def safe_disambiguation_iteration():
from invenio.bibauthorid_tortoise import tortoise
from invenio.bibauthorid_rabbit import rabbit
from invenio.bibauthorid_personid_maintenance import check_author_paper_associations \
, duplicated_tortoise_results_exist \
, repair_author_paper_associations
if not check_author_paper_associations():
rabbit([])
repair_author_paper_associations()
rabbit([])
assert check_author_paper_associations()
tortoise()
assert duplicated_tortoise_results_exist()
def safe_disambiguation_iteration():
from invenio.bibauthorid_tortoise import tortoise
from invenio.bibauthorid_rabbit import rabbit
from invenio.bibauthorid_personid_maintenance import check_author_paper_associations \
, duplicated_tortoise_results_exist \
, repair_author_paper_associations
if not check_author_paper_associations():
rabbit([])
repair_author_paper_associations()
rabbit([])
assert check_author_paper_associations()
tortoise()
assert duplicated_tortoise_results_exist()
def run_tortoise(from_scratch,
last_names_thresholds=None,
single_threaded=False):
_prepare_tortoise_cache()
from invenio.bibauthorid_tortoise import tortoise, \
tortoise_from_scratch, tortoise_last_name, tortoise_last_names
if single_threaded and last_names_thresholds:
for last_name, threshold in last_names_thresholds.items():
tortoise_last_name(last_name,
wedge_threshold=threshold,
from_mark=from_scratch)
elif last_names_thresholds:
names_with_args = list()
for last_name, threshold in last_names_thresholds.items():
kwargs = dict()
if from_scratch:
kwargs['from_mark'] = from_scratch
else:
kwargs['pure'] = from_scratch
if threshold:
args = (last_name, threshold)
else:
args = (last_name, )
names_with_args.append((args, kwargs))
tortoise_last_names(names_with_args)
elif from_scratch:
tortoise_from_scratch()
else:
start_time = get_db_time()
tortoise_db_name = 'tortoise'
last_run = get_user_logs(userinfo=tortoise_db_name,
only_most_recent=True)
if last_run:
modified = get_modified_papers_since(last_run[0][2])
else:
modified = []
tortoise(modified)
insert_user_log(tortoise_db_name,
'-1',
'',
'',
'',
timestamp=start_time)
def safe_disambiguation_iteration():
from invenio.bibauthorid_tortoise import tortoise
from invenio.bibauthorid_rabbit import rabbit
from invenio.bibauthorid_personid_maintenance import check_personid_papers \
, check_results \
, repair_personid
if not check_personid_papers():
rabbit([])
repair_personid()
rabbit([])
assert check_personid_papers()
tortoise()
assert check_results()
def safe_disambiguation_iteration():
from invenio.bibauthorid_tortoise import tortoise
from invenio.bibauthorid_rabbit import rabbit
from invenio.bibauthorid_personid_maintenance import check_personid_papers \
, check_results \
, repair_personid
if not check_personid_papers():
rabbit([])
repair_personid()
rabbit([])
assert check_personid_papers()
tortoise()
assert check_results()
def run_tortoise(from_scratch):
from invenio.bibauthorid_tortoise import tortoise, tortoise_from_scratch
if from_scratch:
tortoise_from_scratch()
else:
start_time = get_sql_time()
tortoise_db_name = 'tortoise'
last_run = get_user_log(userinfo=tortoise_db_name, only_most_recent=True)
if last_run:
modified = get_recently_modified_record_ids(last_run[0][2])
else:
modified = []
tortoise(modified)
insert_user_log(tortoise_db_name, '-1', '', '', '', timestamp=start_time)
def test_accuracy():
from invenio.bibauthorid_tortoise import tortoise
from invenio.bibauthorid_rabbit import rabbit
from invenio.bibauthorid_personid_maintenance import check_author_paper_associations \
, duplicated_tortoise_results_exist \
, repair_author_paper_associations
from invenio.bibauthorid_merge import matched_claims
if not check_author_paper_associations():
rabbit([])
repair_author_paper_associations()
rabbit([])
assert check_author_paper_associations()
tortoise(pure=True)
assert duplicated_tortoise_results_exist()
return matched_claims()
def run_tortoise(from_scratch):
from invenio.bibauthorid_tortoise import tortoise, tortoise_from_scratch
if from_scratch:
tortoise_from_scratch()
else:
start_time = get_db_time()
tortoise_db_name = 'tortoise'
last_run = get_user_logs(userinfo=tortoise_db_name,
only_most_recent=True)
if last_run:
modified = get_modified_papers_since(last_run[0][2])
else:
modified = []
tortoise(modified)
insert_user_log(tortoise_db_name, '-1', '', '', '', timestamp=start_time)
def test_accuracy():
from invenio.bibauthorid_tortoise import tortoise
from invenio.bibauthorid_rabbit import rabbit
from invenio.bibauthorid_personid_maintenance import check_personid_papers \
, check_results \
, repair_personid
from invenio.bibauthorid_merge import matched_claims
if not check_personid_papers():
rabbit([])
repair_personid()
rabbit([])
assert check_personid_papers()
tortoise(pure=True)
assert check_results()
return matched_claims()
def test_accuracy():
from invenio.bibauthorid_tortoise import tortoise
from invenio.bibauthorid_rabbit import rabbit
from invenio.bibauthorid_personid_maintenance import check_personid_papers \
, check_results \
, repair_personid
from invenio.bibauthorid_merge import matched_claims
if not check_personid_papers():
rabbit([])
repair_personid()
rabbit([])
assert check_personid_papers()
tortoise(pure=True)
assert check_results()
return matched_claims()
def test_accuracy():
from invenio.bibauthorid_tortoise import tortoise
from invenio.bibauthorid_rabbit import rabbit
from invenio.bibauthorid_personid_maintenance import check_author_paper_associations \
, duplicated_tortoise_results_exist \
, repair_author_paper_associations
from invenio.bibauthorid_merge import matched_claims
if not check_author_paper_associations():
rabbit([])
repair_author_paper_associations()
rabbit([])
assert check_author_paper_associations()
tortoise(pure=True)
assert duplicated_tortoise_results_exist()
return matched_claims()