def test_series_shallow_pull(size, direction, shallow):
label = "LABEL"
local_repo = Repo()
remote_repo = Repo()
local_coll = local_repo.create_collection(schema, "a_collection")
series = local_coll / label
series.write({"timestamp": arange(size), "value": arange(size)})
series.write({"timestamp": arange(size), "value": arange(size) * 2})
if direction == "pull":
remote_repo.pull(local_repo, shallow=shallow)
else:
local_repo.push(remote_repo, shallow=shallow)
remote_clc = remote_repo / "a_collection"
assert len(remote_clc.changelog.log()) == (1 if shallow else 2)
remote_series = remote_clc / label
expected = series.frame()
assert remote_series.frame() == expected
def test_pull(threaded, large):
c_label = "a_collection"
s_label = "a_series"
remote_repo = Repo()
remote_coll = remote_repo.create_collection(schema, c_label)
rseries = remote_coll / s_label
# Test support of both small dataset (where data is embedded in
# commits) and large one (arrays are save on their own)
N = 100_000 if large else 10
for i in range(10):
# Create 10 series of size N
rseries.write({
"timestamp": range(i, i + N),
"value": range(i + 100, i + 100 + N),
})
nb_items = len(remote_repo.pod.ls())
if large:
assert nb_items > 2
else:
# for small arrays we have only two folder (one for the repo
# registry one for the collection)
assert nb_items == 2
expected = rseries.frame()
# Test pull
local_repo = Repo()
local_coll = local_repo.create_collection(schema, c_label)
local_coll.pull(remote_coll)
lseries = local_coll / s_label
assert lseries.frame() == expected
# Test push
other_repo = Repo()
other_coll = other_repo.create_collection(schema, c_label)
remote_coll.push(other_coll)
oseries = other_coll / s_label
assert oseries.frame() == expected
# Test with existing series
local_repo = Repo()
local_coll = local_repo.create_collection(schema, c_label)
local_coll.pull(remote_coll)
lseries = (
other_repo.create_collection(schema, c_label, raise_if_exists=False) /
s_label)
assert oseries.frame() == expected
# Test with existing series with existing data
local_repo = Repo()
local_coll = local_repo.create_collection(schema, c_label)
lseries = local_coll / s_label
frm = Frame(
schema,
{
"timestamp": range(0, 20),
"value": range(0, 20),
},
)
lseries.write(frm)
local_coll.pull(remote_coll)
assert lseries.frame() == frm
# Test with existing series with other schema
local_repo = Repo()
other_schema = Schema(timestamp="int*", value="int")
local_coll = local_repo.create_collection(other_schema, c_label)
lseries = local_coll / s_label
with pytest.raises(ValueError):
local_repo.pull(remote_repo)