def single_evaluation(extractor_cls, results, dataset_type, dataset_name):
logger.info('started evaluating extractor %s', extractor_cls.NAME)
results.set_extractor(extractor_cls.SLUG)
storage = LocalResultStorage(dataset_name, extractor_cls)
loader = LocalDatasetLoader(dataset_name)
for doc in loader:
logger.debug('doc: %s', doc.id)
format_clean = from_document_factory(doc, slug=dataset_type)
try:
result_string = storage.fetch_result(doc)
except DataError:
logger.info('no stored result for %s at %s extractor', doc.id,
extractor_cls.NAME)
continue
else:
format_result = extractor_cls.formatted_result(result_string)
evaluator = TextOnlyEvaluator(retrieved=format_result,
relevant=format_clean,
id=doc.id)
results.add_result(evaluator.get_eval_results())
def single_evaluation(extractor_cls, results, dataset_type, dataset_name):
logger.info('started evaluating extractor %s', extractor_cls.NAME)
results.set_extractor(extractor_cls.SLUG)
storage = LocalResultStorage(dataset_name, extractor_cls)
loader = LocalDatasetLoader(dataset_name)
for doc in loader:
logger.debug('doc: %s', doc.id)
format_clean = from_document_factory(doc, slug = dataset_type)
try:
result_string = storage.fetch_result(doc)
except DataError:
logger.info('no stored result for %s at %s extractor',
doc.id, extractor_cls.NAME)
continue
else:
format_result = extractor_cls.formatted_result(result_string)
evaluator = TextOnlyEvaluator(
retrieved = format_result,
relevant = format_clean,
id = doc.id)
results.add_result(evaluator.get_eval_results())
def local_extract(dataset_name, extractor_slug, timeout, retry_failed, skip_existing):
# init storage and loader
ex = get_extractor_cls(extractor_slug)
failed_slug = extractor_slug if retry_failed else None
skip_slug = extractor_slug if skip_existing else None
loader = LocalDatasetLoader(dataset_name,
load_failed=failed_slug,
skip_existing=skip_slug)
storage = LocalResultStorage(dataset_name, ex)
logger.info('started extracting content from %s dataset using %s', dataset_name, ex.NAME)
for doc in loader:
storage.push_result(doc)
if timeout:
time.sleep(timeout)
storage.dump_summary()
logger.info('finished with %s dataset', dataset_name)
