Read a NanoSIMS file (possibly corrupted header line)

nanosims.R

# Read a NanoSIMS file (possibly corrupted header line)
# into a data frame (tibble).  Possibly overwrite the orginal with the 
# correct header line.
#
# Requires [readr](https://CRAN.R-project.org/package=readr) package
# 
# Usage within R session:
# > source("/path/to/nanosims.R")
# # make resave FALSE to preserve the orginal file
# > x <- read_nanosims("/path/to/nanosims/file.csv", resave = TRUE)
# # process a directory filled with NanoSIMS .csv files
# > x <- read_nanosims("/path/to/nanosims/directory", resave = TRUE)
#
# Usage from shell prompt
# $ Rscript /path/to/nanosims.R /path/to/nanosims/file.csv --resave TRUE
# or 
# $ Rscript /path/to/nanosims.R /path/to/nanosims/directory --resave TRUE



library(readr)

#' Replace text in a character vector
#' 
#' @param x character, a vector of strings possibly witht he pattern to replace
#' @param pat list of one or more 2 element character patterns stored in a
#'        character vector ala c(find_pattern, replace_pattern)
#' @param ... further arguments for gsub
#' @return character vector of same length as input with pattern replaced as specified.
replace_text <- function(x, 
                         pat = list(c("\nmean", "mean"),
                                    c("\nstddev", "stddev") ),
                         ...){
  
  for (i in seq_along(pat)){
    x <- gsub(pat[[i]][1], pat[[i]][2], x, ...)
  }
  return(x)
}

#' Read a NanoSIMS file as a text stream with replacement of specified patterns.
#' 
#' @param  x character, the name (with path) of the file or a directory of 
#'         NanoSIMS CSV files
#' @param ... arguments for \code{replace_text}
#' @param resave logical, if TRUE resave the file (overwrites original)
#' @param data.frame or a list of data frames if x is a directory with one
#'        element (a data frame) per file found
read_nanosims <- function(x, ..., resave = FALSE){
  fi <- file.info(x[1])
  if (fi$isdir[1]){
    ff <- list.files(x[1], pattern = glob2rx("*.csv"), full.names = TRUE)
    r <- lapply(ff, read_nanosims, ..., resave = resave)
  } else {
    stopifnot(file.exists(x[1]))
    s <- readChar(x[1], file.info(x[1])$size)
    s <- replace_text(s, ...)
    if (resave) writeLines(s, con = x[1], sep = "")
    r <- readr::read_csv(s)
  }
  r  
}


# This only runs if called as a script from the command line
if (!interactive()){
  arg <- commandArgs(trailingOnly = TRUE)
  # caller might have provded no arguments - print help
  # Rscript nanosims.R
  #
  # caller might have provided one argument ala
  # Rscript nanosims.R filename
  # or two
  # Rscript nanosims.R filename --resave TRUE 
  # or 
  # Rscript nanosims.R --resave TRUE filename
  resave <- FALSE
  i <- 1
  if (length(arg) == 0){
    cat("Usage: Rscript /path/to/nanosims.R /path/to/file/or/directory [--resave TRUE|FALSE]\n")
    quit(save = "no", status = 0)
  }
  if (length(arg) > 1){
    ix <- grep("-resave", arg, fixed = TRUE)
    if (length(ix) >= 1){
      resave <- tolower(arg[ix+1]) %in% c("true", "yes", "1")
      if (ix[1] == 1){
        i <- 3
      }
    }
  }
  x <- try(read_nanosims(arg[i], resave = resave))
  ok <- inherits(x, "try-error")
  quit(save = "no", status = as.numeric(ok))
}