dbt Server errors when returning Array types in Postgres

[drewbanin]

Describe the bug

There's some sort of agate-internal issue with converting array type columns to an Agate dataframe from a Postgres response. TBD what the exact culprit is, but it manifests as tuple.index(x): x not in tuple called via agate_table.table_from_data (specifically in table.select(column_names)). There's a solid reproduction case attached below.

Steps To Reproduce

Run the query:

select ARRAY[10000, 10000, 10000, 10000]

against a running rpc server. Poll for a response and see the error:

Traceback (most recent call last):
  File \"/Users/drew/fishtown/dbt/core/dbt/node_runners.py\", line 227, in safe_run
    result = self.compile_and_execute(manifest, ctx)
  File \"/Users/drew/fishtown/dbt/core/dbt/node_runners.py\", line 170, in compile_and_execute
    result = self.run(ctx.node, manifest)
  File \"/Users/drew/fishtown/dbt/core/dbt/node_runners.py\", line 272, in run
    return self.execute(compiled_node, manifest)
  File \"/Users/drew/fishtown/dbt/core/dbt/rpc/node_runners.py\", line 86, in execute
    compiled_node.injected_sql, fetch=True
  File \"/Users/drew/fishtown/dbt/core/dbt/adapters/base/impl.py\", line 227, in execute
    fetch=fetch
  File \"/Users/drew/fishtown/dbt/core/dbt/adapters/sql/connections.py\", line 119, in execute
    table = self.get_result_from_cursor(cursor)
  File \"/Users/drew/fishtown/dbt/core/dbt/adapters/sql/connections.py\", line 110, in get_result_from_cursor
    return dbt.clients.agate_helper.table_from_data(data, column_names)
  File \"/Users/drew/fishtown/dbt/core/dbt/clients/agate_helper.py\", line 84, in table_from_data
    return table.select(column_names)
  File \"/Users/drew/fishtown/dbt/env/lib/python3.7/site-packages/agate/table/select.py\", line 21, in select
    indexes = tuple(self._column_names.index(k) for k in key)
  File \"/Users/drew/fishtown/dbt/env/lib/python3.7/site-packages/agate/table/select.py\", line 21, in <genexpr>
    indexes = tuple(self._column_names.index(k) for k in key)
ValueError: tuple.index(x): x not in tuple

Expected behavior

I'd expect this query to return data successfully. The Array field should be json-ified, as with non-primitive types on other datbases (BQ, Snowflake).

System information

Which database are you using dbt with?

• postgres

The output of dbt --version:

0.16.0

[drewbanin]

Update:

select ARRAY[10000, 10000, 10000, 10000]

ipdb> table.print_table()
| array/0 | array/1 | array/2 | array/3 |
| ------- | ------- | ------- | ------- |
|  10,000 |  10,000 |  10,000 |  10,000 |

Thanks, Agate!