LibWeb: Pull out larger parsing parts from Parser::parse_simple_selector

This lowers its cognitive complexity from 271 to under 100.
The new `parse_pseudo_simple_selector` still has a complexity of 114.

Userland/Libraries/LibWeb/CSS/Parser/Parser.cpp

@@ -307,70 +307,8 @@ Optional<Selector::Combinator> Parser::parse_selector_combinator(TokenStream<Sty
     return {};
 }
 
-Result<Selector::SimpleSelector, Parser::ParsingResult> Parser::parse_simple_selector(TokenStream<StyleComponentValueRule>& tokens)
+Result<Selector::SimpleSelector, Parser::ParsingResult> Parser::parse_attribute_simple_selector(StyleComponentValueRule const& first_value)
 {
-    auto peek_token_ends_selector = [&]() -> bool {
-        auto const& value = tokens.peek_token();
-        return (value.is(Token::Type::EndOfFile) || value.is(Token::Type::Whitespace) || value.is(Token::Type::Comma));
-    };
-
-    if (peek_token_ends_selector())
-        return ParsingResult::Done;
-
-    auto const& first_value = tokens.next_token();
-
-    if (first_value.is(Token::Type::Delim)) {
-        u32 delim = first_value.token().delim();
-        switch (delim) {
-        case '*':
-            return Selector::SimpleSelector {
-                .type = Selector::SimpleSelector::Type::Universal
-            };
-        case '.': {
-            if (peek_token_ends_selector())
-                return ParsingResult::SyntaxError;
-
-            auto const& class_name_value = tokens.next_token();
-            if (!class_name_value.is(Token::Type::Ident)) {
-                dbgln_if(CSS_PARSER_DEBUG, "Expected an ident after '.', got: {}", class_name_value.to_debug_string());
-                return ParsingResult::SyntaxError;
-            }
-            return Selector::SimpleSelector {
-                .type = Selector::SimpleSelector::Type::Class,
-                .value = class_name_value.token().ident()
-            };
-        }
-        case '>':
-        case '+':
-        case '~':
-        case '|':
-            // Whitespace is not required between the compound-selector and a combinator.
-            // So, if we see a combinator, return that this compound-selector is done, instead of a syntax error.
-            tokens.reconsume_current_input_token();
-            return ParsingResult::Done;
-        default:
-            dbgln_if(CSS_PARSER_DEBUG, "!!! Invalid simple selector!");
-            return ParsingResult::SyntaxError;
-        }
-    }
-
-    if (first_value.is(Token::Type::Hash)) {
-        if (first_value.token().hash_type() != Token::HashType::Id) {
-            dbgln_if(CSS_PARSER_DEBUG, "Selector contains hash token that is not an id: {}", first_value.to_debug_string());
-            return ParsingResult::SyntaxError;
-        }
-        return Selector::SimpleSelector {
-            .type = Selector::SimpleSelector::Type::Id,
-            .value = first_value.token().hash_value()
-        };
-    }
-    if (first_value.is(Token::Type::Ident)) {
-        return Selector::SimpleSelector {
-            .type = Selector::SimpleSelector::Type::TagName,
-            .value = first_value.token().ident()
-        };
-    }
-    if (first_value.is_block() && first_value.block().is_square()) {
     auto attribute_tokens = TokenStream { first_value.block().values() };
 
     attribute_tokens.skip_whitespace();
@@ -461,8 +399,15 @@ Result<Selector::SimpleSelector, Parser::ParsingResult> Parser::parse_simple_sel
 
     // FIXME: Handle case-sensitivity suffixes. https://www.w3.org/TR/selectors-4/#attribute-case
     return simple_selector;
-    }
+}
-    if (first_value.is(Token::Type::Colon)) {
+
+Result<Selector::SimpleSelector, Parser::ParsingResult> Parser::parse_pseudo_simple_selector(TokenStream<StyleComponentValueRule>& tokens)
+{
+    auto peek_token_ends_selector = [&]() -> bool {
+        auto const& value = tokens.peek_token();
+        return (value.is(Token::Type::EndOfFile) || value.is(Token::Type::Whitespace) || value.is(Token::Type::Comma));
+    };
+
     if (peek_token_ends_selector())
         return ParsingResult::SyntaxError;
 
@@ -571,7 +516,6 @@ Result<Selector::SimpleSelector, Parser::ParsingResult> Parser::parse_simple_sel
         return simple_selector;
     }
     if (pseudo_class_token.is_function()) {
-
         auto parse_nth_child_pattern = [this](Selector::SimpleSelector& simple_selector, StyleFunctionRule const& pseudo_function, bool allow_of = false) -> bool {
             auto function_values = TokenStream<StyleComponentValueRule>(pseudo_function.values());
             auto nth_child_pattern = parse_a_n_plus_b_pattern(function_values, allow_of ? AllowTrailingTokens::Yes : AllowTrailingTokens::No);
@@ -649,7 +593,76 @@ Result<Selector::SimpleSelector, Parser::ParsingResult> Parser::parse_simple_sel
     }
     dbgln_if(CSS_PARSER_DEBUG, "Unexpected Block in pseudo-class name, expected a function or identifier. '{}'", pseudo_class_token.to_debug_string());
     return ParsingResult::SyntaxError;
+}
+
+Result<Selector::SimpleSelector, Parser::ParsingResult> Parser::parse_simple_selector(TokenStream<StyleComponentValueRule>& tokens)
+{
+    auto peek_token_ends_selector = [&]() -> bool {
+        auto const& value = tokens.peek_token();
+        return (value.is(Token::Type::EndOfFile) || value.is(Token::Type::Whitespace) || value.is(Token::Type::Comma));
+    };
+
+    if (peek_token_ends_selector())
+        return ParsingResult::Done;
+
+    auto const& first_value = tokens.next_token();
+
+    if (first_value.is(Token::Type::Delim)) {
+        u32 delim = first_value.token().delim();
+        switch (delim) {
+        case '*':
+            return Selector::SimpleSelector {
+                .type = Selector::SimpleSelector::Type::Universal
+            };
+        case '.': {
+            if (peek_token_ends_selector())
+                return ParsingResult::SyntaxError;
+
+            auto const& class_name_value = tokens.next_token();
+            if (!class_name_value.is(Token::Type::Ident)) {
+                dbgln_if(CSS_PARSER_DEBUG, "Expected an ident after '.', got: {}", class_name_value.to_debug_string());
+                return ParsingResult::SyntaxError;
             }
+            return Selector::SimpleSelector {
+                .type = Selector::SimpleSelector::Type::Class,
+                .value = class_name_value.token().ident()
+            };
+        }
+        case '>':
+        case '+':
+        case '~':
+        case '|':
+            // Whitespace is not required between the compound-selector and a combinator.
+            // So, if we see a combinator, return that this compound-selector is done, instead of a syntax error.
+            tokens.reconsume_current_input_token();
+            return ParsingResult::Done;
+        default:
+            dbgln_if(CSS_PARSER_DEBUG, "!!! Invalid simple selector!");
+            return ParsingResult::SyntaxError;
+        }
+    }
+
+    if (first_value.is(Token::Type::Hash)) {
+        if (first_value.token().hash_type() != Token::HashType::Id) {
+            dbgln_if(CSS_PARSER_DEBUG, "Selector contains hash token that is not an id: {}", first_value.to_debug_string());
+            return ParsingResult::SyntaxError;
+        }
+        return Selector::SimpleSelector {
+            .type = Selector::SimpleSelector::Type::Id,
+            .value = first_value.token().hash_value()
+        };
+    }
+    if (first_value.is(Token::Type::Ident)) {
+        return Selector::SimpleSelector {
+            .type = Selector::SimpleSelector::Type::TagName,
+            .value = first_value.token().ident()
+        };
+    }
+    if (first_value.is_block() && first_value.block().is_square())
+        return parse_attribute_simple_selector(first_value);
+
+    if (first_value.is(Token::Type::Colon))
+        return parse_pseudo_simple_selector(tokens);
 
     dbgln_if(CSS_PARSER_DEBUG, "!!! Invalid simple selector!");
     return ParsingResult::SyntaxError;
@@ -4901,7 +4914,6 @@ TimePercentage Parser::Dimension::time_percentage() const
         return percentage();
     VERIFY_NOT_REACHED();
 }
-
 }
 
 namespace Web {

Userland/Libraries/LibWeb/CSS/Parser/Parser.h

@@ -315,6 +315,9 @@ private:
     Result<NonnullRefPtr<Selector>, ParsingResult> parse_complex_selector(TokenStream<StyleComponentValueRule>&, SelectorType);
     Result<Selector::CompoundSelector, ParsingResult> parse_compound_selector(TokenStream<StyleComponentValueRule>&);
     Optional<Selector::Combinator> parse_selector_combinator(TokenStream<StyleComponentValueRule>&);
+
+    Result<Selector::SimpleSelector, ParsingResult> parse_attribute_simple_selector(StyleComponentValueRule const&);
+    Result<Selector::SimpleSelector, ParsingResult> parse_pseudo_simple_selector(TokenStream<StyleComponentValueRule>&);
     Result<Selector::SimpleSelector, ParsingResult> parse_simple_selector(TokenStream<StyleComponentValueRule>&);
 
     NonnullRefPtr<MediaQuery> parse_media_query(TokenStream<StyleComponentValueRule>&);