us2020.R

read_nyt_vote_table <- function(state, force = FALSE) {
  library(rvest)  ## CRAN package 'rvest'
  states <- c(AZ = "arizona", GA = "georgia", NC = "north-carolina", NV = "nevada", PA = "pennsylvania")
  stopifnot(state %in% names(states))
  file <- sprintf("results-%s-president.html", states[state])
  url <- file.path("https://www.nytimes.com/interactive/2020/11/03/us/elections", file)
  if (force || !file_test("-f", file)) {
    message("Downloading: ", file)
    download.file(url, destfile = file, quiet = TRUE)
  }
  stopifnot(utils::file_test("-f", file))
  id <- sprintf("%s-G-P-2020-11-03", state)
  html <- read_html(file)
  xpath <- sprintf('//*[@id="%s-results-table-container"]', id)
  data <- html_node(html, xpath = xpath)
  table <- html_children(data)[[2]]
  table <- html_table(table, header = FALSE)
  table <- table[, -1L]
  colnames(table) <- c("candidate", "party", "votes", "fraction")
  idx <- grep("View", table[,"candidate"])
  if (length(idx) > 0) table <- table[-idx,]
  table[,"candidate"] <- gsub("[*[:space:]].*", "", table[,"candidate"])
  table[,"votes"] <- as.integer(gsub(",", "", table[,"votes"]))
  table[,"fraction"] <- as.numeric(gsub("[<%]", "", table[,"fraction"]))/100
  table <- table[, c("candidate", "votes", "fraction")]
  table
}