Will CLI support a parquet query language or alike?

[rmannibucau]

Hi

What is the plan to filter a file with a query language like in the CLI? Would be awesome to get it wired in show command (#163)

[gunnarmorling]

Good question; my first reaction would be "No, probably not". There's already DuckDB which is doing an amazing job at querying Parquet files using SQL AFAIK. I don't think there's much point in reinventing the wheel here.

The idea for the CLI really is more about providing a tool for inspecting Parquet files, their metadata, schema, statistics, maybe bloom filters, things like that. In that sense, I'm even a bit on the fence in regards to those show/tail/head commands (or a single print command, as suggested in the show command PR), as they are somewhat out of scope. But I can see the value of having a quick way to also examine the actual data, so it's fine to have them. Whether we should provide a query language though, I'm not so sure about that.

[rmannibucau]

my idea was mainly to enable simple filters:

• equals
• like (or startswith/endswith/contains/regex to make it easier)
• <= >= < >

in terms of ergonomy it is mainly about avoiding grep and supporting the transpose mode which is often more readable

For files with millions of lines it is really neat and helpful without entering into something complex

it is fine if not a query language but a set of options (like find can get -name foo -o -name bar for ex), goal is really just to filter the file for print command without loosing the formatting