html_node -> html_element, match_player return cleanup, revert roxygen

DESCRIPTION

@@ -54,4 +54,4 @@ Suggests:
     rmarkdown,
     testthat
 Encoding: UTF-8
-RoxygenNote: 7.3.1
+RoxygenNote: 7.2.3

NAMESPACE

@@ -118,9 +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)

R/understat_match_players.R

@@ -8,6 +8,11 @@
 #' @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
 
@@ -16,12 +21,11 @@ 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_nodes("script") %>%
+    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('[', . , ']') %>%
@@ -35,36 +39,36 @@ understat_match_players <- function(match_url) {
   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 <- dplyr::bind_rows(match_player_data_home,match_player_data_away) %>%
-
-    dplyr::mutate(match_id = as.integer(match_id),
-                  team_id = as.integer(team_id),
-                  team_status = as.character(team_status),
-                  player_id = as.integer(player_id),
-                  swap_id = as.integer(id),
-                  player = as.character(player),
-                  position = as.character(position),
-                  positionOrder = as.integer(positionOrder),
-                  time_played =  as.integer(time),
-                  goals = as.integer(goals),
-                  own_goals = as.integer(own_goals),
-                  shots = as.integer(shots),
-                  xG = as.numeric(xG),
-                  yellow_card = as.integer(yellow_card),
-                  red_card = as.integer(red_card),
-                  roster_in = as.integer(roster_in),
-                  roster_out = as.integer(roster_out),
-                  key_passes = as.integer(key_passes),
-                  assists = as.integer(assists),
-                  xA = as.numeric(xA),
-                  xGChain = as.numeric(xGChain),
-                  XGBuildup = as.numeric(XGBuildup))  %>%
-
-    dplyr::select(-c(time,id)) %>%
-
-    dplyr::mutate(team_status = ifelse(team_status=="h","home","away"))
 
-  return(match_player_data)
+  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

@@ -10,14 +10,15 @@
 #' @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_nodes("div.scheme-block.is-hide[data-scheme='stats']") %>%
-    rvest::html_nodes(".progress-value") %>%
+    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)]

tests/testthat/test-understat.R

@@ -52,7 +52,7 @@ 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), 22)
+  expect_equal(ncol(match_players), 23)
   expect_gt(nrow(match_players), 0)
 })