Read one gazetteer layer, pinned and verified

Downloads a collection's places.parquet from the oceanmetrics gazetteer and returns it as an sf. The read is pinned: the manifest row (index/layers.json, matched on slug) names the version and the parquet's parquet_sha256, and the downloaded bytes must hash to it, else an error (a silently changed layer must never reach a release). The manifest's newer fields are optional, so this works against older manifests:

Usage

read_gazetteer_layer(
  slug,
  version = NULL,
  manifest = GAZETTEER_MANIFEST,
  cache_dir = tools::R_user_dir("calcofi4db", "cache")
)

Arguments

slug

the gazetteer collection, e.g. "noaa_aoa_socal".

version

the collection version to read; NULL (default) = the manifest's version.

manifest

URL or local path of the gazetteer's index/layers.json.

cache_dir

directory for verified downloads (default: the user cache of calcofi4db).

Value

An sf in EPSG:4326 with geometry column geometry, the columns place_id, name, status, status_date, status_source, status_upstream, status_history first (NA when the parquet lacks one), then the native attributes (the GeoParquet bbox covering column dropped). Attribute gazetteer = list(slug, version, sha256, url).

Details

• no parquet: the URL is <base><slug>/places.parquet (base from the manifest, default https://storage.oceanmetrics.io/gazetteer/);

• no parquet_sha256, or a version other than the manifest's: the hash comes from <base><slug>/versions.json (versions[] entry for version, assets["places.parquet"].sha256, and its href when given).

A verified copy is kept in cache_dir/<slug>/<version>/places.parquet and reused while its hash still matches. manifest, base, parquet and versions.json may also be local paths.

Examples

if (FALSE) { # \dontrun{
aoa <- read_gazetteer_layer("noaa_aoa_socal")
attr(aoa, "gazetteer")
} # }