make_data.R

soccer_field_length = 105
soccer_field_width = 68
soccer_goal_width = 14.02
get_referee <- function(rcl) {
    referee <- rcl |>
        dplyr::filter(command == "referee") %>%
        dplyr::select(
            step,
            judge=args,
        ) %>%
        tidyr::drop_na() %>%
        distinct(step, .keep_all = TRUE)
    return(referee)
}

read_goal <- function(referee, name) {
    goal <- referee |>
        dplyr::filter(stringr::str_detect(judge, "goal_[rl]_[0-9]+")) |>
        dplyr::mutate(
            side = judge |>
                stringr::str_remove("goal_") |>
                stringr::str_extract("[rl]"),
            score = judge |>
                stringr::str_remove("goal_[rl]_") |>
                stringr::str_extract("[0-9]+"),
        ) |>
        dplyr::inner_join(name, by = "side") |>
        dplyr::select(
            step,
            judge,
            side,
            name,
            score,
        ) %>%
        tidyr::drop_na()

    return(goal)
}

read_goal_no_judge <- function(goal) {
    goal <- goal |>
        dplyr::select(
            step,
            side,
            name,
            score,
        )

    return(goal)
}


get_player <- function(rcg, name) {
    output <- rcg |>
        dplyr::inner_join(name, by = "side") |>
        dplyr::select(
            step,
            team = name,
            side,
            unum,
            px = x,
            py = y,
            pvx = vx,
            pvy = vy,
            body,
            neck,
        )

    return(output)
}

get_ball <- function(rcg) {
    ball <- rcg |>
        dplyr::select(step, ball_x, ball_y, ball_vx, ball_vy) |>
        dplyr::distinct(step, .keep_all = TRUE) %>%
        dplyr::mutate(before_ball_x = lag(ball_x)) %>%
        dplyr::mutate(before_ball_y = lag(ball_y)) %>%
        dplyr::mutate(before_bvx = lag(ball_vx)) %>%
        dplyr::mutate(before_bvy = lag(ball_vy)) %>%
        dplyr::mutate(next_ball_x = lead(ball_x)) %>%
        dplyr::mutate(next_ball_y = lead(ball_y)) %>%
        dplyr::mutate(next_ball_vx = lead(ball_vx)) %>%
        dplyr::mutate(next_ball_vy = lead(ball_vy)) %>%
        dplyr::mutate(kick_speed= (ball_vx^2+ball_vy^2)) %>%
        dplyr::mutate(a_kick_speed= (next_ball_vx^2+next_ball_vy^2)) %>%
        dplyr::mutate(target_ball_x = (ball_x + ball_vx)) %>%
        dplyr::mutate(target_ball_y = (ball_y + ball_vy)) %>%
        dplyr::mutate(shoot_ball = (kick_speed > 1)&((soccer_field_length/2)<abs(target_ball_x))&((soccer_goal_width/2)>abs(target_ball_y))) 

    return(ball)
}


get_action <- function(rcl,ball,players) {
    referee <- get_referee(rcl)
    output <- rcl |>
        dplyr::filter(command == "kick" | command == "tackle" | command == "catch") |>
        dplyr::mutate(
            unum = as.numeric(unum)
        ) |>
        dplyr::select(
            step,
            team,
            unum,
            command,
        ) |>
        dplyr::inner_join(players, by = c("step", "team", "unum"))%>%
        dplyr::inner_join(ball, by = "step") %>%
        dplyr::distinct(step,team, .keep_all = TRUE) %>%
        dplyr::full_join(referee,by="step") %>%
        dplyr::arrange(step) %>%
        dplyr::mutate(next_judge = lead(judge)) %>%
        dplyr::filter(command == "kick" | command == "tackle" | command == "catch") %>%
        dplyr::mutate(a_tackle = (lead(command)=="tackle")) %>%
        dplyr::mutate(next_team = lead(team)) %>%
        dplyr::mutate(a_sameteam = (next_team == team)) %>%
        dplyr::mutate(tackle_scc = (judge == "NULL" & command == "tackle" & a_sameteam)) %>%
        dplyr::mutate(a_tackle_scc = lead(tackle_scc))

    return(output)
}

get_kick_log <- function(action,goal) {
    output <- action %>%
        dplyr::filter(command != "tackle") %>%
        dplyr::mutate(next_catch = (lead(command) == "catch")) %>%
        dplyr::filter(command == "kick") %>%
        dplyr::mutate(before_team = lag(team)) %>%
        dplyr::mutate(before_unum = lag(unum)) %>%
        dplyr::mutate(next_team = lead(team)) %>%
        dplyr::mutate(next_unum = lead(unum)) %>%
        dplyr::mutate(b_sameteam = (before_team == team)) %>%
        dplyr::mutate(b_sameunum = (before_unum == unum)) %>%
        dplyr::mutate(a_sameteam = (next_team == team)) %>%
        dplyr::mutate(a_sameunum = (next_unum == unum))

    output$b_sameteam[1] <- TRUE
    output$b_sameunum[1] <- FALSE
    output <- dplyr::full_join(output,goal,by=c("step","judge","side")) %>%
        dplyr::arrange(step) %>%
        dplyr::mutate(next_score = lead(score)) %>%
        dplyr::mutate(next_judge = lead(judge)) %>%
        dplyr::mutate(is_goal = (!is.na(next_score)&(side==lead(side)))) %>%
        dplyr::filter(command == "kick") %>%
        dplyr::mutate(shoot_scc = (judge == "NULL") & is_goal) %>%
        dplyr::mutate(dribble_scc = (
            (judge == "NULL") & (next_judge == "NULL") & a_sameteam & a_sameunum)) %>%
        dplyr::mutate(pass_scc = (
            a_sameteam & !a_sameunum)) %>%
        dplyr::mutate(shoot = (
            shoot_scc |
            abs(target_ball_x) > (soccer_field_length/2) & abs(target_ball_y) < 12)) %>%
        dplyr::mutate(dribble = (
            (!shoot)&
            (!pass_scc)&
            (dribble_scc |
                (!shoot) & (a_tackle_scc)|
                (!shoot) & sqrt((ball_x - next_ball_x)^2 + (ball_y - next_ball_y)^2) < 1.0 )
        )) %>%
        dplyr::mutate(pass = (
            (pass_scc | (!dribble & !shoot))
        )) %>%
        group_by(grc = cumsum(!dribble)) %>%
        mutate(touch = row_number()) %>%
        ungroup()
    return(output)
}

get_kick <- function(kick_log) {
    output <- action %>%
        dplyr::filter(command == "kick") %>%
        select(-c(command, pvx, pvy, body, neck))
    return(output)
}

get_kick_Allplayer <- function(players,kick){
    kick <- kick %>%
        dplyr::select(step,action_unum=unum,action_team = team,ball_x,ball_y,ball_vx,ball_vy,next_ball_x,next_ball_y,next_ball_vx,next_ball_vy)
    output <- players %>%
        dplyr::inner_join(kick, by = "step") %>%
        dplyr::distinct(step,team,unum, .keep_all = TRUE)
    return(output)
}


get_action_Allplayer <- function(players, action) {
    action <- action %>%
        dplyr::select(step, action_team = team, ball_x, ball_y, pass, dribble)
    output <- players %>%
        dplyr::inner_join(action, by = "step")
    return(output)
}

get_pass_Allplayer <- function(players, pass) {
    pass <- pass %>%
        dplyr::select(step,pass_team = team,ball_x,ball_y,ball_vx,ball_vy,next_ball_x,next_ball_y,next_ball_vx,next_ball_vy,pass,pass_scc)
    output <- players %>%
        dplyr::inner_join(pass, by = "step") %>%
        dplyr::distinct(step,team,unum, .keep_all = TRUE)
    return(output)
}

get_dribble_Allplayer <- function(players, dribble) {
    dribble <- dribble %>%
        dplyr::select(step,dribble_team = team,ball_x,ball_y,ball_vx,ball_vy,next_ball_x,next_ball_y,next_ball_vx,next_ball_vy,dribble,dribble_scc)
    output <- players %>%
        dplyr::inner_join(dribble, by = "step") %>%
        dplyr::distinct(step,team,unum, .keep_all = TRUE)
    return(output)
}

get_shoot_Allplayer <- function(players, shoot) {
    shoot <- shoot %>%
        dplyr::select(step,shoot_team = team,ball_x,ball_y,ball_vx,ball_vy,next_ball_x,next_ball_y,next_ball_vx,next_ball_vy,shoot,shoot_scc,ball_goal_dist,ball_goal_angle)
    output <- players %>%
        dplyr::inner_join(shoot, by = "step") %>%
        dplyr::distinct(step,team,unum, .keep_all = TRUE)%>%
        dplyr::mutate(ball_player_dist = sqrt((px-ball_x)^2+(py-ball_y)^2))
    return(output)
}

get_pass <- function(kick) {
    output <- kick |>
        dplyr::filter(pass == TRUE) %>%
        distinct(step, .keep_all = TRUE) %>%
        dplyr::select(step,team,unum,px,py,pvx,pvy,ball_x,ball_y,ball_vx,ball_vy,next_ball_vx,next_ball_vy,pass,pass_scc) %>%
        dplyr::mutate(next_ball_x = lead(ball_x)) %>%
        dplyr::mutate(next_ball_y = lead(ball_y)) %>%
        drop_na()
    return(output)
}

get_dribble <- function(kick) {
    output <- kick %>%
        dplyr::filter(dribble == TRUE) %>%
        distinct(step, .keep_all = TRUE) %>%
        dplyr::select(step,team,unum,px,py,pvx,pvy,ball_x,ball_y,ball_vx,ball_vy,next_ball_x,next_ball_y,next_ball_vx,next_ball_vy,dribble,dribble_scc)
    return(output)
}

get_shoot <- function(kick) {
    output <- kick %>%
        dplyr::filter(shoot == TRUE) %>%
        distinct(step, .keep_all = TRUE) %>%
        dplyr::select(step,team,ball_x,ball_y,ball_vx,ball_vy,next_ball_x,next_ball_y,next_ball_vx,next_ball_vy,shoot,shoot_scc) %>%
        dplyr::mutate(ball_goal_dist = sqrt(((soccer_field_length/2)-abs(ball_x))^2+abs((ball_y))^2)) %>%
        dplyr::mutate(ball_goal_angle = atan2(((soccer_field_length/2)-abs(ball_x)),ball_y))
    return(output)
}

get_xg <- function(kick){
    output <- kick |> 
        dplyr::filter(shoot == TRUE) %>%
        distinct(step, .keep_all = TRUE) %>%
        dplyr::mutate(ball_goal_dist = sqrt(((soccer_field_length/2)-abs(ball_x))^2+abs((ball_y))^2)) %>%
        dplyr::mutate(ball_goal_angle = atan2(((soccer_field_length/2)-abs(ball_x)),ball_y)) %>%
        dplyr::select(step,team,px,py,ball_x,ball_y,ball_goal_dist,ball_goal_angle)
        return(output)
}
get_xg_label <- function(kick){
    output <- kick |> 
        dplyr::filter(shoot == TRUE) %>%
        distinct(step, .keep_all = TRUE) %>%
        dplyr::select(step,team,shoot_scc)
        return(output)
}

get_xg_l <- function(kick){
    output <- kick |> 
        dplyr::filter(shoot == TRUE) %>%
        dplyr::filter(side == "l") %>%
        distinct(step, .keep_all = TRUE) %>%
        dplyr::mutate(ball_goal_dist = sqrt(((soccer_field_length/2)-abs(ball_x))^2+abs((ball_y))^2)) %>%
        dplyr::mutate(ball_goal_angle = atan2(((soccer_field_length/2)-abs(ball_x)),ball_y)) %>%
        dplyr::select(px,py,ball_x,ball_y,ball_goal_dist,ball_goal_angle)
        return(output)
}

get_xg_l_label <- function(kick){
    output <- kick |> 
        dplyr::filter(shoot == TRUE) %>%
        dplyr::filter(side == "l") %>%
        distinct(step, .keep_all = TRUE) %>%
        dplyr::select(shoot_scc)
        return(output)
}

get_xg_r <- function(kick){
    output <- kick |> 
        dplyr::filter(shoot == TRUE) %>%
        dplyr::filter(side == "r") %>%
        distinct(step, .keep_all = TRUE) %>%
        dplyr::mutate(px = px * -1,py = py *-1,ball_x = -1 * ball_x,ball_y = ball_y * -1) %>%
        dplyr::mutate(ball_goal_dist = sqrt(((soccer_field_length/2)-abs(ball_x))^2+abs((ball_y))^2)) %>%
        dplyr::mutate(ball_goal_angle = atan2(((soccer_field_length/2)-abs(ball_x)),ball_y)) %>%
        dplyr::select(px,py,ball_x,ball_y,ball_goal_dist,ball_goal_angle)
        return(output)
}

get_xg_r_label <- function(kick){
    output <- kick |> 
        dplyr::filter(shoot == TRUE) %>%
        dplyr::filter(side == "r") %>%
        distinct(step, .keep_all = TRUE) %>%
        dplyr::select(shoot_scc)
        return(output)
}

get_dp <- function(dribbleAll){
    output <- dribbleAll |>
        dplyr::mutate(p_speed = sqrt(pvx^2 + pvy^2)) %>%
        dplyr::mutate(theta = atan2(pvx,pvy)) %>%
        dplyr::mutate(cos = cos(theta),sin = sin(theta))
    return(output)
}

get_as <- function(kick){
    output <- kick |>
        dplyr::select(step,action_team=team,shoot,dribble,pass)
    return(output)
}

get_goalkeeper_action<- function(actionAll){
    output <- actionAll |>
        dplyr::filter(unum == 1)  %>%
        dplyr::mutate(ball_keeper_dist_x = sqrt((px-ball_x)^2)) %>%
        dplyr::mutate(player_goal_dist = sqrt(((soccer_field_length/2)-abs(px))^2+abs((py))^2)) %>%
        dplyr::mutate(opp_team =(team != shoot_team))  %>% 
        dplyr::mutate(ball_closer_goal_goalkeeper = opp_team & (ball_goal_dist < player_goal_dist))
    return(output)
}

get_shoot_epv <- function(players,shoot){
    shootAll <- get_shoot_Allplayer(players,shoot)
    goalkeeper <- get_goalkeeper_action(shootAll) |>
        dplyr::filter(team != shoot_team) |>
        dplyr::select(step,side,ball_keeper_dist_x,keeper_goal_dist=player_goal_dist,ball_closer_goal_goalkeeper)
    opponent_team <- shootAll |>
        dplyr::filter(team != shoot_team) %>%
        dplyr::mutate(r_post_x = (ball_x + (py - ball_y)*((soccer_goal_width/2)-ball_x)/(-(soccer_field_length/2)-ball_y))) %>%
        dplyr::mutate(l_post_x = (ball_x + (py - ball_y)*(-(soccer_goal_width/2)-ball_x)/(-(soccer_field_length/2)-ball_y))) %>%
        dplyr::mutate(triangle_in_player = (px > l_post_x)&(px < r_post_x)) %>%
        dplyr::mutate(dist_ball_in_3m = (ball_player_dist < 3))
    opp_triangle_in_player_num <- opponent_team |>
        dplyr::filter(triangle_in_player) %>% 
        dplyr::group_by(step) %>%
        dplyr::summarise(triangle_in_player_num=n())
    opp_dist_ball_in_3m_num <- opponent_team |>
        dplyr::filter(dist_ball_in_3m) %>% 
        dplyr::group_by(step) %>%
        dplyr::summarise(dist_ball_in_3m_num=n())
    output <- shoot |>
        dplyr::left_join(goalkeeper,by = "step") %>%
        dplyr::left_join(opp_dist_ball_in_3m_num,by="step") %>%
        dplyr::left_join(opp_triangle_in_player_num,by = "step") %>%
        replace_na(list(triangle_in_player_num = 0, dist_ball_in_3m_num = 0))
    return(output)
}

before_goal <- function(kick_log, goal_step, goal_team) {
    playlist <- kick_log %>%
        dplyr::arrange(desc(step)) %>%
        dplyr::filter(
            step > goal_step - 100,
            step < goal_step,
            team == goal_team,
        )
}

get_goal_playlist <- function(kick_log, goal) {
    if (length(goal$step) < 1) {
        return(NA)
    }
    i <- 1
    output <- before_goal(kick_log, goal$step[i], goal$name[i])
    i <- i + 1
    while (i <= length(goal$step)) {
        # print(i)
        i_list <- before_goal(kick_log, goal$step[i], goal$name[i])
        output <- output |> bind_rows(i_list)
        i <- i + 1
    }
    # for(i in goal$step){
    #   i_list <- before_goal(kick_log,goal$step[i],goal$name[i])
    #   output <- output |> bind_rows(i_list)
    #   print(i)
    # }
    return(output)
}

convert_by_step <- function(rcg, rcl) {
    output <- dplyr::inner_join(rcg, rcl, by = "step")
    return(output)
}

get_dribble_point <- function(log_data) {
    output <- log_data %>% filter(dribble == TRUE, pass == FALSE)
    return(output)
}

get_pass_point <- function(log_data) {
    output <- log_data %>% filter(pass == TRUE)
    return(output)
}

get_tackle_point <- function(log_data) {
    output <- log_data %>% filter(tackle == TRUE)
    return(output)
}

get_rcl <- function(path_rcl) {
    output <- path_rcl |> read_rcl()
    return(output)
}

get_name <- function(path_rcg) {
    team_name <- path_rcg |>
        stringr::str_extract("[a-zA-Z0-9_-]*.rcg") |>
        stringr::str_remove("[0-9]*-") |>
        stringr::str_remove(".rcl")
    l_team <- team_name |>
        stringr::str_extract("[a-zA-Z0-9_]*-vs-") |>
        stringr::str_remove("_[0-9]*-vs-")
    r_team <- path_rcg |>
        stringr::str_extract("[a-zA-Z0-9_]*.rcg") |>
        stringr::str_remove("_[0-9]*.rcg")
    name_table <- tibble::tribble(
        ~side, ~name,
        "l", l_team,
        "r", r_team,
    )

    return(name_table)
}

replase_rcg_to_rcl <- function(path_rcg) {
    output <- path_rcg |> stringr::str_replace(pattern = ".rcg", replacement = ".rcl")
}

replase_rcl_to_rcg <- function(path_rcl) {
    output <- path_rcl |> stringr::str_replace(pattern = ".rcl", replacement = ".rcg")
}


kick_dist <- function(rcg) {
    ball_path <- rcg |>
        dplyr::group_nest(step, ball_x, ball_y) |>
        dplyr::mutate(
            move_dist_x = ball_x - dplyr::lag(ball_x),
            move_dist_y = ball_y - dplyr::lag(ball_y),
            move_dist = sqrt(move_dist_x^2 + move_dist_y^2),
            move_dist = dplyr::if_else(is.na(move_dist), 0, move_dist)
        ) |>
        dplyr::filter(move_dist != 0 & move_dist < 40)
}

select_name <- function(data, name) {
    output <- data %>%
        dplyr::filter(team %in% name)
    return(output)
}

get_AttackLine <- function(pass) {
    output <- pass %>%
        dplyr::group_by(step) %>%
        filter(px == max(px)) %>%
        dplyr::select(
            step,
            pass_team,
            pass_scc,
            AL = px
        )
    return(output)
}

get_DefendLine <- function(pass) {
    output <- pass %>%
        dplyr::group_by(step) %>%
        filter(px == min(px)) %>%
        dplyr::select(step, DL = px)
    return(output)
}

get_MedianLine <- function(pass) {
    output <- pass %>%
        dplyr::group_by(step) %>%
        dplyr::summarize(ML = median(px)) %>%
        dplyr::select(step, ML)
    return(output)
}

get_DynamicPressureLine <- function(pass) {
    pass <- pass %>%
        filter(unum != 1)
    ad <- get_AttackLine(pass)
    dd <- get_DefendLine(pass)
    md <- get_MedianLine(pass)
    output <- dplyr::inner_join(ad, dd, by = "step") %>%
        dplyr::inner_join(md, by = "step") %>%
        distinct(step, .keep_all = TRUE)
    return(output)
}

get_AttackLine_D <- function(dribble) {
    output <- dribble %>%
        dplyr::group_by(step) %>%
        filter(px == max(px)) %>%
        dplyr::select(
            step,
            dribble_team,
            dribble_scc,
            AL = px
        )
    return(output)
}

get_DynamicPressureLine_D <- function(dribble) {
    dribble <- dribble %>%
        filter(unum != 1)
    ad <- get_AttackLine_D(dribble)
    dd <- get_DefendLine(dribble)
    md <- get_MedianLine(dribble)
    output <- dplyr::inner_join(ad, dd, by = "step") %>%
        dplyr::inner_join(md, by = "step") %>%
        distinct(step, .keep_all = TRUE) %>%
        dplyr::select(step,AL,ML,DL)
    return(output)
}

add_DynamicPressureLine <- function(action_All,name){
    action_l <- select_name(action_All,name[1,2])
    action_r <- select_name(action_All,name[2,2])
    dpl_l <- get_DynamicPressureLine_D(action_l)
    dpl_r <- get_DynamicPressureLine_D(action_r)
    
    output <- dplyr::inner_join(action_All,dpl_l,by="step")  %>% 
        dplyr::inner_join(dpl_r,by="step",suffix = c("_l","_r")) 
    return(output)
}

get_AttackLine_AS <- function(dribble) {
    output <- dribble %>%
        dplyr::group_by(step) %>%
        filter(px == max(px)) %>%
        dplyr::select(
            step,
            AL = px
        )
    return(output)
}

get_DynamicPressureLine_AS <- function(dribble) {
    dribble <- dribble %>%
        filter(unum != 1)
    ad <- get_AttackLine_AS(dribble)
    dd <- get_DefendLine(dribble)
    md <- get_MedianLine(dribble)
    output <- dplyr::inner_join(ad, dd, by = "step") %>%
        dplyr::inner_join(md, by = "step") %>%
        distinct(step, .keep_all = TRUE) %>%
        dplyr::select(step,AL,ML,DL)
    return(output)
}

add_DynamicPressureLine_AS <- function(action_All,name){
    action_l <- select_name(action_All,name[1,2])
    action_r <- select_name(action_All,name[2,2])
    dpl_l <- get_DynamicPressureLine_AS(action_l)
    dpl_r <- get_DynamicPressureLine_AS(action_r)
    
    output <- dplyr::inner_join(action_All,dpl_l,by="step")  %>% 
        dplyr::inner_join(dpl_r,by="step",suffix = c("_l","_r")) 
    return(output)
}

get_movie <- function(rcl,ball,players) {
    referee <- get_referee(rcl)
    output <- rcl |>
        dplyr::mutate(
            unum = as.numeric(unum)
        ) |>
        dplyr::select(
            step,
            team,
            unum,
            command,
        ) |>
        dplyr::inner_join(players, by = c("step", "team", "unum"))%>%
        dplyr::inner_join(ball, by = "step") %>%
        dplyr::distinct(step,team, .keep_all = TRUE) %>%
        dplyr::full_join(referee,by="step") %>%
        dplyr::arrange(step)

    return(output)
}

get_DPL <- function(movie) {
    movie <- movie %>%
        filter(unum != 1)
    ad <- movie %>%
        dplyr::group_by(step) %>%
        filter(px == max(px)) %>%
        dplyr::select(
        step,
        AL = px
        )
    md <- movie %>%
        dplyr::group_by(step) %>%
        dplyr::summarize(ML = median(px)) %>%
        dplyr::select(step, ML)
    
    dd <- movie %>%
        dplyr::group_by(step) %>%
        filter(px == min(px)) %>%
        dplyr::select(step, DL = px)
    
    output <- dplyr::inner_join(ad, dd, by = "step") %>%
        dplyr::inner_join(md, by = "step") %>%
        distinct(step, .keep_all = TRUE)
    return(output)
}

get_movie <- function(ball,players) {
    output <- players |>
        dplyr::inner_join(ball, by = "step") %>%
        dplyr::distinct(step,team,unum, .keep_all = TRUE) %>%
        dplyr::arrange(step) %>%
        dplyr::select(step,team,unum,px,py,pvx,pvy,ball_x,ball_y,ball_vx,ball_vy,next_ball_x,next_ball_y,next_ball_vx,next_ball_vy)
    
    return(output)
}