yeast: Add location heuristic for synthetic nodes with no location set

For such nodes, we assign them the zero-length location at the beginning
of the matched input node.

Author: tausbn

shared/yeast/doc/yeast.md

@@ -238,9 +238,10 @@ yeast::trees!(ctx,
 ### Source locations
 
 Captured nodes keep the locations assigned by their own translations. New
-nodes in an output template derive their locations from their children. After
-the transform completes, the matched input node's range is added only to
-locally-created nodes returned as rule results:
+nodes in an output template derive their locations from their children. A
+source-less nested node receives an empty location at the start of the matched
+input node. After the transform completes, the full matched range is added only
+to locally-created nodes returned as rule results:
 
 ```rust
 rule!(
@@ -251,10 +252,12 @@ rule!(
 ```
 
 Here `inner` derives its range from `child`, while the returned `outer` node
-also includes the full `wrapper` range. This lets replacement roots include
-elided keywords or delimiters without assigning the same broad range to every
-synthetic descendant. A transform that simply returns a translated capture
-does not widen that capture to the wrapper's range.
+also includes the full `wrapper` range. A nested node with no located children
+would instead receive an empty range at the start of `wrapper`. This lets
+replacement roots include elided keywords or delimiters without assigning the
+same broad range to every synthetic descendant. A transform that simply
+returns a translated capture does not widen that capture to the wrapper's
+range.
 
 When the desired range belongs to another node, `tree_at!` assigns that range
 to the template's root. Nested nodes still derive their own locations normally:

shared/yeast/src/build.rs

@@ -40,7 +40,8 @@ pub struct BuildCtx<'a, C: 'a = ()> {
     ///
     /// The `rule!` macro applies this range to locally-created result roots
     /// after the transform completes. Nested synthetic nodes derive their
-    /// ranges from their children unless `source_range` is explicitly set.
+    /// ranges from their children, falling back to an empty range at this
+    /// range's start.
     pub matched_source_range: Option<Range>,
     /// User-supplied context, accessible directly via `ctx.field` (via Deref).
     pub user_ctx: &'a mut C,
@@ -128,6 +129,16 @@ impl<'a, C> BuildCtx<'a, C> {
         let id = self
             .ast
             .create_node_with_range(kind, content, fields, is_named, source_range);
+        if self
+            .ast
+            .get_node(id)
+            .is_some_and(|node| node.source_range().is_none())
+        {
+            if let Some(source_range) = self.matched_source_range {
+                self.ast
+                    .extend_source_range(id, source_range.empty_at_start());
+            }
+        }
         if !has_explicit_source_range {
             self.created_nodes.insert(id);
         }
@@ -153,6 +164,10 @@ impl<'a, C> BuildCtx<'a, C> {
         source_range: Option<Range>,
     ) -> Id {
         let has_explicit_source_range = source_range.is_some();
+        let source_range = source_range.or_else(|| {
+            self.matched_source_range
+                .map(|range| range.empty_at_start())
+        });
         let id = self
             .ast
             .create_named_token_with_range(kind, content, source_range);

shared/yeast/tests/test.rs

@@ -1709,6 +1709,34 @@ fn test_nested_synthetic_node_uses_child_location() {
     assert_eq!(arguments.byte_range(), 0..3);
 }
 
+/// A nested node with no explicit range or located children gets an empty
+/// location at the start of the rule's matched node.
+#[test]
+fn test_source_less_nested_node_uses_empty_match_start() {
+    let rule: Rule = rule!(
+        (call
+            method: (identifier) @name
+            receiver: (identifier) @recv
+        )
+        =>
+        (call
+            method: {name}
+            receiver: {recv}
+            arguments: (argument_list)
+        )
+    );
+
+    let ast = run_and_ast("foo.bar()", vec![rule]);
+    let arguments = ast
+        .reachable_node_ids()
+        .into_iter()
+        .filter_map(|id| ast.get_node(id))
+        .find(|node| node.kind_name() == "argument_list")
+        .expect("argument list exists");
+    let range = arguments.source_range().unwrap();
+    assert_eq!(range.start_byte..range.end_byte, 0..0);
+}
+
 /// An explicit empty range at byte zero is a real location, not the sentinel
 /// for an absent location, and therefore contributes to parent ranges.
 #[test]

unified/extractor/tests/corpus/swift/collections/empty-array-literal-with-type.output

@@ -46,6 +46,6 @@ top_level source="⟨body⟩"
           pattern: identifier "xs" source="xs"
           type:
             generic_type_expr source="[⟨type_argument⟩]"
-              base: identifier "Array" source=<no location>
+              base: identifier "Array" source=""
               type_argument: identifier "Int" source="Int"
           value: array_literal "[]" source="[]"

unified/extractor/tests/corpus/swift/control-flow/guard-let.output

@@ -42,18 +42,18 @@ top_level source="⟨body⟩"
       stmt:
         guard_if_stmt source="guard ⟨condition⟩ else ⟨else⟩"
           condition:
-            pattern_guard_expr source="let ⟨pattern⟩ = ⟨value⟩"
+            pattern_guard_expr source="⟨pattern⟩ = ⟨value⟩"
               pattern:
                 call_expr source="⟨argument⟩"
                   callee:
-                    member_access_expr source=<no location>
-                      base: identifier "Optional" source=<no location>
-                      member_name_node: identifier "some" source=<no location>
+                    member_access_expr source=""
+                      base: identifier "Optional" source=""
+                      member_name_node: identifier "some" source=""
                   argument:
                     argument source="⟨value⟩"
                       value:
-                        expr_pattern source="⟨expr⟩"
-                          modifier: modifier "let" source=<no location>
+                        expr_pattern source="let ⟨expr⟩"
+                          modifier: modifier "let" source=""
                           expr: identifier "value" source="value"
               value: identifier "optional" source="optional"
           else:

unified/extractor/tests/corpus/swift/control-flow/if-let-optional-binding.output

@@ -55,18 +55,18 @@ top_level source="⟨body⟩"
       stmt:
         if_expr source="if ⟨condition⟩ ⟨then⟩"
           condition:
-            pattern_guard_expr source="let ⟨pattern⟩ = ⟨value⟩"
+            pattern_guard_expr source="⟨pattern⟩ = ⟨value⟩"
               pattern:
                 call_expr source="⟨argument⟩"
                   callee:
-                    member_access_expr source=<no location>
-                      base: identifier "Optional" source=<no location>
-                      member_name_node: identifier "some" source=<no location>
+                    member_access_expr source=""
+                      base: identifier "Optional" source=""
+                      member_name_node: identifier "some" source=""
                   argument:
                     argument source="⟨value⟩"
                       value:
-                        expr_pattern source="⟨expr⟩"
-                          modifier: modifier "let" source=<no location>
+                        expr_pattern source="let ⟨expr⟩"
+                          modifier: modifier "let" source=""
                           expr: identifier "value" source="value"
               value: identifier "optional" source="optional"
           then:

unified/extractor/tests/corpus/swift/control-flow/switch-expression-pattern.output

@@ -344,29 +344,29 @@ top_level source="⟨body⟩"
                     map_literal "[key: value]" source="[key: value]"
                     call_expr source="⟨argument⟩?"
                       callee:
-                        member_access_expr source=<no location>
-                          base: identifier "Optional" source=<no location>
-                          member_name_node: identifier "some" source=<no location>
+                        member_access_expr source=""
+                          base: identifier "Optional" source=""
+                          member_name_node: identifier "some" source=""
                       argument:
                         argument source="⟨value⟩"
                           value: identifier "optional" source="optional"
                     unary_expr source="try ⟨operand⟩"
                       operand: identifier "value" source="value"
-                      operator: prefix_operator "try" source=<no location>
+                      operator: prefix_operator "try" source=""
                     unary_expr source="⟨operand⟩!"
                       operand: identifier "value" source="value"
-                      operator: postfix_operator "!" source=<no location>
+                      operator: postfix_operator "!" source=""
                     type_cast_expr source="⟨expr⟩ as ⟨type⟩"
                       expr: identifier "value" source="value"
-                      operator: infix_operator "as" source=<no location>
+                      operator: infix_operator "as" source=""
                       type: identifier "Target" source="Target"
                     type_test_expr source="⟨expr⟩ is ⟨type⟩"
                       expr: identifier "value" source="value"
-                      operator: infix_operator "is" source=<no location>
+                      operator: infix_operator "is" source=""
                       type: identifier "Target" source="Target"
                     unary_expr source="await ⟨operand⟩"
                       operand: identifier "value" source="value"
-                      operator: prefix_operator "await" source=<no location>
+                      operator: prefix_operator "await" source=""
               body:
                 block source="⟨stmt⟩"
                   stmt:

unified/extractor/tests/corpus/swift/desugar/import-with-deeply-nested-path-three-parts.output

@@ -27,7 +27,7 @@ top_level source="⟨body⟩"
   body:
     block source="⟨stmt⟩"
       stmt:
-        import_declaration source="import ⟨imported_expr⟩⟨pattern⟩"
+        import_declaration source="⟨pattern⟩⟨imported_expr⟩"
           imported_expr:
             member_access_expr source="⟨base⟩.⟨member_name_node⟩"
               base:
@@ -36,6 +36,6 @@ top_level source="⟨body⟩"
                   member_name_node: identifier "Networking" source="Networking"
               member_name_node: identifier "URLSession" source="URLSession"
           pattern:
-            named_pattern source="⟨name_node⟩"
+            named_pattern source="import Foundation.Networking.⟨name_node⟩"
               name_node: identifier "URLSession" source="URLSession"
-              sub_pattern: bulk_importing_pattern source=<no location>
+              sub_pattern: bulk_importing_pattern source=""

unified/extractor/tests/corpus/swift/desugar/import-with-dotted-path-two-parts.output

@@ -24,12 +24,12 @@ top_level source="⟨body⟩"
   body:
     block source="⟨stmt⟩"
       stmt:
-        import_declaration source="import ⟨imported_expr⟩⟨pattern⟩"
+        import_declaration source="⟨pattern⟩⟨imported_expr⟩"
           imported_expr:
             member_access_expr source="⟨base⟩.⟨member_name_node⟩"
               base: identifier "Foundation" source="Foundation"
               member_name_node: identifier "Networking" source="Networking"
           pattern:
-            named_pattern source="⟨name_node⟩"
+            named_pattern source="import Foundation.⟨name_node⟩"
               name_node: identifier "Networking" source="Networking"
-              sub_pattern: bulk_importing_pattern source=<no location>
+              sub_pattern: bulk_importing_pattern source=""

unified/extractor/tests/corpus/swift/desugar/simple-import-with-single-name.output

@@ -21,9 +21,9 @@ top_level source="⟨body⟩"
   body:
     block source="⟨stmt⟩"
       stmt:
-        import_declaration source="import ⟨imported_expr⟩⟨pattern⟩"
+        import_declaration source="⟨pattern⟩⟨imported_expr⟩"
           imported_expr: identifier "Foundation" source="Foundation"
           pattern:
-            named_pattern source="⟨name_node⟩"
+            named_pattern source="import ⟨name_node⟩"
               name_node: identifier "Foundation" source="Foundation"
-              sub_pattern: bulk_importing_pattern source=<no location>
+              sub_pattern: bulk_importing_pattern source=""