############################################################################
#
#	File:     datmerge.icn
#
#	Subject:  Program to merge data files
#
#	Author:   Gregg M. Townsend
#
#	Date:     November 16, 2001
#
############################################################################
#
#	This file is in the public domain.
#
############################################################################
#
#	Datmerge reads and combines arbitrary text-based data files that
#	contain whitespace-separated data.  For each data field, a single
#	value is written to standard output after applying a selected
#	operator (such as median or minimum) to the corresponding values
#	from all the input files.
#
#	Usage:  datmerge [-operator] filename...
#
#	Operators:
#	    -min or -minimum
#	    -max or -maximum
#	    -med or -median   (this is the default)
#	    -mean
#
#	Values convertible to numeric are treated as such.
#	All others are treated as strings.
#
############################################################################
#
#  Links:  numbers, strings
#
############################################################################

link numbers, strings



procedure main(args)
   local a, opr, files, lines

   if args[1][1] == '-' then {
      a := get(args)
      opr := case a of {
         "-min" | "-minimum":  minimum
         "-max" | "-maximum":  maximum
         "-med" | "-median":   median
         "-mean":              mean
         default:              stop(&progname, ": unrecognized operator: ", a)
         }
      }
   else
      opr := median

   if *args < 1 then
      stop("usage: ", &progname, " [-operator] filename...")

   files := []
   while a := get(args) do
      put(files, open(a)) | stop("cannot open ", a)

   repeat {
      lines := []
      every put(lines, read(!files))
      if *lines = 0 then break
      merge(lines, opr)
      }

end



#  merge(lines, opr) -- output the result of merging a list of lines.

procedure merge(lines, opr)
   local a, s, w, fields, ws

   fields := []
   every s := !lines do {
      put(fields, a := [])
      every w := words(s) do
         put(a, numeric(w) | w)
      }

   ws := ""
   repeat {
      a := []
      every put(a, get(!fields))
      if *a = 0 then break
      writes(ws, opr(a))
      ws := " "
      }

   write()
end



#  Operator Procedures
#
#  These procedures take a list and return a value.
#  They must always return something regardless of the data.
#  Those that involve arithmetic need to tolerate string data somehow.

procedure minimum(a)
   a := sort(a)
   return a[1]
end

procedure maximum(a)
   a := sort(a)
   return a[-1]
end

procedure mean(a)
   return (amean ! nsubset(a)) | median(a)
end

procedure median(a)
   a := sort(a)
   return a[(*a + 1) / 2]
end



#  nsubset(a) -- return subset of array a that contains numeric values

procedure nsubset(a)
   local b
   b := []
   every put(b, numeric(!a))
   if *b > 0 then
      return b
   else
      fail
end