draft/getMoma.R

getMoma <- function(){  
  url      <- 'https://www.momaclub.com.tw/store.php'
  htmldoc1 <- content(GET(url, encoding='utf8'), as='text', encoding='utf8')
  htmldoc2 <- htmlParse(htmldoc1, encoding='utf8')
  data1    <- cssApply(htmldoc2,"div > dl  > dd > p" ,cssCharacter)
  
  store_nm <- gsub(" ","",str_sub(data1,1,4))
  tel      <- str_extract(data1,'[0-9]{2}-[0-9]{3,4}-[0-9]{3,4}')
  addr1    <- gsub("\n|地圖| |\\[|\\]","",str_replace(str_replace(data1,'[0-9]{2}-[0-9]{3,4}-[0-9]{3,4}',''),str_sub(data1,1,4),''))
  addr     <- gsub(' ','',addr1)
  
  datas    <- data.frame('MOMA',store_nm,tel,addr)
  names(datas) <- c('brand_nm','store_nm','tel_no','addr')
  datas
}
leoluyi/CRMaddress documentation built on May 21, 2019, 5:08 a.m.