R/piece.R

#SUFC Reports
#Set Piece Numbers Statistics

set_piece <- function(match_report){
  library(pdftools, quietly = TRUE)
  library(stringr, quietly = TRUE)
  
  test <- match_report
  file <- pdf_text(test)
  file_page_1 <- file[1]
  
  keyword <- "SET PIECE"
  file_page_1_sub <- sub(paste0(".*",keyword), "", file_page_1)
  file_page_1_sub <- gsub("^ *|(?<= ) | *$", "", file_page_1_sub, perl = TRUE)
  file_page_1_sub <- strsplit(file_page_1_sub, "\n", fixed = TRUE)
  for(i in 1:length(file_page_1_sub[[1]])){
    file_page_1_sub[[1]][i] <- trimws(file_page_1_sub[[1]][i], "both")
  }
  
  set_piece <- NULL
  for(i in 1:length(file_page_1_sub[[1]])){
    if(grepl("LINEOUT", file_page_1_sub[[1]][i])){
      set_piece[i] <- file_page_1_sub[[1]][i]
    } else if(grepl("START HALF", file_page_1_sub[[1]][i])){
      set_piece[i] <- file_page_1_sub[[1]][i]
    } else if(grepl("START 22", file_page_1_sub[[1]][i])){
      set_piece[i] <- file_page_1_sub[[1]][i]
    } else if(grepl("TAP PENALTY", file_page_1_sub[[1]][i])){
      set_piece[i] <- file_page_1_sub[[1]][i]
    } else if(grepl("PENALTY KICK", file_page_1_sub[[1]][i])){
      set_piece[i] <- file_page_1_sub[[1]][i]
    } else if(grepl("FREE KICK", file_page_1_sub[[1]][i])){
      set_piece[i] <- file_page_1_sub[[1]][i]
    } else {
      set_piece[i] <- "Empty"
    }
  }
  
  set_piece <- set_piece[set_piece != "Empty"]
  set_piece <- gsub("START HALF", "START-HALF", set_piece)
  set_piece <- gsub("START 22", "START-22", set_piece)
  set_piece <- gsub("TAP PENALTY", "TAP-PENALTY", set_piece)
  set_piece <- gsub("PENALTY KICK", "PENALTY-KICK", set_piece)
  set_piece <- gsub("FREE KICK", "FREE-KICK", set_piece)
  set_piece <- strsplit(set_piece, " ")
  
  set_piece[[4]] <- c(set_piece[[4]][1], 0, set_piece[[4]][2], 0, set_piece[[4]][3])
  set_piece[[5]] <- c(set_piece[[5]][1], 0, set_piece[[5]][2], 0, set_piece[[5]][3])
  set_piece[[6]] <- c(set_piece[[6]][1], 0, set_piece[[6]][2], 0, set_piece[[6]][3])
  
  set_piece <- data.frame(matrix(unlist(set_piece), ncol = 5, byrow = TRUE), stringsAsFactors = FALSE)
  
  file_page_1 <- gsub("^ *|(?<= ) | *$", "", file_page_1, perl = TRUE)
  file_page_1 <- strsplit(file_page_1, "\n", fixed = TRUE)
  file_page_1[[1]][2] <- trimws(file_page_1[[1]][2], "both")
  
  names(set_piece) <- c("home_retain", "home_turnover", file_page_1[[1]][2], "away_turnover", "away_retain")
  set_piece <- set_piece[,c(3, 1, 2, 5, 4)]
  
  set_piece
}
AndrewFerris/sufc1863 documentation built on May 5, 2019, 5:59 a.m.