# ---------------------
# 
# Mark the plays that signal the end of a drive
# 
# Currently, the data has no easy way to determine when a
# drive has ended. Weather the drive ends on a punt, turnover,
# or score.
# 
# This R Script will read in the SORTED .csv file, check to see if the
# offensive and defensive teams switch, or if there was a PlayType 
# that would result in a change of possession, or if there was a turnover 
# on the play, and then make a bool variable based on the result. This 
# would be used to help with marking how many points a drive resulted in.
# 
# ---------------------

# Libraries to Use
library(tidyverse)

# Set the working directory to match where the files are pulled from
setwd("C:/Users/Ryant/Downloads/Capstone/Cleaned Data Files")

# Read in the data
Plays <- read_csv("pbpSorted2021-2024.csv")

# Set aside the rows that aren't a play
NonPlays <- Plays %>% 
  filter(is.na(OffenseTeam) |
           is.na(DefenseTeam) |
           (
             is.na(PreviousPlayID) &
               is.na(NextPlayID)
           )) %>% 
  mutate(DriveEnd = FALSE) %>% 
  select(1:10,45,11:44) # Shift to match the other set later

# Grab only the rows of actual plays
OnlyPlays <- Plays %>% 
  filter(!is.na(OffenseTeam) &
           !is.na(DefenseTeam) &
           (
             !is.na(PreviousPlayID) |
               !is.na(NextPlayID)
           ))

# Case 1: The "Standard" Case
# Check for if the Offense & Defense Swap

# Compare the Offense and Defense Teams with the next row
EndDrive <- OnlyPlays %>% 
  mutate(Case1 = case_when(
    OffenseTeam == lead(OffenseTeam) & DefenseTeam == lead(DefenseTeam) ~ FALSE,
    TRUE ~ TRUE
  )) %>% 
  select(1:10,45,11:44) # Shift up for easier check


# Case 2: End of Half/Game

# Check for if Quarter == 2 && lead(Quarter) == 3 or...
# Check for if GameId != lead(GameId)
EndDrive <- EndDrive %>% 
  mutate(Case2 = case_when(
    Quarter == 2 & lead(Quarter) == 3 ~ TRUE,
    GameId != lead(GameId) ~ TRUE,
    TRUE ~ FALSE
  )) %>% 
  select(1:11,46,12:45) # Shift up for easier check

# Case 3: Pick 6/Fumble Recovery for a TD

# Check if IsFunble or IsInterception == 1
EndDrive <- EndDrive %>% 
  mutate(Case3 = case_when(
    # If there is an interception, 
    # that is not listed as "No Play", 
    # doesn't have an accepted penalty,
    # and doesn't have replay assist reverse the play, 
    # then it is counted as the end of the drive
    IsInterception == 1 & 
      IsNoPlay == 0 & 
      IsPenaltyAccepted == 0 &
      !str_detect(Description, "PLAY WAS REVERSED") ~ TRUE,
    # If there is a fumble, 
    # that is not listed as "No Play", 
    # doesn't have an accepted penalty,
    # isn't a botched snap,
    # isn't fumbled out of bounds,
    # isn't recovered by the same player on offence,
    # and doesn't have replay assist reverse the play, 
    # then it is counted as the end of the drive
    IsFumble == 1 &
      IsNoPlay == 0 & 
      IsPenaltyAccepted == 0 &
      !str_detect(Description, "ABORTED") &
      !str_detect(Description, "OUT OF BOUNDS") &
      !str_detect(Description, "AND RECOVERS") &
      !str_detect(Description, "REVERSED") ~ TRUE,
    TRUE ~ FALSE
  )) %>% 
  select(1:12,47,13:46) # Shift up for easier check

# Case 4: Multiple Turnovers resulting in original offense maintaining possession

# Covered by previous cases
# If Case 3 == True (Fumble/Interception) & Case 1 == False (Same team following play)

# Case 5: Successful Onside Kicks

# Check for the offence and defense matching following the extra-point/two-point conversion/kickoff
EndDrive <- EndDrive %>% 
  mutate(Case5 = case_when(
    (str_detect(PlayTypeUpdate, "EXTRA POINT") |
     str_detect(PlayTypeUpdate, "TWO-POINT CONVERSION")) &
      OffenseTeam == lead(OffenseTeam) &
      DefenseTeam == lead(DefenseTeam) &
      !Case2 ~ TRUE, # make sure it wasn't the last play of the half
    TRUE ~ FALSE
  )) %>% select(1:13,48,14:47) # Shift up for easier check

# Combine the Cases
EndDrive <- EndDrive %>% 
  mutate(DriveEnd = case_when(
    Case1 |
    Case2 |
    Case3 |
    Case5 ~ TRUE,
    TRUE ~ FALSE
  )) %>% select(1:10,49,15:48) # Shift up for easier check, removing the case checks

# Get all the data back together
PlaysFinal <- rbind(EndDrive, NonPlays) %>% 
  arrange(GameId, PlayID)


# Save the CSV
write_csv(PlaysFinal, "pbpDriveEnd2021-2024.csv")

print("Done!")