draft/getGolfshop.R

getGolfshop <- function(){
  wantURL <- 'http://www.golfshop.com.tw/golfcourse/menu41.asp'
  
  res=htmlParse(wantURL, encoding='big5')
  store_nm <- xpathSApply(res, '/html/body/table/tr/td/div/table/tr/td/div/table/tr/td/table/tr/td/table/tbody/tr/td[1]', xmlValue)
  store_nm=store_nm[regexpr("練習場名稱",store_nm)==-1]
  addr <- xpathSApply(res, '/html/body/table/tr/td/div/table/tr/td/div/table/tr/td/table/tr/td/table/tbody/tr/td[2]', xmlValue)
  addr=addr[regexpr("地址",addr)==-1]
  addr <- str_replace_all(addr, '(\r|\n| |\t)+', '')
  tel_no <- xpathSApply(res, '/html/body/table/tr/td/div/table/tr/td/div/table/tr/td/table/tr/td/table/tbody/tr/td[3]', xmlValue)
  tel_no=tel_no[regexpr("電話",tel_no)==-1]
  
  FINISH= data.frame(brand_nm='邁達康',store_nm=store_nm,addr=addr,tel_no=tel_no, data_dt=gsub('-','',Sys.Date()) ,stringsAsFactors=FALSE )
  return(FINISH)
}
leoluyi/CRMaddress documentation built on May 21, 2019, 5:08 a.m.