Build an RDB table from multiple header variants

Usage

build_rdb_table(
  table_name,
  year,
  TABLE.HEADERS,
  con,
  cc_file,
  post_to_s3 = FALSE,
  selection = c("rdb_table", "header"),
  output = c("csv", "parquet", "both"),
  ...
)

Arguments

table_name: Character table id.
year: Integer year.
TABLE.HEADERS: Named list of header xpaths per table. Used only when selection = "header"; the default path ignores it.
con: DBI connection.
cc_file: Concordance crosswalk.
post_to_s3: Logical export flag.
selection: Row-selection strategy. "rdb_table" (default) filters on the RDB_TABLE column assigned during flattening – exact, and immune to the header-prefix collisions described in EF2-6. "header" reproduces the legacy unanchored regex, for diffing old against new output only.
output: One of "csv" (default, unchanged behaviour), "parquet", or "both". Both formats are written from one materialised temp table, so the Parquet file is never a re-parse of the CSV. See write_table_output().
...: Further arguments passed to write_table_output(), e.g. normalize_empty, sort_key, row_group_size.

Value

Invisibly the lazy tibble.