From f812fae922eec06235c9e566b78c7f0fb46a709b Mon Sep 17 00:00:00 2001
From: Přemysl Eric Janouch <p@janouch.name>
Date: Wed, 13 Oct 2021 23:57:36 +0200
Subject: Add the GNU/FDL German-Czech dictionary to dicts

But only build it with WANT_BAD_DICTS set to non-null.
---
 dicts/gnu-fdl-de-cz.sh | 34 ++++++++++++++++++++++++++++++++++
 1 file changed, 34 insertions(+)
 create mode 100755 dicts/gnu-fdl-de-cz.sh

(limited to 'dicts')

diff --git a/dicts/gnu-fdl-de-cz.sh b/dicts/gnu-fdl-de-cz.sh
new file mode 100755
index 0000000..ffe56a4
--- /dev/null
+++ b/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");
+	}'
-- 
cgit v1.2.3-70-g09d2