feat(css_semantic): build semantic model for css (biomejs#3546)

.github/workflows/pull_request.yml

@@ -5,13 +5,13 @@ on:
     branches:
       - main
     paths: # Only run when changes are made to rust code or root Cargo
-      - 'crates/**'
-      - 'fuzz/**'
-      - 'xtask/**'
-      - 'Cargo.toml'
-      - 'Cargo.lock'
-      - 'rust-toolchain.toml'
-      - 'rustfmt.toml'
+      - "crates/**"
+      - "fuzz/**"
+      - "xtask/**"
+      - "Cargo.toml"
+      - "Cargo.lock"
+      - "rust-toolchain.toml"
+      - "rustfmt.toml"
 
 # Cancel jobs when the PR is updated
 concurrency:
@@ -184,5 +184,6 @@ jobs:
         run: |
           if [[ `git status --porcelain` ]]; then
             git status
+            git diff
             exit 1
           fi

Cargo.toml

@@ -97,6 +97,7 @@ biome_css_analyze            = { version = "0.5.7", path = "./crates/biome_css_a
 biome_css_factory            = { version = "0.5.7", path = "./crates/biome_css_factory" }
 biome_css_formatter          = { version = "0.5.7", path = "./crates/biome_css_formatter" }
 biome_css_parser             = { version = "0.5.7", path = "./crates/biome_css_parser" }
+biome_css_semantic           = { version = "0.0.0", path = "./crates/biome_css_semantic" }
 biome_css_syntax             = { version = "0.5.7", path = "./crates/biome_css_syntax" }
 biome_deserialize            = { version = "0.6.0", path = "./crates/biome_deserialize" }
 biome_deserialize_macros     = { version = "0.6.0", path = "./crates/biome_deserialize_macros" }

crates/biome_css_semantic/Cargo.toml

@@ -0,0 +1,23 @@
+
+[package]
+authors.workspace    = true
+categories.workspace = true
+description          = "Biome's semantic model for CSS"
+edition.workspace    = true
+homepage.workspace   = true
+keywords.workspace   = true
+license.workspace    = true
+name                 = "biome_css_semantic"
+repository.workspace = true
+version              = "0.0.0"
+
+[dependencies]
+biome_css_syntax = { workspace = true }
+biome_rowan      = { workspace = true }
+rustc-hash       = { workspace = true }
+
+[dev-dependencies]
+biome_css_parser = { path = "../biome_css_parser" }
+
+[lints]
+workspace = true

crates/biome_css_semantic/src/events.rs

@@ -0,0 +1,105 @@
+use std::collections::VecDeque;
+
+use biome_css_syntax::{AnyCssSelector, CssRelativeSelector, CssSyntaxKind::*};
+use biome_rowan::{AstNode, TextRange};
+
+use crate::semantic_model::model::Specificity;
+
+#[derive(Debug)]
+pub enum SemanticEvent {
+    RuleStart(TextRange),
+    RuleEnd,
+    SelectorDeclaration {
+        name: String,
+        range: TextRange,
+        specificity: Specificity,
+    },
+    PropertyDeclaration {
+        property: String,
+        value: String,
+        property_range: TextRange,
+        value_range: TextRange,
+    },
+}
+
+#[derive(Default, Debug)]
+pub struct SemanticEventExtractor {
+    stash: VecDeque<SemanticEvent>,
+    current_rule_stack: Vec<TextRange>,
+}
+
+impl SemanticEventExtractor {
+    pub fn enter(&mut self, node: &biome_css_syntax::CssSyntaxNode) {
+        match node.kind() {
+            kind if kind == CSS_QUALIFIED_RULE || kind == CSS_NESTED_QUALIFIED_RULE => {
+                let range = node.text_range();
+                self.stash.push_back(SemanticEvent::RuleStart(range));
+                self.current_rule_stack.push(range);
+            }
+            CSS_SELECTOR_LIST => {
+                node.children()
+                    .filter_map(AnyCssSelector::cast)
+                    .for_each(|s| self.process_selector(s));
+            }
+            CSS_RELATIVE_SELECTOR_LIST => {
+                node.children()
+                    .filter_map(CssRelativeSelector::cast)
+                    .filter_map(|s| s.selector().ok())
+                    .for_each(|s| self.process_selector(s));
+            }
+            CSS_DECLARATION => {
+                if let Some(property_name) = node.first_child().and_then(|p| p.first_child()) {
+                    if let Some(value) = property_name.next_sibling() {
+                        self.stash.push_back(SemanticEvent::PropertyDeclaration {
+                            property: property_name.text_trimmed().to_string(),
+                            value: value.text_trimmed().to_string(),
+                            property_range: property_name.text_range(),
+                            value_range: value.text_range(),
+                        });
+                    }
+                }
+            }
+            _ => {}
+        }
+    }
+
+    fn process_selector(&mut self, selector: AnyCssSelector) {
+        match selector {
+            AnyCssSelector::CssComplexSelector(s) => {
+                if let Ok(l) = s.left() {
+                    self.add_selector_event(l.text(), l.range());
+                }
+                if let Ok(r) = s.right() {
+                    self.add_selector_event(r.text(), r.range());
+                }
+            }
+            AnyCssSelector::CssCompoundSelector(selector) => {
+                self.add_selector_event(selector.text().to_string(), selector.range());
+            }
+            _ => {}
+        }
+    }
+
+    fn add_selector_event(&mut self, name: String, range: TextRange) {
+        self.stash.push_back(SemanticEvent::SelectorDeclaration {
+            name,
+            range,
+            specificity: Specificity(0, 0, 0), // TODO: Implement this
+        });
+    }
+
+    pub fn leave(&mut self, node: &biome_css_syntax::CssSyntaxNode) {
+        if matches!(
+            node.kind(),
+            biome_css_syntax::CssSyntaxKind::CSS_QUALIFIED_RULE
+                | biome_css_syntax::CssSyntaxKind::CSS_NESTED_QUALIFIED_RULE
+        ) {
+            self.current_rule_stack.pop();
+            self.stash.push_back(SemanticEvent::RuleEnd);
+        }
+    }
+
+    pub fn pop(&mut self) -> Option<SemanticEvent> {
+        self.stash.pop_front()
+    }
+}

crates/biome_css_semantic/src/lib.rs

@@ -0,0 +1,5 @@
+mod events;
+mod semantic_model;
+
+pub use events::*;
+pub use semantic_model::*;

crates/biome_css_semantic/src/semantic_model/builder.rs

@@ -0,0 +1,96 @@
+use biome_css_syntax::{CssRoot, CssSyntaxKind, CssSyntaxNode};
+use biome_rowan::TextRange;
+use rustc_hash::FxHashMap;
+
+use super::model::{Declaration, Rule, Selector, SemanticModel, SemanticModelData};
+use crate::events::SemanticEvent;
+
+pub struct SemanticModelBuilder {
+    root: CssRoot,
+    node_by_range: FxHashMap<TextRange, CssSyntaxNode>,
+    rules: Vec<Rule>,
+    current_rule_stack: Vec<Rule>,
+}
+
+impl SemanticModelBuilder {
+    pub fn new(root: CssRoot) -> Self {
+        Self {
+            root,
+            node_by_range: FxHashMap::default(),
+            rules: Vec::new(),
+            current_rule_stack: Vec::new(),
+        }
+    }
+
+    pub fn build(self) -> SemanticModel {
+        let data = SemanticModelData {
+            root: self.root,
+            node_by_range: self.node_by_range,
+            rules: self.rules,
+        };
+        SemanticModel::new(data)
+    }
+
+    #[inline]
+    pub fn push_node(&mut self, node: &CssSyntaxNode) {
+        use CssSyntaxKind::*;
+        if matches!(
+            node.kind(),
+            CSS_SELECTOR_LIST | CSS_DECLARATION | CSS_DECLARATION_OR_RULE_LIST | CSS_QUALIFIED_RULE
+        ) {
+            self.node_by_range.insert(node.text_range(), node.clone());
+        }
+    }
+
+    #[inline]
+    pub fn push_event(&mut self, event: SemanticEvent) {
+        match event {
+            SemanticEvent::RuleStart(range) => {
+                let new_rule = Rule {
+                    selectors: Vec::new(),
+                    declarations: Vec::new(),
+                    children: Vec::new(),
+                    range,
+                };
+                self.current_rule_stack.push(new_rule);
+            }
+            SemanticEvent::RuleEnd => {
+                if let Some(completed_rule) = self.current_rule_stack.pop() {
+                    if let Some(parent_rule) = self.current_rule_stack.last_mut() {
+                        parent_rule.children.push(completed_rule);
+                    } else {
+                        self.rules.push(completed_rule);
+                    }
+                }
+            }
+            SemanticEvent::SelectorDeclaration {
+                name,
+                range,
+                specificity,
+            } => {
+                if let Some(current_rule) = self.current_rule_stack.last_mut() {
+                    current_rule.selectors.push(Selector {
+                        name,
+                        range,
+                        specificity,
+                    });
+                }
+            }
+            SemanticEvent::PropertyDeclaration {
+                property,
+                value,
+                property_range,
+                value_range,
+            } => {
+                if let Some(current_rule) = self.current_rule_stack.last_mut() {
+                    current_rule.declarations.push(Declaration {
+                        property,
+                        value,
+                        property_range,
+                        value_range,
+                    });
+                }
+            }
+        }
+    }
+}

crates/biome_css_semantic/src/semantic_model/mod.rs

@@ -0,0 +1,88 @@
+mod builder;
+pub(crate) mod model;
+
+use biome_css_syntax::CssRoot;
+use biome_rowan::AstNode;
+use builder::SemanticModelBuilder;
+use model::SemanticModel;
+
+use crate::events::SemanticEventExtractor;
+
+pub fn semantic_model(root: &CssRoot) -> SemanticModel {
+    let mut extractor = SemanticEventExtractor::default();
+    let mut builder = SemanticModelBuilder::new(root.clone());
+
+    let root = root.syntax();
+    for node in root.preorder() {
+        match node {
+            biome_css_syntax::WalkEvent::Enter(node) => {
+                builder.push_node(&node);
+                extractor.enter(&node);
+            }
+            biome_css_syntax::WalkEvent::Leave(node) => extractor.leave(&node),
+        }
+    }
+
+    while let Some(e) = extractor.pop() {
+        builder.push_event(e);
+    }
+
+    builder.build()
+}
+
+#[cfg(test)]
+mod tests {
+    use biome_css_parser::parse_css;
+    use biome_css_parser::CssParserOptions;
+
+    #[test]
+    fn test_simple_ruleset() {
+        let parse = parse_css(
+            r#"p {
+  font-family: verdana;
+  font-size: 20px;
+}"#,
+            CssParserOptions::default(),
+        );
+
+        let root = parse.tree();
+        let model = super::semantic_model(&root);
+        let rule = model.rules().first().unwrap();
+
+        assert_eq!(rule.selectors.len(), 1);
+        assert_eq!(rule.declarations.len(), 2);
+    }
+    #[test]
+    fn test_nested_selector() {
+        let parse = parse_css(
+            r#".parent {
+  color: blue;
+
+  .child {
+    color: red;
+  }
+}"#,
+            CssParserOptions::default(),
+        );
+
+        let root = parse.tree();
+        let model = super::semantic_model(&root);
+        let rule = model.rules().first().unwrap();
+
+        assert_eq!(rule.selectors.len(), 1);
+        assert_eq!(rule.declarations.len(), 1);
+        assert_eq!(rule.children.len(), 1);
+    }
+
+    #[test]
+    fn debug() {
+        let parse = parse_css(
+            r#"[a="b"i], [ a="b"i], [ a ="b"i], [ a = "b"i], [ a = "b" i], [ a = "b" i ] {}"#,
+            CssParserOptions::default(),
+        );
+
+        let root = parse.tree();
+        let model = super::semantic_model(&root);
+        dbg!(&model.rules());
+    }
+}