feat(grit): implement disregarded snippet nodes (#4084)

.gitattributes

@@ -26,6 +26,8 @@
 /crates/biome_js_analyze/src/{lint,assists,syntax}/*.rs linguist-generated=true text=auto eol=lf
 /crates/biome_js_analyze/src/options.rs linguist-generated=true text=auto eol=lf
 /crates/biome_js_analyze/src/registry.rs linguist-generated=true text=auto eol=lf
+# Grit
+/crates/biome_grit_patterns/src/grit_target_language/*/constants.rs linguist-generated=true text=auto eol=lf
 
 # Other
 /crates/biome_unicode_table/src/tables.rs linguist-generated=true text=auto eol=lf

crates/biome_grit_patterns/src/grit_node_patterns.rs

@@ -85,10 +85,9 @@ impl Matcher<GritQueryContext> for GritNodePattern {
             let mut cur_state = running_state.clone();
 
             let res = pattern.execute(
-                &if let Some(child) = node.child_by_slot_index(*slot_index) {
-                    GritResolvedPattern::from_node_binding(child)
-                } else {
-                    GritResolvedPattern::from_empty_binding(node.clone(), *slot_index)
+                &match node.child_by_slot_index(*slot_index) {
+                    Some(child) => GritResolvedPattern::from_node_binding(child),
+                    None => GritResolvedPattern::from_empty_binding(node.clone(), *slot_index),
                 },
                 &mut cur_state,
                 context,

crates/biome_grit_patterns/src/grit_target_language.rs

@@ -74,6 +74,17 @@ macro_rules! generate_target_language {
                 }
             }
 
+            pub fn is_disregarded_snippet_field(
+                &self,
+                kind: GritTargetSyntaxKind,
+                slot_index: u32,
+                node: Option<GritTargetNode<'_>>,
+            ) -> bool {
+                match self {
+                    $(Self::$language(lang) => lang.is_disregarded_snippet_field(kind, slot_index, node)),+
+                }
+            }
+
             pub fn get_equivalence_class(
                 &self,
                 kind: GritTargetSyntaxKind,
@@ -266,6 +277,44 @@ trait GritTargetLanguageImpl {
         false
     }
 
+    /// Ordinarily, we want to match on all possible fields, including the absence of nodes within a field.
+    /// e.g., `my_function()` should not match `my_function(arg)`.
+    ///
+    /// However, some fields are trivial or not expected to be part of the snippet, and should be disregarded.
+    /// For example, in JavaScript, we want to match both `function name() {}` and `async function name() {}` with the same snippet.
+    ///
+    /// You can still match on the presence/absence of the field in the snippet by including a metavariable and checking its value.
+    /// For example, in JavaScript:
+    /// ```grit
+    /// `$async func name(args)` where $async <: .
+    /// ```
+    ///
+    /// This method allows you to specify fields that should be (conditionally) disregarded in snippets.
+    /// The actual value of the field from the snippet, if any, is passed in as the third argument.
+    ///
+    /// Note that if a field is always disregarded, you can still switch to ast_node syntax to match on these fields.
+    /// For example, in react_to_hooks we match on `arrow_function` and capture `$parenthesis` for inspection.
+    ///
+    /// ```grit
+    /// arrow_function(parameters=$props, $body, $parenthesis) where {
+    ///     $props <: contains or { `props`, `inputProps` },
+    ///     $body <: not contains `props`,
+    ///    if ($parenthesis <: .) {
+    ///         $props => `()`
+    ///     } else {
+    ///         $props => .
+    ///     }
+    /// }
+    /// ```
+    fn is_disregarded_snippet_field(
+        &self,
+        _kind: GritTargetSyntaxKind,
+        _slot_index: u32,
+        _node: Option<GritTargetNode<'_>>,
+    ) -> bool {
+        false
+    }
+
     /// Returns an optional "equivalence class" for the given syntax kind.
     ///
     /// Equivalence classes allow leaf nodes to be classified as being equal,
@@ -353,3 +402,9 @@ fn normalize_quoted_string(string: &str) -> Option<&str> {
     // Strip the quotes, regardless of type:
     (string.len() >= 2).then(|| &string[1..string.len() - 1])
 }
+
+#[derive(Debug, Clone)]
+enum DisregardedSlotCondition {
+    Always,
+    OnlyIf(&'static [&'static str]),
+}

crates/biome_grit_patterns/src/grit_target_language/js_target_language.rs

@@ -1,9 +1,16 @@
+mod constants;
+
 use super::{
-    normalize_quoted_string, GritTargetLanguageImpl, LeafEquivalenceClass, LeafNormalizer,
+    normalize_quoted_string, DisregardedSlotCondition, GritTargetLanguageImpl,
+    LeafEquivalenceClass, LeafNormalizer,
+};
+use crate::{
+    grit_target_node::{GritTargetNode, GritTargetSyntaxKind},
+    CompileError,
 };
-use crate::{grit_target_node::GritTargetSyntaxKind, CompileError};
 use biome_js_syntax::{JsLanguage, JsSyntaxKind};
 use biome_rowan::{RawSyntaxKind, SyntaxKindSet};
+use constants::DISREGARDED_SNIPPET_SLOTS;
 
 const COMMENT_KINDS: SyntaxKindSet<JsLanguage> =
     SyntaxKindSet::from_raw(RawSyntaxKind(JsSyntaxKind::COMMENT as u16)).union(
@@ -135,6 +142,30 @@ impl GritTargetLanguageImpl for JsTargetLanguage {
         })
     }
 
+    fn is_disregarded_snippet_field(
+        &self,
+        kind: GritTargetSyntaxKind,
+        slot_index: u32,
+        node: Option<GritTargetNode<'_>>,
+    ) -> bool {
+        DISREGARDED_SNIPPET_SLOTS.iter().any(
+            |(disregarded_kind, disregarded_slot_index, condition)| {
+                if GritTargetSyntaxKind::from(*disregarded_kind) != kind
+                    || *disregarded_slot_index != slot_index
+                {
+                    return false;
+                }
+
+                match condition {
+                    DisregardedSlotCondition::Always => true,
+                    DisregardedSlotCondition::OnlyIf(node_texts) => node_texts.iter().any(|text| {
+                        *text == node.as_ref().map(|node| node.text()).unwrap_or_default()
+                    }),
+                }
+            },
+        )
+    }
+
     fn get_equivalence_class(
         &self,
         kind: GritTargetSyntaxKind,

crates/biome_grit_patterns/src/pattern_compiler/snippet_compiler.rs

@@ -213,7 +213,7 @@ fn pattern_from_node(
         return Ok(metavariable);
     }
 
-    if node.slots().is_none() {
+    let Some(slots) = node.slots() else {
         let content = node.text();
         let lang = &context.compilation.lang;
         let pattern = if let Some(regex_pattern) = lang
@@ -227,22 +227,19 @@ fn pattern_from_node(
         };
 
         return Ok(pattern);
-    }
+    };
 
     let kind = node.kind();
-    let args = node
-        .slots()
-        .map(|slots| {
-            // TODO: Implement filtering for disregarded snippet fields.
-            // Implementing this will make it more convenient to match
-            // CST nodes without needing to match all the trivia in the
-            // snippet (if I understand correctly).
-            slots
-                .map(|slot| pattern_arg_from_slot(slot, context_range, range_map, context, is_rhs))
-                .collect::<Result<Vec<GritNodePatternArg>, CompileError>>()
+    let args = slots
+        .filter(|slot| {
+            !context.compilation.lang.is_disregarded_snippet_field(
+                kind,
+                slot.index(),
+                node.child_by_slot_index(slot.index()),
+            )
         })
-        .transpose()?
-        .unwrap_or_default();
+        .map(|slot| pattern_arg_from_slot(slot, context_range, range_map, context, is_rhs))
+        .collect::<Result<Vec<GritNodePatternArg>, CompileError>>()?;
 
     Ok(Pattern::AstNode(Box::new(GritNodePattern { kind, args })))
 }

crates/biome_grit_patterns/tests/specs/ts/asyncFunctionDoesntMatchNonAsync.grit

@@ -0,0 +1 @@
+`async function foo() {}`

crates/biome_grit_patterns/tests/specs/ts/asyncFunctionDoesntMatchNonAsync.snap

@@ -0,0 +1,12 @@
+---
+source: crates/biome_grit_patterns/tests/spec_tests.rs
+expression: asyncFunctionDoesntMatchNonAsync
+---
+SnapshotResult {
+    messages: [],
+    matched_ranges: [
+        "3:1-3:23",
+    ],
+    rewritten_files: [],
+    created_files: [],
+}

crates/biome_grit_patterns/tests/specs/ts/asyncFunctionDoesntMatchNonAsync.ts

@@ -0,0 +1,3 @@
+function foo() {}
+function bar(){}
+async function foo(){}

crates/biome_grit_patterns/tests/specs/ts/functionMatchesAsync.grit

@@ -0,0 +1 @@
+`function foo() {}`

crates/biome_grit_patterns/tests/specs/ts/functionMatchesAsync.snap

@@ -0,0 +1,13 @@
+---
+source: crates/biome_grit_patterns/tests/spec_tests.rs
+expression: functionMatchesAsync
+---
+SnapshotResult {
+    messages: [],
+    matched_ranges: [
+        "1:1-1:18",
+        "3:1-3:23",
+    ],
+    rewritten_files: [],
+    created_files: [],
+}

crates/biome_grit_patterns/tests/specs/ts/functionMatchesAsync.ts

@@ -0,0 +1,3 @@
+function foo() {}
+function bar(){}
+async function foo(){}

crates/biome_grit_patterns/tests/specs/ts/typeAlias.grit

@@ -0,0 +1 @@
+`type $T = $U`

crates/biome_grit_patterns/tests/specs/ts/typeAlias.snap

@@ -0,0 +1,13 @@
+---
+source: crates/biome_grit_patterns/tests/spec_tests.rs
+expression: typeAlias
+---
+SnapshotResult {
+    messages: [],
+    matched_ranges: [
+        "1:1-1:19",
+        "2:1-2:19",
+    ],
+    rewritten_files: [],
+    created_files: [],
+}

crates/biome_grit_patterns/tests/specs/ts/typeAlias.ts

@@ -0,0 +1,2 @@
+type Foo = string;
+type Bar = number;

xtask/codegen/src/ast.rs

@@ -11,6 +11,7 @@ use super::{
 use crate::generate_node_factory::generate_node_factory;
 use crate::generate_nodes_mut::generate_nodes_mut;
 use crate::generate_syntax_factory::generate_syntax_factory;
+use crate::generate_target_language_constants::generate_target_language_constants;
 use crate::js_kinds_src::{
     AstEnumSrc, AstListSeparatorConfiguration, AstListSrc, AstNodeSrc, TokenKind,
 };
@@ -76,6 +77,9 @@ pub(crate) fn generate_syntax(ast: AstSrc, mode: &Mode, language_kind: LanguageK
         .join("crates")
         .join(language_kind.factory_crate_name())
         .join("src/generated");
+    let target_language_path = project_root()
+        .join("crates/biome_grit_patterns/src/grit_target_language")
+        .join(language_kind.grit_target_language_module_name());
 
     let kind_src = language_kind.kinds();
 
@@ -103,6 +107,12 @@ pub(crate) fn generate_syntax(ast: AstSrc, mode: &Mode, language_kind: LanguageK
     let contents = generate_macros(&ast, language_kind)?;
     update(ast_macros_file.as_path(), &contents, mode)?;
 
+    if language_kind.supports_grit() {
+        let target_language_constants_file = target_language_path.join("constants.rs");
+        let contents = generate_target_language_constants(&ast, language_kind)?;
+        update(target_language_constants_file.as_path(), &contents, mode)?;
+    }
+
     Ok(())
 }
 

xtask/codegen/src/generate_target_language_constants.rs

@@ -0,0 +1,49 @@
+use crate::{
+    js_kinds_src::{AstSrc, Field},
+    language_kind::LanguageKind,
+};
+use biome_string_case::Case;
+use xtask::Result;
+
+pub fn generate_target_language_constants(
+    ast: &AstSrc,
+    _language_kind: LanguageKind,
+) -> Result<String> {
+    let disregarded_slots: Vec<String> = ast
+        .nodes
+        .iter()
+        .flat_map(|node| {
+            let node_kind = Case::Constant.convert(node.name.as_str());
+            node.fields
+                .iter()
+                .enumerate()
+                .filter_map(|(index, field)| match field {
+                    Field::Token { name, optional, .. } => Some((index, name, optional)),
+                    Field::Node { .. } => None,
+                })
+                // TODO: We might want to move this to `js_kinds_src.rs` when we
+                //       start supporting other languages with Grit.
+                .filter_map(|(index, name, optional)| match (name.as_str(), optional) {
+                    ("async", true) => Some(format!("({node_kind}, {index}, OnlyIf(&[\"\"])),")),
+                    (";", true) => Some(format!("({node_kind}, {index}, Always),")),
+                    _ => None,
+                })
+                .collect::<Vec<_>>()
+        })
+        .collect();
+    let disregarded_slots = disregarded_slots.join("\n    ");
+
+    let result = format!(
+        "use crate::grit_target_language::DisregardedSlotCondition::{{self, *}};
+use biome_js_syntax::JsSyntaxKind::{{self, *}};
+
+pub(crate) const DISREGARDED_SNIPPET_SLOTS: &[(JsSyntaxKind, u32, DisregardedSlotCondition)] = &[
+    {disregarded_slots}
+];
+"
+    );
+
+    let pretty = xtask::reformat(result)?;
+
+    Ok(pretty)
+}

xtask/codegen/src/language_kind.rs

@@ -131,6 +131,10 @@ impl LanguageKind {
         format!("biome_{self}_factory")
     }
 
+    pub fn grit_target_language_module_name(&self) -> String {
+        format!("{self}_target_language")
+    }
+
     pub fn kinds(&self) -> KindsSrc {
         match self {
             LanguageKind::Js => JS_KINDS_SRC,
@@ -156,4 +160,8 @@ impl LanguageKind {
             LanguageKind::Markdown => include_str!("../markdown.ungram"),
         }
     }
+
+    pub fn supports_grit(&self) -> bool {
+        matches!(self, Self::Js)
+    }
 }