draft/getYamaha.R

In leoluyi/CRMaddress: Crawling address in Taiwan

getYamaha <- function(){
  # 縣市別
  url      <- 'http://www.yamaha-motor.com.tw/service.aspx'
  htmldoc1 <- content(GET(url, encoding='utf8'), as='text', encoding='utf8')
  htmldoc2 <- htmlParse(htmldoc1, encoding='utf8')
  page     <- xpathSApply(htmldoc2,cssToXpath("#ctl00_ContentPlaceHolder1_selCity > option ") ,xmlAttrs)
  page1    <- na.exclude(str_extract(unlist(page),'C.+'))
  page2    <- as.vector(page1)
  
  
  # 所有頁
  
  DATA3 <- list()
  for (i in 1:length(page2)){
    aa         <- '&selVillage=&selType=&_=1435839451710.htm'
    URL        <- sprintf('http://www.yamaha-motor.com.tw/service_dealer.aspx?selCity=%s%s',page2[i],aa)
    HTML1      <- content(GET(URL, encoding='utf8'), as='text', encoding='utf8')
    HTML2      <- htmlParse(HTML1, encoding='utf8')
    DATA1      <- cssApply(HTML2," ul.body" , cssCharacter)
    DATA2      <- unlist(str_split(DATA1,"\r\n\t\t\t"))
    DATA3[[i]] <- as.data.frame(matrix(DATA2, ncol=5, byrow=TRUE), stringsAsFactors = FALSE)
  }
  
  
  data1 <- do.call(rbind,DATA3)
  data2 <- data.frame('YAMAHA',data1)
  names(data2) <- c('brand_nm','store_nm', 'addr', 'tel_no','aa','aa')
  data2$aa <- NULL 
  data2$aa <- NULL 
  data2
}