opendatasus_processar: Process large OpenDataSUS CSV resources in chunks

In datasus: An Interface to Brazilian Public Health Data

Process large OpenDataSUS CSV resources in chunks

Description

Downloads the selected resource using the regular atomic cache, expands multipart resources, and invokes a callback for bounded-size chunks. Returning small summaries or 'NULL' from the callback keeps memory use independent of the complete dataset size.

Usage

opendatasus_processar(
  conjunto,
  FUN,
  recurso = NULL,
  ano = "last",
  formato = "CSV",
  destino = NULL,
  cache = TRUE,
  atualizar = FALSE,
  colunas = NULL,
  tamanho_bloco = 100000L,
  sistema = NULL,
  ...
)

Arguments

conjunto	Dataset slug returned by [opendatasus_catalogo()].
FUN	Function called as 'FUN(dados, posicao, arquivo)', where 'posicao' is the first row position within a file and 'arquivo' is the one-based file-part index.
recurso	Optional exact resource name or stable resource identifier.
ano	Resource year, '"last"' for the latest dated resource, or 'NULL'. Ignored when 'recurso' is supplied.
formato	Resource format. When both 'recurso' and 'formato' are 'NULL', CSV is preferred.
destino	Optional destination file or existing directory. 'NULL' uses the package cache, or a temporary file when 'cache = FALSE'.
cache	Whether to cache portal metadata for the duration controlled by option 'datasus.cache_ttl'.
atualizar	Whether to ignore cached metadata and query the portal again.
colunas	Optional character vector selecting columns while CSV files are parsed. JSON columns are selected after parsing.
tamanho_bloco	Positive number of rows parsed per chunk.
sistema	Optional dictionary system accepted by [datasus_padronizar()]. When supplied, each chunk is standardized before 'FUN' is called.
...	Additional arguments passed to [readr::read_delim()] or [jsonlite::fromJSON()].

Value

An object of class '"datasus_processamento"' containing callback results, row, chunk and file counts, and provenance for every file.

Examples

old_options <- options(
  datasus.timeout = 5,
  datasus.download_timeout = 15,
  datasus.max_tries = 1
)
try({
summary <- opendatasus_processar(
  "doses-aplicadas-pelo-programa-de-nacional-de-imunizacoes-pni-2026",
  recurso = "Vacinação - Julho 2026",
  ano = NULL,
  formato = "CSV",
  FUN = function(dados, posicao, arquivo) nrow(dados),
  tamanho_bloco = 50000
)
})
options(old_options)

datasus documentation built on Oct. 7, 2026, 1:07 a.m.