'
+ '<div class="inner">\n'
+ '<em>foo</em>\n'
+ '</div>\n'
+ '</div>',
+ extensions=['md_in_html']
+ )
+
+ def test_md1_oneliner_span_block_start(self):
+ self.assertMarkdownRenders(
+ self.dedent(
+ """
+ <div class="outer" markdown="span"><div class="inner" markdown="block">
+ *foo*
+ </div>
+ *foo*
+ </div>
+ """
+ ),
+ '<div class="outer">\n'
+ '<div class="inner">\n'
+ '<em>foo</em>\n'
+ '</div>\n\n'
+ '<em>foo</em></div>',
+ extensions=['md_in_html']
+ )
 
  def test_md1_code_comment(self):