input parquet got decoding panic

[fearfate]

config:

input:
  parquet:
    paths:
      - '*.parquet'
    auto_replay_nacks: false
    # batch_count: 0

output:
  drop: {}

run process like this:

the parquet file:

example.zip

[mihaitodor]

Hey @fearfate thanks for raising this! The issue seems to be in the parquet-go library (this code panics for certain rows). I raised an issue upstream: parquet-go/parquet-go#204

Note: I was able to read the file successfully using https://github.com/apache/arrow-go/tree/249ec029ad4c02488c73eee2dfcbc2aec89ff464/parquet/cmd/parquet_reader. Might be worth having this library as an alternative implementation in the future.