Throw dedicated exceptions in MDNodaRepository

src/enums/MDNodaRepository.php

@@ -65,9 +65,11 @@ enum MDNodaRepository implements MDValueEnumInterface, JsonSerializable {
             'GND'       => self::gnd,
             'o-gnd'     => self::gnd,
             'O-GND'     => self::gnd,
+            'pnd'       => self::gnd,
             'http://d-nb.info/gnd'  => self::gnd,
             'http://d-nb.info/gnd/' => self::gnd,
             'd-nb.info'             => self::gnd,
+            'https://portal.dnb.de' => self::gnd,
             'grobsystematik' => self::grobsystematik,
             'iconclass' => self::iconclass,
             'klbb'      => self::klbb,
@@ -92,6 +94,7 @@ enum MDNodaRepository implements MDValueEnumInterface, JsonSerializable {
             'pleiades'  => self::pleiades,
             'rkd'       => self::rkd,
             'ulan'      => self::ulan,
+            'ULAN'      => self::ulan,
             'viaf'      => self::viaf,
             'wikidata'  => self::wikidata,
             'Wikidata'  => self::wikidata,
@@ -377,7 +380,10 @@ enum MDNodaRepository implements MDValueEnumInterface, JsonSerializable {
         else if (substr($id, 0, 1) === 'n') {
             if (filter_var(trim(substr($id, 1), '0'), FILTER_VALIDATE_INT) === false) return false;
         }
-        else throw new MDgenericInvalidInputsException("LOC IDs must start with n or nr or nb (provided: " . strip_tags($id) . ")");
+        else if (substr($id, 0, 2) === 'sh') {
+            throw new MDInvalidNodaLinkLocIdIsSh("The link to the Library of Congress authority files refers to the LOC Subject Headings. The subject headings are listed as a separate vocabulary.");
+        }
+        else throw new MDInvalidNodaLinkException("LOC IDs must start with n or nr or nb (provided: " . strip_tags($id) . ")");
 
         return (string)$id;
 
@@ -396,11 +402,14 @@ enum MDNodaRepository implements MDValueEnumInterface, JsonSerializable {
             $id = strtr($id, [
                 'http://id.loc.gov/authorities/names/' => '',
                 'https://id.loc.gov/authorities/names/' => '',
+                'http://id.loc.gov/authorities/subjects/' => '',
+                'https://id.loc.gov/authorities/subjects/' => '',
+                '.html' => '',
             ]);
         }
 
         if (substr($id, 0, 2) !== 'sh') {
-            throw new MDgenericInvalidInputsException("LCSH IDs must start with sh");
+            throw new MDInvalidNodaLinkException("LCSH IDs must start with sh");
         }
 
         if (filter_var(ltrim(substr($id, 2), "0"), FILTER_VALIDATE_INT) === false) {
@@ -443,6 +452,31 @@ enum MDNodaRepository implements MDValueEnumInterface, JsonSerializable {
 
     }
 
+    /**
+     * Validates an NPG ID, returning a string or false.
+     *
+     * @param string $id ID to validate.
+     *
+     * @return string|false
+     */
+    private static function validateNpgId(string $id):string|false {
+
+        if (filter_var($id, FILTER_VALIDATE_URL) !== false) {
+            $id = strtr($id, ['https://www.npg.org.uk/collections/search/person/' => '']);
+        }
+
+        if (substr($id, 0, 2) === 'mp' && is_numeric(substr($id, 2))) {
+            return $id;
+        }
+
+        if (filter_var($id, FILTER_VALIDATE_INT) === false) {
+            return false;
+        }
+
+        return $id;
+
+    }
+
     /**
      * Validates a wikidata ID, returning a string or false.
      *
@@ -498,10 +532,17 @@ enum MDNodaRepository implements MDValueEnumInterface, JsonSerializable {
     public static function validateWikipediaId(string $id):string|false {
 
         $validation = strtr($id, [
+            'http://de.wikipedia.org/wiki/' => '',
             'https://de.wikipedia.org/wiki/' => '',
+            'http://en.wikipedia.org/wiki/' => '',
             'https://en.wikipedia.org/wiki/' => '',
+            'http://fr.wikipedia.org/wiki/' => '',
             'https://fr.wikipedia.org/wiki/' => '',
+            'http://nl.wikipedia.org/wiki/' => '',
+            'https://nl.wikipedia.org/wiki/' => '',
+            'http://sv.wikipedia.org/wiki/' => '',
             'https://sv.wikipedia.org/wiki/' => '',
+            'http://ru.wikipedia.org/wiki/' => '',
             'https://ru.wikipedia.org/wiki/' => '',
         ]);
 
@@ -522,7 +563,12 @@ enum MDNodaRepository implements MDValueEnumInterface, JsonSerializable {
 
         return match($this) {
             self::allgemein => filter_var($id, FILTER_VALIDATE_URL),
-            self::aat       => self::validateNumericId($id, ['https://vocab.getty.edu/page/aat/']),
+            self::aat       => self::validateNumericId($id, [
+                'http://vocab.getty.edu/aat/',
+                'http://vocab.getty.edu/page/aat/',
+                'https://vocab.getty.edu/aat/',
+                'https://vocab.getty.edu/page/aat/',
+            ]),
             self::ackerbau  => self::validateNumericId($id, ['https://term.museum-digital.de/ackerbau/tag/']),
             self::bne       => self::validateBneId($id, ['http://datos.bne.es/persona/']),
             self::bnf       => self::validateBnfId($id),
@@ -537,7 +583,7 @@ enum MDNodaRepository implements MDValueEnumInterface, JsonSerializable {
             self::mbl       => self::validateNumericId($id, ['http://www.uni-magdeburg.de/mbl/PHP_Skripte/mbl_verwaltung/mbl_verw_anzeige_biog.php?auswahl=3&liste_biog_name=']),
             self::mindatorg => self::validateNumericId($id, ['https://www.mindat.org/min-', '.html']),
             self::moebeltypologie => self::validateNumericId($id, ['https://term.museum-digital.de/moebel/tag/']),
-            self::ndb_adb   => self::validateGndId($id, ['https://www.deutsche-biographie.de/pnd', '.html']),
+            self::ndb_adb   => self::validateGndId($id, ['https://www.deutsche-biographie.de/pnd', '.html', '#adbcontent', '#ndbcontent']),
             self::ndl       => self::validateNumericId($id, [
                 'http://id.ndl.go.jp/auth/ndlna/',
                 'https://id.ndl.go.jp/auth/ndlna/',
@@ -545,17 +591,20 @@ enum MDNodaRepository implements MDValueEnumInterface, JsonSerializable {
             self::ndp_ikmk  => self::validateNumericId($id, ['https://ikmk.smb.museum/ndp/land/']),
             self::ndp_ikmk_persons => self::validateNumericId($id, ['https://ikmk.smb.museum/ndp/person/']),
             self::nomisma   => str_replace('http://nomisma.org/id/', '', $id),
-            self::npg       => self::validateNumericId($id, [
+            self::npg       => self::validateNpgId($id),
-                'https://www.npg.org.uk/collections/search/person/',
-                'https://www.npg.org.uk/collections/search/person/mp',
-            ]),
             self::oberbegriffsdatei => self::validateNumericId($id, ['https://term.museum-digital.de/oberbegriffsdatei/tag/']),
             self::orcid     => preg_match('/^[0-9]{4}-[0-9]{4}-[0-9]{4}-[0-9]{4}$/', $id) ? $id : false,
             self::osm       => self::validateNumericId($id, ['https://www.openstreetmap.org/relation/']),
             self::pim       => self::validatePimId($id),
             self::pleiades  => self::validateNumericId($id, ['https://pleiades.stoa.org/places/']),
             self::rkd       => self::validateNumericId($id, ['http://rkd.nl/explore/artists/', 'https://rkd.nl/explore/artists/']),
-            self::ulan      => self::validateNumericId($id, ['http://vocab.getty.edu/ulan/', 'http://vocab.getty.edu/page/ulan/', 'https://vocab.getty.edu/page/ulan/']),
+            self::ulan      => self::validateNumericId($id, [
+                'http://vocab.getty.edu/ulan/',
+                'http://vocab.getty.edu/page/ulan/',
+                'https://vocab.getty.edu/page/ulan/',
+                'http://www.getty.edu/vow/ULANFullDisplay?find=&role=&nation=&subjectid=',
+                'https://www.getty.edu/vow/ULANFullDisplay?find=&role=&nation=&subjectid=',
+            ]),
             self::viaf      => self::validateNumericId($id, [
                 'https://viaf.org/viaf/',
                 'http://viaf.org/viaf/',

tests/MDNodaRepositoryTest.php

@@ -35,7 +35,7 @@ final class MDNodaRepositoryTest extends TestCase {
         self::assertEquals("00967046", MDNodaRepository::ndl->validateId("http://id.ndl.go.jp/auth/ndlna/00967046"));
 
         // NPG: National Portrait Gallery
-        self::assertEquals('01751', MDNodaRepository::npg->validateId("https://www.npg.org.uk/collections/search/person/mp01751"));
+        self::assertEquals('mp01751', MDNodaRepository::npg->validateId("https://www.npg.org.uk/collections/search/person/mp01751"));
 
         // Library of Congress
         self::assertEquals("n2022014604", MDNodaRepository::loc->validateId("https://id.loc.gov/authorities/names/n2022014604"));
@@ -45,6 +45,7 @@ final class MDNodaRepositoryTest extends TestCase {
         self::assertEquals("sh2022014604", MDNodaRepository::lcsh->validateId("https://id.loc.gov/authorities/names/sh2022014604"));
         self::assertEquals("sh2022014604", MDNodaRepository::lcsh->validateId("http://id.loc.gov/authorities/names/sh2022014604"));
         self::assertEquals("sh2022014604", MDNodaRepository::lcsh->validateId("sh2022014604"));
+        self::assertEquals("sh85081569", MDNodaRepository::lcsh->validateId("http://id.loc.gov/authorities/subjects/sh85081569.html"));
 
         self::assertEquals("PIM71684", MDNodaRepository::pim->validateId("PIM71684"));
         self::assertEquals("PIM71684", MDNodaRepository::pim->validateId("https://opac-nevter.pim.hu/en/record/-/record/PIM71684"));