fix(scanner): compare model roots by identity, not by spelling (#1146)

ComfyUI hands its model folders over in the host's own spelling --
`utils/extra_config.py` runs every configured path through `os.path.normpath()`,
which is all backslashes on Windows, and `main.py` registers `<output>/loras`
unconditionally for its save nodes. `Config` kept that spelling verbatim while
`_prepare_*_paths()` normalized the live root lists to forward slashes, and the
scanner compared the two as raw strings. On Windows every configured root
therefore looked like a root that was not in the live list: the Refresh menu
listed each directory twice (the twin showing "0 models"), the completion toast
claimed reachable roots were "currently inaccessible" and kept their whole cache
count as `kept_unreachable`, and the settings sync persisted host-registered
directories that were never created.

The duplication was not only cosmetic. `_UnreachablePaths.add()` normalizes the
prefix it is given, so the configured twin of a live root registered exactly the
live root's prefix as unreadable: in the prune phase every cached-but-missing
entry under it matched and was "kept", so models deleted from disk were never
removed from the cache again.

* `py/utils/root_paths.py` holds the single `normalize_root_identity()`
  (separator style, redundant/trailing separators, and case where the filesystem
  folds it: whole paths on Windows, drive letters everywhere). `config.py` and
  `settings_manager.py` each carried a private copy and `model_scanner.py`
  compared raw strings.
* `ModelScanner` compares root identities in `get_configured_model_roots()`,
  `describe_model_roots()` (dedupe + `available`), the reconcile's
  `configured_only` list and the `ReconcileScope.roots` membership test, so one
  directory is one root whatever the host called it. `_normalized_root_prefix()`
  is built on the same identity, and an empty prefix no longer matches every
  path.
* A configured root that never made it into the live list is reported only while
  it still holds cached entries -- in the menu and in `skipped_roots` /
  `unavailable_paths` alike. A directory that neither exists nor was ever scanned
  has nothing to keep, so `<output>/loras` stops being an unavailable root.
* The startup sync drops saved roots that neither exist nor hold cached entries
  (`PersistentModelCache.roots_with_cached_entries()`, one scan per model type,
  separator-agnostic and wildcard-escaped). A root named by the library's
  `default_*_root` is kept even when empty: dropping it would let the next boot
  "repair" the default onto another root. When the cache cannot be consulted,
  every root is kept.
* `GET /scan?roots=` validates requested roots by identity too, so the browser
  extension or a script passing the spelling settings.json holds selects the root
  instead of answering 400 "Unknown model root(s)".

Verified: 8 of the new tests fail against the pre-fix source (checked with the
fix stashed); 3798 passed, 7 skipped, 11 deselected. The probe costs 71ms for a
100k-model library with seven missing roots (357ms when probing root by root).
This commit is contained in:
Will Miao
2026-10-12 08:43:06 +08:00
parent e056142669
commit 6d0d7cb15b
11 changed files with 822 additions and 57 deletions
+194
View File
@@ -968,3 +968,197 @@ def test_save_paths_keeps_configured_root_of_a_switched_off_drive(
assert config_instance.loras_roots == [readable.replace("\\", "/")]
# The user's default root survives because it is merely unavailable.
assert fake_settings.payload["default_lora_root"] == switched_off.replace("\\", "/")
# --- roots the host registered but never created (issue #1146) --------------
class _FakePersistentCache:
"""Stand-in for PersistentModelCache in the startup sync."""
def __init__(self, cached_roots=None, error: Exception | None = None):
self._cached_roots = cached_roots or {}
self._error = error
self.queries: List[tuple] = []
def roots_with_cached_entries(self, model_type: str, roots):
self.queries.append((model_type, list(roots)))
if self._error is not None:
raise self._error
known = {path.replace("\\", "/") for path in self._cached_roots.get(model_type, [])}
return {
root for root in roots if root.replace("\\", "/") in known
}
def _stub_persistent_cache(monkeypatch: pytest.MonkeyPatch, cache) -> None:
from py.services import persistent_model_cache as persistent_model_cache_module
monkeypatch.setattr(
persistent_model_cache_module, "get_persistent_cache", lambda: cache
)
class _LibrarySettingsService:
"""Minimal settings service recording what the startup sync persists."""
active_library = "comfyui"
def __init__(self, library: Dict[str, Any]):
self._library = library
self.payload: Dict[str, Any] = {}
def get_libraries(self):
return {"comfyui": self._library}
def rename_library(self, *_): # pragma: no cover - defensive guard
raise AssertionError("rename_library should not be invoked")
def delete_library(self, *_): # pragma: no cover - defensive guard
raise AssertionError("delete_library should not be invoked")
def get_active_library_name(self):
return self.active_library
def upsert_library(self, name: str, **payload):
self.payload = payload
def test_save_paths_drops_registered_directories_that_never_existed(
monkeypatch: pytest.MonkeyPatch, tmp_path
):
"""ComfyUI registers `<output>/<type>` for its save nodes unconditionally.
Those directories are not library paths when they do not exist and hold no
cached models: persisting them brings them back as "unavailable" roots on
every boot.
"""
folder_paths = _setup_config_environment(monkeypatch, tmp_path)
unet_dir = tmp_path / "unet"
unet_dir.mkdir()
folder_paths["unet"] = [str(unet_dir)]
output_loras = str(tmp_path / "output" / "loras")
output_checkpoints = str(tmp_path / "output" / "checkpoints")
output_unet = str(tmp_path / "output" / "diffusion_models")
output_embeddings = str(tmp_path / "output" / "embeddings")
folder_paths["loras"].append(output_loras)
folder_paths["checkpoints"].append(output_checkpoints)
folder_paths["unet"].append(output_unet)
folder_paths["embeddings"].append(output_embeddings)
fake_settings = _LibrarySettingsService(
{
"folder_paths": {key: list(value) for key, value in folder_paths.items()},
"default_lora_root": folder_paths["loras"][0],
"default_checkpoint_root": folder_paths["checkpoints"][0],
"default_embedding_root": folder_paths["embeddings"][0],
}
)
monkeypatch.setattr(settings_manager_module, "settings", fake_settings)
cache = _FakePersistentCache()
_stub_persistent_cache(monkeypatch, cache)
config_module.Config()
persisted = fake_settings.payload["folder_paths"]
for key, never_created in (
("loras", output_loras),
("checkpoints", output_checkpoints),
("unet", output_unet),
("embeddings", output_embeddings),
):
assert never_created.replace("\\", "/") not in persisted[key], key
# The real root of that key is untouched.
assert folder_paths[key][0].replace("\\", "/") in persisted[key], key
# Lora and checkpoint roots are looked up in their own model type.
assert [model_type for model_type, _roots in cache.queries] == [
"lora",
"checkpoint",
"checkpoint",
"embedding",
]
def test_save_paths_keeps_an_unavailable_root_that_has_cached_models(
monkeypatch: pytest.MonkeyPatch, tmp_path
):
"""A switched-off drive keeps its path: its models are still cached."""
folder_paths = _setup_config_environment(monkeypatch, tmp_path)
readable = folder_paths["loras"][0]
switched_off = str(tmp_path / "drive-Z" / "loras")
folder_paths["loras"] = [readable, switched_off]
fake_settings = _LibrarySettingsService(
{
"folder_paths": {"loras": [readable]},
"default_lora_root": readable,
}
)
monkeypatch.setattr(settings_manager_module, "settings", fake_settings)
_stub_persistent_cache(
monkeypatch, _FakePersistentCache({"lora": [switched_off]})
)
config_module.Config()
persisted = fake_settings.payload["folder_paths"]["loras"]
assert switched_off.replace("\\", "/") in persisted
assert readable.replace("\\", "/") in persisted
def test_save_paths_keeps_every_root_when_the_cache_cannot_be_read(
monkeypatch: pytest.MonkeyPatch, tmp_path
):
"""No answer from the cache means no root is dropped."""
folder_paths = _setup_config_environment(monkeypatch, tmp_path)
readable = folder_paths["loras"][0]
unknown = str(tmp_path / "drive-Z" / "loras")
folder_paths["loras"] = [readable, unknown]
fake_settings = _LibrarySettingsService(
{
"folder_paths": {"loras": [readable]},
"default_lora_root": readable,
}
)
monkeypatch.setattr(settings_manager_module, "settings", fake_settings)
_stub_persistent_cache(
monkeypatch, _FakePersistentCache(error=FileNotFoundError("no cache"))
)
config_module.Config()
persisted = fake_settings.payload["folder_paths"]["loras"]
assert unknown.replace("\\", "/") in persisted
def test_save_paths_keeps_the_root_a_default_setting_points_at(
monkeypatch: pytest.MonkeyPatch, tmp_path
):
"""An empty unavailable default root must not be dropped.
Dropping it would let the next boot "repair" the default onto another root and
silently retarget downloads. ``default_unet_root`` is only known from the
stored library (the sync resolves the lora/checkpoint/embedding ones itself),
so this covers the fallback to the stored value.
"""
folder_paths = _setup_config_environment(monkeypatch, tmp_path)
unet_dir = tmp_path / "unet"
unet_dir.mkdir()
default_target = str(tmp_path / "drive-Z" / "unet")
folder_paths["unet"] = [str(unet_dir), default_target]
fake_settings = _LibrarySettingsService(
{
"folder_paths": {"unet": [str(unet_dir)]},
"default_unet_root": default_target,
}
)
monkeypatch.setattr(settings_manager_module, "settings", fake_settings)
_stub_persistent_cache(monkeypatch, _FakePersistentCache())
config_module.Config()
persisted = fake_settings.payload["folder_paths"]["unet"]
assert default_target.replace("\\", "/") in persisted
assert str(unet_dir).replace("\\", "/") in persisted