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_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 dolt_sql_to_csv(db: dolt.Dolt,
sql: str,
filename: str,
load_args: dict = None):
if load_args is None:
load_args = {}
db.sql(query=sql, result_file=filename, result_format="csv", **load_args)
return
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 __call__(self, db: dolt.Dolt):
starting_head = db.head
starting_branch = db.active_branch
try:
self.checkout(db)
yield db
finally:
self.merge(db, starting_branch)
db.checkout(starting_branch, error=False)
def checkout(self, db: dolt.Dolt):
res = db.sql(
f"select * from dolt_branches where name = '{self.branch}'",
result_format="csv")
if len(res) == 0:
db.sql(f"select dolt_checkout('-b', '{self.branch}')",
result_format="csv")
else:
db.sql(f"select dolt_checkout('{self.branch}')",
result_format="csv")
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_config_local(init_empty_test_repo):
repo = init_empty_test_repo
current_global_config = Dolt.config_global(list=True)
test_username, test_email = "test_user", "test_email"
repo.config_local(add=True, name="user.name", value=test_username)
repo.config_local(add=True, name="user.email", value=test_email)
local_config = repo.config_local(list=True)
global_config = Dolt.config_global(list=True)
assert (local_config["user.name"] == test_username
and local_config["user.email"] == test_email)
assert global_config["user.name"] == current_global_config["user.name"]
assert global_config["user.email"] == current_global_config["user.email"]
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_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 doltdb():
db_path = os.path.join(os.path.dirname(__file__), "foo")
try:
db = Dolt.init(db_path)
df_v1 = pd.DataFrame({"A": [1, 1, 1], "B": [1, 1, 1]})
write_pandas(
dolt=db,
table="bar",
df=df_v1.reset_index(),
primary_key=["index"],
import_mode="create",
)
db.add("bar")
db.commit("Initialize bar")
df_v2 = pd.DataFrame({"A": [2, 2, 2], "B": [2, 2, 2]})
write_pandas(
dolt=db,
table="bar",
df=df_v2.reset_index(),
primary_key=["index"],
import_mode="create",
)
db.add("bar")
db.commit("Edit bar")
yield db_path
finally:
if os.path.exists(db_path):
shutil.rmtree(db_path)
def dolt_import_csv(db: dolt.Dolt,
tablename: str,
filename: str,
save_args: dict = None):
mode = "-c"
tables = db.ls()
for t in tables:
if t.name == tablename:
mode = "-r"
break
filetype = "--file-type csv"
imp = ["table", "import", filetype, mode, tablename]
if save_args and "primary_key" in save_args:
pks = ",".join(save_args["primary_key"])
imp.append(f"--pk {pks}")
imp.append(filename)
db.execute(imp)
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 doltdb():
db_path = os.path.join(os.path.dirname(__file__), "foo")
try:
db = Dolt.init(db_path)
db.sql("create table t1 (a bigint primary key, b bigint, c bigint)")
db.sql("insert into t1 values (1,1,1), (2,2,2)")
db.sql("select dolt_add('t1')")
db.sql("select dolt_commit('-m', 'initialize t1')")
db.sql("insert into t1 values (3,3,3)")
db.sql("select dolt_add('t1')")
db.sql("select dolt_commit('-m', 'edit t1')")
yield db_path
finally:
if os.path.exists(db_path):
shutil.rmtree(db_path)
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 _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 dolt_export_csv(db: dolt.Dolt,
tablename: str,
filename: str,
load_args: dict = None):
exp = ["table", "export", "-f", "--file-type", "csv", tablename, filename]
db.execute(exp)
def test_get_clone_dir_no_remote(tmp_path):
new_dir = os.path.join(tmp_path, "new_dir")
res = Dolt._get_clone_dir(new_dir)
assert new_dir == res
def _init_helper(path: str, ext: str = None):
repo_path, repo_data_dir = get_repo_path_tmp_path(path, ext)
return Dolt.init(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 checkout(self, db: dolt.Dolt):
# branch must exist
db.checkout(branch=self.branch, error=False)
def test_repo_name_trailing_slash(tmp_path):
repo_path, repo_data_dir = get_repo_path_tmp_path(tmp_path)
assert Dolt.init(str(repo_path) +
"/").repo_name == "test_repo_name_trailing_slash0"
shutil.rmtree(repo_data_dir)
def test_get_clone_dir_remote_only(tmp_path):
new_dir = os.path.join(os.getcwd(), "remote")
res = Dolt._get_clone_dir(remote_url="some/remote")
assert new_dir == res
def test_get_clone_dir_new_dir_only(tmp_path):
res = Dolt._get_clone_dir("new_dir")
assert "new_dir" == res
def test_bad_repo_path(tmp_path):
bad_repo_path = tmp_path
with pytest.raises(ValueError):
Dolt(bad_repo_path)
def test_working(doltdb):
db = Dolt(doltdb)
assert db.head != db.working
def test_init(tmp_path):
repo_path, repo_data_dir = get_repo_path_tmp_path(tmp_path)
assert not os.path.exists(repo_data_dir)
Dolt.init(repo_path)
assert os.path.exists(repo_data_dir)
shutil.rmtree(repo_data_dir)
def test_get_clone_dir_new_dir_and_remote(tmp_path):
new_dir = os.path.join("foo/bar", "remote")
res = Dolt._get_clone_dir(new_dir="foo/bar", remote_url="some/remote")
assert new_dir == res
