#!/usr/bin/perl

# pg_restore must be in the path
#
# we use pg_restore to get the toc (requires the backup be made with pg_dump -Fc) from a backup
# create a newtoc with all the things that look like tsearch2 stuff stripped out
# so you can pg_restore -Fc -L newtoc ...

# syntax:  xtract dumpfile.bak > newtoc

use strict;

my %list = ('gtsq' => 1,
'gtsq_in' => 1,
'gtsq_out' => 1,
'gtsvector' => 1,
'gtsvector_in' => 1,
'gtsvector_out' => 1,
'tsquery' => 1,
'tsquery_in' => 1,
'tsquery_out' => 1,
'tsvector' => 1,
'tsvector_in' => 1,
'tsvector_out' => 1,
'statinfo' => 1,
'tokenout' => 1,
'tokentype' => 1,
'tsdebug' => 1,
'_get_parser_from_curcfg' => 1,
'concat' => 1,
'dex_init' => 1,
'dex_lexize' => 1,
'exectsq' => 1,
'get_covers' => 1,
'gin_extract_tsquery' => 1,
'gin_extract_tsvector' => 1,
'gin_ts_consistent' => 1,
'gtsq_compress' => 1,
'gtsq_consistent' => 1,
'gtsq_decompress' => 1,
'gtsq_penalty' => 1,
'gtsq_picksplit' => 1,
'gtsq_same' => 1,
'gtsq_union' => 1,
'gtsvector_compress' => 1,
'gtsvector_consistent' => 1,
'gtsvector_decompress' => 1,
'gtsvector_penalty' => 1,
'gtsvector_picksplit' => 1,
'gtsvector_same' => 1,
'gtsvector_union' => 1,
'headline' => 1,
'length' => 1,
'lexize' => 1,
'numnode' => 1,
'parse' => 1,
'plainto_tsquery' => 1,
'prsd_end' => 1,
'prsd_getlexeme' => 1,
'prsd_headline' => 1,
'prsd_lextype' => 1,
'prsd_start' => 1,
'querytree' => 1,
'rank' => 1,
'rank_cd' => 1,
'reset_tsearch' => 1,
'rewrite' => 1,
'rewrite_accum' => 1,
'rewrite_finish' => 1,
'rexectsq' => 1,
'set_curcfg' => 1,
'set_curdict' => 1,
'set_curprs' => 1,
'setweight' => 1,
'show_curcfg' => 1,
'snb_en_init' => 1,
'snb_lexize' => 1,
'snb_ru_init_koi8' => 1,
'snb_ru_init_utf8' => 1,
'spell_init' => 1,
'spell_lexize' => 1,
'stat' => 1,
'strip' => 1,
'syn_init' => 1,
'syn_lexize' => 1,
'thesaurus_init' => 1,
'thesaurus_lexize' => 1,
'to_tsquery' => 1,
'to_tsvector' => 1,
'token_type' => 1,
'ts_debug' => 1,
'tsearch2' => 1,
'tsq_mcontained' => 1,
'tsq_mcontains' => 1,
'tsquery_and' => 1,
'tsquery_cmp' => 1,
'tsquery_eq' => 1,
'tsquery_ge' => 1,
'tsquery_gt' => 1,
'tsquery_le' => 1,
'tsquery_lt' => 1,
'tsquery_ne' => 1,
'tsquery_not' => 1,
'tsquery_or' => 1,
'tsvector_cmp' => 1,
'tsvector_eq' => 1,
'tsvector_ge' => 1,
'tsvector_gt' => 1,
'tsvector_le' => 1,
'tsvector_lt' => 1,
'tsvector_ne' => 1,
'rewrite' => 1,
'gin_tsvector_ops' => 1,
'gist_tp_tsquery_ops' => 1,
'gist_tsvector_ops' => 1,
'tsquery_ops' => 1,
'tsvector_ops' => 1,
'pg_ts_cfg' => 1,
'pg_ts_cfg_pkey' => 1,
'pg_ts_cfgmap' => 1,
'pg_ts_cfgmap_pkey' => 1,
'pg_ts_dict' => 1,
'pg_ts_dict_pkey' => 1,
'pg_ts_parser' => 1,
'pg_ts_parser_pkey' => 1
);


my @list = qx/pg_restore -Fc --list $ARGV[0]/;
#open(F,'>tmp') or die;
#print F @list;
#close(F);

foreach (@list)
{
	if (/(FUNCTION|AGGREGATE)\s+(public\s+)?(.*)\(/)
	{
		#print "$1 | $2 | $3";
		if (!exists($list{$3}))
		{
			print;
		}
	}
	elsif (/(TYPE|ACL|CONSTRAINT|TABLE|TABLE DATA|OPERATOR CLASS)\s+(public)?\s+(.*?)\s/)
	{
		#print "$1 | $2 | $3";
		if (!exists($list{$3}))
		{
			print;
		}
	}
	elsif (/(OPERATOR)\s+(public)?\s+(.*?)\s/)
	{
		#print "$1 | $2 | $3";
		open(F, '>item.tmp') or die;
		print F $_;
		close(F);
		my $buff = qx/pg_restore -Fc -L item.tmp $ARGV[0]/;
		unlink('item.tmp');

		#open(F, '>item.txt') or die;
		#print F $buff;
		#close(F);

		if ($buff =~ /PROCEDURE = (\w+)/)
		{
			if (!exists($list{$1}))
			{
				print;
			}
		}
		else {
			print STDERR "cannot find PROCEDURE\n";
		}

	}
	else {
		print;
	}
}