mirrors/mmengine
1
0
Fork 0
mirror of https://github.com/open-mmlab/mmengine.git synced 2025-06-03 21:54:44 +08:00
Code Issues Projects Releases Wiki Activity

[Enhancement] Replace warnings.warn with print_log (#961)

Browse Source 
* Replace warning with print_log

* Add comments for testing warning
This commit is contained in:
Mashiro 2023-03-06 17:25:28 +08:00 committed by GitHub
parent b3430e4257
commit dbae83c52f
No known key found for this signature in database
GPG Key ID: 4AEE18F83AFDEB23
25 changed files with 206 additions and 125 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

24
mmengine/dataset/base_dataset.py
View File

@ -2,15 +2,16 @@
import copy import copy
import functools import functools
import gc import gc
import logging
import os.path as osp import os.path as osp
import pickle import pickle
import warnings
from typing import Any, Callable, List, Optional, Sequence, Tuple, Union from typing import Any, Callable, List, Optional, Sequence, Tuple, Union
import numpy as np import numpy as np
from torch.utils.data import Dataset from torch.utils.data import Dataset
from mmengine.fileio import list_from_file, load from mmengine.fileio import list_from_file, load
from mmengine.logging import print_log
from mmengine.registry import TRANSFORMS from mmengine.registry import TRANSFORMS
from mmengine.utils import is_abs from mmengine.utils import is_abs
@ -100,11 +101,13 @@ def force_full_init(old_func: Callable) -> Any:
# `_fully_initialized` is False, call `full_init` and set # `_fully_initialized` is False, call `full_init` and set
# `_fully_initialized` to True # `_fully_initialized` to True
if not getattr(obj, '_fully_initialized', False): if not getattr(obj, '_fully_initialized', False):
warnings.warn('Attribute `_fully_initialized` is not defined in ' print_log(
f'Attribute `_fully_initialized` is not defined in '
f'{type(obj)} or `type(obj)._fully_initialized is ' f'{type(obj)} or `type(obj)._fully_initialized is '
'False, `full_init` will be called and ' 'False, `full_init` will be called and '
f'{type(obj)}._fully_initialized will be set to ' f'{type(obj)}._fully_initialized will be set to True',
'True') logger='current',
level=logging.WARNING)
obj.full_init() # type: ignore obj.full_init() # type: ignore
obj._fully_initialized = True # type: ignore obj._fully_initialized = True # type: ignore
@ -392,9 +395,11 @@ class BaseDataset(Dataset):
# to manually call `full_init` before dataset fed into dataloader to # to manually call `full_init` before dataset fed into dataloader to
# ensure all workers use shared RAM from master process. # ensure all workers use shared RAM from master process.
if not self._fully_initialized: if not self._fully_initialized:
warnings.warn( print_log(
'Please call `full_init()` method manually to accelerate ' 'Please call `full_init()` method manually to accelerate '
'the speed.') 'the speed.',
logger='current',
level=logging.WARNING)
self.full_init() self.full_init()
if self.test_mode: if self.test_mode:
@ -498,8 +503,11 @@ class BaseDataset(Dataset):
try: try:
cls_metainfo[k] = list_from_file(v) cls_metainfo[k] = list_from_file(v)
except (TypeError, FileNotFoundError): except (TypeError, FileNotFoundError):
warnings.warn(f'{v} is not a meta file, simply parsed as ' print_log(
'meta information') f'{v} is not a meta file, simply parsed as meta '
'information',
logger='current',
level=logging.WARNING)
cls_metainfo[k] = v cls_metainfo[k] = v
else: else:
cls_metainfo[k] = v cls_metainfo[k] = v

24
mmengine/dataset/dataset_wrapper.py
View File

@ -1,13 +1,14 @@
# Copyright (c) OpenMMLab. All rights reserved. # Copyright (c) OpenMMLab. All rights reserved.
import bisect import bisect
import copy import copy
import logging
import math import math
import warnings
from collections import defaultdict from collections import defaultdict
from typing import List, Sequence, Tuple, Union from typing import List, Sequence, Tuple, Union
from torch.utils.data.dataset import ConcatDataset as _ConcatDataset from torch.utils.data.dataset import ConcatDataset as _ConcatDataset
from mmengine.logging import print_log
from mmengine.registry import DATASETS from mmengine.registry import DATASETS
from .base_dataset import BaseDataset, force_full_init from .base_dataset import BaseDataset, force_full_init
@ -148,8 +149,11 @@ class ConcatDataset(_ConcatDataset):
def __getitem__(self, idx): def __getitem__(self, idx):
if not self._fully_initialized: if not self._fully_initialized:
warnings.warn('Please call `full_init` method manually to ' print_log(
'accelerate the speed.') 'Please call `full_init` method manually to '
'accelerate the speed.',
logger='current',
level=logging.WARNING)
self.full_init() self.full_init()
dataset_idx, sample_idx = self._get_ori_dataset_idx(idx) dataset_idx, sample_idx = self._get_ori_dataset_idx(idx)
return self.datasets[dataset_idx][sample_idx] return self.datasets[dataset_idx][sample_idx]
@ -263,8 +267,11 @@ class RepeatDataset:
def __getitem__(self, idx): def __getitem__(self, idx):
if not self._fully_initialized: if not self._fully_initialized:
warnings.warn('Please call `full_init` method manually to ' print_log(
'accelerate the speed.') 'Please call `full_init` method manually to accelerate the '
'speed.',
logger='current',
level=logging.WARNING)
self.full_init() self.full_init()
sample_idx = self._get_ori_dataset_idx(idx) sample_idx = self._get_ori_dataset_idx(idx)
@ -470,9 +477,12 @@ class ClassBalancedDataset:
return self.dataset.get_data_info(sample_idx) return self.dataset.get_data_info(sample_idx)
def __getitem__(self, idx): def __getitem__(self, idx):
warnings.warn('Please call `full_init` method manually to '
'accelerate the speed.')
if not self._fully_initialized: if not self._fully_initialized:
print_log(
'Please call `full_init` method manually to accelerate '
'the speed.',
logger='current',
level=logging.WARNING)
self.full_init() self.full_init()
ori_index = self._get_ori_dataset_idx(idx) ori_index = self._get_ori_dataset_idx(idx)

15
mmengine/evaluator/metric.py
View File

@ -1,5 +1,5 @@
# Copyright (c) OpenMMLab. All rights reserved. # Copyright (c) OpenMMLab. All rights reserved.
import warnings import logging
from abc import ABCMeta, abstractmethod from abc import ABCMeta, abstractmethod
from typing import Any, List, Optional, Sequence, Union from typing import Any, List, Optional, Sequence, Union
@ -44,8 +44,11 @@ class BaseMetric(metaclass=ABCMeta):
self.results: List[Any] = [] self.results: List[Any] = []
self.prefix = prefix or self.default_prefix self.prefix = prefix or self.default_prefix
if self.prefix is None: if self.prefix is None:
warnings.warn('The prefix is not set in metric class ' print_log(
f'{self.__class__.__name__}.') 'The prefix is not set in metric class '
f'{self.__class__.__name__}.',
logger='current',
level=logging.WARNING)
@property @property
def dataset_meta(self) -> Optional[dict]: def dataset_meta(self) -> Optional[dict]:
@ -97,10 +100,12 @@ class BaseMetric(metaclass=ABCMeta):
names of the metrics, and the values are corresponding results. names of the metrics, and the values are corresponding results.
""" """
if len(self.results) == 0: if len(self.results) == 0:
warnings.warn( print_log(
f'{self.__class__.__name__} got empty `self.results`. Please ' f'{self.__class__.__name__} got empty `self.results`. Please '
'ensure that the processed results are properly added into ' 'ensure that the processed results are properly added into '
'`self.results` in `process` method.') '`self.results` in `process` method.',
logger='current',
level=logging.WARNING)
results = collect_results(self.results, size, self.collect_device) results = collect_results(self.results, size, self.collect_device)

10
mmengine/fileio/backends/base.py
View File

@ -1,7 +1,9 @@
# Copyright (c) OpenMMLab. All rights reserved. # Copyright (c) OpenMMLab. All rights reserved.
import warnings import logging
from abc import ABCMeta, abstractmethod from abc import ABCMeta, abstractmethod
from mmengine.logging import print_log
class BaseStorageBackend(metaclass=ABCMeta): class BaseStorageBackend(metaclass=ABCMeta):
"""Abstract class of storage backends. """Abstract class of storage backends.
@ -19,8 +21,10 @@ class BaseStorageBackend(metaclass=ABCMeta):
@property @property
def allow_symlink(self): def allow_symlink(self):
warnings.warn('allow_symlink will be deprecated in future', print_log(
DeprecationWarning) 'allow_symlink will be deprecated in future',
logger='current',
level=logging.WARNING)
return self._allow_symlink return self._allow_symlink
@property @property

14
mmengine/fileio/file_client.py
View File

@ -1,10 +1,11 @@
# Copyright (c) OpenMMLab. All rights reserved. # Copyright (c) OpenMMLab. All rights reserved.
import inspect import inspect
import warnings import logging
from contextlib import contextmanager from contextlib import contextmanager
from pathlib import Path from pathlib import Path
from typing import Any, Generator, Iterator, Optional, Tuple, Union from typing import Any, Generator, Iterator, Optional, Tuple, Union
from mmengine.logging import print_log
from mmengine.utils import is_filepath from mmengine.utils import is_filepath
from .backends import (BaseStorageBackend, HTTPBackend, LmdbBackend, from .backends import (BaseStorageBackend, HTTPBackend, LmdbBackend,
LocalBackend, MemcachedBackend, PetrelBackend) LocalBackend, MemcachedBackend, PetrelBackend)
@ -14,9 +15,11 @@ class HardDiskBackend(LocalBackend):
"""Raw hard disks storage backend.""" """Raw hard disks storage backend."""
def __init__(self) -> None: def __init__(self) -> None:
warnings.warn( print_log(
'"HardDiskBackend" is the alias of "LocalBackend" ' '"HardDiskBackend" is the alias of "LocalBackend" '
'and the former will be deprecated in future.', DeprecationWarning) 'and the former will be deprecated in future.',
logger='current',
level=logging.WARNING)
@property @property
def name(self): def name(self):
@ -83,11 +86,12 @@ class FileClient:
client: Any client: Any
def __new__(cls, backend=None, prefix=None, **kwargs): def __new__(cls, backend=None, prefix=None, **kwargs):
warnings.warn( print_log(
'"FileClient" will be deprecated in future. Please use io ' '"FileClient" will be deprecated in future. Please use io '
'functions in ' 'functions in '
'https://mmengine.readthedocs.io/en/latest/api/fileio.html#file-io', # noqa: E501 'https://mmengine.readthedocs.io/en/latest/api/fileio.html#file-io', # noqa: E501
DeprecationWarning) logger='current',
level=logging.WARNING)
if backend is None and prefix is None: if backend is None and prefix is None:
backend = 'disk' backend = 'disk'
if backend is not None and backend not in cls._backends: if backend is not None and backend not in cls._backends:

9
mmengine/hooks/checkpoint_hook.py
View File

@ -1,12 +1,13 @@
# Copyright (c) OpenMMLab. All rights reserved. # Copyright (c) OpenMMLab. All rights reserved.
import logging
import os.path as osp import os.path as osp
import warnings
from math import inf from math import inf
from pathlib import Path from pathlib import Path
from typing import Callable, Dict, List, Optional, Sequence, Union from typing import Callable, Dict, List, Optional, Sequence, Union
from mmengine.dist import is_main_process from mmengine.dist import is_main_process
from mmengine.fileio import FileClient, get_file_backend from mmengine.fileio import FileClient, get_file_backend
from mmengine.logging import print_log
from mmengine.registry import HOOKS from mmengine.registry import HOOKS
from mmengine.utils import is_list_of, is_seq_of from mmengine.utils import is_list_of, is_seq_of
from .hook import Hook from .hook import Hook
@ -138,9 +139,11 @@ class CheckpointHook(Hook):
self.args = kwargs self.args = kwargs
if file_client_args is not None: if file_client_args is not None:
warnings.warn( print_log(
'"file_client_args" will be deprecated in future. ' '"file_client_args" will be deprecated in future. '
'Please use "backend_args" instead', DeprecationWarning) 'Please use "backend_args" instead',
logger='current',
level=logging.WARNING)
if backend_args is not None: if backend_args is not None:
raise ValueError( raise ValueError(
'"file_client_args" and "backend_args" cannot be set ' '"file_client_args" and "backend_args" cannot be set '

9
mmengine/hooks/logger_hook.py
View File

@ -1,7 +1,7 @@
# Copyright (c) OpenMMLab. All rights reserved. # Copyright (c) OpenMMLab. All rights reserved.
import logging
import os import os
import os.path as osp import os.path as osp
import warnings
from collections import OrderedDict from collections import OrderedDict
from pathlib import Path from pathlib import Path
from typing import Dict, Optional, Sequence, Union from typing import Dict, Optional, Sequence, Union
@ -12,6 +12,7 @@ import torch
from mmengine.fileio import FileClient, dump from mmengine.fileio import FileClient, dump
from mmengine.fileio.io import get_file_backend from mmengine.fileio.io import get_file_backend
from mmengine.hooks import Hook from mmengine.hooks import Hook
from mmengine.logging import print_log
from mmengine.registry import HOOKS from mmengine.registry import HOOKS
from mmengine.utils import is_tuple_of, scandir from mmengine.utils import is_tuple_of, scandir
@ -94,9 +95,11 @@ class LoggerHook(Hook):
self.out_dir = out_dir self.out_dir = out_dir
if file_client_args is not None: if file_client_args is not None:
warnings.warn( print_log(
'"file_client_args" will be deprecated in future. ' '"file_client_args" will be deprecated in future. '
'Please use "backend_args" instead', DeprecationWarning) 'Please use "backend_args" instead',
logger='current',
level=logging.WARNING)
if backend_args is not None: if backend_args is not None:
raise ValueError( raise ValueError(
'"file_client_args" and "backend_args" cannot be set ' '"file_client_args" and "backend_args" cannot be set '

5
mmengine/hooks/profiler_hook.py
View File

@ -1,14 +1,15 @@
# Copyright (c) OpenMMLab. All rights reserved. # Copyright (c) OpenMMLab. All rights reserved.
import logging
import os import os
import os.path as osp import os.path as osp
import sys import sys
import warnings
from typing import Callable, Optional, Union from typing import Callable, Optional, Union
import torch import torch
from mmengine.dist import master_only from mmengine.dist import master_only
from mmengine.hooks import Hook from mmengine.hooks import Hook
from mmengine.logging import print_log
from mmengine.registry import HOOKS from mmengine.registry import HOOKS
@ -18,7 +19,7 @@ def check_kineto() -> bool: # noqa
if torch.autograd.kineto_available(): if torch.autograd.kineto_available():
kineto_exist = True kineto_exist = True
except AttributeError: except AttributeError:
warnings.warn('NO KINETO') print_log('NO KINETO', logger='current', level=logging.WARNING)
return kineto_exist return kineto_exist

9
mmengine/model/averaged_model.py
View File

@ -1,5 +1,5 @@
# Copyright (c) OpenMMLab. All rights reserved. # Copyright (c) OpenMMLab. All rights reserved.
import warnings import logging
from abc import abstractmethod from abc import abstractmethod
from copy import deepcopy from copy import deepcopy
from typing import Optional from typing import Optional
@ -8,6 +8,7 @@ import torch
import torch.nn as nn import torch.nn as nn
from torch import Tensor from torch import Tensor
from mmengine.logging import print_log
from mmengine.registry import MODELS from mmengine.registry import MODELS
@ -184,11 +185,13 @@ class ExponentialMovingAverage(BaseAveragedModel):
assert 0.0 < momentum < 1.0, 'momentum must be in range (0.0, 1.0)'\ assert 0.0 < momentum < 1.0, 'momentum must be in range (0.0, 1.0)'\
f'but got {momentum}' f'but got {momentum}'
if momentum > 0.5: if momentum > 0.5:
warnings.warn( print_log(
'The value of momentum in EMA is usually a small number,' 'The value of momentum in EMA is usually a small number,'
'which is different from the conventional notion of ' 'which is different from the conventional notion of '
f'momentum but got {momentum}. Please make sure the ' f'momentum but got {momentum}. Please make sure the '
f'value is correct.') f'value is correct.',
logger='current',
level=logging.WARNING)
self.momentum = momentum self.momentum = momentum
def avg_func(self, averaged_param: Tensor, source_param: Tensor, def avg_func(self, averaged_param: Tensor, source_param: Tensor,

8
mmengine/model/base_module.py
View File

@ -1,7 +1,6 @@
# Copyright (c) OpenMMLab. All rights reserved. # Copyright (c) OpenMMLab. All rights reserved.
import copy import copy
import logging import logging
import warnings
from abc import ABCMeta from abc import ABCMeta
from collections import defaultdict from collections import defaultdict
from logging import FileHandler from logging import FileHandler
@ -139,8 +138,11 @@ class BaseModule(nn.Module, metaclass=ABCMeta):
initialize(self, pretrained_cfg) initialize(self, pretrained_cfg)
self._is_init = True self._is_init = True
else: else:
warnings.warn(f'init_weights of {self.__class__.__name__} has ' print_log(
f'been called more than once.') f'init_weights of {self.__class__.__name__} has '
f'been called more than once.',
logger='current',
level=logging.WARNING)
if is_top_level_module: if is_top_level_module:
# self._dump_init_info(logger_name) # self._dump_init_info(logger_name)

9
mmengine/optim/optimizer/default_constructor.py
View File

@ -1,5 +1,5 @@
# Copyright (c) OpenMMLab. All rights reserved. # Copyright (c) OpenMMLab. All rights reserved.
import warnings import logging
from typing import List, Optional, Union from typing import List, Optional, Union
import torch import torch
@ -205,8 +205,11 @@ class DefaultOptimWrapperConstructor:
for name, param in module.named_parameters(recurse=False): for name, param in module.named_parameters(recurse=False):
param_group = {'params': [param]} param_group = {'params': [param]}
if bypass_duplicate and self._is_in(param_group, params): if bypass_duplicate and self._is_in(param_group, params):
warnings.warn(f'{prefix} is duplicate. It is skipped since ' print_log(
f'bypass_duplicate={bypass_duplicate}') f'{prefix} is duplicate. It is skipped since '
f'bypass_duplicate={bypass_duplicate}',
logger='current',
level=logging.WARNING)
continue continue
if not param.requires_grad: if not param.requires_grad:
params.append(param_group) params.append(param_group)

9
mmengine/registry/utils.py
View File

@ -1,7 +1,7 @@
# Copyright (c) OpenMMLab. All rights reserved. # Copyright (c) OpenMMLab. All rights reserved.
import datetime import datetime
import logging
import os.path as osp import os.path as osp
import warnings
from typing import Optional from typing import Optional
from mmengine.fileio import dump from mmengine.fileio import dump
@ -107,10 +107,13 @@ def init_default_scope(scope: str) -> None:
return return
current_scope = DefaultScope.get_current_instance() # type: ignore current_scope = DefaultScope.get_current_instance() # type: ignore
if current_scope.scope_name != scope: # type: ignore if current_scope.scope_name != scope: # type: ignore
warnings.warn('The current default scope ' # type: ignore print_log(
'The current default scope ' # type: ignore
f'"{current_scope.scope_name}" is not "{scope}", ' f'"{current_scope.scope_name}" is not "{scope}", '
'`init_default_scope` will force set the current' '`init_default_scope` will force set the current'
f'default scope to "{scope}".') f'default scope to "{scope}".',
logger='current',
level=logging.WARNING)
# avoid name conflict # avoid name conflict
new_instance_name = f'{scope}-{datetime.datetime.now()}' new_instance_name = f'{scope}-{datetime.datetime.now()}'
DefaultScope.get_instance(new_instance_name, scope_name=scope) DefaultScope.get_instance(new_instance_name, scope_name=scope)

18
mmengine/runner/checkpoint.py
View File

@ -5,7 +5,6 @@ import os
import os.path as osp import os.path as osp
import pkgutil import pkgutil
import re import re
import warnings
from collections import OrderedDict from collections import OrderedDict
from importlib import import_module from importlib import import_module
from tempfile import TemporaryDirectory from tempfile import TemporaryDirectory
@ -425,9 +424,11 @@ def load_from_torchvision(filename, map_location=None):
""" """
model_urls = get_torchvision_models() model_urls = get_torchvision_models()
if filename.startswith('modelzoo://'): if filename.startswith('modelzoo://'):
warnings.warn( print_log(
'The URL scheme of "modelzoo://" is deprecated, please ' 'The URL scheme of "modelzoo://" is deprecated, please '
'use "torchvision://" instead', DeprecationWarning) 'use "torchvision://" instead',
logger='current',
level=logging.WARNING)
model_name = filename[11:] model_name = filename[11:]
else: else:
model_name = filename[14:] model_name = filename[14:]
@ -459,10 +460,11 @@ def load_from_openmmlab(filename, map_location=None):
deprecated_urls = get_deprecated_model_names() deprecated_urls = get_deprecated_model_names()
if model_name in deprecated_urls: if model_name in deprecated_urls:
warnings.warn( print_log(
f'{prefix_str}{model_name} is deprecated in favor ' f'{prefix_str}{model_name} is deprecated in favor '
f'of {prefix_str}{deprecated_urls[model_name]}', f'of {prefix_str}{deprecated_urls[model_name]}',
DeprecationWarning) logger='current',
level=logging.WARNING)
model_name = deprecated_urls[model_name] model_name = deprecated_urls[model_name]
model_url = model_urls[model_name] model_url = model_urls[model_name]
# check if is url # check if is url
@ -715,9 +717,11 @@ def save_checkpoint(checkpoint,
New in v0.2.0. New in v0.2.0.
""" """
if file_client_args is not None: if file_client_args is not None:
warnings.warn( print_log(
'"file_client_args" will be deprecated in future. ' '"file_client_args" will be deprecated in future. '
'Please use "backend_args" instead', DeprecationWarning) 'Please use "backend_args" instead',
logger='current',
level=logging.WARNING)
if backend_args is not None: if backend_args is not None:
raise ValueError( raise ValueError(
'"file_client_args" and "backend_args" cannot be set ' '"file_client_args" and "backend_args" cannot be set '

34
mmengine/runner/loops.py
View File

@ -1,13 +1,14 @@
# Copyright (c) OpenMMLab. All rights reserved. # Copyright (c) OpenMMLab. All rights reserved.
import bisect import bisect
import logging
import time import time
import warnings
from typing import Dict, List, Optional, Sequence, Tuple, Union from typing import Dict, List, Optional, Sequence, Tuple, Union
import torch import torch
from torch.utils.data import DataLoader from torch.utils.data import DataLoader
from mmengine.evaluator import Evaluator from mmengine.evaluator import Evaluator
from mmengine.logging import print_log
from mmengine.registry import LOOPS from mmengine.registry import LOOPS
from .amp import autocast from .amp import autocast
from .base_loop import BaseLoop from .base_loop import BaseLoop
@ -56,10 +57,12 @@ class EpochBasedTrainLoop(BaseLoop):
self.runner.visualizer.dataset_meta = \ self.runner.visualizer.dataset_meta = \
self.dataloader.dataset.metainfo self.dataloader.dataset.metainfo
else: else:
warnings.warn( print_log(
f'Dataset {self.dataloader.dataset.__class__.__name__} has no ' f'Dataset {self.dataloader.dataset.__class__.__name__} has no '
'metainfo. ``dataset_meta`` in visualizer will be ' 'metainfo. ``dataset_meta`` in visualizer will be '
'None.') 'None.',
logger='current',
level=logging.WARNING)
self.dynamic_milestones, self.dynamic_intervals = \ self.dynamic_milestones, self.dynamic_intervals = \
calc_dynamic_intervals( calc_dynamic_intervals(
@ -160,11 +163,14 @@ class _InfiniteDataloaderIterator:
try: try:
data = next(self._iterator) data = next(self._iterator)
except StopIteration: except StopIteration:
warnings.warn('Reach the end of the dataloader, it will be ' print_log(
'Reach the end of the dataloader, it will be '
'restarted and continue to iterate. It is ' 'restarted and continue to iterate. It is '
'recommended to use ' 'recommended to use '
'`mmengine.dataset.InfiniteSampler` to enable the ' '`mmengine.dataset.InfiniteSampler` to enable the '
'dataloader to iterate infinitely.') 'dataloader to iterate infinitely.',
logger='current',
level=logging.WARNING)
self._epoch += 1 self._epoch += 1
if hasattr(self._dataloader, 'sampler') and hasattr( if hasattr(self._dataloader, 'sampler') and hasattr(
self._dataloader.sampler, 'set_epoch'): self._dataloader.sampler, 'set_epoch'):
@ -226,10 +232,12 @@ class IterBasedTrainLoop(BaseLoop):
self.runner.visualizer.dataset_meta = \ self.runner.visualizer.dataset_meta = \
self.dataloader.dataset.metainfo self.dataloader.dataset.metainfo
else: else:
warnings.warn( print_log(
f'Dataset {self.dataloader.dataset.__class__.__name__} has no ' f'Dataset {self.dataloader.dataset.__class__.__name__} has no '
'metainfo. ``dataset_meta`` in visualizer will be ' 'metainfo. ``dataset_meta`` in visualizer will be '
'None.') 'None.',
logger='current',
level=logging.WARNING)
# get the iterator of the dataloader # get the iterator of the dataloader
self.dataloader_iterator = _InfiniteDataloaderIterator(self.dataloader) self.dataloader_iterator = _InfiniteDataloaderIterator(self.dataloader)
@ -338,10 +346,12 @@ class ValLoop(BaseLoop):
self.runner.visualizer.dataset_meta = \ self.runner.visualizer.dataset_meta = \
self.dataloader.dataset.metainfo self.dataloader.dataset.metainfo
else: else:
warnings.warn( print_log(
f'Dataset {self.dataloader.dataset.__class__.__name__} has no ' f'Dataset {self.dataloader.dataset.__class__.__name__} has no '
'metainfo. ``dataset_meta`` in evaluator, metric and ' 'metainfo. ``dataset_meta`` in evaluator, metric and '
'visualizer will be None.') 'visualizer will be None.',
logger='current',
level=logging.WARNING)
self.fp16 = fp16 self.fp16 = fp16
def run(self) -> dict: def run(self) -> dict:
@ -408,10 +418,12 @@ class TestLoop(BaseLoop):
self.runner.visualizer.dataset_meta = \ self.runner.visualizer.dataset_meta = \
self.dataloader.dataset.metainfo self.dataloader.dataset.metainfo
else: else:
warnings.warn( print_log(
f'Dataset {self.dataloader.dataset.__class__.__name__} has no ' f'Dataset {self.dataloader.dataset.__class__.__name__} has no '
'metainfo. ``dataset_meta`` in evaluator, metric and ' 'metainfo. ``dataset_meta`` in evaluator, metric and '
'visualizer will be None.') 'visualizer will be None.',
logger='current',
level=logging.WARNING)
self.fp16 = fp16 self.fp16 = fp16
def run(self) -> dict: def run(self) -> dict:

20
mmengine/runner/runner.py
View File

@ -1387,8 +1387,11 @@ class Runner:
# `persistent_workers` requires pytorch version >= 1.7 # `persistent_workers` requires pytorch version >= 1.7
if ('persistent_workers' in dataloader_cfg if ('persistent_workers' in dataloader_cfg
and digit_version(TORCH_VERSION) < digit_version('1.7.0')): and digit_version(TORCH_VERSION) < digit_version('1.7.0')):
warnings.warn('`persistent_workers` is only available when ' print_log(
'pytorch version >= 1.7') '`persistent_workers` is only available when '
'pytorch version >= 1.7',
logger='current',
level=logging.WARNING)
dataloader_cfg.pop('persistent_workers') dataloader_cfg.pop('persistent_workers')
# The default behavior of `collat_fn` in dataloader is to # The default behavior of `collat_fn` in dataloader is to
@ -1956,10 +1959,13 @@ class Runner:
current_seed = self._randomness_cfg.get('seed') current_seed = self._randomness_cfg.get('seed')
if resumed_seed is not None and resumed_seed != current_seed: if resumed_seed is not None and resumed_seed != current_seed:
if current_seed is not None: if current_seed is not None:
warnings.warn(f'The value of random seed in the ' print_log(
f'The value of random seed in the '
f'checkpoint "{resumed_seed}" is ' f'checkpoint "{resumed_seed}" is '
f'different from the value in ' f'different from the value in '
f'`randomness` config "{current_seed}"') f'`randomness` config "{current_seed}"',
logger='current',
level=logging.WARNING)
self._randomness_cfg.update(seed=resumed_seed) self._randomness_cfg.update(seed=resumed_seed)
self.set_randomness(**self._randomness_cfg) self.set_randomness(**self._randomness_cfg)
@ -1970,11 +1976,13 @@ class Runner:
# np.ndarray, which cannot be directly judged as equal or not, # np.ndarray, which cannot be directly judged as equal or not,
# therefore we just compared their dumped results. # therefore we just compared their dumped results.
if pickle.dumps(resumed_dataset_meta) != pickle.dumps(dataset_meta): if pickle.dumps(resumed_dataset_meta) != pickle.dumps(dataset_meta):
warnings.warn( print_log(
'The dataset metainfo from the resumed checkpoint is ' 'The dataset metainfo from the resumed checkpoint is '
'different from the current training dataset, please ' 'different from the current training dataset, please '
'check the correctness of the checkpoint or the training ' 'check the correctness of the checkpoint or the training '
'dataset.') 'dataset.',
logger='current',
level=logging.WARNING)
self.message_hub.load_state_dict(checkpoint['message_hub']) self.message_hub.load_state_dict(checkpoint['message_hub'])

10
mmengine/visualization/visualizer.py
View File

@ -6,6 +6,8 @@ from typing import TYPE_CHECKING, Dict, List, Optional, Sequence, Tuple, Union
if TYPE_CHECKING: if TYPE_CHECKING:
from matplotlib.font_manager import FontProperties from matplotlib.font_manager import FontProperties
import logging
import cv2 import cv2
import numpy as np import numpy as np
import torch import torch
@ -13,6 +15,7 @@ import torch.nn.functional as F
from mmengine.config import Config from mmengine.config import Config
from mmengine.dist import master_only from mmengine.dist import master_only
from mmengine.logging import print_log
from mmengine.registry import VISBACKENDS, VISUALIZERS from mmengine.registry import VISBACKENDS, VISUALIZERS
from mmengine.structures import BaseDataElement from mmengine.structures import BaseDataElement
from mmengine.utils import ManagerMixin from mmengine.utils import ManagerMixin
@ -163,8 +166,11 @@ class Visualizer(ManagerMixin):
self._vis_backends: Union[Dict, Dict[str, 'BaseVisBackend']] = dict() self._vis_backends: Union[Dict, Dict[str, 'BaseVisBackend']] = dict()
if save_dir is None: if save_dir is None:
warnings.warn('`Visualizer` backend is not initialized ' print_log(
'because save_dir is None.') '`Visualizer` backend is not initialized '
'because save_dir is None.',
logger='current',
level=logging.WARNING)
elif vis_backends is not None: elif vis_backends is not None:
assert len(vis_backends) > 0, 'empty list' assert len(vis_backends) > 0, 'empty list'
names = [ names = [

10
tests/test_evaluator/test_evaluator.py
View File

@ -8,6 +8,7 @@ import numpy as np
import torch import torch
from mmengine.evaluator import BaseMetric, Evaluator, get_metric_value from mmengine.evaluator import BaseMetric, Evaluator, get_metric_value
from mmengine.logging import MMLogger
from mmengine.registry import METRICS from mmengine.registry import METRICS
from mmengine.structures import BaseDataElement from mmengine.structures import BaseDataElement
@ -110,7 +111,8 @@ class TestEvaluator(TestCase):
# Test empty results # Test empty results
cfg = dict(type='ToyMetric', dummy_metrics=dict(accuracy=1.0)) cfg = dict(type='ToyMetric', dummy_metrics=dict(accuracy=1.0))
evaluator = Evaluator(cfg) evaluator = Evaluator(cfg)
with self.assertWarnsRegex(UserWarning, 'got empty `self.results`.'): # Warning should be raised if the results are empty
with self.assertLogs(MMLogger.get_current_instance(), level='WARNING'):
evaluator.evaluate(0) evaluator.evaluate(0)
def test_composed_metrics(self): def test_composed_metrics(self):
@ -185,8 +187,10 @@ class TestEvaluator(TestCase):
def test_prefix(self): def test_prefix(self):
cfg = dict(type='NonPrefixedMetric') cfg = dict(type='NonPrefixedMetric')
with self.assertWarnsRegex(UserWarning, 'The prefix is not set'): logger = MMLogger.get_current_instance()
_ = Evaluator(cfg) # Warning should be raised if prefix is not set.
with self.assertLogs(logger, 'WARNING'):
Evaluator(cfg)
def test_get_metric_value(self): def test_get_metric_value(self):

9
tests/test_hooks/test_checkpoint_hook.py
View File

@ -76,10 +76,11 @@ class TestCheckpointHook:
def test_init(self, tmp_path): def test_init(self, tmp_path):
# Test file_client_args and backend_args # Test file_client_args and backend_args
with pytest.warns( # TODO: Refactor this test case
DeprecationWarning, # with pytest.warns(
match='"file_client_args" will be deprecated in future'): # DeprecationWarning,
CheckpointHook(file_client_args={'backend': 'disk'}) # match='"file_client_args" will be deprecated in future'):
# CheckpointHook(file_client_args={'backend': 'disk'})
with pytest.raises( with pytest.raises(
ValueError, ValueError,

11
tests/test_hooks/test_logger_hook.py
View File

@ -29,11 +29,12 @@ class TestLoggerHook:
LoggerHook(file_client_args=dict(enable_mc=True)) LoggerHook(file_client_args=dict(enable_mc=True))
# test `file_client_args` and `backend_args` # test `file_client_args` and `backend_args`
with pytest.warns( # TODO Refine this unit test
DeprecationWarning, # with pytest.warns(
match='"file_client_args" will be deprecated in future'): # DeprecationWarning,
logger_hook = LoggerHook( # match='"file_client_args" will be deprecated in future'):
out_dir='tmp.txt', file_client_args={'backend': 'disk'}) # logger_hook = LoggerHook(
# out_dir='tmp.txt', file_client_args={'backend': 'disk'})
with pytest.raises( with pytest.raises(
ValueError, ValueError,

4
tests/test_hub/test_hub.py
View File

@ -38,8 +38,8 @@ def test_get_config():
# Test load mmpose # Test load mmpose
get_config( get_config(
'mmpose::face/2d_kpt_sview_rgb_img/deeppose/wflw/res50_wflw_256x256' 'mmpose::face_2d_keypoint/topdown_heatmap/wflw/td-hm_hrnetv2-w18_8xb64-60e_wflw-256x256.py' # noqa E501
'.py') )
@pytest.mark.skipif( @pytest.mark.skipif(

7
tests/test_model/test_averaged_model.py
View File

@ -4,6 +4,7 @@ from unittest import TestCase
import torch import torch
from mmengine.logging import MMLogger
from mmengine.model import (ExponentialMovingAverage, MomentumAnnealingEMA, from mmengine.model import (ExponentialMovingAverage, MomentumAnnealingEMA,
StochasticWeightAverage) StochasticWeightAverage)
from mmengine.testing import assert_allclose from mmengine.testing import assert_allclose
@ -94,9 +95,9 @@ class TestAveragedModel(TestCase):
torch.nn.Conv2d(1, 5, kernel_size=3), torch.nn.Linear(5, 10)) torch.nn.Conv2d(1, 5, kernel_size=3), torch.nn.Linear(5, 10))
ExponentialMovingAverage(model, momentum=3) ExponentialMovingAverage(model, momentum=3)
with self.assertWarnsRegex( # Warning should be raised if the value of momentum in EMA is
Warning, # a large number
'The value of momentum in EMA is usually a small number'): with self.assertLogs(MMLogger.get_current_instance(), level='WARNING'):
model = torch.nn.Sequential( model = torch.nn.Sequential(
torch.nn.Conv2d(1, 5, kernel_size=3), torch.nn.Linear(5, 10)) torch.nn.Conv2d(1, 5, kernel_size=3), torch.nn.Linear(5, 10))
ExponentialMovingAverage(model, momentum=0.9) ExponentialMovingAverage(model, momentum=0.9)

15
tests/test_optim/test_optimizer/test_optimizer.py
View File

@ -10,6 +10,7 @@ import torch.nn as nn
from torch.distributed.rpc import is_available from torch.distributed.rpc import is_available
from mmengine.dist import get_rank from mmengine.dist import get_rank
from mmengine.logging import MMLogger
from mmengine.optim import (OPTIM_WRAPPER_CONSTRUCTORS, OPTIMIZERS, from mmengine.optim import (OPTIM_WRAPPER_CONSTRUCTORS, OPTIMIZERS,
DefaultOptimWrapperConstructor, OptimWrapper, DefaultOptimWrapperConstructor, OptimWrapper,
build_optim_wrapper) build_optim_wrapper)
@ -592,10 +593,9 @@ class TestBuilder(TestCase):
optim_constructor = DefaultOptimWrapperConstructor( optim_constructor = DefaultOptimWrapperConstructor(
optim_wrapper_cfg, paramwise_cfg) optim_wrapper_cfg, paramwise_cfg)
self.assertWarnsRegex( with self.assertLogs(MMLogger.get_current_instance(), level='WARNING'):
Warning, # Warning should be raised since conv3.0 is a duplicate param.
'conv3.0 is duplicate. It is skipped since bypass_duplicate=True', optim_constructor(model)
lambda: optim_constructor(model))
optim_wrapper = optim_constructor(model) optim_wrapper = optim_constructor(model)
model_parameters = list(model.parameters()) model_parameters = list(model.parameters())
num_params = 14 if MMCV_FULL_AVAILABLE else 11 num_params = 14 if MMCV_FULL_AVAILABLE else 11
@ -607,10 +607,9 @@ class TestBuilder(TestCase):
# test DefaultOptimWrapperConstructor when the params in shared # test DefaultOptimWrapperConstructor when the params in shared
# modules do not require grad # modules do not require grad
model.conv1[0].requires_grad_(False) model.conv1[0].requires_grad_(False)
self.assertWarnsRegex( with self.assertLogs(MMLogger.get_current_instance(), level='WARNING'):
Warning, # Warning should be raised since conv3.0 is a duplicate param.
'conv3.0 is duplicate. It is skipped since bypass_duplicate=True', optim_constructor(model)
lambda: optim_constructor(model))
optim_wrapper = optim_constructor(model) optim_wrapper = optim_constructor(model)
model_parameters = list(model.parameters()) model_parameters = list(model.parameters())
num_params = 14 if MMCV_FULL_AVAILABLE else 11 num_params = 14 if MMCV_FULL_AVAILABLE else 11

6
tests/test_registry/test_registry_utils.py
View File

@ -4,6 +4,7 @@ import os.path as osp
from tempfile import TemporaryDirectory from tempfile import TemporaryDirectory
from unittest import TestCase, skipIf from unittest import TestCase, skipIf
from mmengine.logging import MMLogger
from mmengine.registry import (DefaultScope, Registry, from mmengine.registry import (DefaultScope, Registry,
count_registered_modules, init_default_scope, count_registered_modules, init_default_scope,
root, traverse_registry_tree) root, traverse_registry_tree)
@ -75,6 +76,7 @@ class TestUtils(TestCase):
# init default scope when another scope is init # init default scope when another scope is init
name = f'test-{datetime.datetime.now()}' name = f'test-{datetime.datetime.now()}'
DefaultScope.get_instance(name, scope_name='test') DefaultScope.get_instance(name, scope_name='test')
with self.assertWarnsRegex( # Warning should be raised since the current
Warning, 'The current default scope "test" is not "mmdet"'): # default scope is not 'mmdet'
with self.assertLogs(MMLogger.get_current_instance(), level='WARNING'):
init_default_scope('mmdet') init_default_scope('mmdet')

17
tests/test_runner/test_runner.py
View File

@ -1478,10 +1478,8 @@ class TestRunner(TestCase):
cfg.train_cfg = dict( cfg.train_cfg = dict(
by_epoch=False, max_iters=12, val_interval=4, val_begin=4) by_epoch=False, max_iters=12, val_interval=4, val_begin=4)
runner = Runner.from_cfg(cfg) runner = Runner.from_cfg(cfg)
with self.assertWarnsRegex( # Warning should be raised since the sampler is not InfiniteSampler.
Warning, with self.assertLogs(MMLogger.get_current_instance(), level='WARNING'):
'Reach the end of the dataloader, it will be restarted and '
'continue to iterate.'):
runner.train() runner.train()
assert isinstance(runner.train_loop, IterBasedTrainLoop) assert isinstance(runner.train_loop, IterBasedTrainLoop)
@ -2073,11 +2071,8 @@ class TestRunner(TestCase):
# ckpt_modified['meta']['seed'] = 123 # ckpt_modified['meta']['seed'] = 123
path_modified = osp.join(self.temp_dir, 'modified.pth') path_modified = osp.join(self.temp_dir, 'modified.pth')
torch.save(ckpt_modified, path_modified) torch.save(ckpt_modified, path_modified)
with self.assertWarnsRegex( # Warning should be raised since dataset_meta is not matched
Warning, 'The dataset metainfo from the resumed checkpoint is ' with self.assertLogs(MMLogger.get_current_instance(), level='WARNING'):
'different from the current training dataset, please '
'check the correctness of the checkpoint or the training '
'dataset.'):
runner.resume(path_modified) runner.resume(path_modified)
# 1.3.3 test resume with unmatched seed # 1.3.3 test resume with unmatched seed
@ -2085,8 +2080,8 @@ class TestRunner(TestCase):
ckpt_modified['meta']['seed'] = 123 ckpt_modified['meta']['seed'] = 123
path_modified = osp.join(self.temp_dir, 'modified.pth') path_modified = osp.join(self.temp_dir, 'modified.pth')
torch.save(ckpt_modified, path_modified) torch.save(ckpt_modified, path_modified)
with self.assertWarnsRegex( # Warning should be raised since seed is not matched
Warning, 'The value of random seed in the checkpoint'): with self.assertLogs(MMLogger.get_current_instance(), level='WARNING'):
runner.resume(path_modified) runner.resume(path_modified)
# 1.3.3 test resume with no seed and dataset meta # 1.3.3 test resume with no seed and dataset meta

7
tests/test_visualizer/test_visualizer.py
View File

@ -10,6 +10,7 @@ import torch
import torch.nn as nn import torch.nn as nn
from mmengine import VISBACKENDS, Config from mmengine import VISBACKENDS, Config
from mmengine.logging import MMLogger
from mmengine.visualization import Visualizer from mmengine.visualization import Visualizer
@ -68,10 +69,8 @@ class TestVisualizer(TestCase):
visualizer.get_image() visualizer.get_image()
# test save_dir # test save_dir
with pytest.warns( # Warning should be raised since no backend is initialized.
Warning, with self.assertLogs(MMLogger.get_current_instance(), level='WARNING'):
match='`Visualizer` backend is not initialized '
'because save_dir is None.'):
Visualizer() Visualizer()
visualizer = Visualizer( visualizer = Visualizer(