Kotlin: receiver-typed member calls, and park them for cross-repo merges

graphify/cross_repo_calls.py

@@ -42,6 +42,9 @@
     "cpp": frozenset({".cpp", ".cc", ".cxx", ".hpp", ".hh", ".hxx", ".h", ".cu", ".cuh"}),
     "csharp": frozenset({".cs"}),
     "java": frozenset({".java"}),
+    # A Kotlin call may be answered by any JVM-family declaration: the classpath is one
+    # namespace, and a Kotlin module over a Java library is the common Android shape.
+    "kotlin": frozenset({".kt", ".kts", ".java", ".scala", ".groovy", ".gradle"}),
     "swift": frozenset({".swift"}),
 }
 

graphify/extract.py

@@ -146,7 +146,7 @@
 
 from graphify.symbol_resolution import resolve_bash_source_edges  # noqa: E402
 
-from graphify.extractors.engine import REFERENCE_CONTEXTS, _CSHARP_TYPE_PARAMETER_SCOPE_DECLARATIONS, _C_PRIMITIVE_TYPE_NODES, _JAVA_BUILTIN_TYPES, _JAVA_TYPE_PARAMETER_SCOPE_DECLARATIONS, _JS_FUNCTION_VALUE_TYPES, _JS_SCOPE_BOUNDARY, _PYTHON_ANNOTATION_NOISE, _PYTHON_TYPE_CONTAINERS, _RUBY_CLASS_FACTORIES, _c_collect_type_refs, _cpp_collect_type_refs, _cpp_declarator_name, _cpp_local_var_types, _csharp_attribute_names, _csharp_classify_base, _csharp_collect_type_refs, _csharp_extra_walk, _csharp_namespace_id, _csharp_namespace_name, _csharp_pre_scan_interfaces, _csharp_type_parameters_in_scope, _dynamic_import_js, _extract_generic, _find_body, _find_require_call, _get_cpp_func_name, _java_annotation_names, _java_collect_type_refs, _java_extra_walk, _java_type_parameters_in_scope, _js_collect_pattern_idents, _js_dispatch_value_idents, _js_extra_walk, _js_local_bound_names, _js_member_assignment_target, _js_module_bound_names, _kotlin_collect_type_refs, _kotlin_function_return_type_node, _kotlin_property_type_node, _kotlin_user_type_name, _php_collect_type_refs, _php_method_return_type_node, _php_name_text, _python_collect_assignment_targets, _python_collect_param_refs, _python_collect_type_refs, _python_local_bound_names, _python_module_bound_names, _python_param_names, _read_csharp_type_name, _require_imports_js, _ruby_const_last_name, _ruby_extra_walk, _ruby_local_class_bindings, _ruby_new_class_name, _scala_collect_type_refs, _semantic_reference_edge, _source_location, _swift_classify_base, _swift_collect_type_refs, _swift_constructor_type, _swift_declaration_keyword, _swift_extra_walk, _swift_local_var_types, _swift_pre_scan, _swift_property_name, _swift_property_type_node, _swift_receiver_name, _swift_user_type_name, _ts_decorator_name, _ts_descendant_decorators, _ts_emit_decorator_edges, _ts_extra_walk, _ts_method_name, _ts_receiver_type_table  # noqa: E402,F401
+from graphify.extractors.engine import REFERENCE_CONTEXTS, _CSHARP_TYPE_PARAMETER_SCOPE_DECLARATIONS, _C_PRIMITIVE_TYPE_NODES, _JAVA_BUILTIN_TYPES, _JAVA_TYPE_PARAMETER_SCOPE_DECLARATIONS, _JS_FUNCTION_VALUE_TYPES, _JS_SCOPE_BOUNDARY, _KOTLIN_BUILTIN_TYPES, _PYTHON_ANNOTATION_NOISE, _PYTHON_TYPE_CONTAINERS, _RUBY_CLASS_FACTORIES, _c_collect_type_refs, _cpp_collect_type_refs, _cpp_declarator_name, _cpp_local_var_types, _csharp_attribute_names, _csharp_classify_base, _csharp_collect_type_refs, _csharp_extra_walk, _csharp_namespace_id, _csharp_namespace_name, _csharp_pre_scan_interfaces, _csharp_type_parameters_in_scope, _dynamic_import_js, _extract_generic, _find_body, _find_require_call, _get_cpp_func_name, _java_annotation_names, _java_collect_type_refs, _java_extra_walk, _java_type_parameters_in_scope, _js_collect_pattern_idents, _js_dispatch_value_idents, _js_extra_walk, _js_local_bound_names, _js_member_assignment_target, _js_module_bound_names, _kotlin_collect_type_refs, _kotlin_function_return_type_node, _kotlin_property_type_node, _kotlin_user_type_name, _php_collect_type_refs, _php_method_return_type_node, _php_name_text, _python_collect_assignment_targets, _python_collect_param_refs, _python_collect_type_refs, _python_local_bound_names, _python_module_bound_names, _python_param_names, _read_csharp_type_name, _require_imports_js, _ruby_const_last_name, _ruby_extra_walk, _ruby_local_class_bindings, _ruby_new_class_name, _scala_collect_type_refs, _semantic_reference_edge, _source_location, _swift_classify_base, _swift_collect_type_refs, _swift_constructor_type, _swift_declaration_keyword, _swift_extra_walk, _swift_local_var_types, _swift_pre_scan, _swift_property_name, _swift_property_type_node, _swift_receiver_name, _swift_user_type_name, _ts_decorator_name, _ts_descendant_decorators, _ts_emit_decorator_edges, _ts_extra_walk, _ts_method_name, _ts_receiver_type_table  # noqa: E402,F401
 
 from graphify.extractors.pascal import _PAS_BEGIN_END_TOKEN_RE, _PAS_CALL_RE, _PAS_END_SEMI_RE, _PAS_IMPL_HEADER_RE, _PAS_KEYWORDS, _PAS_METHOD_DECL_RE, _PAS_MODULE_RE, _PAS_TOKEN_RE, _PAS_TYPE_HEADER_RE, _PAS_USES_RE, _extract_pascal_regex, _pascal_find_body, _pascal_split_bases, _pascal_split_sections, _pascal_split_uses, _pascal_strip_comments, extract_pascal  # noqa: E402,F401
 
@@ -4414,6 +4414,110 @@ def _resolve_csharp_qualified_calls(
         })
 
 
+def _resolve_kotlin_member_calls(
+    per_file: list[dict],
+    all_nodes: list[dict],
+    all_edges: list[dict],
+) -> None:
+    """Resolve Kotlin member calls (``greeter.greet()``) through the receiver's type.
+
+    The shared cross-file pass skips member calls, so a call on a typed receiver whose
+    method is declared in another file resolved to nothing at all. The per-file
+    ``kotlin_type_table`` names the declared type of every property, primary-constructor
+    parameter, function parameter and local binding; this pass looks the receiver up
+    there, takes the single class/object declaring that type, and emits the ``calls``
+    edge to its member — EXTRACTED when the receiver names the type in source
+    (``Registry.register()``), INFERRED when the type came from the table.
+
+    A receiver typed to a class this corpus declares nowhere is parked on the caller for
+    a merged graph to finish (#3152).
+    """
+    raw = [
+        rc
+        for result in per_file
+        for rc in result.get("raw_calls", [])
+        if rc.get("lang") == "kotlin" and rc.get("is_member_call")
+        and rc.get("receiver") and rc.get("callee") and rc.get("caller_nid")
+    ]
+    if not raw:
+        return
+    type_table_by_file: dict[str, dict[str, str]] = {}
+    for result in per_file:
+        tt = result.get("kotlin_type_table")
+        if tt and tt.get("path"):
+            type_table_by_file[tt["path"]] = tt.get("table", {})
+
+    def _key(label: str) -> str:
+        return re.sub(r"[^a-zA-Z0-9]+", "", str(label)).lower()
+
+    # A genuine declaration is the target of a `contains` edge from its file node; a bare
+    # type reference mints a same-label stub that would otherwise make a real name ambiguous.
+    # The whole JVM family counts, `.java` included: one classpath makes interop a real
+    # answer, while a same-named class outside the family is a collision.
+    contained = {e.get("target") for e in all_edges if e.get("relation") == "contains"}
+    type_def_nids: dict[str, list[str]] = {}
+    node_by_id: dict[str, dict] = {}
+    for n in all_nodes:
+        node_by_id[n.get("id")] = n
+        if (_lang_family(n.get("source_file")) == "jvm"
+                and n.get("id") in contained and _is_type_like_definition(n)):
+            type_def_nids.setdefault(_key(n.get("label", "")), []).append(n["id"])
+
+    method_index: dict[tuple[str, str], str] = {}
+    for e in all_edges:
+        if e.get("relation") != "method":
+            continue
+        tnode = node_by_id.get(e.get("target"))
+        if tnode is not None:
+            method_index[(e.get("source"), _key(tnode.get("label", "")))] = e["target"]
+
+    existing_pairs = {(e.get("source"), e.get("target")) for e in all_edges}
+    for rc in raw:
+        receiver, callee, caller = rc["receiver"], rc["callee"], rc["caller_nid"]
+        # A capitalized receiver is the type itself — a companion/static call or an
+        # `object` singleton. Kotlin imports the class name into scope rather than a
+        # module alias, so unlike TS/JS this spelling is not also the namespace idiom.
+        if receiver[:1].isupper():
+            type_name = receiver
+            type_qualified = True
+        else:
+            type_name = type_table_by_file.get(rc.get("source_file", ""), {}).get(receiver)
+            type_qualified = False
+        if not type_name:
+            continue
+        if (type_name in _LANGUAGE_BUILTIN_GLOBALS or type_name in _KOTLIN_BUILTIN_TYPES
+                or type_name in _JAVA_BUILTIN_TYPES):
+            continue
+        type_defs = type_def_nids.get(_key(type_name), [])
+        if not type_defs:
+            # Declared nowhere here — usually "in a repo this build does not contain",
+            # so park it for the merge (#3152). The extractor's `lang` tag already says
+            # who is asking, so no suffix sniff is needed.
+            _park_unresolved_member_call(
+                node_by_id.get(caller), callee, type_name, "kotlin", rc,
+            )
+            continue
+        if len(type_defs) != 1:  # ambiguous -> bail (god-node guard)
+            continue
+        target = method_index.get((type_defs[0], _key(callee)))
+        if not target or target == caller or (caller, target) in existing_pairs:
+            continue
+        existing_pairs.add((caller, target))
+        all_edges.append({
+            "source": caller,
+            "target": target,
+            "relation": "calls",
+            "context": "call",
+            "confidence": "EXTRACTED" if type_qualified else "INFERRED",
+            # The rubric's discrete INFERRED scale (references/extraction-spec.md):
+            # a single-definition type-table hit is the high-confidence rung.
+            "confidence_score": 1.0 if type_qualified else 0.85,
+            "source_file": rc.get("source_file", ""),
+            "source_location": rc.get("source_location"),
+            "weight": 1.0,
+        })
+
+
 def _resolve_kotlin_qualified_calls(
     per_file: list[dict],
     all_nodes: list[dict],
@@ -4591,6 +4695,12 @@ def _resolve_kotlin_qualified_calls(
         resolve_pascal_inherited_calls,
     )
 )
+# Kotlin receiver-typed member-call resolution: `greeter.greet()` where the method is
+# declared in another file. The shared pass skips member calls, so these had no edge.
+register_language_resolver(
+    LanguageResolver("kotlin_member_calls", frozenset({".kt", ".kts"}),
+                     _resolve_kotlin_member_calls)
+)
 # Kotlin fully-qualified call resolution (#2550): `com.pkg.Fn()` /
 # `com.pkg.Object.method()` raw_calls the shared pass skips (member calls with
 # no receiver). Runs in the tail registry like the other member-call resolvers;