🐛 FIX: attrs: unterminated groups, class accumulation, closing-token merge

mdit_py_plugins/attrs/index.py

@@ -41,8 +41,14 @@ def attrs_plugin(
     - `key="value"` or `key=value` specifies a key-value attribute.
        Quotes are not needed when the value consists entirely of
        ASCII alphanumeric characters or `_` or `:` or `-`.
-       Backslash escapes may be used inside quoted values.
-    - `%` begins a comment, which ends with the next `%` or the end of the attribute (`}`).
+       Backslash escapes may be used inside quoted values,
+       to allow a ``"`` character within the value.
+       Note that the backslash is retained in the value:
+       ``{k="a\\"b"}`` gives ``k`` the value ``a\\"b``.
+    - `%` begins a comment, which ends with the next `%`.
+      If no further `%` occurs in the rest of the line (for a block attribute)
+      or of the paragraph (for an inline attribute),
+      the comment instead ends at the next `}`, which also ends the attribute.
 
     Multiple attribute blocks are merged.
 
@@ -175,8 +181,8 @@ def _attr_inline_rule(
     state.pos += new_pos + 1
     if not silent:
         attr_token = state.tokens[token_index]
-        if "class" in attrs and "class" in token.attrs:
-            attrs["class"] = f"{token.attrs['class']} {attrs['class']}"
+        if "class" in attrs and "class" in attr_token.attrs:
+            attrs["class"] = f"{attr_token.attrs['class']} {attrs['class']}"
         _add_attrs(attr_token, attrs, allowed)
     return True
 
@@ -238,7 +244,8 @@ def _attr_resolve_block_rule(state: StateCore, *, allowed: set[str] | None) -> N
             i += 1
             continue
 
-        if i + 1 < len_tokens:
+        # skip closing tokens; attributes would be rendered into the closing tag
+        if i + 1 < len_tokens and state.tokens[i + 1].nesting >= 0:
             next_token = state.tokens[i + 1]
 
             # classes are appended

mdit_py_plugins/attrs/parse.py

@@ -100,17 +100,28 @@ def parse(string: str) -> tuple[int, dict[str, str]]:
     """Parse attributes from start of string.
 
     :returns: (length of parsed string, dict of attributes)
+    :raises ParseError: if the attributes are malformed,
+        or the string ends before the closing `}`
     """
     pos = 0
     state: State = State.START
     tokens = TokenState()
     while pos < len(string):
-        state = HANDLERS[state](string[pos], pos, tokens)
+        if (
+            state == State.SCANNING_COMMENT
+            and string[pos] == "}"
+            and "%" not in string[pos + 1 :]
+        ):
+            # a comment ends at the next `%`, but there is none left to close it,
+            # so this `}` ends both the comment and the attributes
+            state = State.DONE
+        else:
+            state = HANDLERS[state](string[pos], pos, tokens)
         if state == State.DONE:
             return pos, tokens.compile(string)
         pos = pos + 1
 
-    return pos, tokens.compile(string)
+    raise ParseError("Attributes not terminated", pos)
 
 
 def handle_start(char: str, pos: int, tokens: TokenState) -> State: