Add wdd cmc test case
Change-Id: I9641c128f3c84d36dcf3d23822c1b14b2e406e76
diff --git a/t/real/wdd.t b/t/real/wdd.t
index 886b8f4..5116dda 100644
--- a/t/real/wdd.t
+++ b/t/real/wdd.t
@@ -331,5 +331,44 @@
is($output->{data}->{stream}->[0]->[4], 'opennlp/p:ADV$<b>129<b>242', 'data');
+
+
+$path = catdir(dirname(__FILE__), 'corpus','WDD24','O0131','01268');
+
+ok($doc = KorAP::XML::Krill->new( path => $path . '/' ), 'Load Korap::Document');
+ok($doc->parse, 'Parse document');
+
+is($doc->text_sigle, 'WDD24/O0131/01268', 'Correct text sigle');
+is($doc->doc_sigle, 'WDD24/O0131', 'Correct document sigle');
+is($doc->corpus_sigle, 'WDD24', 'Correct corpus sigle');
+
+# Tokenization
+use_ok('KorAP::XML::Tokenizer');
+
+($token_base_foundry, $token_base_layer) = (qw/Base Tokens/);
+
+# Get tokenization
+$tokens = KorAP::XML::Tokenizer->new(
+ path => $doc->path,
+ doc => $doc,
+ foundry => $token_base_foundry,
+ layer => $token_base_layer,
+ name => 'tokens'
+);
+ok($tokens, 'Token Object is fine');
+ok($tokens->parse, 'Token parsing is fine');
+
+$tokens->add('CMC', 'Morpho');
+
+$output = decode_json( $tokens->to_json );
+is($output->{data}->{foundries},
+ 'cmc cmc/morpho',
+ 'Foundries');
+is($output->{data}->{layerInfos}, 'cmc/l=tokens cmc/p=tokens', 'layerInfos');
+
+is($output->{data}->{stream}->[94]->[1], 'cmc/p:URL', 'data');
+is($output->{data}->{stream}->[94]->[2], 'i:https://scrcguides.libraries.wm.edu/repositories/2/digital_objects/2606', 'data');
+
+
done_testing;
__END__