diff options
| author | Přemysl Eric Janouch <p@janouch.name> | 2021-10-07 03:37:12 +0200 |
|---|---|---|
| committer | Přemysl Eric Janouch <p@janouch.name> | 2021-10-07 14:06:57 +0200 |
| commit | ed8b1bcdad7c430af1eef5fbe78b6ec4eb3eb60e (patch) | |
| tree | 4d864fe34d2102eb1ef4812d4613c2eca7006336 /dicts/gnu-fdl-en-cz.sh | |
| parent | 3881725904473cd9fdbd3e60cd1de2010f14d767 (diff) | |
| download | tdv-ed8b1bcdad7c430af1eef5fbe78b6ec4eb3eb60e.tar.gz tdv-ed8b1bcdad7c430af1eef5fbe78b6ec4eb3eb60e.tar.xz tdv-ed8b1bcdad7c430af1eef5fbe78b6ec4eb3eb60e.zip | |
Add sample dictionary downloaders/builders
Diffstat (limited to 'dicts/gnu-fdl-en-cz.sh')
| -rwxr-xr-x | dicts/gnu-fdl-en-cz.sh | 21 |
1 files changed, 21 insertions, 0 deletions
diff --git a/dicts/gnu-fdl-en-cz.sh b/dicts/gnu-fdl-en-cz.sh new file mode 100755 index 0000000..8feeea9 --- /dev/null +++ b/dicts/gnu-fdl-en-cz.sh @@ -0,0 +1,21 @@ +#!/bin/sh -e +# GNU/FDL English-Czech dictionary, see https://www.svobodneslovniky.cz/ +curl -Lo- https://www.svobodneslovniky.cz/data/en-cs.txt.gz | \ +zcat | grep -v ^# | sed 's/\\//g' | perl -CSD -F\\t -le ' + sub e { shift =~ s/\\/\\\\/gr =~ s/\n/\\n/gr =~ s/\t/\\t/gr } + sub w { + open(my $f, "|-", "tabfile gnu-fdl-$_[0]") or die $!; + print $f e($k) . "\t" . e(join("\n", @$v)) + while ($k, $v) = each %{$_[1]}; + close($f); + } + my ($en, $cz, $notes, $special, $translator) = @F; + if ($cz) { + $notes =~ s/\w+:\s?//g; # remove word classes + $notes =~ s/(\w+\.)(?!])/($1)/; # quote "pl." + push(@{$encz{$en}}, $notes ? "$cz " . $notes : $cz); + push(@{$czen{$cz}}, $notes ? "$en " . $notes : $en); + } END { + w("en-cz", \%encz); + w("cz-en", \%czen); + }' |