# ---------------------
# 
# Mark the plays with how many points are scored at the end of a drive
# 
# Currently, the data has no way to determine how many points are
# scored at the end of a drive.
# 
# This R Script will read in the SORTED .csv file, check to see what the 
# result of a drive is, and then mark the whole drive with the value
# of 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("pbpDriveEnd2021-2024.csv")

# Flip the order to make running the loop to add variable easier
Points <- Plays %>% 
  arrange(desc(GameId), desc(PlayID))

# Run a loop to add point value
PointsScored <- 0 # Initialize the variable
Points$DrivePoints <- NA_integer_ # Create the new column

# Set aside the rows that aren't a play
NonPlays <- Points %>% 
  filter(is.na(OffenseTeam) |
           is.na(DefenseTeam) |
           (
             is.na(PreviousPlayID) &
               is.na(NextPlayID)
           ))

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

# Loop through rows and assign the value
for (i in 1:nrow(OnlyPlays)) {
  if (OnlyPlays$DriveEnd[i] == TRUE) { # End of Drive
    # Field Goal Successful
    if (OnlyPlays$PlayTypeUpdate[i] == "FIELD GOAL" && 
        str_detect(OnlyPlays$Description[i], "IS GOOD")) {
      PointsScored <- 3 # Assign value of points scored
    }
    # Following Touchdown, going for 1
    else if (OnlyPlays$PlayTypeUpdate[i] == "EXTRA POINT") { 
      # Got the extra point
      if (str_detect(OnlyPlays$Description[i], "IS GOOD")) { 
        PointsScored <- 7
      }
      # Failed the extra point
      else { 
        PointsScored <- 6
      }
    }
    # Following Touchdown, going for 2
    else if (OnlyPlays$PlayTypeUpdate[i] == "TWO-POINT CONVERSION") { 
      # Got the extra points
      if (str_detect(OnlyPlays$Description[i], "SUCCEEDS")) { 
        PointsScored <- 8
      }
      # Failed the extra points
      else { 
        PointsScored <- 6
      }
    }
    # The defense scored a Safety
    else if (str_detect(OnlyPlays$Description[i], "SAFETY")) { 
      PointsScored <- -2
    }
    # Punt / End of Game or Half
    else if (OnlyPlays$PlayTypeUpdate[i] == "PUNT" ||
             (OnlyPlays$GameId[i] != OnlyPlays$GameId[i-1] && i > 1) ||
             i == 1 ||
             (OnlyPlays$Quarter[i] == 2 && OnlyPlays$Quarter[i-1] == 3)) {
      PointsScored <- 0
    }
    # Turnover
    else {
      PointsScored <- PointsScored * -1
    }
  }
  OnlyPlays$DrivePoints[i] <- PointsScored # Assign points scored to dataset
}

# Get all the data back in order
PointsFinal <- rbind(OnlyPlays, NonPlays) %>% 
  arrange(GameId, PlayID) %>% 
  select(1:11, 46, 12:45) # Shift over the new variable


# Save the CSV
write_csv(PointsFinal, "pbpDrivePoints2021-2024.csv")

print("Done!")