Allow data providers to validate multiple XML input files from one command line operation.

Specifically, this refers to the element of , p65 on the spec http://nvlpubs.nist.gov/nistpubs/SpecialPublications/NIST.SP.1500-100.pdf

For a element (xsi:type=ReportingUnit), shall be consistent with the element (ocd-id) provided for the .

Contests which are partisan primaries should use the PrimaryPartyIds field to link to the correct political party. For a NIST Election element of ElectionType primary, partisan-primary-open, or partisan-primary-closed, the Contests in that ContestCollection (type CandidateContest) should have a PrimartyPartyIds that is present and non-empty.

See $SUBJECT.

E.g. the following message is printed when only the Selection element is missing.

1528 Error messages found
1528 Schema Error messages
6254: Element 'BallotSelection': Missing child element(s). Expected is one of ( VoteCountsCollection, Selection ).

1. Add a country->url mapping file
2. Add command line option to select country

Add a version of the XSD that is easier to validate against but is still compatible with the original version.

Manually check that every field of type IDREF actually references a value that exists in a field of type ID.

We've seen several instances where a valid enumeration value is instead stored in the OtherType field. E.g.,

<GpUnit objectId="ru1" xsi:type="ReportingUnit">
  <Name>Virginia</Name>
  <Type>other</Type>
  <OtherType>state</Type>
</GpUnit>

Even though state is a valid ReportingUnitType, it has been represented in the OtherType field. The correct XML should be

<GpUnit objectId="ru1" xsi:type="ReportingUnit">
  <Name>Virginia</Name>
  <Type>state</Type>
</GpUnit>

This is an error since it can cause ingestion to fail.

It would be advantageous to possible future open source projects if we were able to refactor the bulk of the validation code into its own library. Right now it doesn't seem like we're too far from that, but I'd like to take the plunge.

Add a command-line flag to specify a comma-separated list of language codes to make sure are present in the various InternationalizedText fields. If this flag is not specified, use "en" as the default.

Each candidate should really have party data associated with them, even if the referenced Party object is "Unaffiliated", "Non-Partisan" or something similar.

If there are Candidate objects which don't show up in any contests, this could be a problem. Print messages at the WARNING level.

An example of violation would be, BallotMeasureSelection elements all have the label of "bmslabel".

The validator shall be able to identify BallotMeasureContest formatted/created as CandidateContest. For example, "For", "Against", "Yes", "No", or other values (for Ballot Name) indicates that these are not people (Candidates) and they are ballot measure instead.

Update the CONTRIBUTING document to include a review process similar to that here:

https://github.com/votinginfoproject/vip-specification/blob/release/CONTRIBUTING.md

Pre-election mode will require more complete information about the candidates, contests, etc, whereas the election results mode will check information about the VoteCounts objects and ensure that SubUnitsReporting and TotalSubUnits information is present.

For whatever reason, the spec currently allows coalitions to not have Party IDs. It should be an error if this is the case.

If a CandidateContest in a primary does have PrimaryPartyIds, a error is to be reported. However, because PrimaryPartyIds can be None, the error cannot reliably report the sourceline. Instead an unhanded exception will be thrown at line 673.

https://github.com/google/election_results_xml_validator/blob/c02b6fe1536b660ee70fd51cbc14d5d75f16f538/rules.py#L666-L673

The 2016/11/27 update to the pycountry package broke our validator.

16.11.27 (2016-11-27)
Documentation fix: iso639_1_code is not a valid key for languages any more. Fixes #13387, thanks to @jmitzka.

Improve the error messages surrounding OCD-IDs to distinguish between the following cases:

1. GpUnit doesn't have any external identifier
2. GpUnit has an ocd-id but it does not have a valid value
3. GpUnit has an external identifier but the the type is not set as "ocd-id" (could be "OCD-ID", in which case saying that that GpUnit doesn't have a valid ocd-id could be a bit confusing).

Clients which are using older versions of Python are using http/urllib/etc libraries which by default are using SSLv2 or SSLv3 in making HTTPS connections. This is leading to a variety of errors, including 'SNIMissingWarning' and 'InsecurePlatformWarning'.

There are similar errors when using the PyGithub module and it attempts to make a secure connection.

The validator should warn users if the names of contests within a file are not unique. We've seen several feeds where the contest name field is just "Congressional", which isn't particularly helpful.

Currently, validation fails if a default namespace is defined in the XML document.

When a namespace is defined, find fails to locate elements in the document, as element searches must include the namespace prefix.

Support default namespaces that are defined, for example, by including xmlns="http://www.votegis.com/schema/NISTV50.xsd" on the ElectionReport element.

Our validator depends on the python requests package, but does not do any sophisticated checking of SSL certs. If the version of Python or the requests package is sufficiently out-of-date, this can lead to the download failing. Because of the way the current code operates, a download failure still leads to country-us.csv being present, but empty. This means the whitelist of valid OCD-IDs is empty, and every check against the list returns false, regardless of whether the OCD-ID is actually valid.

This feature enables user to set/filter out specific logging levels on the command line - e.g. whether the validator will print out info and worse messages (info, warning, and error), warning and worse (warning and error), or just error.

Ensure that the BallotSelection elements in a CandidateContest are CandidateSelections, PartyContests have PartySelections, etc, etc.

These should be errors.

User feedback:

Your validator reads the local csv file correctly but then does not find the ocd id in the array/object generated from the local csv file.

the problem is that he's comparing the following things here https://github.com/google/election_results_xml_validator/blob/master/rules.py#L454:
ocd-division/country:ee (from the candidates-xml) with

{b'ocd-division/country:ee', b'id'} (from the local .csv file)

And he doesn’t comes in the if-statement. Without the b from {b'ocd-division/country:ee', b'id'} it works well.
Example:

text = "ocd-division/country:ee";

ocds = {'id', 'ocd-division/country:ee'};

if text in ocds:
    print("in")
else:
    print("not in")

There is also a TODO here in your code https://github.com/google/election_results_xml_validator/blob/master/rules.py#L428

Your Code:
with io.open(countries_file, mode="rb") as fd:
    for line in fd:
        if line is not "":
            # TODO use a CSV Reader
            ocd_id_codes.add(line.split(b",")[0])

Replaced with:

with open(countries_file) as csv_file:
    csv_reader = csv.reader(csv_file, delimiter=',')
    line_count = 0
    for row in csv_reader:
        if line_count > 0:
            ocd_id_codes.add(row[0])
        line_count += 1

The NIST entities below shall have a globally unique stable external identifier, that looks like this:

other
stable
globally_unique_value

• candidates
• contests
• political parties and coalitions

We've seen several issues where Name elements of Person, Candidate, or Contest elements are in all caps. These are generally not usable and should be a warning.

A Candidate object should only ever be referenced from one contest. If a Person is running in multiple Contests, then that Person is a Candidate several times over, but a Candida(te|cy) can't span contests.

The code that checks for valid OCD-IDs in GpUnits referenced by Contests will crash if an ExternalIdentifier is missing a Value field. Technically not having a Value field means that the feed will fail XSD validation, but we should still be defensive against this error.

This allows us to do OCD-ID validation in a NIST xml file before the specific set of OCD-IDs are being reviewed/published on github.

NIST is creating a JSON representation of the 1500-100 spec. Our validator should be able to

1. Include an instance of the JSON schema definition as opposed to an XSD.
2. Use the above JSON schema definition to perform schema validation.
3. Perform the same content validation as it currently does on the XML validation.

The validator should warn users if the names of contests within a file are not unique.

Deprecated method warning

$ election_results_xml_validator validate samples/post_election_sample_feed_precincts.xml --xsd election_data_spec.xsd

--------- Results after validating file: samples/post_election_sample_feed_precincts.xml 
Validator version: 0.10.1.1
SHA-512/256 checksum: 0x2d06124ff9213a43c1a1c6c3eae3aba0de5b52f17e5404c273ca9228ed200ba7
/usr/local/lib/python3.7/site-packages/election_results_xml_validator/rules.py:475: DeprecationWarning: Call to deprecated method get_dir_contents. (
        Repository.get_dir_contents() is deprecated, use
        Repository.get_contents() instead.
        )
  dir_contents = self.github_repo.get_dir_contents(self.GITHUB_DIR)
     2 Error messages found
         1 GpUnitsHaveSingleRoot Error message
         1 ElectionEndDates Error message
     1 Warning message found
         1 ElectionStartDates Warning message
    12 Info messages found
         6 DuplicatedPartyAbbreviation Info messages
         6 MissingPartyAbbreviationTranslation Info messages
$ 

Add rules support for NIST 1500-100 v2 spec, currently pending official publishing. A non-exhaustive list of changes is available here.

I believe a great number of rules will work as-is, and others only need to be modified insomuch that they can locate the target nodes.

Make sure that the GpUnit relationships described by ComposingGpUnitIds don't end up creating cycles.

Right now we only provide the option to pull down OCD-IDs for the United States and Germany. We should expand that to the full set of countries for which OCD-IDs exist, even in some form.

It appears that when the validator is running on Windows it isn't always able to download the OCD-ID files from GitHub. This is likely due to us assuming UNIX-style directory names:

https://github.com/google/election_results_xml_validator/blob/master/rules.py#L325

We should make this truly platform-agnostic.

This is to make us compatible with usnistgov/Voting#7