# #!/usr/bin/Rscript
# 
# args <- commandArgs(TRUE)
#https://cran.r-project.org/web/packages/stringr/vignettes/regular-expressions.html
library(digest)
#!/usr/bin/env Rscript
args = commandArgs(trailingOnly=TRUE)

#check if file argument was provided
if (length(args) < 1) stop("Usage: Rscript cagi-validate.R <file>")


infile <- args[[1]]
#check if file exists
if (!file.exists(infile)) stop("File ",infile," does not exist!")
# and if file can be read
if (file.access(infile,mode=4) < 0) stop("No permission to read file ",infile,"!")

#read the lines in the file
lines <- scan(infile,what=character(),sep="\n")
if (length(lines) != 39) stop("File does not have the required 39 lines")

#break lines into data fields
fields <- strsplit(lines,"\t")
#check if each line has six fields
if (any(sapply(fields,length) != 9)) stop("Must have 9 tab-delimited columns including Nucleotide position and Variant!")
#Comments are optional

#check if the header is valid
header <- fields[[1]]
valid.header <- c("Nucleotide_position", "Variant", "P-value_MO", "Standard_deviation", "P-value_MO+WT", "Standard_deviation", "Functional_effect", "Confidence:", "Comments")
if (any(header != valid.header)) stop("Invalid table headers!")

#extract data columns for the remaining lines
body <- fields[-1]
body.cols <- lapply(1:9,function(i) sapply(body,`[[`,i))


#check if the first column contains valid allele descriptors, where ^ marks start and $ marks end
if (any(regexpr("^\\w\\d+\\w(,\\w\\d+\\w)*$",body.cols[[1]]) != 1)) {
  stop("First column must contain valid alleles!")
}

#check if the second (variant) column contains valid allele descriptors
if (any(regexpr("^\\w\\d+\\w(,\\w\\d+\\w)*$",body.cols[[2]]) != 1)) {
  stop("First column must contain valid alleles!")
}

#check if the third column "P-value_MO" is numerical values or stars
if (!all(sapply(body.cols[[3]],function(x) x == "*" || !is.na(as.numeric(x))))) {
  stop("Second column must be numeric or \"*\"!")
}

#check if the fourth column "Standard_deviation" is numerical values or stars
if (!all(sapply(body.cols[[4]],function(x) x == "*" || (!is.na(as.numeric(x))) ))) {
  stop("Third column must be numeric or \"*\"!")
}

#check if the fifth column "P-value_MO+WT" is numerical values or stars
if (!all(sapply(body.cols[[5]],function(x) x == "*" || (!is.na(as.numeric(x))) ))) {
  stop("Third column must be numeric or \"*\"!")
}

#check if the sixth column "Standard_deviation" is numerical values or stars
if (!all(sapply(body.cols[[6]],function(x) x == "*" || (!is.na(as.numeric(x))) ))) {
  stop("Third column must be numeric or \"*\"!")
}

#check if the sixth column "Functional_effect" is numerical values or stars
if (!all(sapply(body.cols[[7]],function(x) x == "*" || (!is.na(as.numeric(x))) ))) {
  stop("Third column must be numeric or \"*\"!")
}

#check if the sixth column "Confidence:" is numerical values or stars
if (!all(sapply(body.cols[[8]],function(x) x == "*" || (!is.na(as.numeric(x))) ))) {
  stop("Third column must be numeric or \"*\"!")
}


cat("\nCongratulations!\nYour submitted file has a validated format.\n")

#MD5 Hash Using Package digest, optional
if (file_test("-x", "/usr/bin/md5sum") && file_test("-x", "/usr/bin/cut")) {
  # cat("MD5 hash for your file is:", digest(body, algo="md5",  serialize=TRUE, file=FALSE,
  #                                          length=Inf, skip="auto", ascii=FALSE, raw=FALSE, seed=0,
  #                                          errormode="warn"), "\n")
  cat("MD5 hash for your file is:")
  Sys.unsetenv("PATH")
  system("/usr/bin/md5sum $infile | /usr/bin/cut -c 1-32")
  cat("\n")
  if (file_test("-r", "cagi.txt") && file_test("-x", "/bin/cat")) {
    cat("Receipt code for your file is: ")
    system("/bin/cat cagi.txt $infile | /usr/bin/md5sum | /usr/bin/cut -c 1-32")
    cat("\n")
  }
  quit()
}