draft/getLungYen.R

In leoluyi/CRMaddress: Crawling address in Taiwan

getLungYen <- function(){
  #--經銷商取分頁------------------------------------------------------------------------------------
  res <- GET("http://www.lungyengroup.com.tw/Location")
  res <- htmlParse(res, encoding = 'utf8')
  page_num=xpathSApply(res, '//*[@class="pager"]/a', xmlAttrs)
  page_num=length(page_num)
  wantURL=list()
  for(i in 1:page_num)
  {
    wantURL[i] <- sprintf('http://www.lungyengroup.com.tw/Location?page=%s', i)
  }
  
  #--經銷商取內頁------------------------------------------------------------------------------------
  OUTPUT=list()
  for(i in 1:page_num)
  {
    res=htmlParse(wantURL[i], encoding='utf8')
    store_nm=xpathSApply(res, '//*[@id="textList"]/li/a/span[@class="infoTxt"]', xmlValue)
    tel_no=xpathSApply(res, '//*[@id="textList"]/li/a/span[@class="phone"]', xmlValue)
    fax_no=xpathSApply(res, '//*[@id="textList"]/li/a/span[@class="fax"]', xmlValue)
    addr=xpathSApply(res, '//*[@id="textList"]/li/a/span[@class="addr"]', xmlValue)
    OUTPUT[[i]]= data.frame(brand_nm='龍巖',type='經銷商',store_nm=store_nm,addr=addr,tel_no=tel_no,fax_no=fax_no, data_dt=gsub('-','',Sys.Date()) ,stringsAsFactors=FALSE )
  }
  FINISH1=do.call(rbind,OUTPUT)
  
  #-----------------------------------------------------------------------------------
  
  
  #--禮儀服務處取內頁------------------------------------------------------------------------------------
  res=htmlParse('http://www.lungyengroup.com.tw/Location?type=1&area=0', encoding='utf8')
  store_nm=xpathSApply(res, '//*[@id="textList"]/li/a/span[@class="infoTxt"]', xmlValue)
  tel_no=xpathSApply(res, '//*[@id="textList"]/li/a/span[@class="phone"]', xmlValue)
  fax_no=xpathSApply(res, '//*[@id="textList"]/li/a/span[@class="fax"]', xmlValue)
  addr=xpathSApply(res, '//*[@id="textList"]/li/a/span[@class="addr"]', xmlValue)
  OUTPUT= list(data.frame(brand_nm='龍巖',type='禮儀服務處',store_nm=store_nm,addr=addr,tel_no=tel_no,fax_no=fax_no, data_dt=gsub('-','',Sys.Date()) ,stringsAsFactors=FALSE ))
  FINISH2=do.call(rbind,OUTPUT)
  
  
  
  #--客服櫃台取內頁------------------------------------------------------------------------------------
  res=htmlParse('http://www.lungyengroup.com.tw/Location?type=2&area=0', encoding='utf8')
  store_nm=xpathSApply(res, '//*[@id="textList"]/li/a/span[@class="infoTxt"]', xmlValue)
  tel_no=xpathSApply(res, '//*[@id="textList"]/li/a/span[@class="phone"]', xmlValue)
  fax_no=xpathSApply(res, '//*[@id="textList"]/li/a/span[@class="fax"]', xmlValue)
  addr=xpathSApply(res, '//*[@id="textList"]/li/a/span[@class="addr"]', xmlValue)
  OUTPUT= list(data.frame(brand_nm='龍巖',type='客服櫃台',store_nm=store_nm,addr=addr,tel_no=tel_no,fax_no=fax_no, data_dt=gsub('-','',Sys.Date()) ,stringsAsFactors=FALSE ))
  FINISH3=do.call(rbind,OUTPUT)
  
  
  
  #--墓園取內頁------------------------------------------------------------------------------------
  res=htmlParse('http://www.lungyengroup.com.tw/Location?type=3&area=0', encoding='utf8')
  store_nm=xpathSApply(res, '//*[@id="textList"]/li/a/span[@class="infoTxt"]', xmlValue)
  tel_no=xpathSApply(res, '//*[@id="textList"]/li/a/span[@class="phone"]', xmlValue)
  fax_no=xpathSApply(res, '//*[@id="textList"]/li/a/span[@class="fax"]', xmlValue)
  addr=xpathSApply(res, '//*[@id="textList"]/li/a/span[@class="addr"]', xmlValue)
  OUTPUT= list(data.frame(brand_nm='龍巖',type='墓園塔位',store_nm=store_nm,addr=addr,tel_no=tel_no,fax_no=fax_no, data_dt=gsub('-','',Sys.Date()) ,stringsAsFactors=FALSE ))
  FINISH4=do.call(rbind,OUTPUT)
  
  
  #--會館取內頁------------------------------------------------------------------------------------
  res=htmlParse('http://www.lungyengroup.com.tw/Location?type=4&area=0', encoding='utf8')
  store_nm=xpathSApply(res, '//*[@id="textList"]/li/a/span[@class="infoTxt"]', xmlValue)
  tel_no=xpathSApply(res, '//*[@id="textList"]/li/a/span[@class="phone"]', xmlValue)
  fax_no=xpathSApply(res, '//*[@id="textList"]/li/a/span[@class="fax"]', xmlValue)
  addr=xpathSApply(res, '//*[@id="textList"]/li/a/span[@class="addr"]', xmlValue)
  OUTPUT= list(data.frame(brand_nm='龍巖',type='會館',store_nm=store_nm,addr=addr,tel_no=tel_no,fax_no=fax_no, data_dt=gsub('-','',Sys.Date()) ,stringsAsFactors=FALSE ))
  FINISH5=do.call(rbind,OUTPUT)
  
  FINISH=rbind(FINISH1,FINISH2,FINISH3,FINISH4,FINISH5)
}