Publish / Append / Ignore / Replace Partitions in the Hive Lake

Usage

db_hive_write(
  data,
  section,
  dataset,
  partition_by = NULL,
  mode = c("overwrite", "append", "ignore", "replace_partitions"),
  compression = NULL,
  filename_pattern = "data_{uuid}"
)

Arguments

data

A data.frame / tibble

section

Your section name

dataset

The name of the dataset

partition_by

Character vector of column names to partition by (e.g. c("year","month"))

mode

One of:

"overwrite": replace target files
"append": add new files (requires unique filenames)
"ignore": write only if target path does not exist (best-effort; still race-prone)
"replace_partitions": delete only affected partition folders, then append fresh files (requires partition_by)

compression

Parquet compression codec (NULL means DuckDB default). Options: "zstd", "snappy", "gzip", "brotli", "lz4", "lz4_raw", "uncompressed"

filename_pattern

Used in append-like modes (default "data_\{uuid\}")

Value

Invisibly returns the output path

Examples

if (FALSE) { # \dontrun{
# Basic overwrite
db_hive_write(my_data, "Trade", "Imports")

# Partitioned write
db_hive_write(my_data, "Trade", "Imports", partition_by = c("year", "month"))

# Append mode
db_hive_write(my_data, "Trade", "Imports", mode = "append")

# Replace only touched partitions
db_hive_write(my_data, "Trade", "Imports", 
              partition_by = c("year", "month"), 
              mode = "replace_partitions")
} # }