draft/getComeBuy.R

getComeBuy <- function(){
  
  #各區
  a<-'http://www.comebuy2002.com.tw/store/store.php?M=0&uID=1&Name=%E5%8C%97%E5%8D%80'
  b<-'http://www.comebuy2002.com.tw/store/store.php?M=0&uID=1&Name=%E4%B8%AD%E5%8D%80'
  c<-'http://www.comebuy2002.com.tw/store/store.php?M=0&uID=1&Name=%E5%8D%97%E5%8D%80'
  d<-'http://www.comebuy2002.com.tw/store/store.php?M=0&uID=1&Name=%E6%9D%B1%E5%8D%80'
  e<-'http://www.comebuy2002.com.tw/store/store.php?M=0&uID=1&Name=%E9%9B%A2%E5%B3%B6'
  
  
  #取得內容
  
  urCity_res <- GET(a)
  urCity_res <- content(urCity_res, encoding = 'utf8')
  area <- xpathSApply(urCity_res, '/html/body/table/tr/td/table[5]/tr/td/table/tr/td[3]/table[2]/tr/td/table/tr/td/table/tr/td/table/tr/td/table[2]/tr/td/table/tr/td[1]', xmlValue)
  store_nm <- xpathSApply(urCity_res, '/html/body/table/tr/td/table[5]/tr/td/table/tr/td[3]/table[2]/tr/td/table/tr/td/table/tr/td/table/tr/td/table[2]/tr/td/table/tr/td[2]', xmlValue)
  addr <- xpathSApply(urCity_res, '/html/body/table/tr/td/table[5]/tr/td/table/tr/td[3]/table[2]/tr/td/table/tr/td/table/tr/td/table/tr/td/table[2]/tr/td/table/tr/td[3]', xmlValue)
  addr <- str_replace_all(addr, '[:space:]', '')
  tel_no <- xpathSApply(urCity_res, '/html/body/table/tr/td/table[5]/tr/td/table/tr/td[3]/table[2]/tr/td/table/tr/td/table/tr/td/table/tr/td/table[2]/tr/td/table/tr/td[4]', xmlValue)
  tel_no<- str_replace_all(tel_no, '-|\\(|\\)| ', '')
  tel_no <- str_extract(tel_no, '[0-9]+')
  data<-data.frame(brand_nm='comebuy',area, store_nm, tel_no, addr, stringsAsFactors = FALSE)
  
  urCity_res1 <- GET(b)
  urCity_res1 <- content(urCity_res1, encoding = 'utf8')
  area1 <- xpathSApply(urCity_res1, '/html/body/table/tr/td/table[5]/tr/td/table/tr/td[3]/table[2]/tr/td/table/tr/td/table/tr/td/table/tr/td/table[2]/tr/td/table/tr/td[1]', xmlValue)
  store_nm1 <- xpathSApply(urCity_res1, '/html/body/table/tr/td/table[5]/tr/td/table/tr/td[3]/table[2]/tr/td/table/tr/td/table/tr/td/table/tr/td/table[2]/tr/td/table/tr/td[2]', xmlValue)
  addr1 <- xpathSApply(urCity_res1, '/html/body/table/tr/td/table[5]/tr/td/table/tr/td[3]/table[2]/tr/td/table/tr/td/table/tr/td/table/tr/td/table[2]/tr/td/table/tr/td[3]', xmlValue)
  addr1 <- str_replace_all(addr1, '[:space:]', '')
  tel_no1 <- xpathSApply(urCity_res1, '/html/body/table/tr/td/table[5]/tr/td/table/tr/td[3]/table[2]/tr/td/table/tr/td/table/tr/td/table/tr/td/table[2]/tr/td/table/tr/td[4]', xmlValue)
  tel_no1<- str_replace_all(tel_no1, '-|\\(|\\)| ', '')
  tel_no1 <- str_extract(tel_no1, '[0-9]+')
  data1<-data.frame(brand_nm='comebuy',area=area1, store_nm=store_nm1, tel_no=tel_no1, addr=addr1, stringsAsFactors = FALSE)
  
  urCity_res2 <- GET(c)
  urCity_res2 <- content(urCity_res2, encoding = 'utf8')
  area2 <- xpathSApply(urCity_res2, '/html/body/table/tr/td/table[5]/tr/td/table/tr/td[3]/table[2]/tr/td/table/tr/td/table/tr/td/table/tr/td/table[2]/tr/td/table/tr/td[1]', xmlValue)
  store_nm2 <- xpathSApply(urCity_res2, '/html/body/table/tr/td/table[5]/tr/td/table/tr/td[3]/table[2]/tr/td/table/tr/td/table/tr/td/table/tr/td/table[2]/tr/td/table/tr/td[2]', xmlValue)
  addr2 <- xpathSApply(urCity_res2, '/html/body/table/tr/td/table[5]/tr/td/table/tr/td[3]/table[2]/tr/td/table/tr/td/table/tr/td/table/tr/td/table[2]/tr/td/table/tr/td[3]', xmlValue)
  addr2 <- str_replace_all(addr2, '[:space:]', '')
  tel_no2 <- xpathSApply(urCity_res2, '/html/body/table/tr/td/table[5]/tr/td/table/tr/td[3]/table[2]/tr/td/table/tr/td/table/tr/td/table/tr/td/table[2]/tr/td/table/tr/td[4]', xmlValue)
  tel_no2<- str_replace_all(tel_no2, '-|\\(|\\)| ', '')
  tel_no2 <- str_extract(tel_no2, '[0-9]+')
  data2<-data.frame(brand_nm='comebuy',area=area2, store_nm=store_nm2, tel_no=tel_no2, addr=addr2, stringsAsFactors = FALSE)
  
  urCity_res3 <- GET(d)
  urCity_res3 <- content(urCity_res3, encoding = 'utf8')
  area3 <- xpathSApply(urCity_res3, '/html/body/table/tr/td/table[5]/tr/td/table/tr/td[3]/table[2]/tr/td/table/tr/td/table/tr/td/table/tr/td/table[2]/tr/td/table/tr/td[1]', xmlValue)
  store_nm3 <- xpathSApply(urCity_res3, '/html/body/table/tr/td/table[5]/tr/td/table/tr/td[3]/table[2]/tr/td/table/tr/td/table/tr/td/table/tr/td/table[2]/tr/td/table/tr/td[2]', xmlValue)
  addr3 <- xpathSApply(urCity_res3, '/html/body/table/tr/td/table[5]/tr/td/table/tr/td[3]/table[2]/tr/td/table/tr/td/table/tr/td/table/tr/td/table[2]/tr/td/table/tr/td[3]', xmlValue)
  addr3 <- str_replace_all(addr3, '[:space:]', '')
  tel_no3 <- xpathSApply(urCity_res3, '/html/body/table/tr/td/table[5]/tr/td/table/tr/td[3]/table[2]/tr/td/table/tr/td/table/tr/td/table/tr/td/table[2]/tr/td/table/tr/td[4]', xmlValue)
  tel_no3<- str_replace_all(tel_no3, '-|\\(|\\)| ', '')
  tel_no3 <- str_extract(tel_no3, '[0-9]+')
  data3<-data.frame(brand_nm='comebuy',area=area3, store_nm=store_nm3, tel_no=tel_no3, addr=addr3, stringsAsFactors = FALSE)
  
  urCity_res4 <- GET(e)
  urCity_res4 <- content(urCity_res4, encoding = 'utf8')
  area4 <- xpathSApply(urCity_res4, '/html/body/table/tr/td/table[5]/tr/td/table/tr/td[3]/table[2]/tr/td/table/tr/td/table/tr/td/table/tr/td/table[2]/tr/td/table/tr/td[1]', xmlValue)
  store_nm4 <- xpathSApply(urCity_res4, '/html/body/table/tr/td/table[5]/tr/td/table/tr/td[3]/table[2]/tr/td/table/tr/td/table/tr/td/table/tr/td/table[2]/tr/td/table/tr/td[2]', xmlValue)
  addr4 <- xpathSApply(urCity_res4, '/html/body/table/tr/td/table[5]/tr/td/table/tr/td[3]/table[2]/tr/td/table/tr/td/table/tr/td/table/tr/td/table[2]/tr/td/table/tr/td[3]', xmlValue)
  addr4 <- str_replace_all(addr4, '[:space:]', '')
  tel_no4 <- xpathSApply(urCity_res4, '/html/body/table/tr/td/table[5]/tr/td/table/tr/td[3]/table[2]/tr/td/table/tr/td/table/tr/td/table/tr/td/table[2]/tr/td/table/tr/td[4]', xmlValue)
  tel_no4<- str_replace_all(tel_no4, '-|\\(|\\)| ', '')
  tel_no4 <- str_extract(tel_no4, '[0-9]+')
  data4<-data.frame(brand_nm='comebuy',area=area4, store_nm=store_nm4, tel_no=tel_no4, addr=addr4, stringsAsFactors = FALSE)
  
  #合併
  data_fin <- rbind(data,data1,data2,data3,data4)
  
}
leoluyi/CRMaddress documentation built on May 21, 2019, 5:08 a.m.