MBS-12665: Widen DNB regex

We took this from Wikidata, but given the valid example added as a test here, it was too strict. phw/outsidecontext got in touch with DNB and got a kinda not very useful answer, but his proposal based on it was to just allow digits plus X and hyphen. It seems that the length should be either 9 or 10 characters. Further assumptions seem a bit dangerous so I'm not making any.
metabrainz · Feb 9, 2023 · 271b14c · 271b14c
1 parent d08dcff
commit 271b14c
Show file tree

Hide file tree

Showing 2 changed files with 12 additions and 5 deletions.
diff --git a/root/static/scripts/edit/URLCleanup.js b/root/static/scripts/edit/URLCleanup.js
@@ -1963,8 +1963,8 @@ const CLEANUPS: CleanupEntries = {
     restrict: [LINK_TYPES.otherdatabases],
     clean: function (url) {
       url = url.replace(/^(?:https?:\/\/)?(?:[^\/]+\.)?d-nb\.info\//, 'http://d-nb.info/');
-      url = url.replace(/^(?:https?:\/\/)?(?:[^\/]+\.)?dnb\.de\/opac(?:\.htm\?)?.*\bquery=nid%3D(1[012]?\d{7}[0-9X]|[47]\d{6}-\d|[1-9]\d{0,7}-[0-9X]|3\d{7}[0-9X]).*$/, 'http://d-nb.info/gnd/$1');
-      url = url.replace(/^(?:https?:\/\/)?(?:[^\/]+\.)?dnb\.de\/opac(?:\.htm\?)?.*\bquery=idn%3D(1[012]?\d{7}[0-9X]|[47]\d{6}-\d|[1-9]\d{0,7}-[0-9X]|3\d{7}[0-9X]).*$/, 'http://d-nb.info/$1');
+      url = url.replace(/^(?:https?:\/\/)?(?:[^\/]+\.)?dnb\.de\/opac(?:\.htm\?)?.*\bquery=nid%3D([0-9X-]{9,10}).*$/, 'http://d-nb.info/gnd/$1');
+      url = url.replace(/^(?:https?:\/\/)?(?:[^\/]+\.)?dnb\.de\/opac(?:\.htm\?)?.*\bquery=idn%3D([0-9X-]{9,10}).*$/, 'http://d-nb.info/$1');
       return url;
     },
     validate: function (url, id) {
@@ -1974,17 +1974,17 @@ const CLEANUPS: CleanupEntries = {
         case LINK_TYPES.otherdatabases.series:
         case LINK_TYPES.otherdatabases.work:
           return {
-            result: /^http:\/\/d-nb\.info\/(?:gnd\/)?(?:1[012]?\d{7}[0-9X]|[47]\d{6}-\d|[1-9]\d{0,7}-[0-9X]|3\d{7}[0-9X])$/.test(url),
+            result: /^http:\/\/d-nb\.info\/(?:gnd\/)?[0-9X-]{9,10}$/.test(url),
             target: ERROR_TARGETS.ENTITY,
           };
         case LINK_TYPES.otherdatabases.label:
           return {
-            result: /^http:\/\/d-nb\.info\/(?:(?:dnbn|gnd)\/)?(?:1[012]?\d{7}[0-9X]|[47]\d{6}-\d|[1-9]\d{0,7}-[0-9X]|3\d{7}[0-9X])$/.test(url),
+            result: /^http:\/\/d-nb\.info\/(?:(?:dnbn|gnd)\/)?[0-9X-]{9,10}$/.test(url),
             target: ERROR_TARGETS.ENTITY,
           };
         case LINK_TYPES.otherdatabases.release:
           return {
-            result: /^http:\/\/d-nb\.info\/(?:1[012]?\d{7}[0-9X]|[47]\d{6}-\d|[1-9]\d{0,7}-[0-9X]|3\d{7}[0-9X])$/.test(url),
+            result: /^http:\/\/d-nb\.info\/[0-9X-]{9,10}$/.test(url),
             target: ERROR_TARGETS.ENTITY,
           };
       }

diff --git a/root/static/scripts/tests/Control/URLCleanup.js b/root/static/scripts/tests/Control/URLCleanup.js
@@ -1849,6 +1849,13 @@ limited_link_type_combinations: [
             expected_clean_url: 'http://d-nb.info/1181136512',
        only_valid_entity_types: ['artist', 'label', 'place', 'release', 'series', 'work'],
   },
+  {
+                     input_url: 'http://d-nb.info/97248485X',
+             input_entity_type: 'release',
+    expected_relationship_type: 'otherdatabases',
+            expected_clean_url: 'http://d-nb.info/97248485X',
+       only_valid_entity_types: ['artist', 'label', 'place', 'release', 'series', 'work'],
+  },
   // Dogmazic
   {
                      input_url: 'https://play.dogmazic.net/artists.php?action=show_all_songs&artist=2283',