def test_import_bare(pg_repo_local):
# Check import without checking anything out, just by manipulating metadata and running LQs against
# source images.
# Create a new schema and import 'fruits'
OUTPUT.init()
# Make sure the existing table is preserved.
OUTPUT.run_sql("CREATE TABLE sentinel (key INTEGER)")
OUTPUT.commit()
pg_repo_local.uncheckout()
OUTPUT.uncheckout()
OUTPUT.import_tables(
tables=["imported_fruits"],
source_repository=pg_repo_local,
image_hash=pg_repo_local.images["latest"].image_hash,
source_tables=["SELECT * FROM fruits WHERE fruit_id = 1"],
parent_hash=OUTPUT.images["latest"].image_hash,
do_checkout=False,
table_queries=[True],
)
assert OUTPUT.head is None
assert pg_repo_local.head is None
assert sorted(OUTPUT.images["latest"].get_tables()) == [
"imported_fruits", "sentinel"
]
assert list(OUTPUT.images["latest"].get_table("imported_fruits").query(
columns=["name"], quals=[])) == [{
"name": "apple"
}]
def test_import_multiple_tables(pg_repo_local):
OUTPUT.init()
head = OUTPUT.head
OUTPUT.import_tables(tables=[],
source_repository=pg_repo_local,
source_tables=[])
for table_name in ["fruits", "vegetables"]:
assert OUTPUT.run_sql("SELECT * FROM %s" %
table_name) == pg_repo_local.run_sql(
"SELECT * FROM %s" % table_name)
new_head = OUTPUT.head
assert new_head != head
assert new_head.parent_id == head.image_hash
def test_import_preserves_pending_changes(pg_repo_local):
OUTPUT.init()
OUTPUT.run_sql("""CREATE TABLE test (id integer, name varchar);
INSERT INTO test VALUES (1, 'test')""")
head = OUTPUT.commit()
OUTPUT.run_sql("INSERT INTO test VALUES (2, 'test2')")
changes = get_engine().get_pending_changes(OUTPUT.to_schema(), "test")
OUTPUT.import_tables(tables=["imported_fruits"],
source_repository=pg_repo_local,
source_tables=["fruits"])
assert OUTPUT.head.parent_id == head.image_hash
assert changes == OUTPUT.engine.get_pending_changes(
OUTPUT.to_schema(), "test")
def test_import_basic(pg_repo_local):
# Create a new schema and import 'fruits' from the mounted PG table.
OUTPUT.init()
head = OUTPUT.head
OUTPUT.import_tables(tables=["imported_fruits"],
source_repository=pg_repo_local,
source_tables=["fruits"])
assert OUTPUT.run_sql("SELECT * FROM imported_fruits"
) == pg_repo_local.run_sql("SELECT * FROM fruits")
new_head = OUTPUT.head
assert new_head != head
assert new_head.parent_id == head.image_hash
def test_import_preserves_existing_tables(pg_repo_local):
# Create a new schema and import 'fruits' from the mounted PG table.
head = _setup_dataset()
OUTPUT.import_tables(tables=["imported_fruits"],
source_repository=pg_repo_local,
source_tables=["fruits"])
new_head = OUTPUT.head
head.checkout()
assert OUTPUT.engine.table_exists(OUTPUT.to_schema(), "test")
assert not OUTPUT.engine.table_exists(OUTPUT.to_schema(),
"imported_fruits")
new_head.checkout()
assert OUTPUT.engine.table_exists(OUTPUT.to_schema(), "test")
assert OUTPUT.engine.table_exists(OUTPUT.to_schema(), "imported_fruits")
def test_import_query_reuses_hash(pg_repo_local):
OUTPUT.init()
base = OUTPUT.head
# Run two imports: one importing all rows from `fruits` (will reuse the original `fruits` object),
# one importing just the first row (new hash, won't be reused).
ih_v1 = OUTPUT.import_tables(
source_repository=pg_repo_local,
source_tables=[
"SELECT * FROM fruits", "SELECT * FROM fruits WHERE fruit_id = 1"
],
tables=["fruits_all", "fruits_one"],
do_checkout=False,
table_queries=[True, True],
)
v1 = OUTPUT.images.by_hash(ih_v1)
assert v1.get_table("fruits_all").objects == pg_repo_local.head.get_table(
"fruits").objects
assert (len(OUTPUT.objects.get_all_objects()) == 3
) # Original fruits and vegetables + the 1-row import
# Run the same set of imports again: this time both query results already exist and will be reused.
base.checkout()
ih_v2 = OUTPUT.import_tables(
source_repository=pg_repo_local,
source_tables=[
"SELECT * FROM fruits", "SELECT * FROM fruits WHERE fruit_id = 1"
],
tables=["fruits_all", "fruits_one"],
do_checkout=False,
table_queries=[True, True],
)
v2 = OUTPUT.images.by_hash(ih_v2)
assert v2.get_table("fruits_all").objects == v1.get_table(
"fruits_all").objects
assert v2.get_table("fruits_one").objects == v1.get_table(
"fruits_one").objects
assert len(OUTPUT.objects.get_all_objects()
) == 3 # No new objects have been created.
