Skip to contents

Parse a brreg bulk JSON download into a flat tibble

Source: R/download.R
parse_bulk_json.Rd

The /enheter/lastned and /underenheter/lastned endpoints return gzipped JSON arrays with nested objects. jsonlite::fromJSON(flatten = TRUE) expands nested objects to dot-notation columns but leaves arrays as list columns. The array is read in batches of complete objects (read_json_objects()) and each batch parsed on its own, so the whole decoded tree is never in memory at once; with a finite n_max only the first n_max elements are read, which is how a sample of the register is parsed. Every list column gets the tidy shape of the live path (collapse_observed()):

Usage

parse_bulk_json(path, type = "enheter", n_max = Inf)

Arguments

path

Path to the gzipped JSON file.

type

Entity type (for column context).

n_max

Maximum number of array elements to read (default: all).

Value

A tibble mapped via field_dict; arrays of objects are nested list-columns, everything else atomic.

Details

  • Character vectors (addresses, activities): line fragments of one value, joined with "\n"

  • Data frames (paategninger, historiskeNavn): kept as a list-column of tibbles, one row per element, NULL where the entity has none

  • Empty lists (HAL links): dropped

  • NULL elements: NA_character_

Known columns are renamed via field_dict. Unknown columns pass through with auto-generated snake_case names (zero-drop policy). The raw .gz file is the provenance fallback for anyone needing the original nesting.