Write a data frame to Parquet (language-neutral interchange)

Intermediate/terminal tables written as Parquet can be read by Python (pandas/pyarrow), DuckDB, and Nextflow processes — unlike the default .rds target store. This is the seam that keeps pipelines orchestrator-agnostic.

Usage

pp_write_parquet(df, path)

Arguments

df: A data frame.
path: Output .parquet path.

Value

path (so the target can be format = "file").

Examples

if (requireNamespace("arrow", quietly = TRUE)) {
  path <- pp_write_parquet(data.frame(a = 1:3, b = 4:6),
                           file.path(tempdir(), "demo.parquet"))
  file.exists(path)
}
#> [1] TRUE