def coerce2unicode(obj):
"""
Takes in an object and coerces every string
contained in a list/dictionary/tuple key/value to unicode.
Returns a copy of obj with coerced entries.
Warning: this is a superficial deep copy with coercion;
only tuples, lists, dicts and string types are recreated,
with other references remaining intact.
"""
if isinstance(obj, six.string_types):
return ensure_text(obj)
elif isinstance(obj, tuple):
return tuple(ObjectWithSchema.coerce2unicode(item) for item in obj)
elif isinstance(obj, list):
return [ObjectWithSchema.coerce2unicode(item) for item in obj]
elif isinstance(obj, dict):
# Coerce both keys and values
return {
ObjectWithSchema.coerce2unicode(k):
ObjectWithSchema.coerce2unicode(v)
for k, v in obj.items()
}
else:
# obj is something else - return as is
return obj
def text_string(obj):
"""
Voluptuous validator that expects text strings and coerces Python 2 string
literals to unicode.
"""
if isinstance(obj, six.string_types):
return ensure_text(obj)
raise Invalid('expected str (or unicode)')
def test_ensure_text():
UNICODE_EMOJI = six.u("\U0001F600")
BINARY_EMOJI = b"\xf0\x9f\x98\x80"
converted_unicode = ensure_text(UNICODE_EMOJI,
encoding='utf-8',
errors='strict')
converted_binary = ensure_text(BINARY_EMOJI,
encoding="utf-8",
errors='strict')
if six.PY2:
# PY2: unicode -> unicode
assert converted_unicode == UNICODE_EMOJI and isinstance(
converted_unicode, unicode)
# PY2: str -> unicode
assert converted_binary == UNICODE_EMOJI and isinstance(
converted_unicode, unicode)
else:
# PY3: str -> str
assert converted_unicode == UNICODE_EMOJI and isinstance(
converted_unicode, str)
# PY3: bytes -> str
assert converted_binary == UNICODE_EMOJI and isinstance(
converted_unicode, str)
def test_ensure_text_raises_error():
msg = "not expecting type '{int}'".format(int=int)
with raises(TypeError, match=msg):
ensure_text(5)