2012 and 2016 Summer Olympics Athlete Numbers

library(rvest)

urls <- c("https://en.wikipedia.org/wiki/2012_Summer_Olympics", "https://en.wikipedia.org/wiki/2016_Summer_Olympics")

dfs <- NULL
for (k in 1:length(urls)) {
  webHTML <- tryCatch({read_html(urls[k])}, error=function(err) "Error") # attmpt to read website
  if (all(webHTML != "Error")) {
    htmlTable <- webHTML %>% html_nodes("table.wikitable:nth-child(1)") # get table
    # countries
    countries <- htmlTable %>% html_nodes(".thumbborder+ a") %>% html_text()
    # number of participants are in the form (###), so remove '(' and ')'
    participants <- gsub("\\(|)", "", htmlTable %>% html_nodes(".column-width span") %>% html_text())
    dfs <- rbind(dfs, data.frame(country=countries, num.par=participants, year=strsplit(tail(strsplit(urls[k], "/")[[1]], 1), "_")[[1]][1], stringsAsFactors=F))
  }
  Sys.sleep(0.5)
}
dfs$num.par <- as.numeric(dfs$num.par)

# Values to add to 2016
dfs <- rbind(dfs, data.frame(country=setdiff(dfs$country[dfs$year == "2012"], dfs$country[dfs$year == "2016"]), num.par=0, year="2016", stringsAsFactors=F))
# Values to add to 2012
dfs <- rbind(dfs, data.frame(country=setdiff(dfs$country[dfs$year == "2016"], dfs$country[dfs$year == "2012"]), num.par=0, year="2012", stringsAsFactors=F))
dfs <- dfs[order(dfs$year, dfs$country),]