Fix HTML: Invalid HTML5 file is generated for glossary (refs: #4611)

Note: It was caused by a glossary having multiple terms for one
description.

Author: tk0miya

CHANGES

@@ -155,6 +155,8 @@ Bugs fixed
   if passed as language option?
 * #5179: LaTeX: (lualatex only) escaping of ``>`` by ``\textgreater{}`` is not
   enough as ``\textgreater{}\textgreater{}`` applies TeX-ligature
+* HTML: Invalid HTML5 file is generated for a glossary having multiple terms for
+  one description (refs: #4611)
 
 Testing
 --------

sphinx/writers/html.py

@@ -393,6 +393,48 @@ def visit_enumerated_list(self, node):
         self.generate_targets_for_listing(node)
         super().visit_enumerated_list(node)
 
+    # overwritten
+    def visit_definition(self, node):
+        # type: (nodes.Element) -> None
+        # don't insert </dt> here.
+        self.body.append(self.starttag(node, 'dd', ''))
+
+    # overwritten
+    def depart_definition(self, node):
+        # type: (nodes.Element) -> None
+        self.body.append('</dd>\n')
+
+    # overwritten
+    def visit_classifier(self, node):
+        # type: (nodes.Element) -> None
+        self.body.append(self.starttag(node, 'span', '', CLASS='classifier'))
+
+    # overwritten
+    def depart_classifier(self, node):
+        # type: (nodes.Element) -> None
+        self.body.append('</span>')
+
+        next_node = node.next_node(descend=False, siblings=True)  # type: nodes.Node
+        if not isinstance(next_node, nodes.classifier):
+            # close `<dt>` tag at the tail of classifiers
+            self.body.append('</dt>')
+
+    # overwritten
+    def visit_term(self, node):
+        # type: (nodes.Element) -> None
+        self.body.append(self.starttag(node, 'dt', ''))
+
+    # overwritten
+    def depart_term(self, node):
+        # type: (nodes.Element) -> None
+        next_node = node.next_node(descend=False, siblings=True)  # type: nodes.Node
+        if isinstance(next_node, nodes.classifier):
+            # Leave the end tag to `self.depart_classifier()`, in case
+            # there's a classifier.
+            pass
+        else:
+            self.body.append('</dt>')
+
     # overwritten
     def visit_title(self, node):
         # type: (nodes.Element) -> None

sphinx/writers/html5.py

@@ -339,6 +339,48 @@ def visit_bullet_list(self, node):
             raise nodes.SkipNode
         super().visit_bullet_list(node)
 
+    # overwritten
+    def visit_definition(self, node):
+        # type: (nodes.Element) -> None
+        # don't insert </dt> here.
+        self.body.append(self.starttag(node, 'dd', ''))
+
+    # overwritten
+    def depart_definition(self, node):
+        # type: (nodes.Element) -> None
+        self.body.append('</dd>\n')
+
+    # overwritten
+    def visit_classifier(self, node):
+        # type: (nodes.Element) -> None
+        self.body.append(self.starttag(node, 'span', '', CLASS='classifier'))
+
+    # overwritten
+    def depart_classifier(self, node):
+        # type: (nodes.Element) -> None
+        self.body.append('</span>')
+
+        next_node = node.next_node(descend=False, siblings=True)  # type: nodes.Node
+        if not isinstance(next_node, nodes.classifier):
+            # close `<dt>` tag at the tail of classifiers
+            self.body.append('</dt>')
+
+    # overwritten
+    def visit_term(self, node):
+        # type: (nodes.Element) -> None
+        self.body.append(self.starttag(node, 'dt', ''))
+
+    # overwritten
+    def depart_term(self, node):
+        # type: (nodes.Element) -> None
+        next_node = node.next_node(descend=False, siblings=True)  # type: nodes.Node
+        if isinstance(next_node, nodes.classifier):
+            # Leave the end tag to `self.depart_classifier()`, in case
+            # there's a classifier.
+            pass
+        else:
+            self.body.append('</dt>')
+
     # overwritten
     def visit_title(self, node):
         # type: (nodes.Element) -> None

tests/test_markup.py

@@ -234,6 +234,29 @@ def get(name):
         None,
         r'\\sphinxhref{http://example.com/~me/}{test}.*',
     ),
+    (
+        # description list: simple
+        'verify',
+        'term\n    description',
+        '<dl class="docutils">\n<dt>term</dt><dd>description</dd>\n</dl>',
+        None,
+    ),
+    (
+        # description list: with classifiers
+        'verify',
+        'term : class1 : class2\n    description',
+        ('<dl class="docutils">\n<dt>term<span class="classifier">class1</span>'
+         '<span class="classifier">class2</span></dt><dd>description</dd>\n</dl>'),
+        None,
+    ),
+    (
+        # glossary (description list): multiple terms
+        'verify',
+        '.. glossary::\n\n   term1\n   term2\n       description',
+        ('<dl class="glossary docutils">\n<dt id="term-term1">term1</dt>'
+         '<dt id="term-term2">term2</dt><dd>description</dd>\n</dl>'),
+        None,
+    ),
 ])
 def test_inline(get_verifier, type, rst, html_expected, latex_expected):
     verifier = get_verifier(type)