commit 8c85e9f978f4cb780c3d8e1b45b7f2a8d076dd1b
author Akron <nils@diewald-online.de> Mon Jan 03 16:27:10 2022 +0100
committer Akron <nils@diewald-online.de> Mon Jan 17 12:34:17 2022 +0100
tree 1d49777887990bec8bf4edaab3c123d2d8deb554
parent 9f37ed753f4119febfa38999d4be332ccdefcdf2

Add more Gingko meta data fields

Change-Id: I3f3f7c53ae46aec52547982dad4c35acda89961d

lib/KorAP/XML/Krill.pm

diff --git a/lib/KorAP/XML/Krill.pm b/lib/KorAP/XML/Krill.pm
index 66f481d..cd53751 100644
--- a/lib/KorAP/XML/Krill.pm
+++ b/lib/KorAP/XML/Krill.pm
@@ -16,7 +16,7 @@
 
 our @EXPORT_OK = qw(get_file_name get_file_name_from_glob);
 
-our $VERSION = '0.43';
+our $VERSION = '0.44';
 
 has 'path';
 has [qw/text_sigle doc_sigle corpus_sigle/];

lib/KorAP/XML/Meta/Base.pm

diff --git a/lib/KorAP/XML/Meta/Base.pm b/lib/KorAP/XML/Meta/Base.pm
index 21e4297..b00d894 100644
--- a/lib/KorAP/XML/Meta/Base.pm
+++ b/lib/KorAP/XML/Meta/Base.pm
@@ -160,9 +160,9 @@
     elsif (index($_, 'T_') == 0) {
       push @fields, _text_field(_k($_), $self->{$_});
     }
-    # elsif (index($_, 'I_') == 0) {
-    #  _int_field(_k($_), $self->{$_});
-    # }
+    elsif (index($_, 'I_') == 0) {
+      push @fields, _int_field(_k($_), $self->{$_});
+    }
     elsif (index($_, 'A_') == 0) {
       push @fields, _attachement_field(_k($_), $self->{$_});
     }
@@ -260,4 +260,14 @@
   };
 };
 
+sub _int_field {
+  return {
+    '@type' => 'koral:field',
+    type    => 'type:integer',
+    key     => $_[0],
+    value   => $_[1]
+  };
+};
+
+
 1;

lib/KorAP/XML/Meta/Gingko.pm

diff --git a/lib/KorAP/XML/Meta/Gingko.pm b/lib/KorAP/XML/Meta/Gingko.pm
index 821b0d3..83ac9ac 100644
--- a/lib/KorAP/XML/Meta/Gingko.pm
+++ b/lib/KorAP/XML/Meta/Gingko.pm
@@ -36,10 +36,25 @@
       };
     };
 
+    if (my $analytic = $dom->at('sourceDesc > biblStruct > analytic')) {
+      if ($temp = $analytic->at('biblNote[n=DOI]')) {
+        $temp = $squish->($temp->all_text);
+        if ($temp) {
+          $self->{A_gingko_article_DOI} = $self->korap_data_uri('https://doi.org/' . $temp, title => 'doi:' . $temp);
+        };
+      };
+    };
+
     if ($temp = $dom->at('correction')) {
       $temp = $squish->($temp->all_text);
       $self->{S_gingko_lemma_corr} = $temp if $temp;
     };
+
+    if ($temp = $dom->at('encodingDesc > tagsDecl > tagUsage[gi=w]')) {
+      if ($temp->attr('occurs')) {
+        $self->{I_gingko_text_tokens} = $temp->attr('occurs');
+      };
+    };
   }
 
   elsif ($type eq 'corpus') {
@@ -53,6 +68,17 @@
         $temp = $squish->($temp->all_text);
         $self->{S_gingko_collection_short} = $temp if $temp;
       };
+
+      if ($temp = $mono->at('biblNote[n="url"]')) {
+        $temp = $squish->($temp->all_text);
+        $self->{A_external_link} = $self->korap_data_uri($temp, title => 'Gingko-Webseite an der Universität Leipzig');
+      };
+
+      if ($temp = $mono->at('biblNote[n="url.ids"]')) {
+        $temp = $squish->($temp->all_text);
+        $self->{A_internal_link} = $self->korap_data_uri($temp, title => 'IDS webpage on Gingko in the DeReKo archive');
+      };
+
     };
   };
 };