def create_table_from_describe(self, table_name, describe_rows):
primary_key_names = self._fetch_primary_key_names(table_name)
table = Table(table_name.lower())
for row in describe_rows:
col_name, data_type = row[:2]
col_type = self.parse_col_desc(data_type)
col = self.create_column(col_name, col_type)
col.is_primary_key = col_name in primary_key_names
table.add_col(col)
return table
def create_table_from_describe(self, table_name, describe_rows):
primary_key_names = self._fetch_primary_key_names(table_name)
table = Table(table_name.lower())
for row in describe_rows:
col_name, data_type = row[:2]
col_type = self.parse_col_desc(data_type)
col = self.create_column(col_name, col_type)
col.is_primary_key = col_name in primary_key_names
table.add_col(col)
return table
def FakeTable(name, fake_columns):
"""
Return a Table consisting of one or more FakeColumns. Because Columns are added via
method, we support nesting here instead.
"""
table = Table(name)
if not fake_columns:
raise Exception('You must supply at least one FakeColumn argument')
for fake_column in fake_columns:
table.add_col(fake_column)
return table
def FakeTable(name, fake_columns, storage_format='TEXTFILE'):
"""
Return a Table consisting of one or more FakeColumns. Because Columns are added via
method, we support nesting here instead.
"""
table = Table(name)
if not fake_columns:
raise Exception('You must supply at least one FakeColumn argument')
for fake_column in fake_columns:
table.add_col(fake_column)
table.storage_format = storage_format
return table
def test_table_model(self, cursor, hive_cursor):
table = Table("some_test_table")
cursor.drop_table(table.name, if_exists=True)
table.storage_format = 'textfile'
table.add_col(Column(table, "bigint_col", BigInt))
table.add_col(Column(table, "string_col", String))
cursor.create_table(table)
try:
other = hive_cursor.describe_table(table.name)
assert other.name == table.name
assert other.cols == table.cols
finally:
cursor.drop_table(table.name)
def test_table_model(self, cursor, hive_cursor):
table = Table("some_test_table")
cursor.drop_table(table.name, if_exists=True)
table.storage_format = 'textfile'
table.add_col(Column(table, "bigint_col", BigInt))
table.add_col(Column(table, "string_col", String))
cursor.create_table(table)
try:
other = hive_cursor.describe_table(table.name)
assert other.name == table.name
assert other.cols == table.cols
finally:
cursor.drop_table(table.name)
def test_hive_create_equality_only_joins():
"""
Tests that QueryGenerator produces a join condition with only equality functions if the
HiveProfile is used.
"""
class FakeHiveQueryProfile(HiveProfile):
"""
A fake QueryProfile that extends the HiveProfile, various weights are modified in
order to ensure that this test is deterministic.
"""
def choose_join_condition_count(self):
"""
There should be only one operator in the JOIN condition
"""
return 1
def choose_conjunct_disjunct_fill_ratio(self):
"""
There should be no AND or OR operators
"""
return 0
def choose_relational_func_fill_ratio(self):
"""
Force all operators to be relational
"""
return 1
query_generator = QueryGenerator(FakeHiveQueryProfile())
# Create two tables that have one joinable Column
right_table_expr_list = TableExprList()
right_table = Table("right_table")
right_table.add_col(Column("right_table", "right_col", Int))
right_table_expr_list.append(right_table)
left_table_expr_list = TableExprList()
left_table = Table("left_table")
left_table.add_col(Column("left_table", "left_col", Int))
left_table_expr_list.append(left_table)
# Validate the root predicate is an Equals funcs
assert isinstance(query_generator._create_relational_join_condition(
right_table_expr_list, left_table_expr_list), Equals)
def test_use_nested_width_subquery():
"""
Tests that setting DefaultProfile.use_nested_with to False works properly. Setting this
method to return False should prevent a WITH clause from being used inside a sub-query.
"""
class MockQueryProfile(DefaultProfile):
"""
A mock QueryProfile that sets use_nested_with to False and forces the
QueryGenerator to created nested queries.
"""
def __init__(self):
super(MockQueryProfile, self).__init__()
# Force the QueryGenerator to create nested queries
self._bounds['MAX_NESTED_QUERY_COUNT'] = (4, 4)
# Force the QueryGenerator to use WITH clauses whenever possible
self._probabilities['OPTIONAL_QUERY_CLAUSES']['WITH'] = 1
# Force the QueryGenerator to create inline views whenever possible
self._probabilities['MISC']['INLINE_VIEW'] = 1
def use_nested_with(self):
return False
mock_query_gen = QueryGenerator(MockQueryProfile())
# Create two tables
table_expr_list = TableExprList()
right_table = Table("right_table")
right_table.add_col(Column("right_table", "right_col", Int))
table_expr_list.append(right_table)
left_table = Table("left_table")
left_table.add_col(Column("left_table", "left_col", Int))
table_expr_list.append(left_table)
# Check that each nested_query doesn't have a with clause
for nested_query in mock_query_gen.create_query(table_expr_list).nested_queries:
assert nested_query.with_clause is None
def test_use_nested_width_subquery():
"""
Tests that setting DefaultProfile.use_nested_with to False works properly. Setting this
method to return False should prevent a WITH clause from being used inside a sub-query.
"""
class MockQueryProfile(DefaultProfile):
"""
A mock QueryProfile that sets use_nested_with to False and forces the
QueryGenerator to created nested queries.
"""
def __init__(self):
super(MockQueryProfile, self).__init__()
# Force the QueryGenerator to create nested queries
self._bounds['MAX_NESTED_QUERY_COUNT'] = (4, 4)
# Force the QueryGenerator to use WITH clauses whenever possible
self._probabilities['OPTIONAL_QUERY_CLAUSES']['WITH'] = 1
# Force the QueryGenerator to create inline views whenever possible
self._probabilities['MISC']['INLINE_VIEW'] = 1
def use_nested_with(self):
return False
mock_query_gen = QueryGenerator(MockQueryProfile())
# Create two tables
table_expr_list = TableExprList()
right_table = Table("right_table")
right_table.add_col(Column("right_table", "right_col", Int))
table_expr_list.append(right_table)
left_table = Table("left_table")
left_table.add_col(Column("left_table", "left_col", Int))
table_expr_list.append(left_table)
# Check that each nested_query doesn't have a with clause
for nested_query in mock_query_gen.generate_statement(table_expr_list).nested_queries:
assert nested_query.with_clause is None
