Validation Errors in ckanext-usmetadata

[nickumia-reisys]

During the py3 upgrade of ckanext-dcat_usmetadata, a call to usmetadata was made passing a null object into one of the regex validators. The error happened on PY2+CKAN2.8, but I don't think that this is version specific, the same behavior would happen on PY3+CKAN2.9.

_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _

regex_candidate = <ckan.lib.navl.dictization_functions.Missing object at 0x7f980c22a8d0>

    def bureau_code_validator(regex_candidate):
        validator = re.compile(r'^\d{3}:\d{2}(\s*,\s*\d{3}:\d{2}\s*)*$')
>       if isinstance(validator.match(regex_candidate), type(re.match("", ""))):
E       TypeError: expected string or buffer

src/ckanext-usmetadata/ckanext/usmetadata/plugin/helper.py:26: TypeError

How to reproduce

This error was hit by running the tests on ckanext-dcat_usmetadata, but because of the py3 upgrade surrounding the extension, I'm not sure the tests can run in a reproducible way right now. To illustrate the issue, the following snippet was run,

$ python
Python 2.7.18 (default, Mar  8 2021, 13:02:45)
[GCC 9.3.0] on linux2
Type "help", "copyright", "credits" or "license" for more information.
>>> import re
>>> validator = re.compile(r'^\d{3}:\d{2}(\s*,\s*\d{3}:\d{2}\s*)*$')
>>> regex_candidate = None
>>> isinstance(validator.match(regex_candidate), type(re.match("", "")))
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
TypeError: expected string or buffer
>>> regex_candidate = ""
>>> isinstance(validator.match(regex_candidate), type(re.match("", "")))
False
>>>

Expected behavior

We would expect that the extension would give a normal validation error. If the parameter was missing, it would give no error, but if the parameter is of the wrong type, it should probably fail with a 'wrong type' error.

Actual behavior

The code exited with a runtime TypeError.

Sketch

Update the validation functions in ckanext/usmetadata/plugin/helper.py to at least check that the input is a str before running the regex validators.