##################################################
################### uxd2mtx ######################
##################################################
uxd2mtx <- function(uxdfile) {
   # Function for reading UXD files 
   # Assumptions: data in two columns
   # Args: uxdfile (filename with extension)
   # Return value: matrix with two columns
   
   cchar <- "[;_]" #regexpr matching the comment characters used in Bruker's UXD
   cdata <- "[0-9]" #regexpr matching one character of any digit
   
   # A new file (datafile) containing only data will be created,
   # extension ".data" appended to uxdfile
   #datafile <- paste(uxdfile,".data",sep="")
   
   ufile <- file(uxdfile, "r")
   f <- readLines(ufile, n=-1) #read _all_ lines from UXD file
   close(ufile)
   
   # This way we identify data rows by looking for numeric characters.
   #wh <- regexpr("[0-9]", f)
   # This way we identify header rows
   # We assume that all other rows are data
   wh <- regexpr(cchar, f)
   
   mh <- wh[1:length(wh)] # this gives you the corresponding index vector
   # the value of each element corresponds to the position of the regexp match.
   # value = 1 means the first character of the row is cchar (row is header)
   # value =-1 means no cchar occur on the row (row is data)
   
   i <- seq(1, length(mh) - 1, 1)
   j <- seq(2, length(mh), 1)
   
   starts <- which(mh[i] == 1 & mh[j] != 1) + 1
   ends   <- length(mh)
   f <- f[starts:ends]
   
   zz <- textConnection(f, "r")
   ff <- matrix(scan(zz, what = numeric()), ncol=2, byrow=T)
   close(zz)
      
   #zz <- file(datafile, "w") #open connection to datafile
   #write.table(ff, file=datafile, row.names=F, sep=",")
   #close(zz)

   # Return matrix
   ff
}