def test_database_insert():
A = Alphabet('ACGT')
S = A.parse('AACT', name='foo')
db = DB(':memory:', A)
db.initialize()
attrs = {'key': 'value'}
rec = db.insert(S, source_file='source.fa', source_pos=10, attrs=attrs)
assert isinstance(rec.id, int)
assert rec.content_id == S.content_id
assert rec.source_pos == 10
assert rec.source_file == 'source.fa'
assert 'key' in rec.attrs and rec.attrs['key'] == 'value', \
'attributes must be populated correctly'
with db.connection() as conn:
cursor = conn.cursor()
cursor.execute('SELECT content_id FROM sequence WHERE id = ?',
(rec.id,))
# NOTE for some reason if we just say next(cursor) == ...
# the cursor remains open after the context is over (which should
# not happen as per docs). This leads to BusyError further down.
assert cursor.fetchall() == [(S.content_id,)], \
'content identifier is properly populated'
# add a second sequence
T = A.parse('GCTG', name='bar')
new_rec = db.insert(T)
assert new_rec.id != rec.id, 'new ids are assigned to new sequences'
with db.connection() as conn:
cursor = conn.cursor()
cursor.execute('SELECT content_id FROM sequence WHERE id = ?',
(new_rec.id,))
assert next(cursor) == (T.content_id,), \
'correct id must be populated'
def test_database_basic():
A = Alphabet('ACGT')
db = DB(':memory:', A)
db.initialize()
db.initialize() # should be able to call it twice
with db.connection() as conn:
# a sequence table should be created
conn.cursor().execute('SELECT * FROM sequence LIMIT 1;')
with pytest.raises(AssertionError):
DB('/cannot/possibly/exist/directory/', A)
def test_database_overwrite():
A = Alphabet('ACGT')
S = A.parse('AACT', name='foo')
db = DB(':memory:', A)
db.initialize()
db.insert(S, source_file='old_source.fa')
db.insert(S, source_file='new_source.fa')
with db.connection() as conn:
cursor = conn.cursor()
cursor.execute(
'SELECT source_file FROM sequence WHERE content_id = ?',
(S.content_id,)
)
res = [x[0] for x in cursor]
assert len(res) == 1 and res[0] == 'old_source.fa', \
'Sequences with observed content id should be ignored'