def get_latest_available_date(model, test_corpus, bucket=EMMAA_BUCKET_NAME):
if not test_corpus:
logger.error('Test corpus is missing, cannot find latest date')
return
model_date = last_updated_date(model,
'model_stats',
extension='.json',
bucket=bucket)
test_date = last_updated_date(model,
'test_stats',
tests=test_corpus,
extension='.json',
bucket=bucket)
if model_date == test_date:
logger.info(f'Latest available date for {model} model and '
f'{test_corpus} is {model_date}.')
return model_date
min_date = min(model_date, test_date)
if is_available(model, test_corpus, min_date, bucket=bucket):
logger.info(f'Latest available date for {model} model and '
f'{test_corpus} is {min_date}.')
return min_date
min_date_obj = datetime.strptime(min_date, "%Y-%m-%d")
for day_count in range(1, 30):
earlier_date = min_date_obj - timedelta(days=day_count)
if is_available(model, test_corpus, earlier_date, bucket=bucket):
logger.info(f'Latest available date for {model} model and '
f'{test_corpus} is {earlier_date}.')
return earlier_date
logger.info(f'Could not find latest available date for {model} model '
f'and {test_corpus}.')
def test_run_model_tests_from_s3():
# Local imports are recommended when using moto
from emmaa.model_tests import run_model_tests_from_s3, ModelManager
from emmaa.model import last_updated_date
client = setup_bucket(add_tests=True, add_mm=True)
# There should not be any results
assert not last_updated_date('test',
'test_results',
tests='simple_tests',
extension='.json',
bucket=TEST_BUCKET_NAME)
mm = run_model_tests_from_s3('test',
'simple_tests',
upload_results=True,
bucket=TEST_BUCKET_NAME)
assert isinstance(mm, ModelManager)
# Results are saved now
assert last_updated_date('test',
'test_results',
tests='simple_tests',
extension='.json',
bucket=TEST_BUCKET_NAME)
def test_last_updated():
# Local imports are recommended when using moto
from emmaa.model import last_updated_date
client = setup_bucket(add_model=True,
add_results=True,
add_model_stats=True,
add_test_stats=True)
# Test for different file types
key_str = last_updated_date('test',
'model',
'datetime',
extension='.pkl',
bucket=TEST_BUCKET_NAME)
assert key_str
assert re.search(RE_DATETIMEFORMAT, key_str).group()
key_str = last_updated_date('test',
'test_results',
'datetime',
'simple_tests',
extension='.json',
bucket=TEST_BUCKET_NAME)
assert key_str
assert re.search(RE_DATETIMEFORMAT, key_str).group()
key_str = last_updated_date('test',
'test_stats',
'datetime',
'simple_tests',
extension='.json',
bucket=TEST_BUCKET_NAME)
assert key_str
assert re.search(RE_DATETIMEFORMAT, key_str).group()
key_str = last_updated_date('test',
'model_stats',
'datetime',
extension='.json',
bucket=TEST_BUCKET_NAME)
assert key_str
assert re.search(RE_DATETIMEFORMAT, key_str).group()
# Test for different date format
key_str = last_updated_date('test',
'model',
'date',
extension='.pkl',
bucket=TEST_BUCKET_NAME)
assert key_str
assert re.search(RE_DATEFORMAT, key_str).group()
# Test with wrong extension
key_str = last_updated_date('test',
'test_stats',
'datetime',
'simple_tests',
extension='.pkl',
bucket=TEST_BUCKET_NAME)
assert not key_str
def get_model_tests_page(model):
model_type = request.args.get('model_type')
test_hash = request.args.get('test_hash')
test_corpus = request.args.get('test_corpus')
if not test_corpus:
abort(Response('Test corpus has to be provided', 404))
date = request.args.get('date')
if model_type not in ALL_MODEL_TYPES:
abort(Response(f'Model type {model_type} does not exist', 404))
test_stats, file_key = get_model_stats(model,
'test',
tests=test_corpus,
date=date)
if not test_stats:
abort(Response(f'Data for {model} for {date} was not found', 404))
try:
current_test = \
test_stats['test_round_summary']['all_test_results'][test_hash]
except KeyError:
abort(Response(f'Result for this test does not exist for {date}', 404))
current_model_types = [
mt for mt in ALL_MODEL_TYPES if mt in test_stats['test_round_summary']
]
test = current_test["test"]
test_status, path_list = current_test[model_type]
correct, incorrect = _label_curations()
if isinstance(path_list, list):
for path in path_list:
for edge in path['edge_list']:
for stmt in edge['stmts']:
cur = ''
url = stmt[0]
if 'stmt_hash' in url:
stmt_hashes = parse.parse_qs(
parse.urlparse(url).query)['stmt_hash']
cur = _set_curation(stmt_hashes, correct, incorrect)
stmt.append(cur)
latest_date = get_latest_available_date(model, test_corpus)
prefix = f'stats/{model}/test_stats_{test_corpus}_'
cur_ix = find_index_of_s3_file(file_key, EMMAA_BUCKET_NAME, prefix)
if test_hash in test_stats['tests_delta']['applied_hashes_delta']['added']:
prev_date = None
elif (cur_ix + 1) < find_number_of_files_on_s3(EMMAA_BUCKET_NAME, prefix,
'.json'):
prev_date = last_updated_date(model,
'test_stats',
'date',
tests=test_corpus,
extension='.json',
n=(cur_ix + 1),
bucket=EMMAA_BUCKET_NAME)
else:
prev_date = None
if cur_ix > 0:
next_date = last_updated_date(model,
'test_stats',
'date',
tests=test_corpus,
extension='.json',
n=(cur_ix - 1),
bucket=EMMAA_BUCKET_NAME)
else:
next_date = None
return render_template('tests_template.html',
link_list=link_list,
model=model,
model_type=model_type,
all_model_types=current_model_types,
test_hash=test_hash,
test=test,
test_status=test_status,
path_list=path_list,
formatted_names=FORMATTED_TYPE_NAMES,
date=date,
latest_date=latest_date,
prev=prev_date,
next=next_date)