def test_strings_from_regex_digit():
generator, size = strings_from_regex('[0-9]')
assert size == 10
assert list(generator) == [
'0', '1', '2', '3', '4', '5', '6', '7', '8', '9'
]
def _get_primary_keys(self, table_name, num_rows):
"""Return the primary key and amount of values for the requested table.
Args:
table_name (str):
Name of the table to get the primary keys from.
num_rows (str):
Number of ``primary_keys`` to generate.
Returns:
tuple (str, pandas.Series):
primary key name and primary key values. If the table has no primary
key, ``(None, None)`` is returned.
Raises:
ValueError:
If the ``metadata`` contains invalid types or subtypes, or if
there are not enough primary keys left on any of the generators.
NotImplementedError:
If the primary key subtype is a ``datetime``.
"""
primary_key = self.metadata.get_primary_key(table_name)
field = self.metadata.get_fields(table_name)[primary_key]
generator = self._primary_key_generators.get(table_name)
if generator is None:
if field['type'] != 'id':
raise ValueError('Only columns with type `id` can be primary keys')
subtype = field.get('subtype', 'integer')
if subtype == 'integer':
generator = itertools.count()
remaining = np.inf
elif subtype == 'string':
regex = field.get('regex', r'^[a-zA-Z]+$')
generator, remaining = utils.strings_from_regex(regex)
elif subtype == 'datetime':
raise NotImplementedError('Datetime ids are not yet supported')
else:
raise ValueError('Only `integer` or `string` id columns are supported.')
self._primary_key_generators[table_name] = generator
self._remaining_primary_keys[table_name] = remaining
else:
remaining = self._remaining_primary_keys[table_name]
if remaining < num_rows:
raise ValueError(
'Not enough unique values for primary key of table {}'
' to generate {} samples.'.format(table_name, num_rows)
)
self._remaining_primary_keys[table_name] -= num_rows
primary_key_values = pd.Series([x for i, x in zip(range(num_rows), generator)])
return primary_key_values
def test_strings_from_regex_repeat_digit():
generator, size = strings_from_regex(r'\d{1,3}')
assert size == 1110
strings = list(generator)
assert strings[0] == '0'
assert strings[-1] == '999'
def _make_ids(cls, field_metadata, length):
field_subtype = field_metadata.get('subtype', 'integer')
if field_subtype == 'string':
regex = field_metadata.get('regex', '[a-zA-Z]+')
generator, max_size = strings_from_regex(regex)
if max_size < length:
raise ValueError(
('Unable to generate {} unique values for regex {}, the '
'maximum number of unique values is {}.').format(
length, regex, max_size))
values = [next(generator) for _ in range(length)]
return pd.Series(list(values)[:length])
else:
return pd.Series(np.arange(length))
def test_strings_from_regex_literal():
generator, size = strings_from_regex('abcd')
assert size == 1
assert list(generator) == ['abcd']
def test_strings_from_regex_repeat_literal():
generator, size = strings_from_regex('a{1,3}')
assert size == 3
assert list(generator) == ['a', 'aa', 'aaa']