Registro Civil por mês...

gistfile1.txt

library(httr2)
library(tibble)
library(dplyr)
library(readr)

`%||%` <- function(x, y) if (!is.null(x)) x else y

# ---- helper: extrai o valor de um cookie do header Set-Cookie
extract_cookie_from_set_cookie <- function(set_cookie_vec, cookie_name) {
  # set_cookie_vec pode ser um vector com várias linhas
  if (length(set_cookie_vec) == 0 || all(is.na(set_cookie_vec))) return(NULL)
  
  # tenta achar a linha que começa com "COOKIE="
  idx <- which(grepl(paste0("^", cookie_name, "="), set_cookie_vec))
  if (length(idx) == 0) return(NULL)
  
  line <- set_cookie_vec[idx[1]]
  # pega tudo entre "COOKIE=" e o primeiro ";"
  sub(paste0("^", cookie_name, "=([^;]*).*"), "\\1", line)
}

registrocivil_start_session <- function(
    landing_url = "https://transparencia.registrocivil.org.br/inicio",
    user_agent = NULL
) {
  ua <- user_agent %||% paste0(
    "Mozilla/5.0 (", Sys.info()[["sysname"]], ") R/", getRversion()
  )
  
  jar <- tempfile(fileext = ".txt")
  
  resp <- request(landing_url) |>
    req_headers(
      "accept" = "text/html,application/xhtml+xml,application/xml;q=0.9,*/*;q=0.8",
      "user-agent" = ua
    ) |>
    req_options(cookiejar = jar, cookiefile = jar) |>
    req_perform()
  
  # Set-Cookie pode vir como múltiplos valores
  hdrs <- resp_headers(resp)
  set_cookie <- hdrs[["set-cookie"]]
  # dependendo do server, isso pode vir como string única; normaliza pra vector
  if (!is.null(set_cookie) && length(set_cookie) == 1) {
    # alguns servidores juntam cookies em uma string; tentar dividir em ", " pode ser perigoso
    # aqui mantemos como 1 linha e ainda funciona se começar com o cookie procurado
    set_cookie_vec <- set_cookie
  } else {
    set_cookie_vec <- set_cookie
  }
  
  xsrf_cookie <- extract_cookie_from_set_cookie(set_cookie_vec, "XSRF-TOKEN")
  session_cookie <- extract_cookie_from_set_cookie(set_cookie_vec, "_session")
  
  x_xsrf_token <- if (!is.null(xsrf_cookie) && nzchar(xsrf_cookie)) {
    utils::URLdecode(xsrf_cookie)
  } else {
    NULL
  }
  
  # Opcional: extrair meta csrf-token do HTML (se pacotes existirem)
  x_csrf_token <- NULL
  html_txt <- tryCatch(resp_body_string(resp), error = function(e) NULL)
  
  if (!is.null(html_txt) &&
      requireNamespace("xml2", quietly = TRUE) &&
      requireNamespace("rvest", quietly = TRUE)) {
    doc <- tryCatch(xml2::read_html(html_txt), error = function(e) NULL)
    if (!is.null(doc)) {
      node <- rvest::html_element(doc, 'meta[name="csrf-token"]')
      val <- rvest::html_attr(node, "content")
      if (!is.na(val) && nzchar(val)) x_csrf_token <- val
    }
  }
  
  list(
    cookiejar = jar,
    xsrf_cookie = xsrf_cookie,
    session_cookie = session_cookie,
    x_xsrf_token = x_xsrf_token,
    x_csrf_token = x_csrf_token,
    user_agent = ua
  )
}

registrocivil_all_name2 <- function(
    session,
    start_date,
    end_date,
    translate = 1,
    base_url = "https://transparencia.registrocivil.org.br/api/record/all-name",
    ...,
    fail_fast = TRUE
) {
  req <- request(base_url) |>
    req_url_query(
      start_date = start_date,
      end_date   = end_date,
      translate  = translate,
      ...
    ) |>
    req_headers(
      "accept" = "application/json, text/plain, */*",
      "x-requested-with" = "XMLHttpRequest",
      "referer" = "https://transparencia.registrocivil.org.br/inicio",
      "user-agent" = session$user_agent
    ) |>
    # reaplica cookies automaticamente
    req_options(cookiejar = session$cookiejar, cookiefile = session$cookiejar)
  
  if (!is.null(session$x_xsrf_token)) req <- req |> req_headers("x-xsrf-token" = session$x_xsrf_token)
  if (!is.null(session$x_csrf_token)) req <- req |> req_headers("x-csrf-token" = session$x_csrf_token)
  
  resp <- req |> req_perform()
  
  if (fail_fast) resp <- resp |> resp_check_status()
  
  payload <- resp |> resp_body_json(simplifyVector = TRUE)
  
  if (is.null(payload$status) || payload$status != 1) {
    stop(
      "Resposta inesperada (status != 1). ",
      "HTTP: ", resp_status(resp), ". ",
      "Corpo (parcial): ",
      substr(resp_body_string(resp), 1, 500)
    )
  }
  
  as_tibble(payload$data) |>
    mutate(
      total = parse_integer(as.character(total)),
      name  = as.character(name)
    )
}

registrocivil_all_name <- function(
    start_date, end_date, translate = 1, ...,
    landing_url = "https://transparencia.registrocivil.org.br/inicio"
) {
  sess <- registrocivil_start_session(landing_url = landing_url)
  registrocivil_all_name2(sess, start_date, end_date, translate = translate, ...)
}

# Exemplo:
df <- registrocivil_all_name("2025-01-01", "2025-01-31", translate = 1)
df