draft/getFishNoodle.R

In leoluyi/CRMaddress: Crawling address in Taiwan

getFishNoodle <- function(){
  
  
  url_link <-'http://www.fishnoodle.com.tw/shopsinfo'
  res_link <- GET(url_link)
  res_link <- content(res_link,'text', encoding = 'utf8')
  html_link<- htmlParse(res_link, encoding = "utf8")
  
  link <- cssApply(html_link, '#food_bg>div>a ', cssLink)
  link <- str_replace_all(link,'/shopsinfo/','')
  
  
  
  result<-list()
  url   <-list()
  
  for(i in 1:length(link)){
    
    url[[i]]<-sprintf('http://www.fishnoodle.com.tw/shopsinfo/%s',link[i])
    res <- GET(url[[i]], encoding= "big5")
    
    url<-'http://www.fishnoodle.com.tw/shopsinfo/taipei'
    res <- GET(url)
    res <- content(res,'text', encoding = 'utf8')
    html <- htmlParse(res, encoding = "utf8")
    data <- cssApply(html, '#add', cssCharacter)
    data <- str_replace_all(data,'\\\r|\\\n|\\\t','')
    data <- unlist(strsplit(data,'：')) #用\n分隔資料
    data <- matrix(data, ncol=3, byrow=TRUE)
    
    store_nm <-data[,1]
    tel_no   <-data[,3]
    addr     <-data[,2]
    addr     <- str_replace_all(addr,'訂餐專線|美食街','')
    
    if(length(store_nm)>0){
      result[[i]]<-data.frame(brand_nm='甘泉魚麵',store_nm,tel_no,addr, stringsAsFactors = FALSE)
    }
    
  }
  
  data_fin <- do.call(rbind, result)
}