#' Download data from AEMO website
#'
#' Trading interval level price and demand data is available on the AEMO
#' website. This function downloads these files for each region and period
#' requested.
#'
#' [AEMO's copyright permissions](https://www.aemo.com.au/privacy-and-legal-notices/copyright-permissions)
#' specify that:
#'
#' "AEMO confirms its general permission for anyone to use AEMO Material
#' for any purpose, but only with accurate and appropriate attribution of the
#' relevant AEMO Material and AEMO as its author".
#'
#' @param regions a case invariant character vector of abbreviated AEMO regions.
#' Must be one of `"nsw"`, `"vic"`, `"sa"`, `"qld"`,
#' `"snowy"` or `"tas"`. Snowy region data is only available up
#' to and including Jun 2008 while Tasmania region data is available from
#' and including May 2005. Otherwise, all regions have data available from and
#' including Dec 1998 up to the end of the month previous to the current month.
#' @param years a numeric vector of years in the form `YYYY`. See the
#' `regions` parameter for more info.
#' @param months a numeric vector of months in the form `MM`.See the
#' `regions` parameter for more info.
#' @param path defaults to `"."` (current working directory) and denotes
#' the path to which the AEMO files are downloaded.
#' @return data files whose number corresponds to the maximum length of one of
#' the parameters. The file names have the pattern: "[REGION][YYYY][MM].csv".
#' @examples
#' \dontrun{
#' get_aemo_data("nsw", 2014, 1)
#' }
#' @references
#' [AEMO website](https://www.aemo.com.au/energy-systems/electricity/national-electricity-market-nem/data-nem/aggregated-data)
#' @export

get_aemo_data <- function(regions, years, months, path = ".") {
  regions <- toupper(regions)
  periods <- as.numeric(paste0(years, formatC(months, width = 2, flag = "0")))
  assertthat::assert_that(
    all(regions %in% aemo_regions()),
    all(periods %in% aemo_periods())
  )
  url <- aemo_data_url(regions, years, months)
  destfile <- file.path(path, aemo_data_file_name(regions, years, months))
  invisible(Map(utils::download.file, url = url, destfile = destfile))
}

#' Collate AEMO data
#'
#' This function collates AEMO data files that are on the `path`. These
#' are identified by having a filename of the form
#' `[REGION][YYYY][MM].csv` (e.g. `"NSW201405.csv"`). If no such
#' file exists on `path`, all available data files are downloaded from the
#' AEMO website using [get_aemo_data()].
#'
#' @param path a string containing the location of the AEMO (CSV) data files
#' that you wish to collate. These must have filenames of the form specified
#' above.
#' @param remove_files a boolean flag indicating whether the AEMO CSV files
#' should be removed once the data sets are collated.
#' @return a data frame containing five fields: `"REGION"` (factor),
#' `"SETTLEMENTDATE"` (POSIXct), `"TOTALDEMAND"` (numeric),
#' `"RRP"` (numeric) and `"PERIODTYPE"` (factor).
#' @examples
#' \dontrun{
#' collate_aemo_data()
#' }
#' @export

collate_aemo_data <- function(path = ".", remove_files = TRUE) {
  aemo_files <- list_aemo_data_files(path)
  if (identical(aemo_files, character())) {
    info <- all_regions_periods_combos()
    message("Downloading data from AEMO....")
    get_aemo_data(info$region, info$years, info$months, path)
    aemo_files <- list_aemo_data_files(path)
  }
  message("Reading AEMO data files...")
  aemo_dfs <- lapply(aemo_files, utils::read.csv, stringsAsFactors = FALSE)
  if (remove_files) {
    message("Removing AEMO data files...")
    clean_up_aemo_data_files(path)
  }
  message('Collating AEMO data...')
  aemo <- dplyr::bind_rows(aemo_dfs)
  message('Formatting data frame...')
  dplyr::mutate(aemo,
    .dots = stats::setNames(
      list(
        as.factor(~REGION),
        lubridate::ymd_hms(~SETTLEMENTDATE, truncated = 1),
        as.factor(~PERIODTYPE)
      ),
      c("REGION", "SETTLEMENTDATE", "PERIODTYPE")
    )
  )
}

#' Remove AEMO CSV files
#'
#' This function removes AEMO data files that are on the `path`. These
#' are identified by having a filename of the form
#' `[REGION][YYYY][MM].csv` (e.g. `"NSW201405.csv"`).
#'
#' @inheritParams collate_aemo_data
#' @return no return value. However, AEMO files that are identified are removed.
#' @export

clean_up_aemo_data_files <- function(path = ".") {
  file.remove(list_aemo_data_files(path))
}

#' AEMO data set May 2009 - May 2014
#'
#' A sample data set containing trading interval price and demand data sourced
#' from the AEMO website on or about 21 June 2014. This was downloaded and
#' collated using [collate_aemo_data()].
#'
#' @format `"REGION"` (factor), `"SETTLEMENTDATE"` (POSIXct),
#' `"TOTALDEMAND"` (numeric), `"RRP"` (numeric) and
#' `"PERIODTYPE"` (factor).
#' @references
#' [AEMO website](https://www.aemo.com.au/energy-systems/electricity/national-electricity-market-nem/data-nem/aggregated-data)
#' @examples
#' data(aemo)
#' head(aemo)
"aemo"

aemo_regions <- function() {
  c("NSW", "QLD", "VIC", "SA", "TAS", "SNOWY")
}

aemo_periods <- function() {
  yy_seq <- 1998:(lubridate::year(lubridate::today()))
  period_min <- 199812
  period_max <- as.numeric(paste0(
    lubridate::year(lubridate::today()),
    formatC(lubridate::month(lubridate::today()) - 1, width = 2, flag = "0")
  ))
  all_periods <- outer(yy_seq, formatC(1:12, width = 2, flag = "0"), paste0)
  all_periods <- sort(as.numeric(all_periods))
  in_period <- all_periods >= period_min & all_periods <= period_max
  all_periods[in_period]
}

aemo_data_url_stub <- function() {
  "https://www.aemo.com.au/aemo/data/nem/priceanddemand/PRICE_AND_DEMAND_"
}

aemo_data_url <- function(regions, years, months) {
  # https://www.aemo.com.au/aemo/data/nem/priceanddemand/PRICE_AND_DEMAND_199911_NSW1.csv
  paste0(
    aemo_data_url_stub(),
    years,
    formatC(months, width = 2, flag = "0"),
    "_",
    toupper(regions),
    "1.csv"
  )
}

aemo_data_file_name <- function(regions, years, months) {
  paste0(
    toupper(regions),
    years,
    formatC(months, width = 2, flag = "0"),
    ".csv"
  )
}

list_aemo_data_files <- function(path = ".") {
  list.files(path, "(NSW|QLD|VIC|SA|TAS|SNOWY)[[:digit:]]{6}\\.csv")
}

all_regions_periods_combos <- function() {
  all_periods <- aemo_periods()
  regions <- rep(aemo_regions(), each = NROW(all_periods))
  years <- as.numeric(rep(
    stringr::str_sub(all_periods, 1, 4),
    NROW(aemo_regions())
  ))
  months <- as.numeric(rep(
    stringr::str_sub(all_periods, 5),
    NROW(aemo_regions())
  ))
  is_invalid_tas <- (years * 100 + months < 200505) & regions == "TAS"
  is_invalid_snowy <- (years * 100 + months > 200806) & regions == "SNOWY"
  is_invalid_tas_snowy <- is_invalid_tas | is_invalid_snowy
  list(
    regions = regions[!is_invalid_tas_snowy],
    years = years[!is_invalid_tas_snowy],
    months = months[!is_invalid_tas_snowy]
  )
}