Merge pull request #386 from jruizcabrejos/main

Understat match stats and players

DESCRIPTION

@@ -1,7 +1,7 @@
 Type: Package
 Package: worldfootballR
 Title: Extract and Clean World Football (Soccer) Data
-Version: 0.6.5.0005
+Version: 0.6.5.0006
 Authors@R: c(
     person("Jason", "Zivkovic", , "[email protected]", role = c("aut", "cre", "cph")),
     person("Tony", "ElHabr", , "[email protected]", role = "ctb"),

NAMESPACE

@@ -63,7 +63,9 @@ export(tm_team_transfers)
 export(understat_available_teams)
 export(understat_league_match_results)
 export(understat_league_season_shots)
+export(understat_match_players)
 export(understat_match_shots)
+export(understat_match_stats)
 export(understat_player_shots)
 export(understat_team_meta)
 export(understat_team_players_stats)
@@ -116,9 +118,11 @@ importFrom(rvest,html_text)
 importFrom(rvest,read_html)
 importFrom(stats,runif)
 importFrom(stats,setNames)
+importFrom(stringi,stri_unescape_unicode)
 importFrom(stringr,str_detect)
 importFrom(stringr,str_replace_all)
 importFrom(stringr,str_squish)
+importFrom(stringr,str_subset)
 importFrom(tibble,tibble)
 importFrom(tidyr,crossing)
 importFrom(tidyselect,vars_select_helpers)

NEWS.md

@@ -11,6 +11,9 @@
 
 * In addressing the issue with `tm_player_injury_history()` in [#375](https://github.com/JaseZiv/worldfootballR/issues/375), the previously names column `club` has been renamed `club_missed_games_for` to better represent that this column will contain the games the player missed games for, as previously this column could have been misunderstood to be who they were playing for when they were injured (0.6.5.0004)
 
+### Improvements
+
+* `understat_match_players` and `understat_match_stats` added. (0.6.5.0006) [#386](https://github.com/JaseZiv/worldfootballR/issues/386)
 
 ***
 

R/understat_match_players.R

@@ -0,0 +1,74 @@
+
+#' Get Understat match player data
+#'
+#' Returns player values for a selected match from Understat.com.
+#'
+#' @param match_url A `character` string with the URL of the match played.
+#'
+#' @return returns a `data.frame` with data for all players for the match.
+#'
+#' @importFrom magrittr %>%
+#' @importFrom rvest html_elements
+#' @importFrom stringi stri_unescape_unicode
+#' @importFrom stringr str_subset
+#' @importFrom jsonlite fromJSON
+#' @importFrom dplyr bind_rows
+#'
+#' @export
+
+understat_match_players <- function(match_url) {
+
+  match_id <- gsub("[^0-9]", "", match_url)
+
+  match_player_data <- .get_understat_json(page_url = match_url)  %>%
+    rvest::html_elements("script") %>%
+    as.character()
+
+  match_player_data <- match_player_data[grep("rostersData\t=", match_player_data)] %>%
+    stringi::stri_unescape_unicode() %>%
+    substr(41,nchar(.)) %>%
+    substr(0,nchar(.)-13) %>%
+    paste0('[', . , ']') %>%
+
+    unlist() %>%
+    stringr::str_subset("\\[\\]", negate = TRUE)
+
+  match_player_data <- lapply(match_player_data, jsonlite::fromJSON) %>%
+    do.call("rbind", .)
+
+  match_player_data_home <- do.call(rbind.data.frame, match_player_data$h)
+  match_player_data_away <- do.call(rbind.data.frame, match_player_data$a)
+
+
+  match_player_data_rebind <- dplyr::bind_rows(match_player_data_home, match_player_data_away)
+
+  match_players <- data.frame(
+    match_id = as.integer(match_id),
+    id = as.integer(match_player_data_rebind[["id"]]),
+    team_id = as.integer(match_player_data_rebind[["team_id"]]),
+    home_away = as.character(match_player_data_rebind[["h_a"]]),
+    player_id = as.integer(match_player_data_rebind[["player_id"]]),
+    swap_id = as.integer(match_player_data_rebind[["id"]]),
+    player = as.character(match_player_data_rebind[["player"]]),
+    position = as.character(match_player_data_rebind[["position"]]),
+    positionOrder = as.integer(match_player_data_rebind[["positionOrder"]]),
+    time_played = as.integer(match_player_data_rebind[["time"]]),
+    goals = as.integer(match_player_data_rebind[["goals"]]),
+    own_goals = as.integer(match_player_data_rebind[["own_goals"]]),
+    shots = as.integer(match_player_data_rebind[["shots"]]),
+    xG = as.numeric(match_player_data_rebind[["xG"]]),
+    yellow_card = as.integer(match_player_data_rebind[["yellow_card"]]),
+    red_card = as.integer(match_player_data_rebind[["red_card"]]),
+    roster_in = as.integer(match_player_data_rebind[["roster_in"]]),
+    roster_out = as.integer(match_player_data_rebind[["roster_out"]]),
+    key_passes = as.integer(match_player_data_rebind[["key_passes"]]),
+    assists = as.integer(match_player_data_rebind[["assists"]]),
+    xA = as.numeric(match_player_data_rebind[["xA"]]),
+    xGChain = as.numeric(match_player_data_rebind[["xGChain"]]),
+    xGBuildup = as.numeric(match_player_data_rebind[["xGBuildup"]])
+  )
+
+  return(match_players)
+}
+
+

R/understat_match_stats.R

@@ -0,0 +1,58 @@
+
+#' Get Understat match stats table data
+#'
+#' Returns the Stats values for a selected match from Understat.com.
+#'
+#' @param match_url A `character` string with the URL of the match played.
+#'
+#' @return returns a `data.frame` with data from the stats table for the match.
+#'
+#' @details For `draw_chances`, `home_chances` and `away_chances`, values below 10% in the browser will be retrieved as NA (e.g. A "5%" chance will be NA in the `data.frame`).
+#'
+#' @importFrom magrittr %>%
+#' @importFrom rvest html_elements html_text
+#'
+#' @export
+
+understat_match_stats <- function(match_url) {
+
+  match_stats <- .get_understat_json(page_url = match_url) %>%
+    rvest::html_elements("div.scheme-block.is-hide[data-scheme='stats']") %>%
+    rvest::html_elements(".progress-value") %>%
+    rvest::html_text()
+
+  away <- match_stats[seq(1, length(match_stats), by=2)]
+  home <- match_stats[seq(2, length(match_stats), by=2)]
+
+  match_stats <- data.frame(
+
+    match_id = as.integer(gsub("[^0-9]", "", match_url)),
+
+    home_team = as.character(away[1]),
+    home_chances = as.integer(gsub("[^0-9]", "", away[2]))/100,
+    home_goals = as.integer(home[3]),
+    home_xG = as.numeric(home[4]),
+    home_shots = as.integer(home[5]),
+    home_shot_on_target = as.integer(home[6]),
+    home_deep = as.integer(home[7]),
+    home_PPDA = as.numeric(home[8]),
+    home_xPTS = as.numeric(home[9]),
+
+    draw_chances = as.integer(gsub("[^0-9]", "", home[2]))/100,
+
+    away_team = home[1],
+    away_chances = as.integer(gsub("[^0-9]", "", away[3]))/100 ,
+    away_goals = as.integer(away[4]),
+    away_xG = as.numeric(away[5]),
+    away_shots = as.integer(away[6]),
+    away_shot_on_target = as.integer(away[7]),
+    away_deep = as.integer(away[8]),
+    away_PPDA = as.numeric(away[9]),
+    away_xPTS = as.numeric(away[10])
+
+  )
+
+  return(match_stats)
+}
+
+

tests/testthat/test-understat.R

@@ -47,3 +47,19 @@ test_that("understat_team_stats_breakdown() works", {
   expect_equal(ncol(team_stats), 11)
   expect_gt(nrow(team_stats), 0)
 })
+
+test_that("understat_match_players() works", {
+  testthat::skip_on_cran()
+  match_players <- understat_match_players(match_url = "https://understat.com/match/14789")
+  expect_true(any("data.frame" == class(match_players)))
+  expect_equal(ncol(match_players), 23)
+  expect_gt(nrow(match_players), 0)
+})
+
+test_that("understat_match_stats() works", {
+  testthat::skip_on_cran()
+  match_stats <- understat_match_stats(match_url = "https://understat.com/match/14789")
+  expect_true(any("data.frame" == class(match_stats)))
+  expect_equal(ncol(match_stats), 20)
+  expect_equal(nrow(match_stats), 1)
+})

vignettes/extract-understat-data.Rmd

@@ -136,6 +136,25 @@ wba_liv_shots <- understat_match_shots(match_url = "https://understat.com/match/
 dplyr::glimpse(wba_liv_shots)
 ```
 
+### Match Stats
+
+To get the data from the stats table for an individual match, use the `understat_match_stats()` function:
+
+```{r match_stats}
+wba_liv_stats <- understat_match_stats(match_url = "https://understat.com/match/14789")
+dplyr::glimpse(wba_liv_stats)
+```
+
+
+### Match Players
+
+To get the data for player in an individual match, use the `understat_match_players()` function:
+
+```{r match_players}
+wba_liv_players <- understat_match_players(match_url = "https://understat.com/match/14789")
+dplyr::glimpse(wba_liv_players)
+```
+
 
 ***