R/breakdown.R

#SUFC Reports
#Breakdown Numbers Statistics

breakdowns <- function(match_report){
  library(pdftools, quietly = TRUE)
  library(stringr, quietly = TRUE)
  
  test <- match_report
  file <- pdf_text(test)
  file_page_1 <- file[1]
  
  keyword <- "BREAKDOWN"
  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")
  }
  
  breakdowns <- NULL
  for(i in 1:length(file_page_1_sub[[1]])){
    if(grepl("RUCKS", file_page_1_sub[[1]][i])){
      breakdowns[i] <- file_page_1_sub[[1]][i]
    } else if(grepl("MAULS", file_page_1_sub[[1]][i])){
      breakdowns[i] <- file_page_1_sub[[1]][i]
    } else {
      breakdowns[i] <- "Empty"
    }
  }
  
  breakdowns <- breakdowns[breakdowns != "Empty"]
  breakdowns <- strsplit(breakdowns, " ")
  breakdowns <- data.frame(matrix(unlist(breakdowns), 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(breakdowns) <- c("home_retain", "home_turnover", file_page_1[[1]][2], "away_turnover", "away_retain")
  breakdowns <- breakdowns[,c(3, 1, 2, 5, 4)]
  
  breakdowns
}
AndrewFerris/sufc1863 documentation built on May 5, 2019, 5:59 a.m.