Crawl Reservoir

library(data.table)
library(stringi)
library(lubridate)
library(foreach)
library(httr)
library(xml2)
library(rvest)
library(pipeR)

# windows才需要先把locale改成C，並用stringi::stri_conv轉成big5
backupLocale <- Sys.getlocale("LC_COLLATE")
Sys.setlocale("LC_ALL", 'C')

timeVec <- seq(ymd("2003/01/01"), ymd("2003/01/11"), 1)
reservoirUrl <-
"http://fhy.wra.gov.tw/ReservoirPage_2011/StorageCapacity.aspx"

tmp <- foreach(time = timeVec, .final = rbindlist) %do% {
  VIEWSTATE <- GET(reservoirUrl) %>>% content %>>%
    xml_find_all("//input[@name='__VIEWSTATE']") %>>% xml_attr("value")

  postBody <- c(list("所有水庫"),
                lapply(c(year, month, day), function(f) f(time)),
                list(VIEWSTATE)) %>>%
    `names<-`(c(paste0("ctl00$cphMain$",
                       c("cboSearch", paste0("ucDate$cbo",
                c("Year", "Month","Day")))),
                "__VIEWSTATE"))
  outTbl <- POST(reservoirUrl, body = postBody) %>>% content %>>%
    xml_find_first("//table[@id='ctl00_cphMain_gvList']") %>>%
    html_table(fill = TRUE) %>>% setDT %>>%
    `[`(j = lapply(.SD, stri_conv, from = "UTF-8", to = "Big5"))
  setnames(outTbl, stri_conv(names(outTbl), "UTF-8", "Big5"))
  fwrite(outTbl, sprintf("result/%s.csv", as.character(time)))
  return(NULL)
}
Sys.setlocale(locale = backupLocale)
fwrite(rbindlist(lapply(list.files("result", "\\.csv$"), fread)), "merge.csv")