Introduce support for CorpusExplorer

Change-Id: I2133463dbf5e851b371e46f4b9c76bba71611532

Changes

@@ -1,3 +1,6 @@
+0.56 2024-06-05
+        - Add support für corpusexplorer.
+
 0.55 2024-06-04
         - Add support for xenodata to i5.
 

Readme.pod

@@ -1,3 +1,5 @@
+__END__
+
 =pod
 
 =encoding utf8
@@ -393,6 +395,9 @@ L<Krill|https://github.com/KorAP/Krill>.
     #NamedEntities
     #Sentences
 
+  CorpusExplorer
+    #Morpho
+
   CMC
     #Morpho
 

lib/KorAP/XML/Annotation/CorpusExplorer/Morpho.pm

@@ -0,0 +1,60 @@
+package KorAP::XML::Annotation::CorpusExplorer::Morpho;
+use KorAP::XML::Annotation::Base;
+use Mojo::Util qw'trim';
+
+sub parse {
+  my $self = shift;
+
+  $$self->add_tokendata(
+    foundry => 'corpusexplorer',
+    layer => 'morpho',
+    cb => sub {
+      my ($stream, $token) = @_;
+      my $mtt = $stream->pos($token->get_pos);
+
+      my $content = $token->get_hash->{fs}->{f} or return;
+
+      $content = ref $content ne 'ARRAY' ? [$content] : $content;
+
+      my $start = $token->get_hash->{-from};
+
+      # Iterate over feature structures
+      foreach my $fs (@$content) {
+        $content = $fs->{fs}->{f} or next;
+
+        foreach (@$content) {
+
+          next unless $_->{'#text'};
+          my $value = trim $_->{'#text'} or next;
+
+          # POS
+          if ($_->{-name} eq 'ctag') {
+            $mtt->add_by_term('cex/p:' . $value);
+          }
+
+          # Lemma
+          elsif ($_->{-name} eq 'lemma') {
+            $mtt->add_by_term('cex/l:' . $value);
+          }
+
+          # Phrase
+          elsif ($_->{-name} eq 'phrase') {
+            $mtt->add_by_term('cex/phrase:' . $value);
+          };
+
+          if ($start == 809) {
+            warn $mtt->to_string;
+            warn $value;
+          };
+        };
+      };
+    }) or return;
+
+  return 1;
+};
+
+sub layer_info {
+  ['cex/p=tokens','cex/l=tokens','cex/phrase=tokens'];
+};
+
+1;

lib/KorAP/XML/Krill.pm

@@ -16,7 +16,7 @@ use Exporter 'import';
 
 our @EXPORT_OK = qw(get_file_name get_file_name_from_glob);
 
-our $VERSION = '0.55';
+our $VERSION = '0.56';
 
 has 'path';
 has [qw/text_sigle doc_sigle corpus_sigle/];

lib/KorAP/XML/Meta/I5.pm

@@ -291,6 +291,10 @@ sub parse {
           $key = 'A_';
           my $title = $_->att('desc');
           $value = $self->korap_data_uri($value, title => ($title // $value));
+        } elsif ($xtype eq 'number') {
+          $self->log->warn('Number currently not supported as xenodata type, treated as string');
+          $key = 'S_';
+          # Maybe render as Integer - but it's understood as a string
         } else {
           $self->log->warn('Unknown xenodata type: ' . $xtype);
           return;

script/korapxml2krill

@@ -178,7 +178,7 @@ use Fcntl qw(:flock SEEK_END);
 # - Improve core count logging.
 # ----------------------------------------------------------
 
-our $LAST_CHANGE = '2024/06/04';
+our $LAST_CHANGE = '2024/06/05';
 our $LOCAL = $FindBin::Bin;
 our $KORAL_VERSION = 0.03;
 our $VERSION_MSG = <<"VERSION";
@@ -460,6 +460,10 @@ push(@layers,
      ['CoreNLP', 'Morpho'],
      ['CoreNLP', 'Constituency']);
 
+# CorpusExplorer
+push(@layers,
+     ['CorpusExplorer', 'Morpho']);
+
 # CMC
 push(@layers, ['CMC', 'Morpho']);
 
@@ -1501,6 +1505,9 @@ L<Krill|https://github.com/KorAP/Krill>.
     #NamedEntities
     #Sentences
 
+  CorpusExplorer
+    #Morpho
+
   CMC
     #Morpho