def test_csv_zipfile(pytestconfig, treedb):
suffix = '-memory' if treedb.engine.file is None else ''
suffix += pytestconfig.option.file_engine_tag
path = treedb.csv_zipfile()
assert path.name == f'treedb{suffix}.zip'
assert_file_size_between(path, 1, 20)
def test_dump_sql(pytestconfig, treedb):
suffix = '-memory' if treedb.engine.file is None else ''
suffix += pytestconfig.option.file_engine_tag
path = treedb.dump_sql()
assert path.name == f'treedb{suffix}.sql.gz'
assert_file_size_between(path, 1, 20)
def test_write_raw_csv(pytestconfig, treedb_raw):
expected = RAW_CSV_SHA256.get(pytestconfig.option.glottolog_tag)
suffix = '-memory' if treedb_raw.engine.file is None else ''
suffix += pytestconfig.option.file_engine_tag
path = treedb_raw.raw.write_raw_csv()
assert path.name == f'treedb{suffix}.raw.csv.gz'
assert_file_size_between(path, 1, 100)
if expected is None:
pass
else:
shasum = treedb_raw.sha256sum(path)
assert shasum == expected
def test_write_csv(pytestconfig, treedb):
expected = QUERY_HASH.get(pytestconfig.option.glottolog_tag)
suffix = '-memory' if treedb.engine.file is None else ''
suffix += pytestconfig.option.file_engine_tag
path = treedb.write_csv()
assert path.name == f'treedb{suffix}.query.csv'
assert_file_size_between(path, 1, 30)
if expected is None:
pass
else:
shasum = treedb.sha256sum(path)
assert shasum == expected
def test_backup(treedb):
path = treedb.engine.file_with_suffix('.backup.sqlite3')
engine = treedb.backup(path.name)
assert engine.url.database == path.name
assert_file_size_between(path, 10, 200)
# SQLiteEngineProxy
engine = treedb.engine.__class__(engine, future=treedb.engine.future)
assert engine.file_exists()
assert engine.file_mtime()
assert engine.file_size() == path.stat().st_size
assert 10 <= engine.file_size(as_megabytes=True) <= 200
assert len(engine.file_sha256()) == 64
def test_write_json_lines(pytestconfig, capsys, treedb, suffix, n=100):
name_suffix = '-memory' if treedb.engine.file is None else ''
name_suffix += pytestconfig.option.file_engine_tag
args = ([f'treedb{name_suffix}.languoids{suffix}']
if suffix != 'jsonl.gz' else [])
filepath, _ = treedb.write_languoids(*args)
assert filepath.name == f'treedb{name_suffix}.languoids{suffix}'
assert_file_size_between(filepath, 1, 200)
if filepath.name.endswith('.jsonl'):
with filepath.open(encoding='utf-8') as f:
for line in get_assert_head(f, n=n):
item = json.loads(line)
assert_nonempty_dict(item)
path = item['__path__']
assert isinstance(path, list)
assert all(isinstance(p, str) for p in path)
assert path
assert all(path)
languoid = item['languoid']
assert_nonempty_dict(languoid)
for key in ('id', 'level', 'name'):
assert_nonempty_string(languoid[key])
assert languoid['parent_id'] or languoid['parent_id'] is None
assert languoid['level'] in ('family', 'language', 'dialect')
out, err = capsys.readouterr()
assert not out
assert not err
expected_checksum = CHECKSUM.get(pytestconfig.option.glottolog_tag)
if expected_checksum is not None:
assert treedb.sha256sum(filepath) == expected_checksum