Fix overly strict validation for npg, ndl

MDNodaLink

src/classes/MDNodaLink.php

@@ -26,7 +26,7 @@ final class MDNodaLink {
 
         $this->source = $source;
         if (($validatedId = $this->source->validateId(trim($id))) === false) {
-            throw new MDInvalidNodaLink("Invalid noda ID / link: " . $id . " [Repository: " . $this->source->toDbName() . "]");
+            throw new MDInvalidNodaLink("Invalid noda ID / link: \"" . $id . "\" [Repository: " . $this->source->toDbName() . "]");
         }
         $this->id = $validatedId;
 

src/enums/MDNodaRepository.php

@@ -251,11 +251,16 @@ enum MDNodaRepository implements MDValueEnumInterface, JsonSerializable {
 
         if (filter_var($id, FILTER_VALIDATE_URL) !== false) {
             $toRemove = [];
-            foreach ($prefixes as $prefix) $toRemove[$prefix] = "";
+            foreach ($prefixes as $prefix) {
+                $toRemove[$prefix] = "";
+            }
             $id = strtr($id, $toRemove);
         }
 
-        if (filter_var($id, FILTER_VALIDATE_INT) === false) {
+        // Strings starting with 0 are quite often linked, notably with the NDL.
+        // PHP's FILTER_VALIDATE_INT does not accept a leading 0 however, so it
+        // is stripped before checking.
+        if (filter_var(ltrim($id, '0'), FILTER_VALIDATE_INT) === false) {
             return false;
         }
 
@@ -320,7 +325,10 @@ enum MDNodaRepository implements MDValueEnumInterface, JsonSerializable {
     private static function validateLocId(string $id):string|false {
 
         if (filter_var($id, FILTER_VALIDATE_URL) !== false) {
-            $id = strtr($id, ['http://id.loc.gov/authorities/names/' => '']);
+            $id = strtr($id, [
+                'http://id.loc.gov/authorities/names/' => '',
+                'https://id.loc.gov/authorities/names/' => '',
+            ]);
         }
 
         if (substr($id, 0, 1) !== 'n') {
@@ -399,7 +407,7 @@ enum MDNodaRepository implements MDValueEnumInterface, JsonSerializable {
             self::aat       => self::validateNumericId($id, ['https://vocab.getty.edu/page/aat/']),
             self::ackerbau  => self::validateNumericId($id, ['https://term.museum-digital.de/ackerbau/tag/']),
             self::bne       => self::validateNumericId($id, ['http://datos.bne.es/persona/']),
-            self::bnf       => self::validateNumericId($id, ["https://catalogue.bnf.fr/ark:/12148/cb"]),
+            self::bnf       => self::validateNumericId(rtrim($id, 't'), ["https://catalogue.bnf.fr/ark:/12148/cb"]), // cb11960399t is a valid entry, too (general)
             self::cona      => self::validateNumericId($id, ['https://vocab.getty.edu/page/cona/']),
             self::editionhumboldtdigital => self::validateGndId($id, ['https://edition-humboldt.de/register/personen/detail.xql?normid=http://d-nb.info/gnd/']),
             self::gnd       => self::validateGndId($id, ['http://d-nb.info/gnd/', 'https://d-nb.info/gnd/']),
@@ -411,11 +419,17 @@ enum MDNodaRepository implements MDValueEnumInterface, JsonSerializable {
             self::mindatorg => self::validateNumericId($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::ndl       => self::validateNumericId($id, ['https://id.ndl.go.jp/auth/ndlna/']),
+            self::ndl       => self::validateNumericId($id, [
+                'http://id.ndl.go.jp/auth/ndlna/',
+                'https://id.ndl.go.jp/auth/ndlna/',
+            ]),
             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, ['https://www.npg.org.uk/collections/search/person/']),
+            self::npg       => self::validateNumericId($id, [
+                'https://www.npg.org.uk/collections/search/person/',
+                'https://www.npg.org.uk/collections/search/person/mp',
+            ]),
             self::oberbegriffsdatei => self::validateNumericId($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/']),

tests/MDNodaRepositoryTest.php

@@ -0,0 +1,48 @@
+<?PHP
+/**
+ * This script contains tests for the home page.
+ *
+ * @author Joshua Ramon Enslin <joshua@museum-digital.de>
+ */
+declare(strict_types = 1);
+use PHPUnit\Framework\TestCase;
+
+require_once __DIR__ . '/../src/enums/MDValueEnumInterface.php';
+require_once __DIR__ . '/../src/enums/MDNodaRepository.php';
+
+/**
+ * Tests for home page.
+ */
+final class MDNodaRepositoryTest extends TestCase {
+    /**
+     * Tests that valid IDs actually validate as valid.
+     *
+     * @return void
+     */
+    public function testValidIdsValidate():void {
+
+        // GND (Germany)
+        self::assertNotFalse(MDNodaRepository::gnd->validateId("https://d-nb.info/gnd/102423008"));
+        self::assertNotFalse(MDNodaRepository::gnd->validateId("http://d-nb.info/gnd/102423008"));
+        self::assertNotFalse(MDNodaRepository::gnd->validateId("102423008"));
+
+        // NDL (Japan)
+        self::assertNotFalse(MDNodaRepository::ndl->validateId("00967046"));
+        self::assertNotFalse(MDNodaRepository::ndl->validateId("https://id.ndl.go.jp/auth/ndlna/00967046"));
+        self::assertNotFalse(MDNodaRepository::ndl->validateId("http://id.ndl.go.jp/auth/ndlna/00967046"));
+
+        // NPG: National Portrait Gallery
+        self::assertNotFalse(MDNodaRepository::npg->validateId("https://www.npg.org.uk/collections/search/person/mp01751"));
+
+        // Library of Congress
+        self::assertNotFalse(MDNodaRepository::loc->validateId("https://id.loc.gov/authorities/names/n2022014604"));
+        self::assertNotFalse(MDNodaRepository::loc->validateId("http://id.loc.gov/authorities/names/n2022014604"));
+        self::assertNotFalse(MDNodaRepository::loc->validateId("n2022014604"));
+
+
+        self::assertNotFalse(MDNodaRepository::loc->validateId("https://id.loc.gov/authorities/names/n2022014604"));
+        self::assertNotFalse(MDNodaRepository::loc->validateId("http://id.loc.gov/authorities/names/n2022014604"));
+        self::assertNotFalse(MDNodaRepository::loc->validateId("n2022014604"));
+
+    }
+}