Removes furrr, future dependencies, adds Rcpp, RcppParallel, …

…and `purrr` dependencies (#74)

Makes data.table's rbindlist_with_attrs() the default loading method.

DESCRIPTION

@@ -35,8 +35,6 @@ Imports:
     cli (>= 3.0.0),
     data.table (>= 1.14.0),
     dplyr,
-    furrr,
-    future,
     glue,
     httr,
     janitor,
@@ -46,8 +44,9 @@ Imports:
     nnet,
     progressr (>= 0.6.0),
     purrr (>= 0.3.0),
+    Rcpp (>= 1.0.7),
+    RcppParallel (>= 5.1.4),
     rlang,
-    stringi,
     stringr (>= 1.3.0),
     tibble (>= 3.0),
     tidyr (>= 1.0.0)
@@ -61,6 +60,7 @@ Suggests:
     rmarkdown,
     RSQLite,
     stats,
+    stringi,
     testthat,
     usethis (>= 1.6.0),
     xgboost (>= 1.1)

NAMESPACE

@@ -82,7 +82,12 @@ import(purrr)
 import(stringr)
 import(tidyr)
 import(utils)
+importFrom(Rcpp,getRcppVersion)
+importFrom(RcppParallel,defaultNumThreads)
 importFrom(cli,cli_abort)
+importFrom(data.table,data.table)
+importFrom(data.table,fread)
+importFrom(data.table,setDT)
 importFrom(dplyr,arrange)
 importFrom(dplyr,as_tibble)
 importFrom(dplyr,between)
@@ -123,7 +128,6 @@ importFrom(purrr,set_names)
 importFrom(rlang,.data)
 importFrom(stats,na.omit)
 importFrom(stats,predict)
-importFrom(stringi,stri_extract_first_regex)
 importFrom(stringr,str_detect)
 importFrom(stringr,str_extract)
 importFrom(stringr,str_length)

NEWS.md

@@ -13,6 +13,8 @@
 * [```load_cfb_schedules()```](https://cfbfastR.sportsdataverse.org/reference/load_cfb_schedules.html)
 * [```load_cfb_teams()```](https://cfbfastR.sportsdataverse.org/reference/load_cfb_teams.html)
 
+- Removes `furrr`, `future` dependencies, adds `Rcpp`, `RcppParallel`, and `purrr` dependencies
+
 # **cfbfastR v1.8.0**
 
 * All functions now default to return tibbles.

R/cfb_pbp.R

@@ -8,48 +8,39 @@ NULL
 #' or writes it into a db using some forwarded arguments in the dots
 #' @param seasons A vector of 4-digit years associated with given College Football seasons.
 #' @param ... Additional arguments passed to an underlying function that writes
-#' the season data into a database (used by [`update_cfb_db()`][update_cfb_db]).
-#' @param qs Whether to use the function [qs::qdeserialize()] for more efficient loading.
+#' the season data into a database (used by [`update_cfb_db()`][update_cfb_db])
+#' @param dbConnection A `DBIConnection` object, as returned by [DBI::dbConnect()]
+#' @param tablename The name of the play by play data table within the database
 #' @export
-load_cfb_pbp <- function(seasons, ..., qs = FALSE) {
+load_cfb_pbp <- function(seasons = most_recent_cfb_season(),...,
+                         dbConnection = NULL, tablename = NULL) {
   dots <- rlang::dots_list(...)
 
-  if (all(c("dbConnection", "tablename") %in% names(dots))) in_db <- TRUE else in_db <- FALSE
+  loader <- rds_from_url
 
-  if (isTRUE(qs) && !is_installed("qs")) {
-    cli::cli_abort("Package {.val qs} required for argument {.val qs = TRUE}. Please install it.")
-  }
+  if (!is.null(dbConnection) && !is.null(tablename)) in_db <- TRUE else in_db <- FALSE
 
-  most_recent <- most_recent_cfb_season()
+  if(isTRUE(seasons)) seasons <- 2014:most_recent_cfb_season()
 
-  if (!all(seasons %in% 2014:most_recent)) {
-    cli::cli_abort("Please pass valid seasons between 2014 and {most_recent}")
-  }
+  stopifnot(is.numeric(seasons),
+            seasons >= 2014,
+            seasons <= most_recent_cfb_season())
 
-  if (length(seasons) > 1 && is_sequential() && isFALSE(in_db)) {
-    cli::cli_alert_info(c(
-      "It is recommended to use parallel processing when trying to load multiple seasons.",
-      "Please consider running {.code future::plan(\"multisession\")}!",
-      "Will go on sequentially..."
-    ))
-  }
+  urls <- paste0("https://raw.githubusercontent.com/sportsdataverse/cfbfastR-data/main/data/rds/pbp_players_pos_",seasons,".rds")
 
+  p <- NULL
+  if (is_installed("progressr")) p <- progressr::progressor(along = seasons)
 
-  p <- progressr::progressor(along = seasons)
-
-  if (isFALSE(in_db)) {
-    out <- furrr::future_map_dfr(seasons, cfb_single_season, p = p, qs = qs)
-  }
-
-  if (isTRUE(in_db)) {
-    purrr::walk(seasons, cfb_single_season, p, ..., qs = qs)
+  out <- lapply(urls, progressively(loader, p))
+  out <- data.table::rbindlist(out, use.names = TRUE, fill = TRUE)
+  if (in_db) {
+    DBI::dbWriteTable(dbConnection, tablename, out, append = TRUE)
     out <- NULL
+  } else {
+    class(out) <- c("cfbfastR_data","tbl_df","tbl","data.table","data.frame")
+
   }
-  # change this later when data in repo has attributes
-  if(is.null(attr(out,"cfbfastR_timestamp"))) {
-    out <- out %>%
-      make_cfbfastR_data("play-by-play data from cfbfastR data repo",Sys.time())
-  }
+  out
   return(out)
 }
 

R/cfbd_betting.R

@@ -149,7 +149,7 @@ cfbd_betting_lines <- function(game_id = NULL,
       df <- res %>%
         httr::content(as = "text", encoding = "UTF-8") %>%
         jsonlite::fromJSON(flatten = TRUE) %>%
-        furrr::future_map_if(is.data.frame, list) %>%
+        purrr::map_if(is.data.frame, list) %>%
         dplyr::as_tibble() %>%
         tidyr::unnest(.data$lines)
 

R/cfbd_coaches.R

@@ -114,7 +114,7 @@ cfbd_coaches <- function(first = NULL,
       df <- res %>%
         httr::content(as = "text", encoding = "UTF-8") %>%
         jsonlite::fromJSON() %>%
-        furrr::future_map_if(is.data.frame, list) %>%
+        purrr::map_if(is.data.frame, list) %>%
         dplyr::as_tibble() %>%
         tidyr::unnest(.data$seasons) %>%
         dplyr::arrange(.data$year)

R/cfbd_games.R

@@ -721,8 +721,8 @@ cfbd_game_box_advanced <- function(game_id, long = FALSE) {
       df <- res %>%
         httr::content(as = "text", encoding = "UTF-8") %>%
         jsonlite::fromJSON(flatten = TRUE) %>%
-        furrr::future_map_if(is.data.frame, list) %>%
-        furrr::future_map_if(is.data.frame, list)
+        purrr::map_if(is.data.frame, list) %>%
+        purrr::map_if(is.data.frame, list)
 
       df <- tibble::enframe(unlist(df$teams, use.names = TRUE))
       team1 <- seq(1, nrow(df) - 1, by = 2)
@@ -969,18 +969,18 @@ cfbd_game_player_stats <- function(year,
       df <- res %>%
         httr::content(as = "text", encoding = "UTF-8") %>%
         jsonlite::fromJSON(flatten = TRUE) %>%
-        furrr::future_map_if(is.data.frame, list) %>%
+        purrr::map_if(is.data.frame, list) %>%
         dplyr::as_tibble() %>%
         dplyr::rename(game_id = .data$id) %>%
         tidyr::unnest(.data$teams) %>%
-        furrr::future_map_if(is.data.frame, list) %>%
+        purrr::map_if(is.data.frame, list) %>%
         dplyr::as_tibble() %>%
         tidyr::unnest(.data$categories) %>%
-        furrr::future_map_if(is.data.frame, list) %>%
+        purrr::map_if(is.data.frame, list) %>%
         dplyr::as_tibble() %>%
         dplyr::rename(category = .data$name) %>%
         tidyr::unnest(.data$types) %>%
-        furrr::future_map_if(is.data.frame, list) %>%
+        purrr::map_if(is.data.frame, list) %>%
         dplyr::as_tibble() %>%
         dplyr::rename(stat_category = .data$name) %>%
         tidyr::unnest(.data$athletes) %>%
@@ -1357,7 +1357,7 @@ cfbd_game_team_stats <- function(year,
       df <- res %>%
         httr::content(as = "text", encoding = "UTF-8") %>%
         jsonlite::fromJSON(flatten = TRUE) %>%
-        furrr::future_map_if(is.data.frame, list) %>%
+        purrr::map_if(is.data.frame, list) %>%
         dplyr::as_tibble()
 
       if (nrow(df) == 0) {

R/cfbd_pbp_data.R

@@ -550,7 +550,7 @@ cfbd_pbp_data <- function(year,
 
     p <- progressr::progressor(along = g_ids)
 
-    play_df <- furrr::future_map_dfr(
+    play_df <- purrr::map_dfr(
       g_ids,
       function(x){
         play_df <- play_df %>%

R/cfbd_play.R

@@ -18,7 +18,7 @@
 #'  year_split <- split(weekly_year_df, weekly_year_df$year)
 #'  for (i in 1:length(year_split)) {
 #'    i <- 1
-#'    future::plan("multisession")
+#'    
 #'    progressr::with_progress({
 #'       year_split[[i]] <- year_split[[i]] %>%
 #'          dplyr::mutate(

R/cfbd_players.R

@@ -395,7 +395,7 @@ cfbd_player_usage <- function(year = 2019,
       df <- res %>%
         httr::content(as = "text", encoding = "UTF-8") %>%
         jsonlite::fromJSON(flatten = TRUE) %>%
-        furrr::future_map_if(is.data.frame, list) %>%
+        purrr::map_if(is.data.frame, list) %>%
         dplyr::as_tibble() %>%
         dplyr::rename(
           athlete_id = .data$id,

R/cfbd_ratings.R

@@ -129,7 +129,7 @@ cfbd_rankings <- function(year, week = NULL, season_type = "regular") {
       polls <- res %>%
         httr::content(as = "text", encoding = "UTF-8") %>%
         jsonlite::fromJSON(flatten = TRUE) %>%
-        furrr::future_map_if(is.data.frame, list) %>%
+        purrr::map_if(is.data.frame, list) %>%
         dplyr::as_tibble() %>%
         tidyr::unnest(.data$polls) %>%
         tidyr::unnest(.data$ranks) %>%

R/create_wpa_naive.R

@@ -132,7 +132,7 @@ create_wpa_naive <- function(df, wp_model) {
     df[kickoff_ind, "wp_before"] <- as.vector(predict(wp_model, new_kick, type = "response"))
   }
   g_ids <- sort(unique(df$game_id))
-  df2 <- furrr::future_map_dfr(
+  df2 <- purrr::map_dfr(
     g_ids,
     function(x) {
       df %>%

R/helper_pbp_add_yardage.R

@@ -36,9 +36,8 @@
 #' }
 #' @keywords internal
 #' @importFrom rlang .data
-#' @importFrom stringi stri_extract_first_regex
-#' @importFrom stringr str_detect str_extract str_remove
 #' @importFrom dplyr mutate arrange case_when
+#' @import stringr
 #' @export
 #'
 

R/load_cfb.R

@@ -29,7 +29,7 @@ load_cfb_schedules <- function(seasons = most_recent_cfb_season()){
   urls <- paste0("https://github.com/sportsdataverse/cfbfastR-data/raw/main/schedules/rds/cfb_schedules_",
                  seasons, ".rds")
 
-  # out <- furrr::future_map_dfr(urls, progressively(rds_from_url, p = p))
+  # out <- purrr::map_dfr(urls, progressively(rds_from_url, p = p))
 
   out <- lapply(urls, progressively(rds_from_url, p))
   out <- rbindlist_with_attrs(out)
@@ -74,7 +74,7 @@ load_cfb_rosters <- function(seasons = most_recent_cfb_season()){
   urls <- paste0("https://github.com/sportsdataverse/cfbfastR-data/raw/main/rosters/rds/cfb_rosters_",
                  seasons, ".rds")
 
-  # out <- furrr::future_map_dfr(urls, progressively(rds_from_url, p = p))
+  # out <- purrr::map_dfr(urls, progressively(rds_from_url, p = p))
 
   out <- lapply(urls, progressively(rds_from_url, p))
   out <- rbindlist_with_attrs(out)