draft/getSampo.R

getSampo <- function(){
  
  wantURL <- 'http://www.sampo.com.tw/storefinder.aspx?storetype=%E7%B6%93%E9%8A%B7%E5%BA%97'
  
  res <- GET(wantURL, encoding='utf8')
  res2 <- htmlParse(res, encoding='utf8')
  
  store_nm=xpathSApply(res2, '//*[@id="StoreList"]/dl/dd/span[1]', xmlValue)
  store_nm <- str_replace_all(store_nm, '(\r|\n| | )+', '')
  addr=xpathSApply(res2, '//*[@id="StoreList"]/dl/dd/span[2]', xmlValue)
  addr <- str_replace_all(addr, '(\r|\n| | )+', '')
  addr <- str_replace_all(addr, '([(]).+$', '')
  tel_no=xpathSApply(res2, '//*[@id="StoreList"]/dl/dd/span[3]', xmlValue)
  tel_no <- str_replace_all(tel_no, '(\r|\n| | )+', '')
  
  OUTPUT= data.frame(brand_nm='聲寶',store_nm=store_nm,addr=addr,tel_no=tel_no, data_dt=gsub('-','',Sys.Date()) ,stringsAsFactors=FALSE )
  
}
leoluyi/CRMaddress documentation built on May 21, 2019, 5:08 a.m.