R/read_icv.R

In rtrim: Trends and Indices for Monitoring Data

read_icv <- function(x,...){
  UseMethod("read_icv")
}

read_icv.character <- function(x, J=0, ...){
  icv_read(filenm=x, J=J)
}


read_icv.trimcommand <- function(x,...){
  basename <- strsplit(x$file,"\\.")[[1]][1]
  filenm <- paste0(basename,".ICV")
  icv_read(filenm, x$ntimes)
}


# Workhorse reader
icv_read <- function(filenm, J=0)
{
  # the covariance matrices might differ in size from site to site due
  # to missing data. So, use autodetection
  covin <- list()
  lines = readLines(filenm)
  nsite <- 0
  while (length(lines)>0) {
    snif = read.table(textConnection(lines), nrows=1)
    J = ncol(snif)
    stopifnot(J>0)
    df <- read.table(textConnection(lines), nrows=J)
    V <- unname(as.matrix(df))
    nsite <- nsite+1
    covin[[nsite]] <- V
    lines <- lines[-(1:J)]
  }
  covin
}

# Workhorse reader
# icv_read <- function(filenm, J=0)
# {
#   m = unname(as.matrix(read.table(filenm)))
#   if (J>0) { # perform a check
#     stopifnot(ncol(m)==J) # COVIN should have J columns
#     stopifnot(nrow(m)%%J == 0) #... and I*J rows
#   } else J <- ncol(m) # autodetect
#   nsite <- nrow(m) %/% J
#   covin <- vector("list", nsite)
#   idx = 1:J
#   for (i in 1:nsite) {
#     covin[[i]] <- m[idx, ]
#     idx = idx + J
#   }
#   covin
# }