Support for empty array literals (ARRAY[])

[bitemyapp]

We had SQL queries that contained fragments that looked like this:

,COALESCE(t.location_id_tokenized, ARRAY []) AS location_id_tokenized

We needed to coalesce the values down to empty arrays. This is the recommended syntax on the AWS Athena docs: https://docs.aws.amazon.com/athena/latest/ug/creating-arrays.html

Oddly, the parser assumed all ARRAY keyword literal values would have a comma separated list of values inside of them. This initially threw me off as I assumed it would be a lexing problem. That's one reason there's an AthenaSqlDialect in here. The other reason is that I wasn't sure if I should bolt the ARRAY[] test onto an existing dialect or carve out an Athena dialect. The Athena SQL dialect tests are a hodgepodge of Hive tests with a single array literal test from PostgreSQL adapted to this specific missing feature.

If you want to migrate the test into an existing dialect and nuke the Athena dialect, that's fine by me. Athena SQL is basically Presto+Hive I think, so it's whatever you want to do. This problem is fixed for all SQL dialects regardless because it's in the parser.

Cheers and thanks for the parser, this thing has saved me countless hours.

[Dandandan]

Hey @bitemyapp , thanks for the PR!

If you want to migrate the test into an existing dialect and nuke the Athena dialect, that's fine by me. Athena SQL is basically Presto+Hive I think, so it's whatever you want to do. This problem is fixed for all SQL dialects regardless because it's in the parser.

Yes please, that sounds like the best way forward.

[coveralls]

Pull Request Test Coverage Report for Build 2625673440

• 12 of 13 (92.31%) changed or added relevant lines in 2 files are covered.
• No unchanged relevant lines lost coverage.
• Overall coverage increased (+0.07%) to 89.88%

---

Changes Missing Coverage	Covered Lines	Changed/Added Lines	%
src/parser.rs	7	8	87.5%

Totals
Change from base Build 2584755255:	0.07%
Covered Lines:	8873
Relevant Lines:	9872

---

💛 - Coveralls

[bitemyapp]

Yes please, that sounds like the best way forward.

Okie dokie artichokie. I'll push it tomorrow.

[bitemyapp]

@Dandandan oh, actually, where should the test specifically for the ARRAY[] syntax go? I nabbed the value-full version from PGSQL but not sure if that's where you actually want it.

[bitemyapp]

Squashed the commits, tests pass locally.

[bitemyapp]

Amended the MR title to fit the reality better.

[Dandandan]

Looks good 👍

[Dandandan]

@bitemyapp could you fix the code style issues? Otherwise this is good to go!

[bitemyapp]

@Dandandan it should be fixed now. FWIW that license template header option seems to be gone in newer versions of cargo fmt.

[Dandandan]

@Dandandan it should be fixed now. FWIW that license template header option seems to be gone in newer versions of cargo fmt.

Thank you @bitemyapp !