Browse Source

Add the GNU/FDL German-Czech dictionary to dicts

But only build it with WANT_BAD_DICTS set to non-null.
master
parent
commit
f812fae922
Signed by: p GPG Key ID: A0420B94F92B9493
  1. 34
      dicts/gnu-fdl-de-cz.sh

34
dicts/gnu-fdl-de-cz.sh

@ -0,0 +1,34 @@
#!/bin/sh -e
# GNU/FDL German-Czech dictionary, see https://gnu.nemeckoceskyslovnik.cz
# Sometimes the domain doesn't resolve, and the contents are close to useless
[ -n "$WANT_BAD_DICTS" ] || exit
curl -Lo- 'https://gnu.nemeckoceskyslovnik.cz/index.php?id=6&sablona=export&format=zcu' | \
grep -v ^# | sed 's/\\//g' | perl -CSD -F\\t -le '
sub tabesc { shift =~ s/\\/\\\\/gr =~ s/\n/\\n/gr =~ s/\t/\\t/gr }
sub w {
my ($name, $dict, $collation) = @_;
open(my $f, "|-", "tabfile", "--pango", "--collation=$collation",
"--website=https://gnu.nemeckoceskyslovnik.cz",
"gnu-fdl-$name") or die $!;
print $f tabesc($keyword) . "\t" . tabesc(join("\n", @$defs))
while ($keyword, $defs) = each %{$dict};
close($f);
}
sub xmlesc { shift =~ s/&/&amp;/gr =~ s/</&lt;/gr =~ s/>/&gt;/gr }
sub entry {
my ($definition, $notes) = map {xmlesc($_)} @_;
$notes ? "$definition <i>$notes</i>" : $definition;
}
next if !$_ .. 0;
my ($de, $cs, $notes, $special, $translator) = @F;
if ($cs) {
$notes =~ s/\w+:\s?//g; # remove word classes
$notes =~ s/(\w+\.)(?!])/($1)/; # quote "pl."
push(@{$decs{$de}}, entry($cs, $notes));
push(@{$csde{$cs}}, entry($de, $notes));
} END {
w("de-cz", \%decs, "de");
w("cz-de", \%csde, "cs");
}'
Loading…
Cancel
Save