#!/bin/sh
# \
	exec tclsh "$0" ${1+"$@"}

proc FindFiles {dir} {
global argv0 files

	if { [catch {glob $dir/* $dir/.*} ents] } { return }
	foreach ent $ents {
		if { [file tail $ent] == "." || [file tail $ent] == ".." } { continue }
		if { [file isdirectory $ent] } {
			FindFiles $ent
		} else {
			lappend files $ent
		}
	}
	return
}

	if { $argc != 1 } {
		puts "Usage: $argv0 <dir>"
		exit 1
	}
	set files ""
	FindFiles [lindex $argv 0]
	set orig_list $files
	foreach file "$files" {
		if { [catch { file size $file } size($file)] } {
			puts stderr "$file does not exist (sym link to nothing?)"
		}
		if { $size($file) == 0 } { puts "$file is zero bytes!" }
	}

	set cnt 0
	foreach file "$orig_list" {
		if { $cnt == 75 } {
			puts -nonewline "."
			flush stdout
			set cnt 0
		}
		incr cnt
		foreach tst_f $files {
			if { $size($tst_f) != $size($file) } { continue }
			if { $tst_f == $file || $size($file) == 0 } { continue }
			catch {eval exec diff -s $file $tst_f} output
			if { [regexp "identical" $output m] == 1 } {
				puts stderr "$file == $tst_f (size: $size($file))"
				if { [regexp {20\d\d\d\d\d\d[_-]} $file] } {
					puts "based on name, I would keep file: $file and delete: $tst_f"
					exec rm $tst_f
				} elseif { [regexp {20\d\d\d\d\d\d[_-]} $tst_f] } {
					puts "based on name, I would keep tst_f: $tst_f and delete: $file"
					exec rm $file
				}
			}
		}
		# only check remainder of files (e.g. lindex 0 was just compared)
		set files [lrange $files 1 end]
	}