# Filter used when gathering a corpus for lang-tables
#	(c) 2003, Robert Spalek <robert@ucw.cz>
#
#	This file is NOT installed anywhere.  It is stored in CVS for
#	re-production purposes only.
#
#	lang-tables-starturls is a companion of this file

switch host {
	case =** "*.centrum.cz",
		=** "*.idnes.cz":
		language = "cz";
		bonus = 50;
		card_bonus = 17;
		accept;
	case =** "*.centrum.sk",
		=** "*.zoznam.sk":
		if path =** "*inglise*" { reject; }
		language = "sk";
		bonus = 30;
		accept;
	case =** "*.index.hu":
		if path =** "/en/*" { reject; }
		language = "hu";
		accept;
	case =** "*.krakow.pl":
		if path =** "/en/*" { reject; }
		if path =~~ ".*[efd]menu\..*" { reject; }
		language = "pl";
		accept;
	case =** "*.msn.de",
		=** "*.berlin.de":
		if path =** "*english*" { reject; }
		if path =** "*_en.*" { reject; }
		language = "de";
		accept;
	case =** "*.planet.nl",
		=** "*.utrecht.nl":
		if host =** "*tucows.planet.nl" { language = "en"; accept; }
		language = "nl";
		accept;
	case =** "*slashdot.org",
		=** "*.microsoft.com",
		=** "*.linux.org":
		language = "en";
		accept;
	case =** "*.paris.fr",
		=** "*.lyon.fr":
		if path =** "/en/*" { reject; }
		if path =** "/sp/*" { reject; }
		language = "fr";
		accept;
	case =** "*.libero.it",
		=** "*.lycos.it":
		if host =** "*tucows.libero.it" { language = "en"; accept; }
		if path =** "/ENG/*" { reject; }
		language = "it";
		accept;
	case =** "*.lycos.es",
		=** "*.bcn.es":
		if path =** "/english/*" { reject; }
		language = "es";
		accept;
}
reject;
