code.R

## import data from websites directly

url <- "http://www.tennis-data.co.uk/2020/ausopen.csv"
tennis_aus <- read.csv(url)
str(tennis_aus)

## webscraping using rvest

require(rvest)
url_elo <- "http://tennisabstract.com/reports/atp_elo_ratings.html"
webpage <- read_html(url_elo)
elo_class <- webpage %>% 
  html_nodes(".tablesorter") %>% 
  html_table()
elo_id <- webpage %>% 
  html_nodes("#reportable") %>% 
  html_table()
identical(elo_class, elo_id)


## Web scraping using RSelenium
require("RSelenium")
rD <- rsDriver(port = 5563L, chromever = "85.0.4183.87")
remDr <- rD$client
url <- "http://www.flashscore.com/match/Cj6I5iL9/#match-statistics;0"
remDr$navigate(url)
webElem <- remDr$findElements(using = 'class', "statBox")
webElem <- unlist(lapply(webElem, function(x){x$getElementText()}))[[1]]
head(unlist(strsplit(webElem, split = '\n')))
remDr$close()





rD <- rsDriver(port = 5572L, chromever = "85.0.4183.87")
remDr <- rD$client
url <- "https://www.flashscore.com/team/connecticut-huskies/8rqVf3Tj/results/"
remDr$navigate(url)


repeat{
  b <- tryCatch({
    suppressMessages({
      webElemMore <- remDr$findElement(using = 'xpath', 
                        '//*[@id="live-table"]/div[1]/div/div/a')
      webElemMore$clickElement()
    })
  }, error = function(e) e)
  if(inherits(b, "error")) break
}


webElemTime <- remDr$findElements(using = 'xpath', 
                              '//*[@class="event__time"]')
webElemTime <- 
  unlist(lapply(webElemTime, function(x){x$getElementText()}))
webElemTime <- gsub("\\n", " ", webElemTime)

webElemHome <- 
  remDr$findElements(using = 'class', 
                     'event__participant')
webElemHome <- 
  unlist(lapply(webElemHome, function(x){x$getElementText()}))

webElemScore <- 
  remDr$findElements(using = 'class', 'event__score')
webElemScore <- 
  unlist(lapply(webElemScore, function(x){x$getElementText()}))

webElemResult <- 
  remDr$findElements(using = 'class', 'wld')
webElemResult <- 
  unlist(lapply(webElemResult, function(x){x$getElementText()}))

n <- length(webElemHome)
basketball <- 
  data.frame(time = webElemTime,
             Home = webElemHome[seq(n) %% 2 == 1],
             Away = webElemHome[seq(n) %% 2 == 0],
             HomeS = webElemScore[seq(n) %% 2 == 1],
             AwayS = webElemScore[seq(n) %% 2 == 0],
             Result = webElemResult)

head(basketball)
remDr$close()