def test_copy(self):
docs = [dict(_id=str(i)) for i in range(10)]
self.c.update(docs)
c2 = Collection(self.c)
self.assertEqual(len(self.c), len(c2))
for doc in c2:
self.assertEqual(len(self.c.find(doc)), 1)
def test_copy(self):
docs = [dict(_id=str(i)) for i in range(10)]
self.c.update(docs)
c2 = Collection(self.c)
assert len(self.c) == len(c2)
for doc in c2:
assert len(self.c.find(doc)) == 1
def test_init_with_list_with_and_without_ids(self):
docs = [{'a': i} for i in range(10)]
for i, doc in enumerate(islice(docs, 5)):
doc.setdefault('_id', str(i))
self.c = Collection(docs)
self.assertEqual(len(self.c), len(docs))
for doc in docs:
self.assertIn(doc['_id'], self.c)
def test_init_with_list_with_and_without_ids(self):
docs = [{"a": i} for i in range(10)]
for i, doc in enumerate(islice(docs, 5)):
doc.setdefault("_id", str(i))
self.c = Collection(docs)
assert len(self.c) == len(docs)
for doc in docs:
assert doc["_id"] in self.c
def test_compression(self):
# Fill with data
docs = [dict(_id=str(i)) for i in range(10)]
self.c.update(docs)
self.c.flush()
# Create uncompressed copy to compare to.
c2 = Collection(self.c)
c2.flush()
assert isinstance(self.c._file, io.BytesIO)
size_compressed = len(self.c._file.getvalue())
assert size_compressed > 0
size_uncompressed = len(c2._file.getvalue().encode("utf-8"))
assert size_uncompressed > 0
compresslevel = size_uncompressed / size_compressed
assert compresslevel > 1.0
def test_compression(self):
# Fill with data
docs = [dict(_id=str(i)) for i in range(10)]
self.c.update(docs)
self.c.flush()
# Create uncompressed copy to compare to.
c2 = Collection(self.c)
c2.flush()
self.assertIsInstance(self.c._file, io.BytesIO)
size_compressed = len(self.c._file.getvalue())
self.assertGreater(size_compressed, 0)
size_uncompressed = len(c2._file.getvalue().encode('utf-8'))
self.assertGreater(size_uncompressed, 0)
compresslevel = size_uncompressed / size_compressed
self.assertGreater(compresslevel, 1.0)
def setUp(self):
self.c = Collection()
def test_init_with_non_serializable(self):
docs = [dict(a=array.array("f", [1, 2, 3])) for i in range(10)]
with pytest.raises(TypeError):
self.c = Collection(docs)
def test_init_with_list_without_ids(self):
docs = [{"a": i} for i in range(10)]
self.c = Collection(docs)
assert len(self.c) == len(docs)
for doc in docs:
assert doc["_id"] in self.c
def test_init_with_list_with_ids_non_sequential(self):
docs = [{"a": i, "_id": "{:032d}".format(i**3)} for i in range(10)]
self.c = Collection(docs)
assert len(self.c) == len(docs)
for doc in docs:
assert doc["_id"] in self.c
def test_init_with_list_with_ids_sequential(self):
docs = [{"a": i, "_id": str(i)} for i in range(10)]
self.c = Collection(docs)
assert len(self.c) == len(docs)
for doc in docs:
assert doc["_id"] in self.c
def test_buffer_size(self):
docs = [{"a": i, "_id": str(i)} for i in range(10)]
self.c = Collection(docs)
assert len(self.c._file.getvalue()) == 0
self.c.flush()
assert len(self.c._file.getvalue()) > 0
def __init__(self) :
from signac import Collection
self.info = {}
self.coll = Collection()
def test_init_with_list_without_ids(self):
docs = [{'a': i} for i in range(10)]
self.c = Collection(docs)
self.assertEqual(len(self.c), len(docs))
for doc in docs:
self.assertIn(doc['_id'], self.c)
def test_init_with_list_with_ids_non_sequential(self):
docs = [{'a': i, '_id': '{:032d}'.format(i**3)} for i in range(10)]
self.c = Collection(docs)
self.assertEqual(len(self.c), len(docs))
for doc in docs:
self.assertIn(doc['_id'], self.c)
def test_init_with_list_with_ids_sequential(self):
docs = [{'a': i, '_id': str(i)} for i in range(10)]
self.c = Collection(docs)
self.assertEqual(len(self.c), len(docs))
for doc in docs:
self.assertIn(doc['_id'], self.c)
def test_buffer_size(self):
docs = [{'a': i, '_id': str(i)} for i in range(10)]
self.c = Collection(docs)
self.assertEqual(len(self.c._file.getvalue()), 0)
self.c.flush()
self.assertGreater(len(self.c._file.getvalue()), 0)
def get_index_collection(self):
c = Collection()
return Mock(spec=c, wraps=c)
def test_init_with_list_with_ids_non_sequential(self):
docs = [{'a': i, '_id': '{:032d}'.format(i**3)} for i in range(10)]
self.c = Collection(docs)
assert len(self.c) == len(docs)
for doc in docs:
assert doc['_id'] in self.c
