fix(resolve): v1.6.1 receiver evidence, store and object-literal bindings, and extraction recall

crates/codegraph-extract/src/lang/c.rs

@@ -57,6 +57,9 @@ impl LanguageSpec for CSpec {
     fn variable_types(&self) -> &'static [&'static str] {
         &["declaration"]
     }
+    fn resolve_name(&self, node: Node<'_>, source: &str) -> Option<String> {
+        crate::lang::cpp::recover_single_arg_macro_defined_name(node, source)
+    }
     fn name_field(&self) -> &'static str {
         "declarator"
     }

crates/codegraph-extract/src/lang/cpp.rs

@@ -70,7 +70,9 @@ impl LanguageSpec for CppSpec {
         "type"
     }
     fn resolve_name(&self, node: Node<'_>, source: &str) -> Option<String> {
-        if let Some(name) = recover_cpp_macro_defined_name(node, source) {
+        if let Some(name) = recover_single_arg_macro_defined_name(node, source)
+            .or_else(|| recover_cpp_macro_defined_name(node, source))
+        {
             return Some(name);
         }
         let qid = declarator_qualified_id(child_by_field(node, "declarator")?)?;
@@ -921,6 +923,108 @@ fn mask_quoted_literal(bytes: &[u8], mask: &mut [bool], start: usize, quote: u8)
     i
 }
 
+/// Recover the function a single-argument macro defines —
+/// `#define NATIVE_FN(name) int name(void)` then `NATIVE_FN(get_version) { … }`
+/// (upstream #1373). Without its definition the shape is ambiguous, so a
+/// preceding local `#define` whose replacement uses its sole parameter as the
+/// function declarator is required; registration, token-pasting and `typedef`
+/// replacements, an `#undef` or redefinition in between, and any conditional
+/// block leave the name alone. C parses the macro as the return type with a
+/// `(name)` declarator; C++ as an implicit-return-type function.
+pub(crate) fn recover_single_arg_macro_defined_name(
+    node: Node<'_>,
+    source: &str,
+) -> Option<String> {
+    if node.kind() != "function_definition" {
+        return None;
+    }
+    let declarator = child_by_field(node, "declarator")?;
+    let (macro_node, argument) = match declarator.kind() {
+        "parenthesized_declarator" if declarator.named_child_count() == 1 => {
+            let macro_node = child_by_field(node, "type")?;
+            let argument = declarator.named_child(0)?;
+            if macro_node.kind() != "type_identifier" || argument.kind() != "identifier" {
+                return None;
+            }
+            (macro_node, argument)
+        }
+        "function_declarator" if child_by_field(node, "type").is_none() => {
+            let macro_node = child_by_field(declarator, "declarator")?;
+            let params = child_by_field(declarator, "parameters")?;
+            let param = params.named_child(0)?;
+            if macro_node.kind() != "identifier"
+                || params.named_child_count() != 1
+                || param.kind() != "parameter_declaration"
+                || param.named_child_count() != 1
+            {
+                return None;
+            }
+            let argument = param.named_child(0)?;
+            if argument.kind() != "type_identifier" {
+                return None;
+            }
+            (macro_node, argument)
+        }
+        _ => return None,
+    };
+    let macro_name = node_text(macro_node, source);
+    static REPLACEMENT: OnceLock<Regex> = OnceLock::new();
+    let replacement_pattern = REPLACEMENT.get_or_init(|| {
+        Regex::new(
+            r"^(?:[A-Za-z_][A-Za-z0-9_:]*\s+)+[*&\s]*([A-Za-z_][A-Za-z0-9_]*)\s*\([^(){};#]*\)\s*$",
+        )
+        .expect("single-argument macro replacement pattern")
+    });
+    static TYPEDEF: OnceLock<Regex> = OnceLock::new();
+    let typedef = TYPEDEF.get_or_init(|| Regex::new(r"(?-u:\b)typedef(?-u:\b)").expect("typedef"));
+    let mut scope = Some(node);
+    while let Some(current) = scope {
+        if current.kind() == "preproc_else" || current.kind().starts_with("preproc_elif") {
+            return None;
+        }
+        let mut previous = current.prev_named_sibling();
+        while let Some(prev) = previous {
+            previous = prev.prev_named_sibling();
+            if prev.kind().starts_with("preproc_if") {
+                return None;
+            }
+            if prev.kind() == "preproc_call"
+                && child_by_field(prev, "directive")
+                    .is_some_and(|directive| node_text(directive, source) == "#undef")
+                && child_by_field(prev, "argument")
+                    .is_some_and(|argument| node_text(argument, source).trim() == macro_name)
+            {
+                return None;
+            }
+            if !matches!(prev.kind(), "preproc_function_def" | "preproc_def")
+                || child_by_field(prev, "name").map(|name| node_text(name, source))
+                    != Some(macro_name.clone())
+            {
+                continue;
+            }
+            let params = child_by_field(prev, "parameters")?;
+            let param = params.named_child(0)?;
+            let value = child_by_field(prev, "value")?;
+            if params.named_child_count() != 1 || param.kind() != "identifier" {
+                return None;
+            }
+            let replacement = node_text(value, source)
+                .replace("\\\r\n", " ")
+                .replace("\\\n", " ");
+            let replacement = replacement.trim();
+            let declared = replacement_pattern
+                .captures(replacement)
+                .and_then(|captures| captures.get(1))?;
+            if typedef.is_match(replacement) || declared.as_str() != node_text(param, source) {
+                return None;
+            }
+            return Some(node_text(argument, source));
+        }
+        scope = current.parent();
+    }
+    None
+}
+
 /// Recover the real function name from the macro-definition idiom
 /// `MACRO_NAME(real_name, typed args…) { body }` (flash-attention's
 /// `DEFINE_FLASH_FORWARD_KERNEL(flash_fwd_kernel, bool Is_dropout, …) { … }`):

crates/codegraph-extract/src/lang/javascript.rs

@@ -43,6 +43,16 @@ impl LanguageSpec for JavaScriptSpec {
         crate::lang::typescript::class_field_is_callable(node, "field_definition")
     }
 
+    fn resolve_name(&self, node: Node<'_>, source: &str) -> Option<String> {
+        // JS `field_definition` names its key the `property` field (TS's
+        // `public_field_definition` uses `name`); without this every JS class
+        // field, handler fields included, is `<anonymous>` (upstream #808).
+        if node.kind() == "field_definition" {
+            return child_by_field(node, "property").map(|property| node_text(property, source));
+        }
+        None
+    }
+
     fn interface_types(&self) -> &'static [&'static str] {
         &[]
     }

crates/codegraph-extract/src/lang/jsx.rs

@@ -33,6 +33,9 @@ impl LanguageSpec for JsxSpec {
     fn method_types(&self) -> &'static [&'static str] {
         JAVASCRIPT_SPEC.method_types()
     }
+    fn class_member_is_method(&self, node: Node<'_>, source: &str) -> bool {
+        JAVASCRIPT_SPEC.class_member_is_method(node, source)
+    }
     fn interface_types(&self) -> &'static [&'static str] {
         &[]
     }
@@ -60,6 +63,9 @@ impl LanguageSpec for JsxSpec {
     fn name_field(&self) -> &'static str {
         JAVASCRIPT_SPEC.name_field()
     }
+    fn resolve_name(&self, node: Node<'_>, source: &str) -> Option<String> {
+        JAVASCRIPT_SPEC.resolve_name(node, source)
+    }
     fn body_field(&self) -> &'static str {
         JAVASCRIPT_SPEC.body_field()
     }

crates/codegraph-extract/src/lang/python.rs

@@ -73,6 +73,10 @@ impl LanguageSpec for PythonSpec {
         Some(signature)
     }
 
+    fn body_docstring(&self, node: Node<'_>, source: &str) -> Option<String> {
+        python_body_docstring(node, source)
+    }
+
     fn is_async(&self, node: Node<'_>) -> bool {
         node.prev_sibling()
             .is_some_and(|prev| prev.kind() == "async")
@@ -96,3 +100,135 @@ impl LanguageSpec for PythonSpec {
         })
     }
 }
+
+/// Python states intent in a docstring — a bare string literal as the first
+/// statement of a module, class or function body (upstream #1905). Reads the
+/// grammar's `string_content`, so `r`/`u` prefixes and both triple-quote forms
+/// work; bytes and f-strings are not docstrings, and a tuple is not a string.
+fn python_body_docstring(node: Node<'_>, source: &str) -> Option<String> {
+    let body = if node.kind() == "module" {
+        node
+    } else {
+        child_by_field(node, "body")?
+    };
+    let first = body
+        .named_children(&mut body.walk())
+        .find(|child| child.kind() != "comment")?;
+    if first.kind() != "expression_statement"
+        || first.named_child_count() != 1
+        || first
+            .children(&mut first.walk())
+            .any(|child| child.kind() == ",")
+    {
+        return None;
+    }
+    let mut literal = first.named_child(0)?;
+    while literal.kind() == "parenthesized_expression" {
+        literal = literal
+            .named_children(&mut literal.walk())
+            .find(|child| child.kind() != "comment")?;
+    }
+    let strings = if literal.kind() == "concatenated_string" {
+        literal
+            .named_children(&mut literal.walk())
+            .filter(|child| child.kind() != "comment")
+            .collect::<Vec<_>>()
+    } else {
+        vec![literal]
+    };
+    let mut raw = String::new();
+    for string in strings {
+        if string.kind() != "string" {
+            return None;
+        }
+        let parts = string
+            .named_children(&mut string.walk())
+            .collect::<Vec<_>>();
+        let start = parts.iter().find(|part| part.kind() == "string_start")?;
+        if node_text(*start, source)
+            .chars()
+            .any(|ch| matches!(ch, 'b' | 'B' | 'f' | 'F'))
+            || !parts.iter().any(|part| part.kind() == "string_end")
+        {
+            return None;
+        }
+        if let Some(content) = parts.iter().find(|part| part.kind() == "string_content") {
+            raw.push_str(&node_text(*content, source));
+        }
+    }
+    let docstring = dedent_docstring(&raw);
+    (!docstring.is_empty()).then_some(docstring)
+}
+
+/// PEP 257 cleaning: tabs expand to 8-column stops, every line after the first
+/// loses the common indentation of the non-blank ones (the first line starts
+/// right after the quotes), and blank edges are dropped.
+fn dedent_docstring(raw: &str) -> String {
+    let lines = raw
+        .replace("\r\n", "\n")
+        .replace('\r', "\n")
+        .split('\n')
+        .map(|line| {
+            let mut column = 0usize;
+            let mut expanded = String::with_capacity(line.len());
+            for ch in line.chars() {
+                if ch == '\t' {
+                    let width = 8 - column % 8;
+                    expanded.extend(std::iter::repeat_n(' ', width));
+                    column += width;
+                } else {
+                    expanded.push(ch);
+                    column += 1;
+                }
+            }
+            expanded
+        })
+        .collect::<Vec<_>>();
+    let leading = |line: &str| line.chars().take_while(|ch| ch.is_whitespace()).count();
+    let indent = lines
+        .iter()
+        .skip(1)
+        .filter(|line| !line.trim().is_empty())
+        .map(|line| leading(line))
+        .min()
+        .unwrap_or(0);
+    let mut out = Vec::with_capacity(lines.len());
+    out.push(
+        lines
+            .first()
+            .map(|line| line.trim().to_string())
+            .unwrap_or_default(),
+    );
+    for line in lines.iter().skip(1) {
+        out.push(
+            line.chars()
+                .skip(indent)
+                .collect::<String>()
+                .trim_end()
+                .to_string(),
+        );
+    }
+    while out.first().is_some_and(|line| line.trim().is_empty()) {
+        out.remove(0);
+    }
+    while out.last().is_some_and(|line| line.trim().is_empty()) {
+        out.pop();
+    }
+    out.join("\n")
+}
+
+#[cfg(test)]
+mod docstring_tests {
+    use super::dedent_docstring;
+
+    #[test]
+    fn dedent_follows_pep_257() {
+        assert_eq!(
+            dedent_docstring("Summary line.\n\n    Details here.\n      indented\n    "),
+            "Summary line.\n\nDetails here.\n  indented"
+        );
+        assert_eq!(dedent_docstring("\n    Only body.\n    "), "Only body.");
+        assert_eq!(dedent_docstring("a\r\n\tb"), "a\nb");
+        assert_eq!(dedent_docstring("   "), "");
+    }
+}

crates/codegraph-extract/src/lang/tsx.rs

@@ -31,6 +31,12 @@ impl LanguageSpec for TsxSpec {
     fn method_types(&self) -> &'static [&'static str] {
         TYPESCRIPT_SPEC.method_types()
     }
+    fn property_types(&self) -> &'static [&'static str] {
+        TYPESCRIPT_SPEC.property_types()
+    }
+    fn class_member_is_method(&self, node: Node<'_>, source: &str) -> bool {
+        TYPESCRIPT_SPEC.class_member_is_method(node, source)
+    }
     fn interface_types(&self) -> &'static [&'static str] {
         TYPESCRIPT_SPEC.interface_types()
     }

crates/codegraph-extract/src/spec.rs

@@ -114,6 +114,13 @@ pub trait LanguageSpec: Sync {
         None
     }
 
+    /// Prose that lives INSIDE the node's body rather than in a preceding
+    /// comment — Python's bare first-statement string (upstream
+    /// `getBodyDocstring`, #1905). Cleaned text, or `None`.
+    fn body_docstring(&self, _node: Node<'_>, _source: &str) -> Option<String> {
+        None
+    }
+
     fn get_return_type(&self, _node: Node<'_>, _source: &str) -> Option<String> {
         None
     }