bobaedream.R

library(rvest)
library(stringr)

dat<-c()
for (i in 1:2005){
  url<-paste0("http://www.bobaedream.co.kr/cyber/CyberCar.php?gubun=K&page=",i)
  usedCar <- read_html(url)
  title <-
    usedCar %>% 
    html_nodes("td.carinfo a.title") %>% 
    html_text
  
  info <-
    usedCar %>% 
    html_nodes("a.sub_01") %>% 
    html_text %>% 
    strsplit("ㅣ") %>% 
    unlist %>% 
    matrix(ncol=3,byrow=T)
  
  transmission <- info[,1]
  fuel <- info[,2]
  distance <- 
    info[,3] %>% 
    str_replace(" [a-z]+","") %>% 
    str_replace_all(",","") %>% 
    as.numeric
  
  price <- 
    usedCar %>% 
    html_nodes("td.price em") %>% 
    html_text %>% 
    str_replace_all(",","") %>% 
    as.numeric
  
  tem<-data.frame(title,transmission,fuel,distance,price)
  dat<-rbind(dat,tem)
  print(i)
}