def test_spark(self):
engine = "spark"
reader = SqlReader.from_connection(None, engine)
assert(isinstance(reader, SqlReader))
assert(isinstance(reader, SparkReader))
assert(isinstance(reader.compare, NameCompare))
assert(isinstance(reader.compare, SparkNameCompare))
assert(isinstance(reader.serializer, Serializer))
assert(isinstance(reader.serializer, SparkSerializer))
def test_presto(self):
engine = "presto"
reader = SqlReader.from_connection(None, engine)
assert(isinstance(reader, SqlReader))
assert(isinstance(reader, PrestoReader))
assert(isinstance(reader.compare, NameCompare))
assert(isinstance(reader.compare, PrestoNameCompare))
assert(isinstance(reader.serializer, Serializer))
assert(isinstance(reader.serializer, PrestoSerializer))
def test_sqlserver(self):
engine = "sqlserver"
reader = SqlReader.from_connection(None, engine)
assert(isinstance(reader, SqlReader))
assert(isinstance(reader, SqlServerReader))
assert(isinstance(reader.compare, NameCompare))
assert(isinstance(reader.compare, SqlServerNameCompare))
assert(isinstance(reader.serializer, Serializer))
assert(isinstance(reader.serializer, SqlServerSerializer))
def test_postgres(self):
engine = "postgres"
reader = SqlReader.from_connection(None, engine)
assert(isinstance(reader, SqlReader))
assert(isinstance(reader, PostgresReader))
assert(isinstance(reader.compare, NameCompare))
assert(isinstance(reader.compare, PostgresNameCompare))
assert(isinstance(reader.serializer, Serializer))
assert(isinstance(reader.serializer, PostgresSerializer))
def test_pandas(self):
engine = "pandas"
meta = Metadata.from_file(meta_path)
df = pd.read_csv(csv_path)
reader = SqlReader.from_connection(df, engine, metadata=meta)
assert(isinstance(reader, SqlReader))
assert(isinstance(reader, PandasReader))
assert(isinstance(reader.compare, NameCompare))
assert(isinstance(reader.compare, PandasNameCompare))
assert(isinstance(reader.serializer, Serializer))
assert(isinstance(reader.serializer, PandasSerializer))
def from_connection(cls, conn, *ignore, privacy, metadata, engine=None, **kwargs):
"""Create a private reader over an established SQL connection. If `engine` is not
passed in, the engine will be automatically detected.
:param conn: An established database connection. Can be pyodbc, psycopg2, SparkSession, Pandas DataFrame, or Presto.
:param privacy: A Privacy object with epsilon, delta, and other privacy properties. Keyword-only.
:param metadata: The metadata describing the database. `Metadata documentation is here <https://github.com/opendp/smartnoise-sdk/blob/new_opendp/sdk/Metadata.md>`_. Keyword-only.
:param engine: Optional keyword-only argument that can be used to specify engine-specific rules if automatic detection fails. This should only be necessary when using an uncommon database or middleware.
:returns: A `PrivateReader` object initialized to process queries against the supplied connection, using the supplied `Privacy` properties.
.. code-block:: python
privacy = Privacy(epsilon=1.0, delta=1/1000)
metadata = 'datasets/PUMS.yaml'
pums = pd.read_csv('datasets/PUMS.csv')
reader = PrivateReader.from_connection(pums, privacy=privacy, metadata=metadata)
"""
_reader = SqlReader.from_connection(conn, engine=engine, metadata=metadata, **kwargs)
return PrivateReader(_reader, metadata, privacy=privacy)
git_root_dir = subprocess.check_output(
"git rev-parse --show-toplevel".split(" ")).decode("utf-8").strip()
meta_path = os.path.join(git_root_dir, os.path.join("datasets",
"PUMS_pid.yaml"))
csv_path = os.path.join(git_root_dir, os.path.join("datasets", "PUMS_pid.csv"))
from snsql.xpath.parse import XPath
p = XPath()
meta = Metadata.from_file(meta_path)
pums = pd.read_csv(csv_path)
query = 'SELECT AVG(age) + 3, STD(age), VAR(age), SUM(age) / 10, COUNT(age) + 2 FROM PUMS.PUMS'
q = QueryParser(meta).query(query)
reader = SqlReader.from_connection(pums, "pandas", metadata=meta)
priv = PrivateReader(reader, meta, privacy=Privacy(epsilon=1.0))
subquery, root = priv._rewrite(query)
class TestXPathExecutionNoRewrite:
def test_all_root_descend(self):
path = '//*' # returns value
xx = p.parse(path)
res = xx.evaluate(q)
assert (len(res) > 40)
assert (str(xx) == path)
def test_all_with_condition(self):
path = '//*[@left]' # returns value
xx = p.parse(path)
def test_reserved(self):
if connection_case is not None and has_postgres:
reader = SqlReader.from_connection(connection_case, "postgres")
res = reader.execute('SELECT "select" FROM nametests')
def test_reserved(self):
if connection is not None and has_sqlserver:
reader = SqlReader.from_connection(connection, "postgres")
res = reader.execute('SELECT "SELECT" FROM nametests')