Build the catalogue manifest by scanning the data share

Walks root, parses every dataset readme it finds, and returns one row per documented layer. Folders holding spatial data with no readme are recorded separately in the undocumented attribute and reported by check_metadata().

Usage

build_catalogue(
  root = spatial_root(),
  files = TRUE,
  refresh = FALSE,
  quiet = FALSE
)

Arguments

root: Character; directory to scan. Defaults to spatial_root().
files: Logical; if TRUE (default), inventory each layer's files to populate n_files, size_mb, and data_type. Set FALSE for a faster, metadata-only scan.
refresh: Logical; if TRUE, rescan even when a cached manifest exists. Default FALSE.
quiet: Logical; if TRUE, suppress progress messages. Default FALSE.

Value

A data.frame with one row per layer — one per variant where a product has them, otherwise one per product. Columns are id (path relative to root), name (the product folder), theme, product_id, variant (NA when the product has no variant folders), the template metadata fields (see as_metadata_row()), n_files, size_mb, data_type, path (the folder holding the data), readme (the record defining the row), and product_readme (equal to readme unless the record is split). Carries the scan root and the undocumented folders as attributes.

Details

The top-level _temp and _deprecated folders are skipped: one is the share's scratch area, the other holds superseded products, so the contents of neither are catalogued as layers or reported as missing readmes.

The result is cached per root for the session, so list_layers() and friends only pay for the scan once.

Examples

if (FALSE) { # \dontrun{
cat_df <- build_catalogue()
nrow(cat_df)
build_catalogue(refresh = TRUE, files = FALSE)
} # }