draft/getFujiBike.R

In leoluyi/CRMaddress: Crawling address in Taiwan

getFujiBike <- function(){
  res <- GET("http://www.fujibikes.com.tw/DealerList.aspx")
  res <- htmlParse(res, encoding = 'utf8')
  
  #金質店
  store_nm=xpathSApply(res, '//*[@id="GridView2"]/tr/td[1]', xmlValue)
  store_nm <- str_replace_all(store_nm, '(\r|\n| |　)+', '')
  addr=xpathSApply(res, '//*[@id="GridView2"]/tr/td[2]', xmlValue)
  addr <- str_replace_all(addr, '(\r|\n| |　)+', '')
  tel_no=xpathSApply(res, '//*[@id="GridView2"]/tr/td[3]', xmlValue)
  tel_no <- str_replace_all(tel_no, '(\r|\n| |　)+', '')
  FINISH1= data.frame(brand_nm='富士自行車',type='金質店',store_nm=store_nm,addr=addr,tel_no=tel_no,data_dt=gsub('-','',Sys.Date()) ,stringsAsFactors=FALSE )
  
  #一級店
  store_nm=xpathSApply(res, '//*[@id="GridView3"]/tr/td[1]', xmlValue)
  store_nm <- str_replace_all(store_nm, '(\r|\n| |　)+', '')
  addr=xpathSApply(res, '//*[@id="GridView3"]/tr/td[2]', xmlValue)
  addr <- str_replace_all(addr, '(\r|\n| |　)+', '')
  tel_no=xpathSApply(res, '//*[@id="GridView3"]/tr/td[3]', xmlValue)
  tel_no <- str_replace_all(tel_no, '(\r|\n| |　)+', '')
  FINISH2= data.frame(brand_nm='富士自行車',type='一級店',store_nm=store_nm,addr=addr,tel_no=tel_no,data_dt=gsub('-','',Sys.Date()) ,stringsAsFactors=FALSE )
  
  #一般店
  store_nm=xpathSApply(res, '//*[@id="GridView4"]/tr/td[1]', xmlValue)
  store_nm <- str_replace_all(store_nm, '(\r|\n| |　)+', '')
  addr=xpathSApply(res, '//*[@id="GridView4"]/tr/td[2]', xmlValue)
  addr <- str_replace_all(addr, '(\r|\n| |　)+', '')
  tel_no=xpathSApply(res, '//*[@id="GridView4"]/tr/td[3]', xmlValue)
  tel_no <- str_replace_all(tel_no, '(\r|\n| |　)+', '')
  FINISH3= data.frame(brand_nm='富士自行車',type='一般店',store_nm=store_nm,addr=addr,tel_no=tel_no,data_dt=gsub('-','',Sys.Date()) ,stringsAsFactors=FALSE )
  
  FINISH=rbind(FINISH1, FINISH2,FINISH3)
  return(FINISH)
}