def init_processors(self):
if not hasattr(self.config, "processors"):
return
extra_params = {"data_dir": self.config.data_dir}
for processor_key, processor_params in self.config.processors.items():
if not processor_params:
continue
reg_key = f"{self._dataset_name}_{processor_key}"
reg_check = registry.get(reg_key, no_warning=True)
if reg_check is None:
processor_object = Processor(processor_params, **extra_params)
setattr(self, processor_key, processor_object)
registry.register(reg_key, processor_object)
else:
setattr(self, processor_key, reg_check)
def build_processors(processors_config: mmf_typings.DictConfig,
registry_key: str = None,
*args,
**kwargs) -> ProcessorDict:
"""Given a processor config, builds the processors present and returns back
a dict containing processors mapped to keys as per the config
Args:
processors_config (mmf_typings.DictConfig): OmegaConf DictConfig describing
the parameters and type of each processor passed here
registry_key (str, optional): If passed, function would look into registry for
this particular key and return it back. .format with processor_key will
be called on this string. Defaults to None.
Returns:
ProcessorDict: Dictionary containing key to
processor mapping
"""
from mmf.datasets.processors.processors import Processor
processor_dict = {}
for processor_key, processor_params in processors_config.items():
if not processor_params:
continue
processor_instance = None
if registry_key is not None:
full_key = registry_key.format(processor_key)
processor_instance = registry.get(full_key, no_warning=True)
if processor_instance is None:
processor_instance = Processor(processor_params, *args, **kwargs)
# We don't register back here as in case of hub interface, we
# want the processors to be instantiate every time. BaseDataset
# can register at its own end
processor_dict[processor_key] = processor_instance
return processor_dict
def test_processor_class_None(self):
config = OmegaConf.create({"type": "UndefinedType"})
with self.assertRaises(ValueError):
Processor(config)