Fix the pattern AST creation function

[chisholm]

Fixes #326

It seems like stix2.pattern_visitor.create_pattern_object() was written based on a copy-paste of a function from the pattern validator. But it wasn't changed enough. I fixed it up:

• The docstring still mentioned validation. That's not what the function does. It was totally wrong. I gave it a better one.
• That pattern[:2] junk was from the validator checking whether the pattern starts with brackets. It doesn't belong in the AST builder. I removed it.
• An error listener was used which produced messages with "FAIL:". That makes sense in the context of validation, but not here. There is no success/fail test happening. I switched to a different error listener which produces a more appropriate message.
• I did an ANTLR parser setup similar to the pattern matcher, with BailErrorStrategy and the appropriate error handling, so that parse errors result in exceptions with sensible error messages.

If we want to make visitor use more convenient, maybe we should give the Pattern class(es) from the pattern validator a visit() method analogous to the walk() method?

An example of error output now (with Python 3):

Traceback (most recent call last):
  File "c:\programming\python\stix\pattern-validator\stix2patterns\v20\grammars\STIXPatternParser.py", line 2090, in primitiveLiteral
    raise NoViableAltException(self)
antlr4.error.Errors.NoViableAltException: None

The above exception was the direct cause of the following exception:

Traceback (most recent call last):
  File "testx_pattern_parse.py", line 3, in <module>
    errs = create_pattern_object(pattern="[ file: name = 'weirdname]")
  File "C:\Programming\python\stix\cti-python-stix2\stix2\pattern_visitor.py", line 363, in create_pattern_object
    six.raise_from(
  File "<string>", line 3, in raise_from
stix2patterns.exceptions.ParseException: 1:15: no viable alternative at input '''

[codecov-io]

Codecov Report

Merging #344 into master will increase coverage by 0.05%.
The diff coverage is 100%.

@@            Coverage Diff             @@
##           master     #344      +/-   ##
==========================================
+ Coverage    98.1%   98.16%   +0.05%     
==========================================
  Files         124      124              
  Lines       14185    14158      -27     
==========================================
- Hits        13916    13898      -18     
+ Misses        269      260       -9

Impacted Files	Coverage Δ
stix2/pattern_visitor.py	72.81% <100%> (-1.75%)	⬇️
stix2/test/v21/test_indicator.py	100% <100%> (ø)	⬆️
stix2/test/v21/test_pattern_expressions.py	100% <100%> (ø)	⬆️
stix2/v21/observables.py	96.93% <0%> (-0.21%)	⬇️
stix2/test/v21/test_observed_data.py	100% <0%> (ø)	⬆️
stix2/base.py	97.8% <0%> (+0.09%)	⬆️
stix2/properties.py	98.36% <0%> (+0.54%)	⬆️
stix2/utils.py	97.72% <0%> (+1.13%)	⬆️
stix2/canonicalization/Canonicalize.py	61.44% <0%> (+1.2%)	⬆️

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 8aca39a...4f00c7c. Read the comment docs.

[clenk]

Looks great! I think we should have a test case to make sure the right exception class is being thrown.

I wouldn't be opposed to also adding a visit() method.

[chisholm]

I added the visit() method and changed create_pattern_object() to take advantage of it. Of course, travis tests won't pass until a new version of the pattern validator is released which has the change. (And we too are suffering from the ANTLR 4.8 issue.)

[clenk]

LGTM!