Handle classes/ids after pseudo-classes

src/parser.py

@@ -24,7 +24,7 @@ class SelectorParseError(Exception): ...
 class InvalidSelectorError(ValueError): ...
 
 
-@dataclass(frozen=True)
+@dataclass
 class SimpleSelector:
     tag: str | None = None
     classes: list[str] = field(default_factory=list)
@@ -61,6 +61,24 @@ class SimpleSelector:
 
         return cls(tag, cls_selectors, id_selectors)
 
+    def merge_selector(self, other: SimpleSelector) -> None:
+        """Merges another selector with this one (mutating this one).
+
+        This function is useful for extending this selector, for cases when we couldn't
+        parse it in one go. (E.g. separated by a pseudo-class).
+        """
+        if other.tag is not None:
+            raise SelectorParseError(f"Can't merge tag-aware selector with current selector ({self=!r}, {other=!r})")
+
+        self.classes.extend(other.classes)
+        self.ids.extend(other.ids)
+
+        if len(self.ids) > 1:
+            warnings.warn(
+                "Simple selector contains multiple IDs. The CSS spec doesn't forbid this, but it will never match.",
+                stacklevel=2,
+            )
+
     @override
     def __str__(self) -> str:
         classes = ".".join(cls_name for cls_name in self.classes)
@@ -73,7 +91,7 @@ class SimpleSelector:
 type ConcretePseudoClassSelector = NotPseudoClassSelector | NthChildPseudoClassSelector
 
 
-@dataclass(frozen=True)
+@dataclass
 class PseudoClassSelector:
     pseudo_class: str
     selector: SimpleSelector | ConcretePseudoClassSelector | None
@@ -137,7 +155,7 @@ class PseudoClassSelector:
         return cls(pseudo_class, selector, arg)
 
 
-@dataclass(frozen=True)
+@dataclass
 class DescendantSelector:
     parent: SimpleSelector | ConcretePseudoClassSelector
     child: SimpleSelector | ConcretePseudoClassSelector | DescendantSelector | SiblingSelector
@@ -149,7 +167,7 @@ class DescendantSelector:
         return f"{self.parent!s}{symbol}{self.child!s}"
 
 
-@dataclass(frozen=True)
+@dataclass
 class SiblingSelector:
     sibling_selector: SimpleSelector | ConcretePseudoClassSelector
     selector: SimpleSelector | ConcretePseudoClassSelector | DescendantSelector | SiblingSelector
@@ -164,7 +182,7 @@ class SiblingSelector:
 type NonMultiSelector = SimpleSelector | ConcretePseudoClassSelector | DescendantSelector | SiblingSelector
 
 
-@dataclass(frozen=True)
+@dataclass
 class MultiSelector:
     selectors: list[NonMultiSelector]
 
@@ -176,7 +194,7 @@ class MultiSelector:
 type AnySelector = NonMultiSelector | MultiSelector
 
 
-@dataclass(frozen=True)
+@dataclass
 class NotPseudoClassSelector:
     selector: SimpleSelector | ConcretePseudoClassSelector | None
     not_selector: AnySelector
@@ -200,7 +218,7 @@ class NotPseudoClassSelector:
         return f"{sel}:not({self.not_selector!s})"
 
 
-@dataclass(frozen=True)
+@dataclass
 class NthChildPseudoClassSelector:
     selector: SimpleSelector | ConcretePseudoClassSelector | None
     n: int
@@ -320,14 +338,29 @@ def parse_tokens(tokens: TokenStream, root: bool = True) -> AnySelector:
                 break
             case TokenType.PSEUDO_ELEMENT:
                 raise NotImplementedError("The parser doesn't (yet) support pseudo-elements")
+            case TokenType.ID | TokenType.CLASS:
+                # it's possible for a pseudo-class to come before class/id, handle this
+                # if we see id/class tokens following anything else, that's an error though
+                if not isinstance(s, (NotPseudoClassSelector, NthChildPseudoClassSelector)):
+                    raise InvalidSelectorError(f"Unexpected token while parsing selector: {tokens!r}")
+
+                # Walk all pseudo-classes in between until we get to the simple selector class
+                # which we're interested in
+                prev_sel: ConcretePseudoClassSelector = s
+                simple_sel: ConcretePseudoClassSelector | SimpleSelector | None = s
+                while not (isinstance(simple_sel, SimpleSelector) or simple_sel is None):
+                    prev_sel = simple_sel
+                    simple_sel = simple_sel.selector
+
+                # Attach the selector appropriately (either merging with existing one, or making
+                # it a new selector)
+                post_sel = SimpleSelector.parse_tokens(tokens)
+                if simple_sel is None:
+                    prev_sel.selector = post_sel
+                else:
+                    simple_sel.merge_selector(post_sel)
             case (
-                TokenType.TAG
-                | TokenType.CLASS
-                | TokenType.ID
-                | TokenType.NUMBER
-                | TokenType.LPARENS
-                | TokenType.RPARENS
-                | TokenType.UNKNOWN
+                TokenType.TAG | TokenType.NUMBER | TokenType.LPARENS | TokenType.RPARENS | TokenType.UNKNOWN
             ):  # we're using an exhaustive case to allow static analysis to catch any missing enum variants
                 raise InvalidSelectorError(f"Unexpected token while parsing selector: {tokens!r}")
 

tests/hypot.py

@@ -65,6 +65,14 @@ selector = st.recursive(
         ),
         # Apply pseudo-suffix
         st.tuples(s, pseudo_suffixes(s)).map("".join),
+        # Optionally add classes / id after the pseudo-suffix
+        st.one_of(
+            s,  # no append, keep as-is
+            st.tuples(
+                s,
+                st.lists(st.one_of(css_class, css_id), min_size=1).map("".join),
+            ).map("".join),
+        ),
     ),
     max_leaves=10,
 )

tests/test_parser.py

@@ -424,6 +424,14 @@ def test_parse_not_with_missing_argument() -> None:
         _ = NotPseudoClassSelector.from_pseudo_cls(pseudo)
 
 
+def test_parse_class_after_pseudo_class() -> None:
+    tokens = TokenStream([Token(TokenType.PSEUDO_CLASS, ":first-child"), Token(TokenType.CLASS, ".bar")])
+    sel = parse_tokens(tokens)
+    assert isinstance(sel, NthChildPseudoClassSelector)
+    assert isinstance(sel.selector, SimpleSelector)
+    assert sel.selector.classes == ["bar"]
+
+
 def test_parse_multiple_combinators() -> None:
     # div .parent > .child + .sibling:not(.bar):first-child
     tokens = TokenStream(

tests/test_qualifier.py

@@ -228,6 +228,7 @@ def test_sibling_selector(selector: str, expected: list[Node]) -> None:
         ("div:last-child", [SECOND_DIV]),
         ("div.nonexistent:last-child", []),
         (".colour-primary:first-child", [INNER_DIV, SECOND_P]),
+        (":first-child.colour-primary", [INNER_DIV, SECOND_P]),
     ],
 )
 def test_nth_child_selector(selector: str, expected: list[Node]) -> None: