museum-digital/MDNodaHelpers
35
0
Fork 0
Code Issues 6 Pull Requests Releases Wiki Activity

Compare commits

base: museum-digital:1cf0f9858a2301ef7736e6060184e97e1629dfa9
Branches Tags
museum-digital:master
..
compare: museum-digital:d1cee17ef5d2aa2ca87947c0011b62229b47cf61
Branches Tags
museum-digital:master

No commits in common. "1cf0f9858a2301ef7736e6060184e97e1629dfa9" and "d1cee17ef5d2aa2ca87947c0011b62229b47cf61" have entirely different histories.
1cf0f9858a ... d1cee17ef5

2 changed files with 21 additions and 207 deletions
Show Stats Expand all files Collapse all files

18
src/NodaWikidataFetcher.php
View File

@ -168,9 +168,8 @@ final class NodaWikidataFetcher {
* *
* @return array<mixed> * @return array<mixed>
*/ */
public static function getWikidataEntity(string $wikidata_id):array { private static function _getWikidataEntity(string $wikidata_id):array {
self::validateWikidataId($wikidata_id);
$data = json_decode(MD_STD::runCurl("https://www.wikidata.org/wiki/Special:EntityData/" . urlencode($wikidata_id) . ".json", 10000), true); $data = json_decode(MD_STD::runCurl("https://www.wikidata.org/wiki/Special:EntityData/" . urlencode($wikidata_id) . ".json", 10000), true);
if ($data === null) { if ($data === null) {
throw new MDhttpFailedException("Failed fetching from Wikidata. Try again later."); throw new MDhttpFailedException("Failed fetching from Wikidata. Try again later.");
@ -1065,7 +1064,8 @@ final class NodaWikidataFetcher {
*/ */
public function retrievePersinstInfoFromWikidataID(string $lang, string $wikidata_id, int $persinst_id, string $erfasst_von) { public function retrievePersinstInfoFromWikidataID(string $lang, string $wikidata_id, int $persinst_id, string $erfasst_von) {
$data = self::getWikidataEntity($wikidata_id); self::validateWikidataId($wikidata_id);
$data = self::_getWikidataEntity($wikidata_id);
// Get links to wikipedia // Get links to wikipedia
@ -1115,7 +1115,8 @@ final class NodaWikidataFetcher {
*/ */
public function retrievePersinstNormDataLinksFromWikidataID(string $wikidata_id, int $persinst_id, string $erfasst_von) { public function retrievePersinstNormDataLinksFromWikidataID(string $wikidata_id, int $persinst_id, string $erfasst_von) {
$data = self::getWikidataEntity($wikidata_id); self::validateWikidataId($wikidata_id);
$data = self::_getWikidataEntity($wikidata_id);
if (!empty($nodaLinks = $this->_getNodaLinksFromWikidataResult('persinst', $wikidata_id, $data))) { if (!empty($nodaLinks = $this->_getNodaLinksFromWikidataResult('persinst', $wikidata_id, $data))) {
NodaBatchInserter::linkNodaForPersinst($this->_mysqli_noda, $persinst_id, $nodaLinks, $erfasst_von); NodaBatchInserter::linkNodaForPersinst($this->_mysqli_noda, $persinst_id, $nodaLinks, $erfasst_von);
} }
@ -1133,7 +1134,8 @@ final class NodaWikidataFetcher {
*/ */
public function retrievePlaceNormDataLinksFromWikidataID(string $wikidata_id, int $onum, string $erfasst_von) { public function retrievePlaceNormDataLinksFromWikidataID(string $wikidata_id, int $onum, string $erfasst_von) {
$data = self::getWikidataEntity($wikidata_id); self::validateWikidataId($wikidata_id);
$data = self::_getWikidataEntity($wikidata_id);
if (!empty($nodaLinks = $this->_getNodaLinksFromWikidataResult('place', $wikidata_id, $data))) { if (!empty($nodaLinks = $this->_getNodaLinksFromWikidataResult('place', $wikidata_id, $data))) {
NodaBatchInserter::linkNodaForPlace($this->_mysqli_noda, $onum, $nodaLinks, $erfasst_von); NodaBatchInserter::linkNodaForPlace($this->_mysqli_noda, $onum, $nodaLinks, $erfasst_von);
} }
@ -1357,7 +1359,8 @@ final class NodaWikidataFetcher {
*/ */
public function retrievePlaceInfoFromWikidataID(string $lang, string $wikidata_id, int $onum, string $erfasst_von) { public function retrievePlaceInfoFromWikidataID(string $lang, string $wikidata_id, int $onum, string $erfasst_von) {
$data = self::getWikidataEntity($wikidata_id); self::validateWikidataId($wikidata_id);
$data = self::_getWikidataEntity($wikidata_id);
$wikilinks = self::_getWikipediaLinksFromWikidataOutput($data); $wikilinks = self::_getWikipediaLinksFromWikidataOutput($data);
@ -1590,7 +1593,8 @@ final class NodaWikidataFetcher {
*/ */
public function retrieveTagInfoFromWikidataID(string $lang, string $wikidata_id, int $tag_id, string $erfasst_von) { public function retrieveTagInfoFromWikidataID(string $lang, string $wikidata_id, int $tag_id, string $erfasst_von) {
$data = self::getWikidataEntity($wikidata_id); self::validateWikidataId($wikidata_id);
$data = self::_getWikidataEntity($wikidata_id);
$wikilinks = self::_getWikipediaLinksFromWikidataOutput($data); $wikilinks = self::_getWikipediaLinksFromWikidataOutput($data);

210
tests/NodaWikidataFetcherTest.php
View File

@ -6,27 +6,19 @@
*/ */
declare(strict_types = 1); declare(strict_types = 1);
use PHPUnit\Framework\TestCase; use PHPUnit\Framework\TestCase;
use PHPUnit\Framework\Attributes\CoversClass;
use PHPUnit\Framework\Attributes\Medium;
use PHPUnit\Framework\Attributes\DataProvider;
require_once __DIR__ . '/../../MDMysqli/test_connections.conf.php';
/** /**
* This script contains tests for the Wikidata fetcher. * This script contains tests for the Wikidata fetcher.
*
* @covers \NodaWikidataFetcher
*/ */
#[medium]
#[CoversClass(\NodaWikidataFetcher::class)]
final class NodaWikidataFetcherTest extends TestCase { final class NodaWikidataFetcherTest extends TestCase {
// Test for getting translations: Telugu
private const TEST_LANG = 'te';
/** /**
* Test to check whether the HTML page is correctly generated. * Test to check whether the HTML page is correctly generated.
* *
* @author Joshua Ramon Enslin <joshua@museum-digital.de> * @author Joshua Ramon Enslin <joshua@museum-digital.de>
* @group ValidOutput * @group ValidOutput
* @small
* *
* @return void * @return void
*/ */
@ -42,6 +34,7 @@ final class NodaWikidataFetcherTest extends TestCase {
* *
* @author Joshua Ramon Enslin <joshua@museum-digital.de> * @author Joshua Ramon Enslin <joshua@museum-digital.de>
* @group ValidOutput * @group ValidOutput
* @small
* *
* @return void * @return void
*/ */
@ -54,7 +47,9 @@ final class NodaWikidataFetcherTest extends TestCase {
/** /**
* Test to check whether the HTML page is correctly generated. * Test to check whether the HTML page is correctly generated.
* *
* @author Joshua Ramon Enslin <joshua@museum-digital.de>
* @group ValidOutput * @group ValidOutput
* @small
* *
* @return void * @return void
*/ */
@ -67,7 +62,9 @@ final class NodaWikidataFetcherTest extends TestCase {
/** /**
* Test for cleaning wikidata info. * Test for cleaning wikidata info.
* *
* @author Joshua Ramon Enslin <joshua@museum-digital.de>
* @group ValidOutput * @group ValidOutput
* @small
* *
* @return void * @return void
*/ */
@ -268,7 +265,9 @@ Transclusion expansion time report (%,ms,calls,template)
/** /**
* Test for cleaning wikidata info. * Test for cleaning wikidata info.
* *
* @author Joshua Ramon Enslin <joshua@museum-digital.de>
* @group ValidOutput * @group ValidOutput
* @small
* *
* @return void * @return void
*/ */
@ -282,193 +281,4 @@ Transclusion expansion time report (%,ms,calls,template)
); );
} }
/**
* Data provider for an actor that has a wikidata link and a Telugu translation.
*
* @return array<string, array{0: int, 1: string}>
*/
public static function actorWithTlAndWikidataLinkProvider():array {
$mysqli = md_main_mysqli_connect();
$result = $mysqli->do_read_query("SELECT `persinst_id`, `noda_nrinsource`
FROM `" . DATABASENAME_NODA . "`.`noda`
WHERE `noda_source` = 'Wikidata'
AND EXISTS (SELECT 1 FROM `" . DATABASENAME_NODA . "`.`persinst_translation`
WHERE `persinst_translation`.`persinst_id` = `noda`.`persinst_id`
AND `trans_language` = '" . $mysqli->escape_string(self::TEST_LANG) . "')");
if (!$cur = $result->fetch_row()) {
throw new Exception("Failed to identify an entry that has a wikidata entry and a translation for language " . self::TEST_LANG);
}
$result->close();
$mysqli->close();
return [
'Actor with wikidata and translation' => [$cur[0], $cur[1]],
];
}
/**
* Data provider for a place that has a wikidata link and a Telugu translation.
*
* @return array<string, array{0: int, 1: string}>
*/
public static function placeWithTlAndWikidataLinkProvider():array {
$mysqli = md_main_mysqli_connect();
$result = $mysqli->do_read_query("SELECT `ort_id`, `noda_nrinsource`
FROM `" . DATABASENAME_NODA . "`.`noda_orte`
WHERE `noda_source` = 'Wikidata'
AND EXISTS (SELECT 1 FROM `" . DATABASENAME_NODA . "`.`ort_translation`
WHERE `ort_translation`.`ort_id` = `noda_orte`.`ort_id`
AND `trans_language` = '" . $mysqli->escape_string(self::TEST_LANG) . "')");
if (!$cur = $result->fetch_row()) {
throw new Exception("Failed to identify an entry that has a wikidata entry and a translation for language " . self::TEST_LANG);
}
$result->close();
$mysqli->close();
return [
'Place with wikidata and translation' => [$cur[0], $cur[1]],
];
}
/**
* Data provider for an tag that has a wikidata link and a Telugu translation.
*
* @return array<string, array{0: int, 1: string}>
*/
public static function tagWithTlAndWikidataLinkProvider():array {
$mysqli = md_main_mysqli_connect();
$result = $mysqli->do_read_query("SELECT `tag_id`, `noda_nrinsource`
FROM `" . DATABASENAME_NODA . "`.`noda_tag`
WHERE `noda_source` = 'Wikidata'
AND EXISTS (SELECT 1 FROM `" . DATABASENAME_NODA . "`.`tag_translation`
WHERE `tag_translation`.`tag_id` = `noda_tag`.`tag_id`
AND `trans_language` = '" . $mysqli->escape_string(self::TEST_LANG) . "')");
if (!$cur = $result->fetch_row()) {
throw new Exception("Failed to identify an entry that has a wikidata entry and a translation for language " . self::TEST_LANG);
}
$result->close();
$mysqli->close();
return [
'Tag with wikidata and translation' => [$cur[0], $cur[1]],
];
}
/**
* Test for fetching and recording translations for an actor.
*
* @param integer $actor_id Actor ID.
* @param string $wikidata_id Wikidata ID.
*
* @return void
*/
#[DataProvider('actorWithTlAndWikidataLinkProvider')]
public function testFetchingTranslationForPersinst(int $actor_id, string $wikidata_id):void {
$mysqli = md_main_mysqli_connect();
$mysqli->do_update_query("DELETE FROM `" . DATABASENAME_NODA . "`.`persinst_translation`
WHERE `persinst_id` = " . $actor_id . "
AND `trans_language` = '" . $mysqli->escape_string(self::TEST_LANG) . "'");
self::assertEquals(0, MDMysqliTesting::queryNumRows($mysqli, "
FROM `" . DATABASENAME_NODA . "`.`persinst_translation`
WHERE `persinst_id` = " . $actor_id . "
AND `trans_language` = '" . $mysqli->escape_string(self::TEST_LANG) . "'"));
$data = NodaWikidataFetcher::getWikidataEntity($wikidata_id);
$fetcher = new NodaWikidataFetcher($mysqli);
$fetcher->getWikidataTranslationsForPersinst($data, $actor_id, [self::TEST_LANG]);
self::assertEquals(1, MDMysqliTesting::queryNumRows($mysqli, "
FROM `" . DATABASENAME_NODA . "`.`persinst_translation`
WHERE `persinst_id` = " . $actor_id . "
AND `trans_language` = '" . $mysqli->escape_string(self::TEST_LANG) . "'"));
$mysqli->close();
}
/**
* Test for fetching and recording translations for an place.
*
* @param integer $place_id Place ID.
* @param string $wikidata_id Wikidata ID.
*
* @return void
*/
#[DataProvider('placeWithTlAndWikidataLinkProvider')]
public function testFetchingTranslationForPlace(int $place_id, string $wikidata_id):void {
$mysqli = md_main_mysqli_connect();
$mysqli->do_update_query("DELETE FROM `" . DATABASENAME_NODA . "`.`ort_translation`
WHERE `ort_id` = " . $place_id . "
AND `trans_language` = '" . $mysqli->escape_string(self::TEST_LANG) . "'");
self::assertEquals(0, MDMysqliTesting::queryNumRows($mysqli, "
FROM `" . DATABASENAME_NODA . "`.`ort_translation`
WHERE `ort_id` = " . $place_id . "
AND `trans_language` = '" . $mysqli->escape_string(self::TEST_LANG) . "'"));
$data = NodaWikidataFetcher::getWikidataEntity($wikidata_id);
$fetcher = new NodaWikidataFetcher($mysqli);
$fetcher->getWikidataTranslationsForPlace($data, $place_id, [self::TEST_LANG]);
self::assertEquals(1, MDMysqliTesting::queryNumRows($mysqli, "
FROM `" . DATABASENAME_NODA . "`.`ort_translation`
WHERE `ort_id` = " . $place_id . "
AND `trans_language` = '" . $mysqli->escape_string(self::TEST_LANG) . "'"));
$mysqli->close();
}
/**
* Test for fetching and recording translations for an tag.
*
* @param integer $tag_id Tag ID.
* @param string $wikidata_id Wikidata ID.
*
* @return void
*/
#[DataProvider('tagWithTlAndWikidataLinkProvider')]
public function testFetchingTranslationForTag(int $tag_id, string $wikidata_id):void {
$mysqli = md_main_mysqli_connect();
$mysqli->do_update_query("DELETE FROM `" . DATABASENAME_NODA . "`.`tag_translation`
WHERE `tag_id` = " . $tag_id . "
AND `trans_language` = '" . $mysqli->escape_string(self::TEST_LANG) . "'");
self::assertEquals(0, MDMysqliTesting::queryNumRows($mysqli, "
FROM `" . DATABASENAME_NODA . "`.`tag_translation`
WHERE `tag_id` = " . $tag_id . "
AND `trans_language` = '" . $mysqli->escape_string(self::TEST_LANG) . "'"));
$data = NodaWikidataFetcher::getWikidataEntity($wikidata_id);
$fetcher = new NodaWikidataFetcher($mysqli);
$fetcher->getWikidataTranslationsForTag($data, $tag_id, [self::TEST_LANG]);
self::assertEquals(1, MDMysqliTesting::queryNumRows($mysqli, "
FROM `" . DATABASENAME_NODA . "`.`tag_translation`
WHERE `tag_id` = " . $tag_id . "
AND `trans_language` = '" . $mysqli->escape_string(self::TEST_LANG) . "'"));
$mysqli->close();
}
} }