Function collect_counts

scripts/mangle-javascript-port-identifiers.py:386–424

(files: list[Path], directory: Path)

def collect_counts(files: list[Path], directory: Path) -> tuple[Counter, frozenset[str]]:
    counts: Counter = Counter()
    for path in files:
        data = path.read_text(encoding="utf-8")
        for match in IDENTIFIER_PATTERN.finditer(data):
            counts[match.group(0)] += 1
    for name in EXCLUDE:
        counts.pop(name, None)

    jso_bridge_classes = _collect_jso_bridge_class_names(files)
    jso_bridge_dispatch_ids = _load_jso_bridge_dispatch_ids(directory)
    # Exclude every ``cn1_<jsoClass>_*`` method id so ``parseJsoBridgeMethod``
    # keeps working against host DOM receivers. Also exclude the class name
    # itself, both because it flows through the runtime as a plain string
    # (the ``className`` argument of ``invokeJsoBridge`` / ``isJsoBridgeClass``
    # / ``classes[...]`` lookup) and so runtime-built ``"cn1_" + className +
    # "_"`` prefixes still match the unmangled method ids we just excluded.
    # In addition, exclude every sig-based dispatch id the translator
    # tagged as a JSO bridge method via ``jso-bridge-dispatch-ids.txt``
    # — those ids reach ``parseJsoBridgeMethod`` for receivers whose
    # ``__class`` resolves through ``isJsoBridgeClass`` and need to keep
    # their original ``cn1_s_<method>_<sig>`` shape so the prefix strip
    # recovers a real DOM member name.
    to_exclude: set[str] = set()
    for name in list(counts.keys()):
        if name in jso_bridge_dispatch_ids:
            to_exclude.add(name)
            continue
        for cls in jso_bridge_classes:
            if name.startswith("cn1_" + cls + "_") or name.startswith("cn1_" + cls + "__"):
                to_exclude.add(name)
                break
        if name in jso_bridge_classes:
            to_exclude.add(name)
    for name in to_exclude:
        counts.pop(name, None)

    preserved = frozenset(to_exclude | set(EXCLUDE) | jso_bridge_classes | jso_bridge_dispatch_ids)
    return counts, preserved


_IMPL_SUFFIX = "__impl"