voculinux/tests/test_rename_sweep.py

"""Ticket #18: offline naming-sweep — no old-name strings on user-visible surfaces.

The fork takes its own name everywhere a user can see it. Anything still
spelling the old name must be on the allowlist below:

- upstream provenance (``VocaHQ/vocalinux`` URLs),
- the retained internal Python module name (``src/vocalinux`` paths,
  ``vocalinux.*`` imports/attributes, ``-m vocalinux.main``),
- retained internal class names (``VocalinuxEngine*``, ``VocalinuxDBusService``),
- one-time-legacy import sources (old config/dictionary/bus literals),
- generated or historical files (``uv.lock``, ``requirements/*.txt``,
  ``CHANGELOG.md``),
- the rest of ``web/`` (owned by ticket #24; only ``page.tsx`` and
  ``package.json`` belong to this ticket).
"""

import os
import re

REPO_ROOT = os.path.dirname(os.path.dirname(os.path.abspath(__file__)))

OLD_NAME = re.compile(r"vocalinux", re.IGNORECASE)

SKIP_DIRS = {".git", "__pycache__", ".venv", "venv", "node_modules", ".next", "build", "dist"}
SKIP_SUFFIXES = (".egg-info",)
SKIP_FILES = {"uv.lock", "CHANGELOG.md"}
SKIP_PREFIXES = ("requirements/",)
BINARY_EXTS = (
    ".wav",
    ".mp3",
    ".ogg",
    ".png",
    ".svg",
    ".ico",
    ".icns",
    ".bin",
    ".onnx",
    ".pt",
    ".zip",
    ".tar",
    ".gz",
    ".whl",
    ".AppImage",
    ".zsync",
    ".pyc",
)

#: web/ belongs to ticket #24 except the two rename-inventory files.
WEB_OWNED = {"web/src/app/page.tsx", "web/package.json"}


def _allowed(path: str, line: str) -> bool:
    """Return True when an old-name match is an allowlisted surface."""
    low = line.lower()
    if "vocahq/" in low or "jatinkrmalik/vocalinux" in low:
        return True
    if "vocalinux" in low and "voculinux" in low:
        # Migration-compat shims that name both installs at once.
        return True
    if "/{name}/vocalinux" in line:
        # Upstream repo-name template (retained provenance).
        return True
    if path == "snap/local/check_runtime.py" and '"vocalinux"' in line:
        # Smoke imports resolve the retained import package.
        return True
    if re.search(r'vocalinux\.["a-z_]', line):
        return True
    if line.strip().rstrip(",").lstrip("/ ") == '"vocalinux"':
        # Retained module-dir path segment element in tests and tooling
        # (single-line and os.path.join element forms).
        return True
    if 'site / "vocalinux"' in line or '"platform/vocalinux"' in line:
        # Fixture module trees named after the retained import package.
        return True
    if '"vocalinux/' in line:
        # Fixture module file paths under the retained package name.
        return True
    if "/vocalinux/" in line or "src/vocalinux" in line:
        return True
    if "-m vocalinux" in line or "import vocalinux" in line:
        return True
    if "VocalinuxEngine" in line or "VocalinuxDBusService" in line:
        return True
    if "cmdline" in line:
        return True
    if " - vocalinux" in line:
        # Log logger names keep the retained module path.
        return True
    if path in (
        "src/vocalinux/legacy_import.py",
        "tests/test_legacy_import.py",
    ):
        return True
    if "LEGACY" in line or "legacy" in line:
        return True
    if "~/.config/vocalinux" in line or '".config" / "vocalinux"' in line:
        # Documented one-time-import source (ticket #18).
        return True
    if "from vocalinux" in line:
        return True
    if '"src"' in line and '"vocalinux"' in line:
        # Retained module-dir path segments in tests and tooling.
        return True
    if path == "tests/test_rename_sweep.py":
        return True
    if path == "pyproject.toml" and re.search(r'"vocalinux[."\'\]]', line):
        return True
    if path == "pyproject.toml" and "vocalinux.version" in line:
        return True
    if 'PKG_NAME="vocalinux"' in line:
        # Installer probes the retained import package, not the distribution.
        return True
    if '== "vocalinux"' in line or '.startswith("vocalinux' in line:
        # Retained module identity comparisons in test isolation helpers.
        return True
    return False


def _iter_files():
    for root, dirs, files in os.walk(REPO_ROOT):
        dirs[:] = [d for d in sorted(dirs) if d not in SKIP_DIRS]
        for name in sorted(files):
            if name in SKIP_FILES or name.endswith(BINARY_EXTS):
                continue
            full = os.path.join(root, name)
            rel = os.path.relpath(full, REPO_ROOT)
            if rel.startswith(SKIP_PREFIXES):
                continue
            if rel.startswith("web/") and rel not in WEB_OWNED:
                continue
            yield rel, full


def test_naming_sweep_no_old_name_outside_allowlist() -> None:
    """Every old-name hit must be an allowlisted retained surface."""
    offenders: list[str] = []
    for rel, full in _iter_files():
        if any(part.endswith(SKIP_SUFFIXES) for part in rel.split("/")):
            continue
        try:
            with open(full, encoding="utf-8") as handle:
                text = handle.read()
        except (UnicodeDecodeError, ValueError):
            continue
        for number, line in enumerate(text.splitlines(), 1):
            if OLD_NAME.search(line) and not _allowed(rel, line):
                offenders.append(f"{rel}:{number}:{line.strip()[:160]}")
    assert (
        offenders == []
    ), f"{len(offenders)} old-name string(s) outside the allowlist:\n" + "\n".join(offenders[:60])


def test_owned_web_files_carry_new_name() -> None:
    """The two web rename surfaces name the fork (URLs aside)."""
    for rel in sorted(WEB_OWNED):
        with open(os.path.join(REPO_ROOT, rel), encoding="utf-8") as handle:
            text = handle.read()
        assert "Voculinux" in text or "voculinux" in text, rel