def test_auditdt_cm_diff(active_run, dolt_audit1, doltdb):
doltdb = Dolt(doltdb)
logs = list(doltdb.log(2).keys())
with DoltDT(run=active_run, audit=dolt_audit1) as dolt:
df = dolt.sql("SELECT * FROM `bar`", as_key="akey")
diff = dolt.diff(from_commit=logs[1], to_commit=logs[0], table="bar")
def test_reset(doltdb):
db = Dolt(doltdb)
db.reset()
db.reset(hard=True)
db.reset(soft=True)
db.reset(tables="t1")
db.reset(tables=["t1"])
def dolt_audit1(doltdb):
db = Dolt(doltdb)
lg = db.log()
commit = lg.popitem(last=False)[0]
yield {
"actions": {
"bar": {
"key": "bar",
"config_id": "dd9f1f38-6802-4657-b869-602dde993180",
"pathspec": "VersioningDemo/1611853111934656/start/1",
"table_name": "bar",
"kind": "read",
"query": "SELECT * FROM `bar`",
"commit": commit,
"artifact_name": None,
"timestamp": 1611853112.794624,
}
},
"configs": {
"dd9f1f38-6802-4657-b869-602dde993180": {
"id": "dd9f1f38-6802-4657-b869-602dde993180",
"database": doltdb,
"branch": "master",
"commit": commit,
"dolthub_remote": False,
"push_on_commit": False,
}
},
}
def test_reset_errors(doltdb):
db = Dolt(doltdb)
with pytest.raises(ValueError):
db.reset(hard=True, soft=True)
with pytest.raises(ValueError):
db.reset(tables="t1", hard=True)
with pytest.raises(ValueError):
db.reset(tables="t1", soft=True)
with pytest.raises(ValueError):
db.reset(tables={"t1": True})
def test_dolt_sql_errors(doltdb):
db = Dolt(doltdb)
with pytest.raises(ValueError):
db.sql(result_parser=lambda x: x, query=None)
with pytest.raises(ValueError):
db.sql(result_parser=2, query="select active_branch()")
with pytest.raises(ValueError):
db.sql(result_file="file.csv", query=None)
with pytest.raises(ValueError):
db.sql(result_format="csv", query=None)
def test_detached_head_cm(doltdb):
db = Dolt(doltdb)
commits = list(db.log().keys())
with detach_head(db, commits[1]):
sum1 = db.sql("select sum(a) as sum from t1", result_format="csv")[0]
with detach_head(db, commits[0]):
sum2 = db.sql("select sum(a) as sum from t1", result_format="csv")[0]
assert sum1["sum"] == "3"
assert sum2["sum"] == "6"
def test_branchdt_diff(inactive_run, dolt_config, doltdb):
doltdb = Dolt(doltdb)
logs = list(doltdb.log(2).keys())
dolt = DoltDT(config=dolt_config)
diff = dolt.diff(from_commit=logs[1], to_commit=logs[0], table="bar")
row = diff["bar"].iloc[0]
assert row.from_A == 1
assert row.from_B == 1
assert row.to_A == 2
assert row.to_B == 2
def test_custom_query_branch(active_run, dolt_config, doltdb):
doltdb = Dolt(doltdb)
logs = list(doltdb.log(2).keys())
dolt_config.commit = logs[1]
with DoltDT(run=active_run, config=dolt_config) as dolt:
df = dolt.sql("SELECT * FROM `bar` LIMIT 2", as_key="akey")
np.testing.assert_array_equal(df.A.values, [1, 1])
audit = active_run.dolt
assert "akey" in audit["actions"]
assert audit["actions"]["akey"]["kind"] == "read"
assert audit["actions"]["akey"]["query"] == "SELECT * FROM `bar` LIMIT 2"
def test_set_dolt_path_error(doltdb):
db = Dolt(doltdb)
set_dolt_path("dolt")
test_cmd = "show tables"
db.sql(test_cmd, result_format="csv")
try:
with pytest.raises(FileNotFoundError):
set_dolt_path("notdolt")
from doltcli.utils import DOLT_PATH
assert DOLT_PATH == "notdolt"
db.sql(test_cmd, result_format="csv")
finally:
set_dolt_path("dolt")
def test_branchdt_cm_write(active_run, dolt_config, doltdb):
input_df = pd.DataFrame({"A": [2, 2, 2], "B": [2, 2, 2]})
with DoltDT(run=active_run, config=dolt_config) as dolt:
dolt.write(df=input_df, table_name="baz")
db = Dolt(doltdb)
output_df = read_pandas_sql(db, "SELECT * from `baz`")
np.testing.assert_array_equal(output_df.A.values, [2, 2, 2])
audit = active_run.dolt
print(audit["actions"]["baz"])
assert "baz" in audit["actions"]
assert audit["actions"]["baz"]["kind"] == "write"
assert audit["actions"]["baz"]["query"] == "SELECT * FROM `baz`"
def _get_db(self, config: DoltConfig):
if config.id in self._dbcache:
return self._dbcache[config.id]
# TODO: clone remote
try:
Dolt.init(repo_dir=config.database)
except DoltException as e:
pass
if ".dolt" not in os.listdir(config.database):
raise ValueError(
f"Passed a path {config.database} that is not a Dolt database directory"
)
doltdb = Dolt(repo_dir=config.database)
current_branch, branches = doltdb.branch()
logger.info(
f"Dolt database in {config.database} at branch {current_branch.name}, using branch {config.branch}"
)
if config.branch == current_branch.name:
pass
elif config.branch not in [branch.name for branch in branches]:
raise ValueError(f"Passed branch '{config.branch}' that does not exist")
else:
doltdb.checkout(config.branch, checkout_branch=False)
if not doltdb.status().is_clean:
raise Exception(
"DoltDT as context manager requires clean working set for transaction semantics"
)
if not config.commit:
config.commit = self._get_latest_commit_hash(doltdb)
self._dbcache[config.id] = doltdb
return doltdb
def test_working(doltdb):
db = Dolt(doltdb)
assert db.head != db.working
def test_bad_repo_path(tmp_path):
bad_repo_path = tmp_path
with pytest.raises(ValueError):
Dolt(bad_repo_path)
def test_clone_new_dir(tmp_path):
target = os.path.join(tmp_path, "state_age")
Dolt.clone("max-hoffman/state-age", new_dir=target)
db = Dolt(target)
assert db.head is not None
def test_checkout_non_existent_branch(doltdb):
repo = Dolt(doltdb)
repo.checkout("main")
