Subsetting data in groups with multiple shiny selectizeInput

Question

I'm trying to make a shinyapp for NBA play-by-play data which can subset the data based on which players are on or off the court for that play. The shiny app is functional, but I'm having trouble with the syntax of choosing when a player is ON the court.

The relevant columns of the base data I'm working looks like this:

pbp_shiny
# A tibble: 20 x 4
# Groups:   team_final [1]
   team_final play_number team_player         description                                           
   <chr>            <dbl> <chr>               <chr>                                                 
 1 DAL             462084 Delon Wright        NA                                                    
 2 DAL             462084 Tim Hardaway Jr.    NA                                                    
 3 DAL             462084 Maxi Kleber         NA                                                    
 4 DAL             462084 Willie Cauley-Stein NA                                                    
 5 DAL             462084 Luka Doncic         NA                                                    
 6 DAL             462083 Delon Wright        Cauley-Stein REBOUND (Off:1 Def:6)                    
 7 DAL             462083 Tim Hardaway Jr.    Cauley-Stein REBOUND (Off:1 Def:6)                    
 8 DAL             462083 Maxi Kleber         Cauley-Stein REBOUND (Off:1 Def:6)                    
 9 DAL             462083 Willie Cauley-Stein Cauley-Stein REBOUND (Off:1 Def:6)                    
10 DAL             462083 Luka Doncic         Cauley-Stein REBOUND (Off:1 Def:6)                    
11 DAL             462082 Delon Wright        MISS Murray 28' 3PT Jump Shot                         
12 DAL             462082 Tim Hardaway Jr.    MISS Murray 28' 3PT Jump Shot                         
13 DAL             462082 Maxi Kleber         MISS Murray 28' 3PT Jump Shot                         
14 DAL             462082 Willie Cauley-Stein MISS Murray 28' 3PT Jump Shot                         
15 DAL             462082 Luka Doncic         MISS Murray 28' 3PT Jump Shot                         
16 DAL             462081 Delon Wright        Wright Bad Pass Turnover (P2.T11): Craig STEAL (1 STL)
17 DAL             462081 Tim Hardaway Jr.    Wright Bad Pass Turnover (P2.T11): Craig STEAL (1 STL)
18 DAL             462081 Maxi Kleber         Wright Bad Pass Turnover (P2.T11): Craig STEAL (1 STL)
19 DAL             462081 Willie Cauley-Stein Wright Bad Pass Turnover (P2.T11): Craig STEAL (1 STL)
20 DAL             462081 Luka Doncic         Wright Bad Pass Turnover (P2.T11): Craig STEAL (1 STL)

There are always 5 players on the floor for a team in basketball. So, play_number identifies which play it is, and team_player shows each player on the floor during that play. I must keep the format like this for other reasons.

The widgets for choosing which players are ON or OFF the court are written as uiOutput in the UI (the player options are dependent on which team is selected in another widget). On the server side, they look like this:

output$player_off <- renderUI({
  selectizeInput(
    "player_off",
    "WITHOUT:",
    choices = c(player_list$team_player[player_list$team_final==input$team_filter]),
    multiple = TRUE
  )
})

output$player_on <- renderUI({
  selectizeInput(
    "player_on",
    "WITH:",
    choices = c(player_list$team_player[player_list$team_final==input$team_filter]),
    multiple = TRUE
  )
})

and now here's the actual code I'm using to subset the data based on who is ON and OFF the court:

subset(pbp_shiny, ave(!(team_player %in% input$player_off), play_number, FUN = all) & ave(team_player %in% input$player_on, play_number, FUN = any))

I know the player_off side is correct -- regardless of how many (even zero) players are chosen to be OFF, it will keep all play_number which DO NOT include those players.

The player_on side is where I'm struggling -- after adding it in, the shiny app only subsets correctly when EXACTLY one player is chosen to be on the court. If 0 or more than 1 are chosen, it subsets pbp_shiny down to 0 observations. I understand why, since FUN = any, but I'm struggling with a solution.