def validate_config(self, config_string):
self.validate_label_config(config_string)
if not hasattr(self, 'summary'):
return
# validate data columns consistency
fields_from_config = get_all_object_tag_names(config_string)
if not fields_from_config:
logger.debug(f'Data fields not found in labeling config')
return
fields_from_data = set(self.summary.common_data_columns)
fields_from_data.discard(settings.DATA_UNDEFINED_NAME)
if fields_from_data and not fields_from_config.issubset(fields_from_data):
different_fields = list(fields_from_config.difference(fields_from_data))
raise ValidationError(f'These fields are not present in the data: {",".join(different_fields)}')
# validate annotations consistency
annotations_from_config = set(get_all_control_tag_tuples(config_string))
if not annotations_from_config:
logger.debug(f'Annotation schema is not found in config')
return
annotations_from_data = set(self.summary.created_annotations)
if annotations_from_data and not annotations_from_data.issubset(annotations_from_config):
different_annotations = list(annotations_from_data.difference(annotations_from_config))
diff_str = []
for ann_tuple in different_annotations:
from_name, to_name, t = ann_tuple.split('|')
diff_str.append(
f'{self.summary.created_annotations[ann_tuple]} '
f'with from_name={from_name}, to_name={to_name}, type={t}')
diff_str = '\n'.join(diff_str)
raise ValidationError(f'Created annotations are incompatible with provided labeling schema, '
f'we found:\n{diff_str}')
# validate labels consistency
labels_from_config = get_all_labels(config_string)
created_labels = self.summary.created_labels
for control_tag_from_data, labels_from_data in created_labels.items():
# Check if labels created in annotations, and their control tag has been removed
if labels_from_data and control_tag_from_data not in labels_from_config:
raise ValidationError(
f'There are {sum(labels_from_data.values(), 0)} annotation(s) created with tag '
f'"{control_tag_from_data}", you can\'t remove it')
labels_from_config_by_tag = set(labels_from_config[control_tag_from_data])
if not set(labels_from_data).issubset(set(labels_from_config_by_tag)):
different_labels = list(set(labels_from_data).difference(labels_from_config_by_tag))
diff_str = '\n'.join(f'{l} ({labels_from_data[l]} annotations)' for l in different_labels)
raise ValidationError(f'These labels still exist in annotations:\n{diff_str}')
def validate_config(self, config_string, strict=False):
self.validate_label_config(config_string)
if not hasattr(self, 'summary'):
return
if self.num_tasks == 0:
logger.debug(f'Project {self} has no tasks: nothing to validate here. Ensure project summary is empty')
self.summary.reset()
return
# validate data columns consistency
fields_from_config = get_all_object_tag_names(config_string)
if not fields_from_config:
logger.debug(f'Data fields not found in labeling config')
return
fields_from_config = {field.split('[')[0] for field in fields_from_config} # Repeater tag support
fields_from_data = set(self.summary.common_data_columns)
fields_from_data.discard(settings.DATA_UNDEFINED_NAME)
if fields_from_data and not fields_from_config.issubset(fields_from_data):
different_fields = list(fields_from_config.difference(fields_from_data))
raise LabelStudioValidationErrorSentryIgnored(
f'These fields are not present in the data: {",".join(different_fields)}'
)
if self.num_annotations == 0:
logger.debug(
f'Project {self} has no annotations: nothing to validate here. '
f'Ensure annotations-related project summary is empty'
)
self.summary.reset(tasks_data_based=False)
return
# validate annotations consistency
annotations_from_config = set(get_all_control_tag_tuples(config_string))
if not annotations_from_config:
logger.debug(f'Annotation schema is not found in config')
return
annotations_from_data = set(self.summary.created_annotations)
if annotations_from_data and not annotations_from_data.issubset(annotations_from_config):
different_annotations = list(annotations_from_data.difference(annotations_from_config))
diff_str = []
for ann_tuple in different_annotations:
from_name, to_name, t = ann_tuple.split('|')
diff_str.append(
f'{self.summary.created_annotations[ann_tuple]} '
f'with from_name={from_name}, to_name={to_name}, type={t}'
)
diff_str = '\n'.join(diff_str)
raise LabelStudioValidationErrorSentryIgnored(
f'Created annotations are incompatible with provided labeling schema, we found:\n{diff_str}'
)
# validate labels consistency
labels_from_config, dynamic_label_from_config = get_all_labels(config_string)
created_labels = self.summary.created_labels
for control_tag_from_data, labels_from_data in created_labels.items():
# Check if labels created in annotations, and their control tag has been removed
if labels_from_data and ((control_tag_from_data not in labels_from_config) and (
control_tag_from_data not in dynamic_label_from_config)):
raise LabelStudioValidationErrorSentryIgnored(
f'There are {sum(labels_from_data.values(), 0)} annotation(s) created with tag '
f'"{control_tag_from_data}", you can\'t remove it'
)
labels_from_config_by_tag = set(labels_from_config[control_tag_from_data])
if not set(labels_from_data).issubset(set(labels_from_config_by_tag)):
different_labels = list(set(labels_from_data).difference(labels_from_config_by_tag))
diff_str = '\n'.join(f'{l} ({labels_from_data[l]} annotations)' for l in different_labels)
if (strict is True) and (control_tag_from_data not in dynamic_label_from_config):
raise LabelStudioValidationErrorSentryIgnored(
f'These labels still exist in annotations:\n{diff_str}')
else:
logger.warning(f'project_id={self.id} inconsistent labels in config and annotations: {diff_str}')
