dicelister/config/example.pl

package config;

use strict;

# To enable use of unicode characters
use utf8;

# You can change minimum word length allowed for a word to be added to list.
$dicelister::minimum_word_length = 4;

@dicelister::data_sources = (
	# Some file with literature text, article body text etc.
	# to extract words from. Add as many random text as you want
	# and edit later if you need.
	#"file://${dicelister::data_dir}/data.txt",
	# You can add any article, text URL, as many as you want.
	# It will extract word from that URL too.
	# Here are some bird related Wikipedia article URLs as an example.
	"https://en.wikipedia.org/wiki/Bird",
	"https://en.wikipedia.org/wiki/Ring_ouzel",
	"https://en.wikipedia.org/wiki/Common_blackbird",
	"https://en.wikipedia.org/wiki/True_thrush",
	"https://en.wikipedia.org/wiki/Song_thrush",
	"https://en.wikipedia.org/wiki/Olive_thrush",
	"https://en.wikipedia.org/wiki/Black-billed_thrush",
	"https://en.wikipedia.org/wiki/Black-headed_bulbul",
	"https://en.wikipedia.org/wiki/Passerine",
	"https://en.wikipedia.org/wiki/Common_cuckoo",
	"https://en.wikipedia.org/wiki/Eurasian_sparrowhawk",
	"https://en.wikipedia.org/wiki/Bird_of_prey",
	"https://en.wikipedia.org/wiki/Eurasian_eagle-owl",
	"https://en.wikipedia.org/wiki/Golden_eagle",
	"https://en.wikipedia.org/wiki/Peregrine_falcon",
	"https://en.wikipedia.org/wiki/Galliformes",
	"https://en.wikipedia.org/wiki/Wild_turkey",
	"https://en.wikipedia.org/wiki/Prairie_chicken",
	"https://en.wikipedia.org/wiki/Northern_goshawk",
	"https://en.wikipedia.org/wiki/Domestic_turkey",
	"https://en.wikipedia.org/wiki/Bald_eagle",
	"https://en.wikipedia.org/wiki/Kingbird",
	"https://en.wikipedia.org/wiki/Old_World_quail",
	"https://en.wikipedia.org/wiki/Malleefowl",
	"https://en.wikipedia.org/wiki/Anseriformes",
	"https://en.wikipedia.org/wiki/Anhimidae",
	"https://en.wikipedia.org/wiki/Magpie_goose",
	"https://en.wikipedia.org/wiki/Duck",
	"https://en.wikipedia.org/wiki/Heritage_turkey",
	"https://en.wikipedia.org/wiki/Ocellated_turkey",
	"https://en.wikipedia.org/wiki/Wattle_(anatomy)",
	"https://en.wikipedia.org/wiki/American_white_pelican",
	"https://en.wikipedia.org/wiki/California_condor",
	# Adding words from string is also possible. Putting all 7000+ words
	# in here isn't probably ideal. A file (shown above) would be more
	# appropriate. So this is just to show you as an example.
	#'The quick brown fox jumps over the lazy dog.'
);

# List symbols, characters or words you don't want to appear in the word list.
# This adds to the default list available in @exclude_strings on dicelister.pl
@dicelister::exclude_strings = (@dicelister::exclude_strings, (
	'`',
	# I've looked through the rough wordlist and found some characters
	# that are not easy to type for everyone.
	# So I listed them here.
	'â',
	'å',
	'ã',
	'Ã',
	'ä',
	'ª',
	'±',
	'¨',
	'Ÿ',
	'–',
	'€',
	'“',
	'ˆ',
	'’',
	'”',
	'œ',
	'†',
	'¾',
	'²',
	'«',
	'¶',
	'¡',
	'¼',
	# This came in somehow. Not a word, so added it.
	's42815',
));

# You can override almost any function used in the dicelister.pl file.
# This is to aid in finetuning the output as you like it.
# This is an example to detect Bangla numbers. Feel free to comment it.
sub dicelister::is_numeric {
	my $val = shift;
	if ( defined $val ) {
		return $val =~ /^\d|১|২|৩|৪|৫|৬|৭|৮|৯|০+$/ ? 1 : 0;
	}
}