[docs] Process Parquet files

[stevhliu]

This PR adds a guide for how to process Parquet files:

• Eager dataframes with pd/pl.read_parquet
• Lazy dataframes with pl.scan_parquet
• Read and query with DuckDB

[HuggingFaceDocBuilderDev]

The documentation is not available anymore as the PR was closed or merged.

[lhoestq]

Nice ! This is super useful :)

Just two comments on the section title:

1. Do you think we can make it clearer in the title that you can query any dataset ? IMO the current title "Process Parquet files" may suggest that it can only be useful if you have parquet files already. Maybe mention that the parquet files are "published", or use a verb that suggest that the data is remote e.g. "access" in addition to "process".
2. (nit) Both polars and duckdb examples run "queries" on parquet files. Maybe say "query" instead of "process" ? I think "process" can be mentioned for data manipulation that involve user defined functions (UDF) - with Spark for example

[mariosasko]

Nice!

IMO it makes sense to remove the read_polars/read_pandas examples from the List Parquet files and instead provide a link to this guide now that we have it.

And maybe we can mention at the end of the guide that Parquet files can also be queried in the browser with DuckDB WASM.

PS: doc-builder generates (empty) CURL version of the examples, would be nice if this can be removed

[stevhliu]

PS: doc-builder generates (empty) CURL version of the examples, would be nice if this can be removed

Oh! I didn't think the doc-builder would generate cURL examples even though I didn't specify them in the <inferencesnippet>. @mishig25 is there a way to remove this?

[mariosasko]

Approving but let's see if we can fix the issue with the cURL version of the examples being displayed before merging!

[stevhliu]

@mariosasko, empty cURL example fixed in huggingface/doc-builder#365!