Use generators for tests for MDNodaRepository validation

src/enums/MDNodaRepository.php

@@ -442,6 +442,9 @@ enum MDNodaRepository implements MDValueEnumInterface, JsonSerializable {
         if (str_contains($id, '(')) {
             $id = trim(explode('(', $id)[0]);
         }
+        if (str_contains($id, PHP_EOL)) {
+            $id = trim(explode(PHP_EOL, $id)[0]);
+        }
 
         // There is an issue with this regex
         if (preg_match("/^[0-9-X]*$/", $id) === false) {
@@ -630,7 +633,10 @@ enum MDNodaRepository implements MDValueEnumInterface, JsonSerializable {
     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/' => '']);
+            $id = strtr($id, [
+                'http://www.npg.org.uk/collections/search/person/' => '',
+                'https://www.npg.org.uk/collections/search/person/' => '',
+            ]);
         }
 
         if (substr($id, 0, 2) === 'mp' && self::_is_numeric(substr($id, 2))) {
@@ -658,18 +664,21 @@ enum MDNodaRepository implements MDValueEnumInterface, JsonSerializable {
             $id = strtr($id, [
                 'http://www.wikidata.org/wiki/' => '',
                 'http://www.wikidata.org/entity/' => '',
+                'http://www.wikidata.org/w/index.php?search=&search=' => '',
                 'https://www.wikidata.org/wiki/' => '',
                 'https://www.wikidata.org/entity/' => '',
                 'https://www.wikidata.org/w/index.php?search=&search=' => '',
             ]);
 
-            if (str_starts_with($id, 'https://www.wikidata.org/w/index.php?title=')) {
+            foreach (['http://www.wikidata.org/w/index.php?title=', 'https://www.wikidata.org/w/index.php?title='] as $urlPrefix) {
-                $id = str_replace('https://www.wikidata.org/w/index.php?title=', '', $id);
+                if (str_starts_with($id, $urlPrefix)) {
+                    $id = str_replace($urlPrefix, '', $id);
                     if (($endPos = strpos($id, '&')) !== false) {
                         $id = substr($id, 0, $endPos);
                     }
                 }
             }
+        }
 
         if (substr($id, 0, 1) !== 'Q') {
             throw new MDInvalidNodaLinkException("Wikidata IDs must be Q IDs - and start with that letter (provided: $id)");

tests/MDNodaRepositoryTest.php

@@ -6,6 +6,9 @@
  */
 declare(strict_types = 1);
 use PHPUnit\Framework\TestCase;
+use PHPUnit\Framework\Attributes\Small;
+use PHPUnit\Framework\Attributes\CoversClass;
+use PHPUnit\Framework\Attributes\DataProvider;
 
 require_once __DIR__ . '/../src/enums/MDValueEnumInterface.php';
 require_once __DIR__ . '/../src/enums/MDNodaRepository.php';
@@ -13,85 +16,353 @@ require_once __DIR__ . '/../src/enums/MDNodaRepository.php';
 /**
  * Tests for home page.
  */
+#[small]
+#[CoversClass(\MDNodaRepository::class)]
 final class MDNodaRepositoryTest extends TestCase {
     /**
-     * Tests that valid IDs actually validate as valid.
+     * Data provider for validating Wikidata IDs.
+     *
+     * @return Generator<string, array{0: string, 1: string}>
+     */
+    public static function wikidataIdForValidationProvider():Generator {
+
+        yield "Full URL, http / Q834961" => ["Q834961", "http://www.wikidata.org/w/index.php?title=Q834961&oldid=2256125706"];
+        yield "Full URL, https / Q834961" => ["Q834961", "https://www.wikidata.org/w/index.php?title=Q834961&oldid=2256125706"];
+        yield "Q-ID / Q834961" => ["Q834961", "Q834961"];
+        yield "Q-ID / Q834961 (trailing space)" => ["Q834961", "Q834961 "];
+
+    }
+
+    /**
+     * Ensure validating references to Wikidata works.
+     *
+     * @param string $expected    Expected output.
+     * @param string $inputString Input string.
      *
      * @return void
      */
-    public function testValidIdsValidate():void {
+    #[DataProvider('wikidataIdForValidationProvider')]
+    public function testWikidataIdValidation(string $expected, string $inputString):void {
 
-        self::assertEquals("Q834961", MDNodaRepository::wikidata->validateId("https://www.wikidata.org/w/index.php?title=Q834961&oldid=2256125706"));
+        self::assertEquals($expected, MDNodaRepository::wikidata->validateId($inputString, ""));
 
-        self::assertEquals("XX5034943", MDNodaRepository::bne->validateId("http://datos.bne.es/persona/XX5034943"));
+    }
-        self::assertEquals("XX5034943", MDNodaRepository::bne->validateId("https://datos.bne.es/persona/XX5034943"));
-        self::assertEquals("XX5034943", MDNodaRepository::bne->validateId("XX5034943"));
-        self::assertEquals(false, MDNodaRepository::bne->validateId("XX503494safdsaf;3"));
 
-        // GND (Germany)
+    /**
-        self::assertEquals("102423008", MDNodaRepository::gnd->validateId("https://d-nb.info/gnd/102423008"));
+     * Data provider for validating BNE IDs.
-        self::assertEquals("102423008", MDNodaRepository::gnd->validateId("http://d-nb.info/gnd/102423008"));
+     *
-        self::assertEquals("102423008", MDNodaRepository::gnd->validateId("http://d-nb.info/gnd/        102423008"));
+     * @return Generator<string, array{0: string|false, 1: string}>
-        self::assertEquals("1037602218", MDNodaRepository::gnd->validateId("http://d-nb.info/gnd/1037602218"));
+     */
-        self::assertEquals("1037602218", MDNodaRepository::gnd->validateId("https://explore.gnd.network/gnd/1037602218"));
+    public static function bneIdForValidationProvider():Generator {
-        self::assertEquals("102423008", MDNodaRepository::gnd->validateId("102423008"));
-        self::assertEquals("102423008", MDNodaRepository::gnd->validateId("102423008,"));
-        self::assertEquals("4114170-2", MDNodaRepository::gnd->validateId("4114170-2"));
-        self::assertEquals("4114170-2", MDNodaRepository::gnd->validateId("4114170-2 (Fenster (Motiv)"));
-        self::assertEquals(false, MDNodaRepository::gnd->validateId("fkld;s102423008"));
 
-        // NDL (Japan)
+        yield "Full URL, http / XX5034943" => ["XX5034943", "http://datos.bne.es/persona/XX5034943"];
-        self::assertEquals("00967046", MDNodaRepository::ndl->validateId("https://id.ndl.go.jp/auth/ndlna/00967046"));
+        yield "Full URL, https / XX5034943" => ["XX5034943", "https://datos.bne.es/persona/XX5034943"];
-        self::assertEquals("00967046", MDNodaRepository::ndl->validateId("http://id.ndl.go.jp/auth/ndlna/00967046"));
+        yield "ID / XX5034943" => ["XX5034943", "XX5034943"];
-        self::assertEquals("00967046", MDNodaRepository::ndl->validateId("00967046"));
+        yield "Broken input / XX503494safdsaf;3" => [false, "XX503494safdsaf;3"];
-        self::assertEquals(false, MDNodaRepository::ndl->validateId("http://id.ndl.go.jp/auth/ndlna/0096704;43s6"));
 
-        // NPG: National Portrait Gallery
+    }
-        self::assertEquals('mp01751', MDNodaRepository::npg->validateId("https://www.npg.org.uk/collections/search/person/mp01751"));
-        self::assertEquals('mp01751', MDNodaRepository::npg->validateId("mp01751"));
-        self::assertEquals(false, MDNodaRepository::npg->validateId("https://www.npg.org.uk/collections/search/person/mp017;51"));
 
-        // Library of Congress
+    /**
-        self::assertEquals("n2022014604", MDNodaRepository::loc->validateId("https://id.loc.gov/authorities/names/n2022014604"));
+     * Ensure validating references to the BNE works.
-        self::assertEquals("n2022014604", MDNodaRepository::loc->validateId("http://id.loc.gov/authorities/names/n2022014604"));
+     *
-        self::assertEquals("n2022014604", MDNodaRepository::loc->validateId("n2022014604"));
+     * @param string|false $expected    Expected output.
-        self::assertEquals(false, MDNodaRepository::loc->validateId("n20220146;04"));
+     * @param string       $inputString Input string.
+     *
+     * @return void
+     */
+    #[DataProvider('bneIdForValidationProvider')]
+    public function testBneIdValidation(string|false $expected, string $inputString):void {
 
-        self::assertEquals("sh2022014604", MDNodaRepository::lcsh->validateId("https://id.loc.gov/authorities/names/sh2022014604"));
+        self::assertEquals($expected, MDNodaRepository::bne->validateId($inputString));
-        self::assertEquals("sh2022014604", MDNodaRepository::lcsh->validateId("http://id.loc.gov/authorities/names/sh2022014604"));
-        self::assertEquals("sh85081569", MDNodaRepository::lcsh->validateId("http://id.loc.gov/authorities/subjects/sh85081569.html"));
-        self::assertEquals("sh2022014604", MDNodaRepository::lcsh->validateId("sh2022014604"));
-        self::assertEquals(false, MDNodaRepository::lcsh->validateId("sh20220146;;04"));
 
-        // PRIM
+    }
-        self::assertEquals("PIM71684", MDNodaRepository::pim->validateId("https://opac-nevter.pim.hu/en/record/-/record/PIM71684"));
-        self::assertEquals("PIM71684", MDNodaRepository::pim->validateId("https://resolver.pim.hu/auth/PIM71684"));
-        self::assertEquals("PIM71684", MDNodaRepository::pim->validateId("PIM71684"));
-        self::assertEquals(false, MDNodaRepository::pim->validateId("PIM7168;;4"));
 
-        self::assertEquals("248941990", MDNodaRepository::viaf->validateId("https://viaf.org/viaf/248941990"));
+    /**
-        self::assertEquals("248941990", MDNodaRepository::viaf->validateId("http://viaf.org/viaf/248941990"));
+     * Data provider for validating GND IDs.
-        self::assertEquals("42893419", MDNodaRepository::viaf->validateId("https://viaf.org/viaf/42893419/"));
+     *
-        self::assertEquals("86145857811423020454", MDNodaRepository::viaf->validateId("86145857811423020454"));
+     * @return Generator<string, array{0: string|false, 1: string}>
-        self::assertEquals("2869150688328112660005", MDNodaRepository::viaf->validateId("2869150688328112660005"));
+     */
-        self::assertEquals(false, MDNodaRepository::viaf->validateId("2869150688328112;;660005"));
+    public static function gndIdForValidationProvider():Generator {
 
-        # NDP-IKMK (Places)
+        yield "Full URL, http / 102423008" => ["102423008", "http://d-nb.info/gnd/102423008"];
-        self::assertEquals("ort/2847", MDNodaRepository::ndp_ikmk->validateId("https://ikmk.smb.museum/ndp/ort/2847"));
+        yield "Full URL, https / 102423008" => ["102423008", "https://d-nb.info/gnd/102423008"];
-        self::assertEquals("land/123", MDNodaRepository::ndp_ikmk->validateId("https://ikmk.smb.museum/ndp/land/123"));
+        yield "Full URL, http, superfluous spaces / 102423008" => ["102423008", "http://d-nb.info/gnd/        102423008"];
-        self::assertEquals("region/123", MDNodaRepository::ndp_ikmk->validateId("https://ikmk.smb.museum/ndp/region/123"));
-        self::assertEquals("land/123", MDNodaRepository::ndp_ikmk->validateId("land/123"));
-        self::assertEquals(false, MDNodaRepository::ndp_ikmk->validateId("123123"));
 
-        # Catalogue of Life
+        yield "Full URL, http / 1037602218" => ["1037602218", "http://d-nb.info/gnd/1037602218"];
-        self::assertEquals(false, MDNodaRepository::col->validateId("123123"));
+        yield "Full URL, GND Explorer / 1037602218" => ["1037602218", "https://explore.gnd.network/gnd/1037602218"];
-        self::assertEquals("C46V", MDNodaRepository::col->validateId("https://www.catalogueoflife.org/data/taxon/C46V"));
+        yield "ID / 1037602218" => ["102423008", "102423008"];
-        self::assertEquals("CRLT8", MDNodaRepository::col->validateId("https://www.catalogueoflife.org/data/taxon/CRLT8"));
+        yield "ID, trailing comma / 102423008" => ["102423008", "102423008,"];
-        self::assertEquals("CRLT8", MDNodaRepository::col->validateId("http://www.catalogueoflife.org/data/taxon/CRLT8"));
+        yield "4114170-2" => ["4114170-2", "4114170-2"];
+        yield "4114170-2, trailing content in brackets" => ["4114170-2", "4114170-2 (Fenster (Motiv)"];
+        yield "4114170-2, trailing content in brackets, with newline" => ["4114170-2", "4114170-2" . PHP_EOL . " (Fenster (Motiv)"];
+        yield "4113617-2, with newline" => ["4113617-2", "4113617-2" . PHP_EOL . "Zauberin (Motiv)"];
+        yield "Broken ID" => [false, "fkld;s102423008"];
 
-        self::assertEquals(false, MDNodaRepository::mindatorg->validateId("adfdasjfklasjl"));
+    }
-        self::assertEquals("2047", MDNodaRepository::mindatorg->validateId("https://www.mindat.org/min-2047.html"));
+
-        self::assertEquals("2047", MDNodaRepository::mindatorg->validateId("http://www.mindat.org/min-2047.html"));
+    /**
+     * Ensure validating references to the GND / Gemeinsame Normdatei works.
+     *
+     * @param string|false $expected    Expected output.
+     * @param string       $inputString Input string.
+     *
+     * @return void
+     */
+    #[DataProvider('gndIdForValidationProvider')]
+    public function testGndIdValidation(string|false $expected, string $inputString):void {
+
+        self::assertEquals($expected, MDNodaRepository::gnd->validateId($inputString));
+
+    }
+
+    /**
+     * Data provider for validating NDL IDs.
+     *
+     * @return Generator<string, array{0: string|false, 1: string}>
+     */
+    public static function ndlIdForValidationProvider():Generator {
+
+        yield "Full URL, http / 00967046" => ["00967046", "http://id.ndl.go.jp/auth/ndlna/00967046"];
+        yield "Full URL, https / 00967046" => ["00967046", "https://id.ndl.go.jp/auth/ndlna/00967046"];
+        yield "ID / 00967046" => ["00967046", "00967046"];
+        yield "Broken input / XX503494safdsaf;3" => [false, "http://id.ndl.go.jp/auth/ndlna/0096704;43s6"];
+
+    }
+
+    /**
+     * Data provider for validating NPG IDs (National Portrait Gallery).
+     *
+     * @return Generator<string, array{0: string|false, 1: string}>
+     */
+    public static function npgIdForValidationProvider():Generator {
+
+        yield "Full URL, http / mp01751" => ["mp01751", "http://www.npg.org.uk/collections/search/person/mp01751"];
+        yield "Full URL, https / mp01751" => ["mp01751", "https://www.npg.org.uk/collections/search/person/mp01751"];
+        yield "ID / mp01751" => ["mp01751", "mp01751"];
+        yield "Broken input / https://www.npg.org.uk/collections/search/person/mp017;51" => [false, "https://www.npg.org.uk/collections/search/person/mp017;51"];
+
+    }
+
+    /**
+     * Ensure validating NPG entries works (National Portrait Gallery).
+     *
+     * @param string|false $expected    Expected output.
+     * @param string       $inputString Input string.
+     *
+     * @return void
+     */
+    #[DataProvider('npgIdForValidationProvider')]
+    public function testNpgIdValidation(string|false $expected, string $inputString):void {
+
+        self::assertEquals($expected, MDNodaRepository::npg->validateId($inputString));
+
+    }
+
+    /**
+     * Data provider for validating references to the Library of Congress works.
+     *
+     * @return Generator<string, array{0: string|false, 1: string}>
+     */
+    public static function locIdForValidationProvider():Generator {
+
+        yield "Full URL, http / n2022014604" => ["n2022014604", "http://id.loc.gov/authorities/names/n2022014604"];
+        yield "Full URL, https / n2022014604" => ["n2022014604", "https://id.loc.gov/authorities/names/n2022014604"];
+        yield "ID / n2022014604" => ["n2022014604", "n2022014604"];
+        yield "Broken input / n20220146;04" => [false, "n20220146;04"];
+
+    }
+
+    /**
+     * Ensure validating LOC entries works (Library of Congress).
+     *
+     * @param string|false $expected    Expected output.
+     * @param string       $inputString Input string.
+     *
+     * @return void
+     */
+    #[DataProvider('locIdForValidationProvider')]
+    public function testLocIdValidation(string|false $expected, string $inputString):void {
+
+        self::assertEquals($expected, MDNodaRepository::loc->validateId($inputString));
+
+    }
+
+    /**
+     * Data provider for validating references to the Library of Congress Subject Headings works.
+     *
+     * @return Generator<string, array{0: string|false, 1: string}>
+     */
+    public static function lcshIdForValidationProvider():Generator {
+
+        yield "Full URL, http / names/sh2022014604" => ["sh2022014604", "http://id.loc.gov/authorities/names/sh2022014604"];
+        yield "Full URL, https / names/sh2022014604" => ["sh2022014604", "https://id.loc.gov/authorities/names/sh2022014604"];
+        yield "Full URL, http / subjects/sh85081569.html" => ["sh85081569", "http://id.loc.gov/authorities/subjects/sh85081569.html"];
+        yield "ID / sh85081569" => ["sh85081569", "sh85081569"];
+        yield "Broken input / sh20220146;;04" => [false, "sh20220146;;04"];
+
+    }
+
+    /**
+     * Ensure validating LCSH entries works (Library of Congress Subject Headings).
+     *
+     * @param string|false $expected    Expected output.
+     * @param string       $inputString Input string.
+     *
+     * @return void
+     */
+    #[DataProvider('lcshIdForValidationProvider')]
+    public function testLcshIdValidation(string|false $expected, string $inputString):void {
+
+        self::assertEquals($expected, MDNodaRepository::lcsh->validateId($inputString));
+
+    }
+
+    /**
+     * Data provider for validating references to the PIM / Petőfi Irodalmi Múzeum.
+     *
+     * @return Generator<string, array{0: string|false, 1: string}>
+     */
+    public static function pimIdForValidationProvider():Generator {
+
+        yield "Full URL, https / opac > en/record/-/record/PIM71684" => ["PIM71684", "https://opac-nevter.pim.hu/en/record/-/record/PIM71684"];
+        yield "Full URL, https / record > auth/PIM71684" => ["PIM71684", "https://resolver.pim.hu/auth/PIM71684"];
+        yield "ID / PIM71684" => ["PIM71684", "PIM71684"];
+        yield "Broken input / PIM7168;;4" => [false, "PIM7168;;4"];
+
+    }
+
+    /**
+     * Ensure validating PIM / Petőfi Irodalmi Múzeum entries works.
+     *
+     * @param string|false $expected    Expected output.
+     * @param string       $inputString Input string.
+     *
+     * @return void
+     */
+    #[DataProvider('pimIdForValidationProvider')]
+    public function testPimIdValidation(string|false $expected, string $inputString):void {
+
+        self::assertEquals($expected, MDNodaRepository::pim->validateId($inputString));
+
+    }
+
+    /**
+     * Data provider for validating references to the VIAF.
+     *
+     * @return Generator<string, array{0: string|false, 1: string}>
+     */
+    public static function viafIdForValidationProvider():Generator {
+
+        yield "Full URL, http / 248941990" => ["248941990", "http://viaf.org/viaf/248941990"];
+        yield "Full URL, https / 248941990" => ["248941990", "https://viaf.org/viaf/248941990"];
+        yield "Full URL, https / 42893419" => ["42893419", "https://viaf.org/viaf/42893419"];
+        yield "Full URL, https / 86145857811423020454" => ["86145857811423020454", "https://viaf.org/viaf/86145857811423020454"];
+        yield "ID / 86145857811423020454" => ["86145857811423020454", "86145857811423020454"];
+        yield "Full URL, https / 2869150688328112660005" => ["2869150688328112660005", "https://viaf.org/viaf/2869150688328112660005"];
+        yield "ID / 2869150688328112660005" => ["2869150688328112660005", "2869150688328112660005"];
+        yield "Broken input / 2869150688328112;;660005" => [false, "2869150688328112;;660005"];
+
+    }
+
+    /**
+     * Ensure validating VIAF works.
+     *
+     * @param string|false $expected    Expected output.
+     * @param string       $inputString Input string.
+     *
+     * @return void
+     */
+    #[DataProvider('viafIdForValidationProvider')]
+    public function testViafIdValidation(string|false $expected, string $inputString):void {
+
+        self::assertEquals($expected, MDNodaRepository::viaf->validateId($inputString));
+
+    }
+
+    /**
+     * Data provider for validating references to the NDP-IKMK.
+     *
+     * @return Generator<string, array{0: string|false, 1: string}>
+     */
+    public static function ndpIkmkIdForValidationProvider():Generator {
+
+        yield "ort/2847 ; url" => ["ort/2847", "https://ikmk.smb.museum/ndp/ort/2847"];
+        yield "land/123 ; url" => ["land/123", "https://ikmk.smb.museum/ndp/land/123"];
+        yield "region/123 ; url" => ["region/123", "https://ikmk.smb.museum/ndp/region/123"];
+        yield "land/123 ; ID" => ["land/123", "land/123"];
+        yield "Broken input / 123123" => [false, "123123"];
+
+    }
+
+    /**
+     * Ensure validating NDP/IKMK works.
+     *
+     * @param string|false $expected    Expected output.
+     * @param string       $inputString Input string.
+     *
+     * @return void
+     */
+    #[DataProvider('ndpIkmkIdForValidationProvider')]
+    public function testNdpIkmkIdValidation(string|false $expected, string $inputString):void {
+
+        self::assertEquals($expected, MDNodaRepository::ndp_ikmk->validateId($inputString));
+
+    }
+
+    /**
+     * Data provider for validating references to the Catalogue of Life.
+     *
+     * @return Generator<string, array{0: string|false, 1: string}>
+     */
+    public static function colIdForValidationProvider():Generator {
+
+        yield "C46V ; url" => ["C46V", "https://www.catalogueoflife.org/data/taxon/C46V"];
+        yield "CRLT8 ; url" => ["CRLT8", "https://www.catalogueoflife.org/data/taxon/CRLT8"];
+        yield "CRLT8 ; url, http" => ["CRLT8", "http://www.catalogueoflife.org/data/taxon/CRLT8"];
+        yield "Broken input / 123123" => [false, "123123"];
+
+    }
+
+    /**
+     * Ensure validating Catalogue of Life works.
+     *
+     * @param string|false $expected    Expected output.
+     * @param string       $inputString Input string.
+     *
+     * @return void
+     */
+    #[DataProvider('colIdForValidationProvider')]
+    public function testColIdValidation(string|false $expected, string $inputString):void {
+
+        self::assertEquals($expected, MDNodaRepository::col->validateId($inputString));
+
+    }
+
+    /**
+     * Data provider for validating references to the Mindat.
+     *
+     * @return Generator<string, array{0: string|false, 1: string}>
+     */
+    public static function mindatIdForValidationProvider():Generator {
+
+        yield "2047 ; url, http" => ["2047", "http://www.mindat.org/min-2047.html"];
+        yield "2047 ; url, https" => ["2047", "https://www.mindat.org/min-2047.html"];
+        yield "Broken input / adfdasjfklasjl" => [false, "adfdasjfklasjl"];
+
+    }
+
+    /**
+     * Ensure validating Mindat references.
+     *
+     * @param string|false $expected    Expected output.
+     * @param string       $inputString Input string.
+     *
+     * @return void
+     */
+    #[DataProvider('mindatIdForValidationProvider')]
+    public function testMindatIdValidation(string|false $expected, string $inputString):void {
+
+        self::assertEquals($expected, MDNodaRepository::mindatorg->validateId($inputString));
 
     }
 }