read_addresses: Download geolocated data of addresses in Brazil

In geobr: Download Official Spatial Data Sets of Brazil

Download geolocated data of addresses in Brazil

Description

This function reads the data of the National Registry of Addresses for Statistical Purposes (Cadastro Nacional de Enderecos para Fins Estatisticos, CNEFE), organized by the Brazilian Institute of Geography and Statistics (IBGE). The data brings the geographical coordinates (lat lon) of every address surveyed in the Population Census, along with the census tract and municipality each address belongs to, its postal code (CEP), the type of building (cod_especie) and the precision level of its coordinates (nv_geo_coord). More information available at https://www.ibge.gov.br/estatisticas/sociais/populacao/38734-cadastro-nacional-de-enderecos-para-fins-estatisticos.html.

Note this is a very large data set: the 2022 CNEFE covers roughly 111 million addresses in a single 1.2 GB file. The file is downloaded in full and only then filtered by code_muni, so the first call in an R session takes a long time even when a single municipality is requested. Subsequent calls reuse the cached file. Passing output = "duckdb" avoids loading the result into memory.

Usage

read_addresses(
  year,
  code_muni,
  output = "sf",
  showProgress = TRUE,
  cache = TRUE,
  verbose = TRUE
)

Arguments

year	Numeric. Year of the data in YYYY format.
code_muni	The 7-digit code of a municipality. Alternatively, if a two-digit state code or a two-letter uppercase abbreviation of a state is passed (e.g. 33 or "RJ"), all data of that state are downloaded. Passing code_muni = "all" downloads the addresses of the whole country. Municipality codes can be consulted with the geobr::lookup_muni() function. Unlike in most geobr functions, this argument is required and has no default: reading the addresses of the whole country may exhaust memory, so the choice is left explicitly to the user.
output	String. Type of object returned by the function. Defaults to "sf", which loads the data into memory as an sf object. Alternatively, "duckdb" returns a lazy spatial table backed by DuckDB via the duckspatial package, and "arrow" returns an Arrow dataset. Both "duckdb" and "arrow" support out-of-memory processing of large data sets.
showProgress	Logical. Defaults to TRUE display progress bar.
cache	Logical. Whether the function should read the data cached locally, which is faster. Defaults to cache = TRUE. By default, geobr stores data files in a temporary directory that exists only within each R session. If cache = FALSE, the function will download the data again and overwrite the local file.
verbose	A logical. If TRUE (the default), the function prints informative messages and shows download progress bar. If FALSE, the function is silent.

Value

An ⁠"sf" "data.frame"⁠ OR an ArrowObject

Examples

# Read all addresses in a given municipality
add <- read_addresses(
  year = 2022,
  code_muni = 3304557
  )

# Read all addresses in a given state, without loading them into memory
add_rj <- read_addresses(
  year = 2022,
  code_muni = "RJ",
  output = "duckdb"
  )

geobr documentation built on Sept. 20, 2026, 9:06 a.m.