Runs clickhouse-obfuscator from docker image, which randomizes your dataset to make its data not sensetive

clickhouse-obfuscator.sh

docker run --rm \
  --volume "$(pwd)"/seeds:/tmp/seeds \
  --entrypoint bash \
  yandex/clickhouse-server:latest \
  -c '/usr/bin/clickhouse obfuscator --seed "$(head -c16 /dev/urandom | base64)" --input-format TSV --output-format TSV --structure "date Date, datetime DateTime, ab_tag String, client_name String" < /tmp/seeds/rows.tsv' \
  > obfuscated_rows.tsv
  
# 1. stdin pipe was created inside container to prevent "Input must be seekable file (it will be read twice)." error
# 2. UUID, LowCardinality(…) and Enum8/16 data-types are not supported in `--structure`.
#    Use FixedString(36) or String instead (See https://github.com/yandex/ClickHouse/blob/881893d/dbms/programs/obfuscator/Obfuscator.cpp#L873-L895)