draft/getEdWin.R

In leoluyi/CRMaddress: Crawling address in Taiwan

getEdWin <- function(){
  mylist <- list()
  for (i in 1:5)
  {
    url<- sprintf("http://www.edwin.com.tw/stores.asp#storeTab%s", i)
    res <- GET(url, encoding='utf8')
    res2 <- content(res,"text",encoding='big5')
    res2 <- toUTF8(res2)
    html <- htmlParse(res2, encoding = "utf8")
    tables <- readHTMLTable(html)
    
    data <- tables[[3]]
    store <- data [,2]
    addr <- data [,3]
    addr1 <-str_replace_all(unlist(addr),'\\(.+\\)', "")
    
    data05=data.frame('edwin', unlist(store), unlist(addr1), stringAsFactors=FALSE) 
    data06 <- data05[,-4]
    
    colnames(data06) = c("brand_nm", "store_nm", "addr")
    data_sum <- data06
    mylist=rbind(mylist,data_sum)
  }
  mylist
}