Detecting references to exluded symbols

Differential Revision: D99417803

Author: Dawid Małecki

scripts/cxx-api/parser/__init__.py

@@ -3,7 +3,16 @@
 # This source code is licensed under the MIT license found in the
 # LICENSE file in the root directory of this source tree.
 
-from .main import build_snapshot
+from .main import (
+    build_snapshot,
+    ExcludedSymbolReference,
+    find_excluded_symbol_references,
+)
 from .path_utils import get_repo_root
 
-__all__ = ["build_snapshot", "get_repo_root"]
+__all__ = [
+    "build_snapshot",
+    "ExcludedSymbolReference",
+    "find_excluded_symbol_references",
+    "get_repo_root",
+]

scripts/cxx-api/parser/__main__.py

@@ -96,6 +96,7 @@ def build_snapshot_for_view(
     input_filter: str = None,
     work_dir: str | None = None,
     exclude_symbols: list[str] | None = None,
+    warn_excluded_refs: bool = False,
 ) -> str:
     if verbose:
         print(f"[{api_view}] Generating API view")
@@ -130,6 +131,23 @@ def build_snapshot_for_view(
     snapshot = build_snapshot(
         os.path.join(work_dir, "xml"), exclude_symbols=exclude_symbols
     )
+
+    if warn_excluded_refs and snapshot.excluded_symbol_references:
+        _YELLOW = "\033[33m"
+        _RESET = "\033[0m"
+        refs = snapshot.excluded_symbol_references
+        print(
+            f"{_YELLOW}[{api_view}] WARNING: Found {len(refs)} reference(s) "
+            f"to excluded symbols:{_RESET}",
+            file=sys.stderr,
+        )
+        for ref in refs:
+            print(
+                f"{_YELLOW}  • {ref.scope}: {ref.context} '{ref.symbol}' "
+                f"matches excluded pattern '{ref.pattern}'{_RESET}",
+                file=sys.stderr,
+            )
+
     snapshot_string = snapshot.to_string()
 
     output_file = os.path.join(output_dir, f"{api_view}Cxx.api")
@@ -151,6 +169,7 @@ def build_snapshots(
     view_filter: str | None = None,
     is_test: bool = False,
     keep_xml: bool = False,
+    warn_excluded_refs: bool = False,
 ) -> None:
     if not is_test:
         configs_to_build = [
@@ -195,6 +214,7 @@ def build_snapshots(
                         input_filter=input_filter if config.input_filter else None,
                         work_dir=work_dir,
                         exclude_symbols=config.exclude_symbols,
+                        warn_excluded_refs=warn_excluded_refs,
                     )
                     futures[future] = config.snapshot_name
 
@@ -285,6 +305,11 @@ def main():
         action="store_true",
         help="Keep the generated Doxygen XML files next to the .api output in a xml/ directory",
     )
+    parser.add_argument(
+        "--warn-excluded-refs",
+        action="store_true",
+        help="Warn when non-excluded symbols reference types matching exclude_symbols patterns",
+    )
     args = parser.parse_args()
 
     verbose = not args.validate
@@ -343,6 +368,7 @@ def main():
             view_filter=args.view,
             is_test=args.test,
             keep_xml=args.xml,
+            warn_excluded_refs=args.warn_excluded_refs,
         )
 
         if args.validate:

scripts/cxx-api/parser/main.py

@@ -10,6 +10,7 @@
 from __future__ import annotations
 
 import os
+from dataclasses import dataclass
 
 from doxmlparser import compound, index
 
@@ -24,8 +25,38 @@
     get_typedef_member,
     get_variable_member,
 )
+from .member import (
+    FriendMember,
+    FunctionMember,
+    PropertyMember,
+    TypedefMember,
+    VariableMember,
+)
+from .scope import Scope, StructLikeScopeKind
+from .scope.extendable import Extendable
 from .snapshot import Snapshot
-from .utils import has_scope_resolution_outside_angles, parse_qualified_path
+from .utils import (
+    format_parsed_type,
+    has_scope_resolution_outside_angles,
+    parse_qualified_path,
+)
+
+
+@dataclass
+class ExcludedSymbolReference:
+    """A reference to an excluded symbol found in the API snapshot."""
+
+    symbol: str
+    """The full text containing the reference (e.g., the type string)."""
+
+    pattern: str
+    """The exclude_symbols pattern that matched."""
+
+    scope: str
+    """The qualified name of the scope containing the reference."""
+
+    context: str
+    """Description of where the reference appears (e.g., 'base class', 'return type')."""
 
 
 def _should_exclude_symbol(name: str, exclude_symbols: list[str]) -> bool:
@@ -170,6 +201,199 @@ def _handle_class_compound(snapshot, compound_object):
 )
 
 
+def _check_text_for_excluded_patterns(
+    text: str,
+    scope_name: str,
+    context: str,
+    exclude_symbols: list[str],
+    results: list[ExcludedSymbolReference],
+) -> None:
+    """Append an ExcludedSymbolReference for each pattern found in *text*."""
+    for pattern in exclude_symbols:
+        if pattern in text:
+            results.append(
+                ExcludedSymbolReference(
+                    symbol=text,
+                    pattern=pattern,
+                    scope=scope_name,
+                    context=context,
+                )
+            )
+
+
+def _check_arguments_for_excluded_patterns(
+    arguments: list,
+    scope_name: str,
+    context_prefix: str,
+    exclude_symbols: list[str],
+    results: list[ExcludedSymbolReference],
+) -> None:
+    """Check every argument's type string for excluded patterns."""
+    for arg in arguments:
+        # Argument is a tuple: (qualifiers, type, name, default_value)
+        arg_type = arg[1]
+        if arg_type:
+            _check_text_for_excluded_patterns(
+                arg_type,
+                scope_name,
+                f"{context_prefix} parameter type",
+                exclude_symbols,
+                results,
+            )
+
+
+def _check_member_for_excluded_patterns(
+    member,
+    scope_name: str,
+    exclude_symbols: list[str],
+    results: list[ExcludedSymbolReference],
+) -> None:
+    """Check a single member for type references matching excluded patterns."""
+    member_name = f"{scope_name}::{member.name}"
+
+    if isinstance(member, FunctionMember):
+        if member.type:
+            _check_text_for_excluded_patterns(
+                member.type,
+                member_name,
+                "return type",
+                exclude_symbols,
+                results,
+            )
+        _check_arguments_for_excluded_patterns(
+            member.arguments,
+            member_name,
+            "function",
+            exclude_symbols,
+            results,
+        )
+
+    elif isinstance(member, VariableMember):
+        type_str = format_parsed_type(member._parsed_type)
+        if type_str:
+            _check_text_for_excluded_patterns(
+                type_str,
+                member_name,
+                "variable type",
+                exclude_symbols,
+                results,
+            )
+        _check_arguments_for_excluded_patterns(
+            member._fp_arguments,
+            member_name,
+            "function pointer",
+            exclude_symbols,
+            results,
+        )
+
+    elif isinstance(member, TypedefMember):
+        value = member.get_value()
+        if value:
+            _check_text_for_excluded_patterns(
+                value,
+                member_name,
+                "typedef target type",
+                exclude_symbols,
+                results,
+            )
+        _check_arguments_for_excluded_patterns(
+            member._fp_arguments,
+            member_name,
+            "function pointer",
+            exclude_symbols,
+            results,
+        )
+
+    elif isinstance(member, PropertyMember):
+        if member.type:
+            _check_text_for_excluded_patterns(
+                member.type,
+                member_name,
+                "property type",
+                exclude_symbols,
+                results,
+            )
+
+    elif isinstance(member, FriendMember):
+        _check_text_for_excluded_patterns(
+            member.name,
+            member_name,
+            "friend declaration",
+            exclude_symbols,
+            results,
+        )
+
+    if member.specialization_args:
+        for arg in member.specialization_args:
+            _check_text_for_excluded_patterns(
+                arg,
+                member_name,
+                "member specialization argument",
+                exclude_symbols,
+                results,
+            )
+
+
+def _walk_scope_for_excluded_patterns(
+    scope: Scope,
+    exclude_symbols: list[str],
+    results: list[ExcludedSymbolReference],
+) -> None:
+    """Recursively walk a scope tree checking for excluded pattern references."""
+    scope_name = scope.get_qualified_name() or "(root)"
+
+    # Check base classes (StructLikeScopeKind, ProtocolScopeKind, InterfaceScopeKind)
+    if isinstance(scope.kind, Extendable):
+        for base in scope.kind.base_classes:
+            _check_text_for_excluded_patterns(
+                base.name,
+                scope_name,
+                "base class",
+                exclude_symbols,
+                results,
+            )
+
+    # Check specialization args
+    if isinstance(scope.kind, StructLikeScopeKind) and scope.kind.specialization_args:
+        for arg in scope.kind.specialization_args:
+            _check_text_for_excluded_patterns(
+                arg,
+                scope_name,
+                "specialization argument",
+                exclude_symbols,
+                results,
+            )
+
+    for member in scope.get_members():
+        _check_member_for_excluded_patterns(
+            member, scope_name, exclude_symbols, results
+        )
+
+    for inner in scope.inner_scopes.values():
+        _walk_scope_for_excluded_patterns(inner, exclude_symbols, results)
+
+
+def find_excluded_symbol_references(
+    snapshot: Snapshot,
+    exclude_symbols: list[str],
+) -> list[ExcludedSymbolReference]:
+    """
+    Walk the snapshot scope tree after it has been finalized and find
+    references to excluded symbols in type strings, base classes, and
+    other type references.
+
+    This detects cases where a non-excluded symbol references an excluded
+    symbol (e.g., a class inherits from an excluded base, a function returns
+    an excluded type, etc.).
+    """
+    if not exclude_symbols:
+        return []
+
+    results: list[ExcludedSymbolReference] = []
+    _walk_scope_for_excluded_patterns(snapshot.root_scope, exclude_symbols, results)
+    return results
+
+
 def build_snapshot(xml_dir: str, exclude_symbols: list[str] | None = None) -> Snapshot:
     """
     Reads the Doxygen XML output and builds a snapshot of the C++ API.
@@ -218,4 +442,9 @@ def build_snapshot(xml_dir: str, exclude_symbols: list[str] | None = None) -> Sn
                 print(f"Unknown compound kind: {kind}")
 
     snapshot.finish()
+
+    snapshot.excluded_symbol_references = find_excluded_symbol_references(
+        snapshot, exclude_symbols
+    )
+
     return snapshot

scripts/cxx-api/parser/snapshot.py

@@ -21,6 +21,7 @@
 class Snapshot:
     def __init__(self) -> None:
         self.root_scope: Scope = Scope(NamespaceScopeKind())
+        self.excluded_symbol_references: list = []
 
     def ensure_scope(self, scope_path: list[str]) -> Scope:
         """