Creating DataFrame based on runtime metadata

[andre2007]

I am using the Apache Arrow C library to read parquet files (https://github.com/apache/arrow/blob/master/c_glib/parquet-glib/).
Based on the columns of the parquet file I want to create a DataFrame (schema not known at compile time).
Is this is possible? I only see examples of creating the DataFrame with schema known at compile time, like

DataFrame!(int, 2, double, 1) df;