LoginSign Up
← Back to all sparks
E

enderecobr

ANALYTICS
Velocity0.0

Padronizador de Endereços Brasileiros (Brazilian Addresses Standardizer)

github.com/ipea/enderecobr ↗

Brazilian address standardisation moves its core to Rust, betting on throughput over pure R

address-standardisationbrazilrustperformancedata-cleaningr-package
◆Current state
enderecobr standardises Brazilian address fields — street types, neighbourhoods, states, postcodes, house numbers — into consistent forms so records from different registries can be compared. It comes out of the ipeaGIT ecosystem and its API is a family of padronizar_* functions plus one padronizar_enderecos() that runs them together. The 0.5.0 release rewrote those standardisation functions in Rust with a documented speedup.
◆Where it's heading
Two years of releases were about control and correctness: new formato arguments letting callers choose state names or abbreviations, integers or characters for house numbers, and a run of fixes for numbers mangled by thousands separators. That work settled the semantics. With the behaviour pinned down, the Rust rewrite becomes the natural next move, and the release pairs it with new matching rules for the two messiest fields, neighbourhoods and street names.
◆Prediction
The rewrite covers the standardisation functions specifically, so the remaining pure-R paths around them are the likely next targets, alongside continued rule additions for street and neighbourhood variants.

◆Recent moves

  1. 7mo ago

    Standardisation functions rewritten in Rust

    ⚡ SPARK

    The package's core moves from R to Rust with a documented performance gain, and new matching rules land for neighbourhoods and street names. It is the largest change in the package's history and rests on the argument semantics the previous two years of releases nailed down.

    View source ↗
  2. 1y ago

    Namespaced calls no longer error on internal warnings

    Fixes two cases where calling padronizar_enderecos() with the namespace prefix turned an internal warning or an invalid postcode into a hard error. Corrections to error propagation, not to how addresses are standardised.

    View source ↗
  3. 1y ago

    House numbers gain output format control, thousands-separator bug fixed

    Adds formato and formato_numeros arguments controlling whether house numbers come back as characters or integers, and fixes a corruption where 1.028 became 1.28. The same argument-plus-fix shape as the state standardisation release before it.

    View source ↗
  4. 1y ago

    States can be standardised to full names or abbreviations

    A formato argument on padronizar_estados(), surfaced through padronizar_enderecos() as formato_estados, lets callers pick full state names or two-letter codes. Part of the steady push to make each field's output form an explicit choice.

    View source ↗
  5. 1y ago

    Author added to package metadata

    Adds Lucas Mation as a package author. Metadata only, with no functional change.

    View source ↗
  6. 1y ago

    Street-type functions added and output columns given a naming convention

    Adds padronizar_tipos_de_logradouro() and padronizar_logradouros_completos(), plus arguments to keep input columns and combine street fields, with results returned in <campo>_padr columns. This release set the output contract that later format arguments refined.

    View source ↗