Raise an error if xhtml-style "/" is ambiguous.

tdom/parser.py

@@ -189,11 +189,9 @@ def make_open_tag(self, tag: str, attrs: Sequence[HTMLAttribute]) -> OpenTag:
         # @NOTE: This must be called when the tag is handled since it is
         # populated based on the most recently finished start tag. Otherwise
         # the value will be out of sync.
-        starttag_text = self.get_starttag_text()
-        if starttag_text is None:
-            raise AssertionError(
-                f"Expected startag_text to be set when parsing component at {i_index}."
-            )
+        starttag_text = self.always_get_starttag_text(
+            f"Expected startag_text to be set when parsing component at {i_index}."
+        )
 
         tattrs = self.make_tattrs(attrs)
 
@@ -371,12 +369,40 @@ def validate_end_tag(self, tag: str, open_tag: OpenTag) -> int | None:
                 # any of this in the parser, instead relying on higher layers.
                 return tag_ref.i_indexes[0]
 
+    def always_get_starttag_text(
+        self, msg: str = "Expecting starttag text to be set."
+    ) -> str:
+        """
+        Wrap get_starttag_text and just raise if None is returned.
+
+        Do this so we don't guard for `None` everywhere.
+        """
+        starttag_text = self.get_starttag_text()
+        if starttag_text is None:
+            raise AssertionError(msg)
+        return starttag_text
+
     # ------------------------------------------
     # HTMLParser tag callbacks
     # ------------------------------------------
 
     def handle_starttag(self, tag: str, attrs: Sequence[HTMLAttribute]) -> None:
         open_tag = self.make_open_tag(tag, attrs)
+        # An unquoted attribute value within a self-closing tag can consume
+        # the "/" as part of the attribute's value if not separated by whitespace.
+        # This is the CORRECT behavior but can can be especially confusing if
+        # preceded by an interpolation, such as `<{Comp} name={value}/>`.
+        # We explicitly dissallow this usage without preceding ascii whitespace.
+        # This applies to all tags (components or not).
+        if self.always_get_starttag_text().endswith("/>"):
+            if isinstance(open_tag, OpenTComponent):
+                error_tag = self.get_source().format_starttag(open_tag.start_i_index)
+            else:
+                error_tag = tag
+            raise ValueError(
+                f'Ambiguous self-closing tag, "<{error_tag}.../>", please precede "/>" with whitespace or apply quotes around the last attribute value.'
+            )
+
         if isinstance(open_tag, OpenTElement) and open_tag.tag in VOID_ELEMENTS:
             final_tag = self.finalize_tag(open_tag)
             self.append_child(final_tag)

tdom/parser_test.py

@@ -602,3 +602,77 @@ def test_extract_with_templated_attr_gt_char(self, Component):
                 strings=("<div>Hello, World!</div>",), i_indexes=()
             ),
         )
+
+
+class TestAmbiguousSelfCloseCheck:
+    @pytest.fixture
+    def comp(self):
+        def component(
+            active: bool = False, title: str = "Title", children: Template = t""
+        ) -> Template:
+            dataset = {"active": active}
+            return t"<div data={dataset} title={title}>{children}</div>"
+
+        return component
+
+    def test_component_ok(self, comp):
+        dynamic = "dynamic"
+        attrs = {"active": True}
+        for template in [
+            t"<{comp}/>",
+            t"<{comp} active/>",  # Still ok because attr name cannot contain /
+            t"<{comp} {attrs}/>",  # Still ok because attr name cannot contain /
+            t"<{comp} />",
+            t"<{comp} title=literal />",
+            t"<{comp} title=literal/ ></{comp}>",  # This is really gross but shouldn't be common.
+            t'<{comp} title="literal"/>',
+            t"<{comp} title={dynamic} />",
+            t'<{comp} title="{dynamic}"/>',
+            t"<{comp} title={dynamic}literal />",
+            t'<{comp} title="{dynamic}literal"/>',
+        ]:
+            tnode = TemplateParser.parse(template)
+            assert isinstance(tnode, TComponent) and tnode.start_i_index == 0
+
+    def test_component_ambiguous_error(self, comp):
+        dynamic = "dynamic"
+        for template in (
+            t"<{comp} title=literal/>",
+            t"<{comp} title={dynamic}/>",
+            t"<{comp} title={dynamic}literal/>",
+            t"<{comp} title=/>",
+            t"<{comp} title=     />",  # WS between = and value is ignored, so title=/
+        ):
+            with pytest.raises(ValueError, match="Ambiguous self-closing tag"):
+                _ = TemplateParser.parse(template)
+
+    def test_element_ok(self):
+        dynamic = "dynamic"
+        attrs = {"active": True}
+        for template in (
+            t"<div/>",
+            t"<div active/>",  # Still ok because attr name cannot contain /
+            t"<div {attrs}/>",  # Still ok because attr name cannot contain /
+            t"<div />",
+            t"<div title=literal />",
+            t"<div title=literal/ ></div>",  # This is really gross but shouldn't be common.
+            t'<div title="literal"/>',
+            t"<div title={dynamic} />",
+            t'<div title="{dynamic}"/>',
+            t"<div title={dynamic}literal />",
+            t'<div title="{dynamic}literal"/>',
+        ):
+            tnode = TemplateParser.parse(template)
+            assert isinstance(tnode, TElement) and tnode.tag == "div"
+
+    def test_element_ambiguous_error(self):
+        dynamic = "dynamic"
+        for template in (
+            t"<div title=literal/>",
+            t"<div title={dynamic}/>",
+            t"<div title={dynamic}literal/>",
+            t"<div title=/>",
+            t"<div title=     />",  # WS between = and value is ignored, so title=/
+        ):
+            with pytest.raises(ValueError, match="Ambiguous self-closing tag"):
+                _ = TemplateParser.parse(template)