convert.pl

来自「memcard is a multiplaform flashcard word」· PL 代码 · 共 53 行

use strict;

binmode STDOUT, ":utf8";

open FILE, "<:utf8", '/tmp/fr_ru_dic/french_audio.utf8';
#open FILE, "<:utf8", '/tmp/fr_popova_utf8.txt';
my @lines=<FILE>;
my %d = ();
foreach my $i (@lines) {
	chomp($i);
	$i =~ s/"/'/g;
	if ($i =~ m/^([^\t]+)\t+([^\t]+)\t+([^\t]+)$/) {
		my ($word, $def, $k) = ($1, $2, $3);
		if (defined($d{$k})) {
			my $listref = $d{$k};
			push @{$listref}, [$word, $def];
		} else {
			my $listref = [];
			push @{$listref}, [$word, $def];
			$d{$k} = $listref;
		}
	} else {
		if ($i =~ m/^([^\t]+)\t+([^\t]+)\t*$/) {
			my $k = 'All';
			my ($word, $def) = ($1, $2);
			if (defined($d{$k})) {
				my $listref = $d{$k};
				push @{$listref}, [$word, $def];
			} else {
				my $listref = [];
				push @{$listref}, [$word, $def];
				$d{$k} = $listref;
			}
		}
	}
}

#print keys %d;
#exit 0;

print '<?xml version="1.0" encoding="UTF-8" standalone="no"?>', "\n";
foreach my $k (keys %d) {
	printf "<Category name=\"%s\">\n<Deck>\n", $k;
	map {
		my @l = @$_;
		map { 
			printf "<Card AmountLearnedBack=\"0\" AmountLearnedFront=\"0\" Backside=\"%s\" DateCreated=\"27-May-2008 16:14:30\" DateExpired=\"28-May-2008 17:00:46\" DateModified=\"27-May-2008 16:14:30\" DateTested=\"27-May-2008 17:00:46\" DateTouched=\"27-May-2008 17:00:55\" Frontside=\"%s\" TestsHit=\"1\" TestsTotal=\"1\">\n<Side/>\n<Side/>\n</Card>\n", $_->[1], $_->[0];
		} @l;
	} $d{$k};
	printf "</Deck>\n</Category>\n";
}