Encoded Data as Values

[friendofasquid]

Hi, I'm trying to Parquet into S3 via your plugin. It looks like the values in the data are somehow being encoded in a way that seems to make the Parquet file unusable with AWS Glue / Athena (Presto). When I use the stdout plugin, the data looks correct.

When I try to open the Parquet file with Sublime (which renders it like JSON), I see data that looks like the following:

{"event_id":"MjI0Yzg4ODc0ZjEzYWJjM2Q4OGI3M2NiYWE5NTcwODQ=","event_timestamp":"MjAxOC0wOS0wNCAxNTozMjoxOC4wMDEwMDAgKzAwMDA="}
{"event_id":"ZjQzNmQxMmNkNmFlNGM5ZmJkMTc3OTExOTJmZGY2MmY=","event_timestamp":"MjAxOC0wOS0wNCAxNTozMjoxNi4xNzIwMDAgKzAwMDA="}
…

Here's the relevant parts of the Embulk file:

in:
  type: command
  command: lib/splunk export …
      
  parser:
    type: jsonl
    columns:
      - {name: "event_id", type: string}
      - {name: "event_timestamp", type: timestamp, format: "%Q"}
      …
out:
  type: parquet
  path_prefix: s3a://…
  extra_configurations:
    fs.s3a.access.key: …
    fs.s3a.secret.key: …