#!/usr/local/bin/perl
# $Id: acecsplit,v 1.1 1995/02/02 22:21:42 rd Exp $
#  File: acecsplit
#  Author: Detlef Wolf (D.Wolf@dkfz-heidelberg.de)
#  Copyright (C) European Data Resource, 1994
#--------------------------------------------------------------------
# This file is part of the IGD project
#
# Description: split a .ace file into parts according to mapping file
# usage: acecsplit [-c <configfile>] [-p <prefix>] [<filename>...]
# Inputs: .ace data from files or stdin
#         if option -c is given, then
#         split according to config file,
#         else split by class name
# Output: .ace data is distributed into files according to config
#         if <prefix> is given then all output files start with it,
#         default for <prefix> is "split."
# Precondition: 
# Postcondition: 
# HISTORY:
# Created: Fri Nov 11 09:49:27 1994 (dok256)
#--------------------------------------------------------------------


# ------------------ evaluate command line options
if ($ARGV[0] =~ /^-h(elp)?/) { 
  print 
"usage: acecsplit [-h] [-c <configfile>] [-p <prefix>] [<filename>...]
purpose: split .ace files into parts
<prefix> is the string prepended to all output file names; default \"split.\"
modes of operation:
- automatic
  put all objects of each class into a file with the classes name
- configured  (option -c)
  a mapping file controls witch class goes to which file;
  Format of the lines in the mapping file: 
    <class_name> <split_file_name>
  A special class name is \"default\", where all non-mapped classes go
";
exit;
};

if ($ARGV[0] eq "-c") { 
  shift;
  $configfile = $ARGV[0];
  shift;
  $byname=0;
} else {$byname=1};

#$byname && die "class name split no yet implemented; use -c option\n";

$prefix="split.";
if ($ARGV[0] eq "-p") { 
  shift;
  $prefix = $ARGV[0];
  shift;
};

if (! $byname) {
# ------------------  parse the config file
open(L,"<$configfile") || die "couldn't open $configfile";
  while (<L>) {
    /^\s*#/ && next;
    /^\s*(\w+)\s+(\w+)\b/ || next;    
    $class2db{$1}=$2;
    $db2class{$2}= $db2class{$2} . " " . $1;
    print "mapping class $1 onto output $2\n";
  };
close(L);
if (! $class2db{"default"}) {die "$configfile must contain a default clause"};

# ---------------------- open the output channels for write

foreach (keys %db2class) {
  &openoutput($_);
};

};


# --------------------- parse the .ace file
# set an intitial target file name
if ($byname) {
#  $targetdb="unclassified";
#  $db2class{$targetdb}= "(lines before first object)"
#  &openoutput($targetdb);  
  $targetdb="";  # ignore the lines before the first object
} else {
  $targetdb=$class2db{"default"};
}

$inobject=0;
while (<>) {
  m|^//| && next;       # skip comments
  if (/^(\w+)\s*:?\s+"?([^"]+)"?/ && !$inobject) {  # a new object starts
    $class=$1;
    $object=$2;
    $inobject=1;

    $targetdb=$class2db{$class};  # determine output destination
    if (! $targetdb) {            # a new/unknown destination
      if ($byname) {              # to a new file or to default?
        $class2db{$class}=$class; # open a new file
        $db2class{$class}= $db2class{$class} . " " . $class;
        &openoutput($class);
        $targetdb=$class;
      }
      else {
        $targetdb=$class2db{"default"}
      }
    }

    if ($class eq "LongText") {
      print $targetdb $_;
      while (<>) {
        print $targetdb $_;
        /^\*\*\*LongTextEnd\*\*\*$/ && last;
      };
      print $targetdb "\n";
      $inobject=0;
      next;
    }
  };
  if (/^\s*$/ && $inobject) {  # an object ends
    $inobject=0;
  }
  print $targetdb $_;

};  # loop over the .ace lines

# -------------------- subroutines ----------------
sub openoutput {
  # input: $_[0]  -- name of output channel
  local($handle,$file);
  $handle=$_[0];
  $file="$prefix$_[0].ace";
  open($handle,">$file") || die "file $file not writeable";
  chop ($date=`date`);
  print $handle "// file $file created by acecsplit on $date
// for class(es) $db2class{$_[0]}\n\n";
  print "opening $file\n";
};