def test_read_from_dbhash_works(self):
dbh = DictDbHash({
"etag:http://boop": "blah",
"last_modified:http://boop": "flarg"
})
assert LastmodInfo.read_from_dbhash("http://boop", dbh) == LastmodInfo(
url="http://boop", etag="blah", last_modified="flarg")
assert LastmodInfo.read_from_dbhash("http://bar", dbh) == LastmodInfo(
"http://bar", None, None)
def test_read_from_dbhash_works(self):
dbh = DictDbHash({
'etag:http://boop': 'blah',
'last_modified:http://boop': 'flarg'
})
assert LastmodInfo.read_from_dbhash('http://boop', dbh) == LastmodInfo(
url="http://boop", etag="blah", last_modified="flarg")
assert LastmodInfo.read_from_dbhash('http://bar', dbh) == LastmodInfo(
"http://bar", None, None)
def test_reset_lastmod_works(db, capsys):
url = HPD_REG_URL
with load_dbhash() as dbhash:
info = LastmodInfo(url,
etag='blah',
last_modified='Tue, 01 Jan 2019 20:56:28 UTC')
info.write_to_dbhash(dbhash)
dbtool.main(['lastmod:reset', 'hpd_registrations'], DATABASE_URL)
with load_dbhash() as dbhash:
info = LastmodInfo.read_from_dbhash(url, dbhash)
assert info == LastmodInfo(url=url, etag=None, last_modified=None)
def list_lastmod(db_url: str, dataset_names: List[str]):
with psycopg2.connect(db_url) as conn:
dbhash = load_dataset.get_dbhash(conn)
for dataset in dataset_names:
print(f"For the dataset {dataset}:")
urls = load_dataset.get_urls_for_dataset(dataset)
for url in urls:
info = LastmodInfo.read_from_dbhash(url, dbhash)
if info.last_modified:
print(
f" The URL {url} was last modified on {info.last_modified}."
)
else:
print(
f" The URL {url} has no metadata about its last modification date."
)