R/getFamily.R

In leoluyi/CRMaddress: Crawling address in Taiwan

#' FamilyMart
#' 
#' @seealso 
#' url: \url{http://www.family.com.tw/marketing/inquiry.aspx} \cr
#' data: \url{http://api.map.com.tw/net/familyShop.aspx}
#' 
#' @return data.table
#' @export
getFamilyMart <- function() {
  url = "http://api.map.com.tw/net/familyShop.aspx"
  res = GET(url,
            add_headers(
              Referer="http://www.family.com.tw/marketing/inquiry.aspx"
            ),
            query = list(
              searchType = "ShopName",
              type = "",
              kw = "%",
              fun = "getByName",
              key = "6F30E8BF706D653965BDE302661D1241F8BE9EBC"))
  res_text = content(res, "text")
  
  jsonDataString <- res_text %>%
    sub("^[^\\[]*","",.) %>%
    sub("[^\\]]*$","",.)
  
  dt_FamilyMart <- jsonlite::fromJSON(jsonDataString) %>% 
    as.data.table()#%>% dplyr::tbl_dt()
  
  # ## check misc items
  # unique(unlist(strsplit(as.character(dt_FamilyMart$all),',')))
  
  # # cooffie
  # COFFEE = ifelse(str_detect(ifelse(
  #   is.na(dt_FamilyMart$all), '', as.character(dt_FamilyMart$all)
  # ), 'COFFEE'), 'Y', 'N')
  # #夯蕃薯
  # SWEETPOTATO = ifelse(str_detect(ifelse(
  #   is.na(temp3$all), '', as.character(temp3$all)
  # ), 'SWEETPOTATO'), 'Y', 'N')
  # #相片咅可得
  # Photo = ifelse(str_detect(ifelse(
  #   is.na(temp3$all), '', as.character(temp3$all)
  # ), 'Photo'), 'Y', 'N')
  # WiFi = ifelse(str_detect(ifelse(
  #   is.na(temp3$all), '', as.character(temp3$all)
  # ), 'WiFi'), 'Y', 'N')
  # Toilet = ifelse(str_detect(ifelse(
  #   is.na(temp3$all), '', as.character(temp3$all)
  # ), 'Toilet'), 'Y', 'N')
  # #休憩區
  # Rest = ifelse(str_detect(ifelse(
  #   is.na(temp3$all), '', as.character(temp3$all)
  # ), 'Rest'), 'Y', 'N')
  # #生鮮蔬菜
  # veg = ifelse(str_detect(ifelse(
  #   is.na(temp3$all), '', as.character(temp3$all)
  # ), 'veg'), 'Y', 'N')
  # #烈日鬆餅
  # muffin = ifelse(str_detect(ifelse(
  #   is.na(temp3$all), '', as.character(temp3$all)
  # ), 'muffin'), 'Y', 'N')
  # #雙口味霜淇淋
  # twoice = ifelse(str_detect(ifelse(
  #   is.na(temp3$all), '', as.character(temp3$all)
  # ), 'twoice'), 'Y', 'N')
  # # 單口味霜淇淋
  # oneice = ifelse(str_detect(ifelse(
  #   is.na(temp3$all), '', as.character(temp3$all)
  # ), 'oneice'), 'Y', 'N')
  # #停車場
  # Parking = ifelse(str_detect(ifelse(
  #   is.na(temp3$all), '', as.character(temp3$all)
  # ), 'Parking'), 'Y', 'N')


  ## rename variables
  setnames(dt_FamilyMart, 
           c("addr", "NAME", "TEL", "px", "py"),
           c("addr", "store_nm", "tel_no", "lon_x", "lat_y"))
  
  dt_FamilyMart[, store_nm := sub("^全家", "", store_nm)]
  
  # add brand name
  dt_FamilyMart[, `:=`(brand_nm = "全家", keyword = "Family Mart")]
  
  ## add url, time, full name
  dt_FamilyMart[, store_url := "http://www.family.com.tw/"][
    , data_time := format(Sys.time(), "%Y-%m-%dT%H:%M:%S%z")][
      , full_nm := paste0(brand_nm, store_nm)]
  
  key_var <- c("full_nm", "brand_nm", "keyword", "store_nm", "addr",
               "lon_x", "lat_y", "store_url")
  ## move brand_nm, key_word to first two colmuns
  setcolorder(dt_FamilyMart, c(key_var, setdiff(names(dt_FamilyMart), key_var)))
  
  dt_FamilyMart
}