script/cosmasvc2koralquery

#!/usr/bin/env perl
use strict;
use warnings;
use KorAP::Def;
use KorAP::DefList;
use lib 'lib';
use Getopt::Long;

binmode(STDERR, ':encoding(UTF-8)');

# 2020-05-20
#   Preliminary support for C2 def-files.
# 2020-05-29
#   Introduce optimizable object system.
# 2024-07-17
#   Add KorAP::Def.

our $VERSION = 0.3;
our @ARGV;

my $cmd = shift @ARGV;


my ($input, $output, $copysrc);

GetOptions (
  "input|i=s" => \$input,
  "output|o=s" => \$output,
  "copy-src|c=s" => \$copysrc,
)
or die("Error in command line arguments\n");

if ($cmd ne 'def' && $cmd ne 'list') {
  print <<'HELP';
Convert a list of C2 VC definitions or a single definition into
KoralQuery VCs.

  $ perl cosmasvc2koralquery def my_vc.txt | gzip -vc > my_vc.jsonld.gz
  $ cat my_vc.txt | perl cosmasvc2koralquery def - | gzip -vc > my_vc.jsonld.gz

Command: def

  Convert a def file or a list of sigles to a KoralQuery VC.

  Takes the list or def from STDIN and exports to STDOUT.

Command: list

  Convert a list with copy or regex instructions to KoralQuery VCs.

  --output: The output directory
  --copy-src: The directory for def files to copy

HELP
exit 1;
};


# Process a list
if ($cmd eq 'list') {
  KorAP::DefList->new(
    file => ($input || $ARGV[0]),
    copy => ($copysrc || '.'),
    output => ($output || '.')
  )->parse;
  exit(0);
};

# Parse a single def
my $def_parser;
if ($ARGV[0] eq '-') {
  $def_parser = KorAP::Def->new(\*STDIN);
}
elsif ($input) {
  $def_parser = KorAP::Def->new($input);
}
else {
  $def_parser = KorAP::Def->new($ARGV[0]);
};

$def_parser->parse;

# Stringify current (extended?) virtual corpus
print $def_parser->to_string;