In deepfriar/dryscrape: Scrapes PxP Data For NHL Games

knitr::opts_chunk$set(
  eval = FALSE,
  include = FALSE,
  error = FALSE,
  warning = FALSE,
  message = FALSE,
  collapse = TRUE,
  comment = "#>"
)

foo <- dplyr::tibble(season = paste0(2010:2018, 2011:2019))
foo <- dplyr::group_by(foo, .data$season)
foo <- dplyr::mutate(foo, players = list(dryscrape:::ds.get_all_players(dplyr::first(.data$season))))
foo <- dplyr::mutate(foo, players = list(dryscrape:::ds.parse_all_players(dplyr::first(.data$players))))
foo <- tidyr::unnest(foo)
oof <- foo

bar <- dplyr::summarise(dplyr::group_by_all(dplyr::select(dplyr::group_by(foo), .data$id)))
bar <- dplyr::group_by_all(bar)
bar <- dplyr::mutate(bar,
                     prof = list(
                       {
                         function(i) {
                           Sys.sleep(2)
                           dryscrape::ds.get_player_profile(i)
                         }
                       }(dplyr::first(.data$id))
                     ))
bar <- dplyr::mutate(bar, prof = list(dryscrape:::ds.parse_profile(dplyr::first(.data$prof))))
bar <- tidyr::unnest(bar)

bar <- dplyr::mutate(bar,
                     first_name = toupper(.data$firstName),
                     last_name  = toupper(.data$lastName))

saveRDS(bar, "profiles.rds")

bar <- readRDS("profiles.rds")

foo <- oof
foo <- dplyr::left_join(foo, bar)
foo <- dplyr::mutate(foo, team = qdap::multigsub(c("NJD", "PHX", "LAK", "TBL", "SJS"),
                                                 c("N.J", "ARI", "L.A", "T.B", "S.J"),
                                                 .data$team))
foo <- dplyr::mutate(foo, player_name = paste0(.data$first_name, ".", .data$last_name))
foo <- dplyr::select(foo, -.data$first_name, -.data$last_name)
foo <- dplyr::mutate(foo, player_name = dryscrape::ds.fix_names(.data$player_name))
foo <- dryscrape::ds.5ebastian(foo)

compiled_games <- dplyr::tibble(season = paste0(2010:2018, 2011:2019))
compiled_games <- dplyr::mutate(compiled_games, filename = paste0("~/nhlgamedata/compiled/", .data$season, ".rds"))
compiled_games <- dplyr::group_by(compiled_games, .data$season)
compiled_games <- dplyr::mutate(compiled_games, games = list(readRDS(dplyr::first(.data$filename))))

compiled_games <- dplyr::transmute(compiled_games, roster = list(getElement(dplyr::first(.data$games), "Roster")))
compiled_games <- tidyr::unnest(compiled_games)

compiled_games <- dplyr::mutate(compiled_games, player_name = dryscrape::ds.mung_names(.data$player_name))
compiled_games <- dryscrape::ds.5ebastian(compiled_games)

setdiff(compiled_games$player_name, foo$player_name)

# TODO: who tf is jamie phillips, the phantom of game 2 for 2017-18 WPG against MNW
saveRDS(foo, "player_ages.rds")

player_ages <- readRDS("player_ages.rds")
# player_ages <- salo::player_ages

player_ages <- dplyr::group_by(player_ages)
player_ages <- dplyr::select(player_ages, -.data$team, -.data$id1, -.data$pos, -.data$season)
player_ages <- dplyr::group_by_all(player_ages)
player_ages <- dplyr::summarise(player_ages)
player_ages <- dplyr::group_by(player_ages, .data$player_name)
player_ages <- dplyr::mutate(player_ages, birthDate = list(data.frame(val = unlist(stringr::str_split(dplyr::first(.data$birthDate), "-")),
                                                                      var = c("birthYear", "birthMonth", "birthDay"),
                                                                      stringsAsFactors = FALSE)))
player_ages <- tidyr::unnest(player_ages)
player_ages <- tidyr::spread(player_ages, .data$var, .data$val)

save(player_ages, file="../../salo/data/player_ages.RData", compress="xz")