R/pasteCsv3.R

In batch: Batching Routines in Parallel and Passing Command-Line Arguments to R

mergeCsv <- function(every=1, outfile="allResults.csv", quote=FALSE) {
  ## List of all the csv files in the current directory
  csvFiles <- setdiff(  sort( dir( pattern=".csv" ) ),  outfile  )
  if( length(csvFiles) == 0 )
    stop("pasteCsv: No csv file in the current directory.")
  print( csvFiles )

  ## Read in and tie the data together
  res <- read.csv( csvFiles[1], stringsAsFactors=FALSE )
  for( i in 2:length(csvFiles) ) {
    temp <- read.csv( csvFiles[i] )
    if(ncol(temp)!=ncol(res) || !all(names(temp)==names(res)) ){
      cat(paste("Names of csv file (", paste(names(temp),collapse=","),") do not match names of current data (", paste(names(res)), ").", sep=""), "\n")

      cat("Current dataset:\n")
      print(res)
      cat("Dataset trying to merge in:\n")
      print(temp)

      if(ncol(temp)!=ncol(res)){
        stop("mergeCsv Error, dimensions not compatible")
      }else{
        names(temp) <- names(res)
        cat("warning, names of new dataset do not match, but were overwritten")
      }
    }
    res <- rbind( res, temp )
  }

  ## Try to sort by seed, if possible
  wh <- which(names(res) == "seed")
  if(length(wh) == 1)
    res <- res[order(res$seed), ]

  ## Now merge pieces of it together
  if( every>1 ) {
    i <- 1
    while( i < nrow(res) ) {
      rows <- i + 1:every - 1

      for( col in 1:ncol(res) ) {
        if( is.numeric(res[rows,col]) ) {
          ## Replace first occurance with mean of the rest
          res[i,col] <- mean(res[rows,col])
        }
      }
      ## destroy non-first occurances
      res <- res[ -rows[-1], ]

      ## and increment i
      i <- i + 1
    }
  }

  ## write the results
  if( nchar(outfile) > 0 )
    write.csv(res, outfile, quote=quote, row.names=FALSE)

  return( res )
}

#mergeCsv( 5 )