def test_convert_with_field_string_with_max_len(self, converter):
self._convert_and_assert_with_one_column(
converter,
{'name': self.col_name,
'type': ['null', 'string'],
'default': None,
AvroMetaDataKeys.MAX_LEN: 16},
SQLColumn(self.col_name, redshift_types.RedshiftVarChar(32))
)
def test_convert_bytes_field_with_exceeded_max_len(self, converter):
self._convert_and_assert_with_one_column(
converter,
{'name': self.col_name,
'type': ['null', 'bytes'],
'default': None,
AvroMetaDataKeys.MAX_LEN: 65536},
SQLColumn(self.col_name, redshift_types.RedshiftVarChar(65535))
)
def test_get_column_def_sql_with_string_type(self, migration):
column = SQLColumn(
'foo',
data_types.RedshiftVarChar(256),
attributes=[SQLAttribute.create_with_value('attr', '')],
aliases='bar')
expected = 'foo varchar(256) attr \'\''
actual = migration.get_column_def_sql(column)
assert expected == actual
def _convert_string_type(self, field, char_bytes=CHAR_BYTES):
"""Only supports char and varchar. If neither fix_len nor max_len
is specified, an exception is thrown.
"""
fix_len = field.props.get(AvroMetaDataKeys.FIX_LEN)
if fix_len:
# Columns with a CHAR data type only accept single-byte UTF-8
# characters. This means we can't support char columns since they
# will not be able to accept multibyte chars. We can instead use
# VARCHAR columns, which accept multibyte UTF-8 characters.
return redshift_data_types.RedshiftVarChar(fix_len)
max_len = field.props.get(AvroMetaDataKeys.MAX_LEN)
if max_len:
return redshift_data_types.RedshiftVarChar(
min(int(max_len) * char_bytes, self.MAX_VARCHAR_BYTES))
raise SchemaConversionException(
"Unable to convert `string` type without metadata {0} or {1}.".
format(AvroMetaDataKeys.FIX_LEN, AvroMetaDataKeys.MAX_LEN))
def test_convert_with_col_varchar(self, converter):
self._convert_and_assert_with_one_column(
converter,
RedshiftSQLColumn(
'col_varchar',
redshift_data_types.RedshiftVarChar(16)
),
{
'name': 'col_varchar',
'type': ['null', 'string'],
'default': None,
AvroMetaDataKeys.MAX_LEN: 16
}
)
def test_convert_with_field_enum(self, converter):
self._convert_and_assert_with_one_column(
converter,
{'name': self.col_name,
'type': {
'type': 'enum',
'name': self.col_name,
'symbols': ['1', '123', '12']}
},
SQLColumn(
self.col_name,
redshift_types.RedshiftVarChar(3),
is_nullable=False
),
)
def _convert_enum_type(self, field):
max_symbol_len = max(len(symbol) for symbol in field.symbols)
return redshift_data_types.RedshiftVarChar(
min(max_symbol_len, self.MAX_VARCHAR_BYTES))
def same_col(self):
return SQLColumn('same_col', data_types.RedshiftVarChar(64))