LibWeb: Implement functional pseudo-element parsing

"Functional" as in "it's a function token" and not "it works", because
the behaviour for these is unimplemented. :^)

This is modeled after the pseudo-class parsing, but with some changes
based on things I don't like about that implementation. I've
implemented the `<pt-name-selector>` parameter used by view-transitions
for now, but nothing else.

Documentation/CSSGeneratedFiles.md

@@ -127,13 +127,15 @@ This generated `PsuedoElement.h` and `PseudoElement.cpp`.
 
 Each entry has the following properties:
 
-| Field                | Required | Default | Description                                                                                                                                                            |
-|----------------------|----------|---------|------------------------------------------------------------------------------------------------------------------------------------------------------------------------|
-| `spec`               | No       | Nothing | Link to the spec definition, for reference. Not used in generated code.                                                                                                |
-| `alias-for`          | No       | Nothing | Use to specify that this should be treated as an alias for the named pseudo-element.                                                                                   |
-| `is-generated`       | No       | `false` | Whether this is a [generated pseudo-element.](https://drafts.csswg.org/css-pseudo-4/#generated-content)                                                                |
-| `is-allowed-in-has`  | No       | `false` | Whether this is a [`:has`-allowed pseudo-element.](https://drafts.csswg.org/selectors/#has-allowed-pseudo-element)                                                     |
-| `property-whitelist` | No       | Nothing | Some pseudo-elements only permit certain properties. If so, name them in an array here. Some special values are allowed here for categories of properties - see below. |
+| Field                | Required | Default        | Description                                                                                                                                                            |
+|----------------------|----------|----------------|------------------------------------------------------------------------------------------------------------------------------------------------------------------------|
+| `alias-for`          | No       | Nothing        | Use to specify that this should be treated as an alias for the named pseudo-element.                                                                                   |
+| `function-syntax`    | No       | Nothing        | Syntax for the function arguments if this is a function-type pseudo-element. Copied directly from the spec.                                                            |
+| `is-generated`       | No       | `false`        | Whether this is a [generated pseudo-element.](https://drafts.csswg.org/css-pseudo-4/#generated-content)                                                                |
+| `is-allowed-in-has`  | No       | `false`        | Whether this is a [`:has`-allowed pseudo-element.](https://drafts.csswg.org/selectors/#has-allowed-pseudo-element)                                                     |
+| `property-whitelist` | No       | Nothing        | Some pseudo-elements only permit certain properties. If so, name them in an array here. Some special values are allowed here for categories of properties - see below. |
+| `spec`               | No       | Nothing        | Link to the spec definition, for reference. Not used in generated code.                                                                                                |
+| `type`               | No       | `"identifier"` | What type of pseudo-element is this. Either "identifier", "function", or "both".                                                                                       |
 
 The generated code provides:
 - A `PseudoElement` enum listing every pseudo-element name

Libraries/LibWeb/Animations/KeyframeEffect.cpp

@@ -778,7 +778,7 @@ Optional<String> KeyframeEffect::pseudo_element() const
 {
     if (!m_target_pseudo_selector.has_value())
         return {};
-    return MUST(String::formatted("::{}", m_target_pseudo_selector->name()));
+    return m_target_pseudo_selector->serialize();
 }
 
 // https://drafts.csswg.org/web-animations-1/#dom-keyframeeffect-pseudoelement

Libraries/LibWeb/CSS/Parser/SelectorParsing.cpp

@@ -401,46 +401,104 @@ Parser::ParseErrorOr<Selector::SimpleSelector> Parser::parse_pseudo_simple_selec
     if (peek_token_ends_selector())
         return ParseError::SyntaxError;
 
-    bool is_pseudo = false;
+    // Note that we already consumed one colon before we entered this function.
+    // FIXME: Don't do that.
+    bool is_pseudo_element = false;
     if (tokens.next_token().is(Token::Type::Colon)) {
-        is_pseudo = true;
+        is_pseudo_element = true;
         tokens.discard_a_token();
         if (peek_token_ends_selector())
             return ParseError::SyntaxError;
     }
 
-    if (is_pseudo) {
+    if (is_pseudo_element) {
         auto const& name_token = tokens.consume_a_token();
-        if (!name_token.is(Token::Type::Ident)) {
-            dbgln_if(CSS_PARSER_DEBUG, "Expected an ident for pseudo-element, got: '{}'", name_token.to_debug_string());
+        bool is_function = false;
+        FlyString pseudo_name;
+
+        if (name_token.is(Token::Type::Ident)) {
+            pseudo_name = name_token.token().ident();
+        } else if (name_token.is_function()) {
+            pseudo_name = name_token.function().name;
+            is_function = true;
+        } else {
+            dbgln_if(CSS_PARSER_DEBUG, "Expected an ident or function token for pseudo-element, got: '{}'", name_token.to_debug_string());
             return ParseError::SyntaxError;
         }
 
-        auto pseudo_name = name_token.token().ident();
+        bool is_aliased_pseudo = false;
+        auto pseudo_element = pseudo_element_from_string(pseudo_name);
+        if (!pseudo_element.has_value()) {
+            pseudo_element = aliased_pseudo_element_from_string(pseudo_name);
+            is_aliased_pseudo = pseudo_element.has_value();
+        }
+
+        if (pseudo_element.has_value()) {
+            auto metadata = pseudo_element_metadata(*pseudo_element);
 
-        if (auto pseudo_element = pseudo_element_from_string(pseudo_name); pseudo_element.has_value()) {
             // :has() is fussy about pseudo-elements inside it
             if (m_pseudo_class_context.contains_slow(PseudoClass::Has) && !is_has_allowed_pseudo_element(*pseudo_element)) {
                 return ParseError::SyntaxError;
             }
 
-            return Selector::SimpleSelector {
-                .type = Selector::SimpleSelector::Type::PseudoElement,
-                .value = Selector::PseudoElementSelector { pseudo_element.release_value() }
-            };
-        }
+            Selector::PseudoElementSelector::Value value = Empty {};
+            if (is_function) {
+                if (!metadata.is_valid_as_function) {
+                    dbgln_if(CSS_PARSER_DEBUG, "Pseudo-element '::{}()' is not valid as a function.", pseudo_name);
+                    return ParseError::SyntaxError;
+                }
 
-        // Aliased pseudo-elements behave like their target pseudo-element, but serialize as themselves. So store their
-        // name like we do for unknown -webkit pseudos below.
-        if (auto pseudo_element = aliased_pseudo_element_from_string(pseudo_name); pseudo_element.has_value()) {
-            // :has() is fussy about pseudo-elements inside it
-            if (m_pseudo_class_context.contains_slow(PseudoClass::Has) && !is_has_allowed_pseudo_element(*pseudo_element)) {
-                return ParseError::SyntaxError;
+                // Parse arguments
+                TokenStream function_tokens { name_token.function().value };
+                function_tokens.discard_whitespace();
+
+                switch (metadata.parameter_type) {
+                case PseudoElementMetadata::ParameterType::None:
+                    if (function_tokens.has_next_token()) {
+                        dbgln_if(CSS_PARSER_DEBUG, "Pseudo-element '::{}()' invalid: Should have no arguments.", pseudo_name);
+                        return ParseError::SyntaxError;
+                    }
+                    break;
+                case PseudoElementMetadata::ParameterType::PTNameSelector: {
+                    // <pt-name-selector> = '*' | <custom-ident>
+                    // https://drafts.csswg.org/css-view-transitions-1/#typedef-pt-name-selector
+                    if (function_tokens.next_token().is_delim('*')) {
+                        function_tokens.discard_a_token(); // *
+                        value = Selector::PseudoElementSelector::PTNameSelector { .is_universal = true };
+                    } else if (auto custom_ident = parse_custom_ident(function_tokens, {}); custom_ident.has_value()) {
+                        value = Selector::PseudoElementSelector::PTNameSelector { .value = custom_ident.release_value() };
+                    } else {
+                        dbgln_if(CSS_PARSER_DEBUG, "Invalid <pt-name-selector> in :{}() - expected `*` or `<custom-ident>`, got `{}`", pseudo_name, function_tokens.next_token().to_debug_string());
+                        return ParseError::SyntaxError;
+                    }
+                    function_tokens.discard_whitespace();
+                    if (function_tokens.has_next_token()) {
+                        dbgln_if(CSS_PARSER_DEBUG, "Invalid <pt-name-selector> in :{}() - trailing tokens", pseudo_name);
+                        return ParseError::SyntaxError;
+                    }
+                    break;
+                }
+                }
+
+            } else {
+                if (!metadata.is_valid_as_identifier) {
+                    dbgln_if(CSS_PARSER_DEBUG, "Pseudo-element '::{}' is not valid as an identifier.", pseudo_name);
+                    return ParseError::SyntaxError;
+                }
+            }
+
+            // Aliased pseudo-elements behave like their target pseudo-element, but serialize as themselves. So store their
+            // name like we do for unknown -webkit pseudos below.
+            if (is_aliased_pseudo) {
+                return Selector::SimpleSelector {
+                    .type = Selector::SimpleSelector::Type::PseudoElement,
+                    .value = Selector::PseudoElementSelector { pseudo_element.release_value(), pseudo_name.to_string().to_ascii_lowercase(), move(value) }
+                };
             }
 
             return Selector::SimpleSelector {
                 .type = Selector::SimpleSelector::Type::PseudoElement,
-                .value = Selector::PseudoElementSelector { pseudo_element.release_value(), pseudo_name.to_string().to_ascii_lowercase() }
+                .value = Selector::PseudoElementSelector { pseudo_element.release_value(), move(value) }
             };
         }
 
@@ -449,7 +507,7 @@ Parser::ParseErrorOr<Selector::SimpleSelector> Parser::parse_pseudo_simple_selec
         // and that are not functional notations must be treated as valid at parse time. (That is, ::-webkit-asdf is
         // valid at parse time, but ::-webkit-jkl() is not.) If they’re not otherwise recognized and supported, they
         // must be treated as matching nothing, and are unknown -webkit- pseudo-elements.
-        if (pseudo_name.starts_with_bytes("-webkit-"sv, CaseSensitivity::CaseInsensitive)) {
+        if (!is_function && pseudo_name.starts_with_bytes("-webkit-"sv, CaseSensitivity::CaseInsensitive)) {
             // :has() only allows a limited set of pseudo-elements inside it, which doesn't include unknown ones.
             if (m_pseudo_class_context.contains_slow(PseudoClass::Has))
                 return ParseError::SyntaxError;

Libraries/LibWeb/CSS/PseudoElements.json

@@ -111,5 +111,28 @@
   },
   "track": {
     "spec": "https://drafts.csswg.org/css-forms-1/#selectordef-track"
+  },
+  "view-transition": {
+    "spec": "https://drafts.csswg.org/css-view-transitions-1/#selectordef-view-transition"
+  },
+  "view-transition-group": {
+    "spec": "https://drafts.csswg.org/css-view-transitions-1/#selectordef-view-transition-group",
+    "type": "function",
+    "function-syntax": "<pt-name-selector>"
+  },
+  "view-transition-image-pair": {
+    "spec": "https://drafts.csswg.org/css-view-transitions-1/#selectordef-view-transition-image-pair",
+    "type": "function",
+    "function-syntax": "<pt-name-selector>"
+  },
+  "view-transition-new": {
+    "spec": "https://drafts.csswg.org/css-view-transitions-1/#selectordef-view-transition-new",
+    "type": "function",
+    "function-syntax": "<pt-name-selector>"
+  },
+  "view-transition-old": {
+    "spec": "https://drafts.csswg.org/css-view-transitions-1/#selectordef-view-transition-old",
+    "type": "function",
+    "function-syntax": "<pt-name-selector>"
   }
 }

Libraries/LibWeb/CSS/Selector.cpp

@@ -1,6 +1,6 @@
 /*
  * Copyright (c) 2018-2020, Andreas Kling <andreas@ladybird.org>
- * Copyright (c) 2021-2024, Sam Atkins <sam@ladybird.org>
+ * Copyright (c) 2021-2025, Sam Atkins <sam@ladybird.org>
  *
  * SPDX-License-Identifier: BSD-2-Clause
  */
@@ -285,6 +285,31 @@ u32 Selector::specificity() const
     return *m_specificity;
 }
 
+String Selector::PseudoElementSelector::serialize() const
+{
+    StringBuilder builder;
+    builder.append("::"sv);
+
+    if (!m_name.is_empty()) {
+        builder.append(m_name);
+    } else {
+        builder.append(pseudo_element_name(m_type));
+    }
+
+    m_value.visit(
+        [&builder](PTNameSelector const& pt_name_selector) {
+            builder.append('(');
+            if (pt_name_selector.is_universal)
+                builder.append('*');
+            else
+                builder.append(pt_name_selector.value);
+            builder.append(')');
+        },
+        [](Empty const&) {});
+
+    return builder.to_string_without_validation();
+}
+
 // https://www.w3.org/TR/cssom/#serialize-a-simple-selector
 String Selector::SimpleSelector::serialize() const
 {
@@ -519,8 +544,7 @@ String Selector::serialize() const
             // 4. If this is the last part of the chain of the selector and there is a pseudo-element,
             // append "::" followed by the name of the pseudo-element, to s.
             if (compound_selector.simple_selectors.last().type == Selector::SimpleSelector::Type::PseudoElement) {
-                s.append("::"sv);
-                s.append(compound_selector.simple_selectors.last().pseudo_element().name());
+                s.append(compound_selector.simple_selectors.last().pseudo_element().serialize());
             }
         }
     }

Libraries/LibWeb/CSS/Selector.h

@@ -25,15 +25,24 @@ class Selector : public RefCounted<Selector> {
 public:
     class PseudoElementSelector {
     public:
-        explicit PseudoElementSelector(PseudoElement type)
+        struct PTNameSelector {
+            bool is_universal { false };
+            FlyString value {};
+        };
+
+        using Value = Variant<Empty, PTNameSelector>;
+
+        explicit PseudoElementSelector(PseudoElement type, Value value = {})
             : m_type(type)
+            , m_value(move(value))
         {
             VERIFY(is_known_pseudo_element_type(type));
         }
 
-        PseudoElementSelector(PseudoElement type, String name)
+        PseudoElementSelector(PseudoElement type, String name, Value value = {})
             : m_type(type)
             , m_name(move(name))
+            , m_value(move(value))
         {
         }
 
@@ -44,19 +53,16 @@ public:
             return to_underlying(type) < to_underlying(PseudoElement::KnownPseudoElementCount);
         }
 
-        StringView name() const
-        {
-            if (!m_name.is_empty())
-                return m_name;
-
-            return pseudo_element_name(m_type);
-        }
+        String serialize() const;
 
         PseudoElement type() const { return m_type; }
 
+        PTNameSelector const& pt_name_selector() const { return m_value.get<PTNameSelector>(); }
+
     private:
         PseudoElement m_type;
         String m_name;
+        Variant<Empty, PTNameSelector> m_value;
     };
 
     struct SimpleSelector {

Libraries/LibWeb/Dump.cpp

@@ -586,7 +586,19 @@ void dump_selector(StringBuilder& builder, CSS::Selector const& selector, int in
             }
 
             if (simple_selector.type == CSS::Selector::SimpleSelector::Type::PseudoElement) {
-                builder.appendff(" pseudo_element={}", simple_selector.pseudo_element().name());
+                auto const& pseudo_element = simple_selector.pseudo_element();
+                builder.appendff(" pseudo_element={}", CSS::pseudo_element_name(pseudo_element.type()));
+                auto pseudo_element_metadata = CSS::pseudo_element_metadata(pseudo_element.type());
+
+                switch (pseudo_element_metadata.parameter_type) {
+                case CSS::PseudoElementMetadata::ParameterType::None:
+                    break;
+                case CSS::PseudoElementMetadata::ParameterType::PTNameSelector: {
+                    auto const& [is_universal, value] = pseudo_element.pt_name_selector();
+                    builder.appendff("(is_universal={}, value='{}')", is_universal, value);
+                    break;
+                }
+                }
             }
 
             if (simple_selector.type == CSS::Selector::SimpleSelector::Type::Attribute) {

Meta/Lagom/Tools/CodeGenerators/LibWeb/GenerateCSSPseudoElement.cpp

@@ -88,6 +88,16 @@ StringView pseudo_element_name(PseudoElement);
 bool is_has_allowed_pseudo_element(PseudoElement);
 bool pseudo_element_supports_property(PseudoElement, PropertyID);
 
+struct PseudoElementMetadata {
+    enum class ParameterType {
+        None,
+        PTNameSelector,
+    } parameter_type;
+    bool is_valid_as_function;
+    bool is_valid_as_identifier;
+};
+PseudoElementMetadata pseudo_element_metadata(PseudoElement);
+
 enum class GeneratedPseudoElement : @generated_pseudo_element_underlying_type@ {
 )~~~");
     pseudo_elements_data.for_each_member([&](auto& name, JsonValue const& value) {
@@ -463,6 +473,65 @@ bool pseudo_element_supports_property(PseudoElement pseudo_element, PropertyID p
     }
 }
 
+PseudoElementMetadata pseudo_element_metadata(PseudoElement pseudo_element)
+{
+    switch (pseudo_element) {
+)~~~");
+    pseudo_elements_data.for_each_member([&](auto& name, JsonValue const& value) {
+        auto& pseudo_element = value.as_object();
+        if (pseudo_element.has("alias-for"sv))
+            return;
+
+        bool is_valid_as_function = false;
+        bool is_valid_as_identifier = false;
+        auto const& type = pseudo_element.get_string("type"sv);
+        if (type == "function"sv) {
+            is_valid_as_function = true;
+        } else if (type == "both"sv) {
+            is_valid_as_function = true;
+            is_valid_as_identifier = true;
+        } else {
+            is_valid_as_identifier = true;
+        }
+
+        String parameter_type = "None"_string;
+        if (is_valid_as_function) {
+            auto const& function_syntax = pseudo_element.get_string("function-syntax"sv).value();
+            if (function_syntax == "<pt-name-selector>"sv) {
+                parameter_type = "PTNameSelector"_string;
+            } else {
+                warnln("Unrecognized pseudo-element parameter type: `{}`", function_syntax);
+                VERIFY_NOT_REACHED();
+            }
+        } else if (pseudo_element.has("function-syntax"sv)) {
+            warnln("Pseudo-element `::{}` has `function-syntax` but is not a function type.", name);
+            VERIFY_NOT_REACHED();
+        }
+
+        auto member_generator = generator.fork();
+        member_generator.set("name:titlecase", title_casify(name));
+        member_generator.set("parameter_type", parameter_type);
+        member_generator.set("is_valid_as_function", is_valid_as_function ? "true"_string : "false"_string);
+        member_generator.set("is_valid_as_identifier", is_valid_as_identifier ? "true"_string : "false"_string);
+
+        member_generator.append(R"~~~(
+    case PseudoElement::@name:titlecase@:
+        return {
+            .parameter_type = PseudoElementMetadata::ParameterType::@parameter_type@,
+            .is_valid_as_function = @is_valid_as_function@,
+            .is_valid_as_identifier = @is_valid_as_identifier@,
+        };
+)~~~");
+    });
+
+    generator.append(R"~~~(
+    case PseudoElement::KnownPseudoElementCount:
+    case PseudoElement::UnknownWebKit:
+        break;
+    }
+    VERIFY_NOT_REACHED();
+}
+
 Optional<GeneratedPseudoElement> to_generated_pseudo_element(PseudoElement pseudo_element)
 {
     switch (pseudo_element) {