draft/getToyota.R

In leoluyi/CRMaddress: Crawling address in Taiwan

getToyota <- function(){
  ##展示中心
  url="http://www.toyota.com.tw/location.aspx"
  res <- GET(url, encoding='utf8')
  res2 <- content(res,"text", encoding='utf8')
  html <- htmlParse(res2, encoding = "utf8")
  tables <- readHTMLTable(html)
  
  data=str_extract_all(res2, 'address: .+ key')
  data01 <-str_replace_all(unlist(data),'address|:|,|key', "")
  data02 <-str_replace_all(unlist(data01),'|[:space:]', "")
  data03 <-str_replace_all(unlist(data02),"'|[:space:]" , "")
  
  data21=str_extract_all(res2, 'store: .+所')
  data22 <-str_replace_all(unlist(data21),"store:|'|[:space:]", "")
  
  data31=str_extract_all(res2, 'tel:.+dealer')
  data32 <-str_replace_all(unlist(data31),"tel:|'|[:space:]|dealer|,", "")
  
  data41=str_extract_all(res2, 'dealer:.+zip')
  data42 <-str_replace_all(unlist(data41),"dealer:|'|[:space:]|zip|,", "")
  
  data51=str_extract_all(res2, 'dat:.+tel')
  data52 <-str_replace_all(unlist(data51),"dat:|'|[:space:]|tel|,", "")
  data53 <-str_replace_all(unlist(data52),"週一至週日", "")
  
  
  data91=data.frame("Toyota_展示中心", data22, data42, data03, data32, data53, stringAsFactors=FALSE)
  data92 <- data91[,-7]
  
  colnames(data92) = c("brand_nm", "store_nm", "desc", "addr", "tel_no","time")
  
  
  ##維修中心
  url="http://www.toyota.com.tw/dealer.aspx"
  res <- GET(url, encoding='utf8')
  res2 <- content(res,"text", encoding='utf8')
  html <- htmlParse(res2, encoding = "utf8")
  tables <- readHTMLTable(html)
  
  data=str_extract_all(res2, 'address: .+ key')
  data01 <-str_replace_all(unlist(data),'address|:|,|key', "")
  data02 <-str_replace_all(unlist(data01),'|[:space:]', "")
  data03 <-str_replace_all(unlist(data02),"'|[:space:]" , "")
  
  data21=str_extract_all(res2, 'store: .+廠')
  data22 <-str_replace_all(unlist(data21),"store:|'|[:space:]|(<imgsrc.+)", "")
  data23 <-str_replace_all(unlist(data22),"<i.+致電服務廠", "")
  
  data31=str_extract_all(res2, 'tel:.+dealer')
  data32 <-str_replace_all(unlist(data31),"tel:|'|[:space:]|dealer|,", "")
  
  data41=str_extract_all(res2, 'dealer:.+zip')
  data42 <-str_replace_all(unlist(data41),"dealer:|'|[:space:]|zip|,", "")
  
  data51=str_extract_all(res2, 'dat:.+tel')
  data52 <-str_replace_all(unlist(data51),"dat:|'|[:space:]|tel|,|<BR>", "")
  
  data93=data.frame("Toyota_維修中心", data23, data42, data03, data32, data52, stringAsFactors=FALSE)
  data94 <- data93[,-7]
  
  colnames(data94) = c("brand_nm", "store_nm", "desc", "addr", "tel_no","time")
  
  data55= rbind(data92 , data94)
  
}