From 8491b62a8330e2e71165900ed87888addb1669fb Mon Sep 17 00:00:00 2001
From: Joshua Ramon Enslin <joshua@museum-digital.de>
Date: Fri, 4 Apr 2025 20:03:59 +0200
Subject: [PATCH] Validate against time errors in autogenerating translations
 for times

Close #30
---
 src/NodaTimeAutotranslater.php       | 78 ++++++++++++++++++++++++++--
 tests/NodaTimeAutotranslaterTest.php | 27 ++++++++--
 2 files changed, 99 insertions(+), 6 deletions(-)

diff --git a/src/NodaTimeAutotranslater.php b/src/NodaTimeAutotranslater.php
index 9ff6e74..1466469 100644
--- a/src/NodaTimeAutotranslater.php
+++ b/src/NodaTimeAutotranslater.php
@@ -13,7 +13,7 @@ final class NodaTimeAutotranslater {
 
     // TODO: Move these to NodaTimeAutotranslaterLocales
 
-    const LANGS_SYLLABLE_CLEANING = [
+    public const LANGS_SYLLABLE_CLEANING = [
         "hu" => [
             "10-as évek" => "10-es évek",
             "40-as évek" => "40-es évek",
@@ -463,13 +463,13 @@ final class NodaTimeAutotranslater {
     }
 
     /**
-     * Gets translations for a given entry type.
+     * Prepares translations for each available language.
      *
      * @param array<integer|string> $timeInfo Time information.
      *
      * @return array<string>
      */
-    public static function getTranslations(array $timeInfo):array {
+    public static function prepareTranslations(array $timeInfo):array {
 
         if (!empty($timeInfo['zeit_name']) and strlen((string)$timeInfo['zeit_name']) > 10 and !empty($timespanDates = NodaTimeSplitter::attempt_splitting_from_till((string)$timeInfo['zeit_name']))) {
 
@@ -604,6 +604,78 @@ final class NodaTimeAutotranslater {
 
     }
 
+    /**
+     * Validates correctness of years in translation strings.
+     *
+     * @param string|integer        $start        Start year.
+     * @param string|integer        $end          End year.
+     * @param array<string, string> $translations Translations.
+     *
+     * @return boolean
+     */
+    public static function validateTranslations(string|int $start, string|int $end, array $translations):bool {
+
+        $start = ltrim((string)$start, ' 0-');
+        $end = ltrim((string)$end, ' 0-');
+
+        // Edge cases: Centuries and decades have special translations
+        // and can thus not be validated properly
+        // Century BCE
+        if (substr($start, -1) === "0" && substr($end, -1) === '1' && $start > $end) {
+            return true;
+        }
+        // Century CE
+        if (substr($start, -1) === "1" && substr($end, -1) === '0' && $start < $end) {
+            return true;
+        }
+        // Decade
+        if (substr($start, -1) === "0" && substr($end, -1) === '9' && $start < $end) {
+            return true;
+        }
+
+        // 1920 + ? can be both Since 1920 and After 1919, so validation
+        // is impossible there, too
+        if ($start === '?' || $end === '?') return true;
+
+        // Unset unvalidatable languages
+        unset($translations['ar'], $translations['fa']);
+
+        if ($start !== '?') {
+            foreach ($translations as $t) {
+                if (!str_contains($t, $start)) {
+                    return false;
+                }
+            }
+        }
+        if ($end !== '?' && $start !== $end) {
+            foreach ($translations as $t) {
+                if (!str_contains($t, $end)) {
+                    return false;
+                }
+            }
+        }
+
+        return true;
+
+    }
+
+    /**
+     * Gets translations for a given entry type.
+     *
+     * @param array<integer|string> $timeInfo Time information.
+     *
+     * @return array<string>
+     */
+    public static function getTranslations(array $timeInfo):array {
+
+        $output = self::prepareTranslations($timeInfo);
+
+        if (self::validateTranslations($timeInfo['zeit_beginn'], $timeInfo['zeit_ende'], $output) === false) return [];
+
+        return $output;
+
+    }
+
     /**
      * Runs autotranslater.
      *
diff --git a/tests/NodaTimeAutotranslaterTest.php b/tests/NodaTimeAutotranslaterTest.php
index d8bfda9..ac0af7f 100644
--- a/tests/NodaTimeAutotranslaterTest.php
+++ b/tests/NodaTimeAutotranslaterTest.php
@@ -6,12 +6,14 @@
  */
 declare(strict_types = 1);
 use PHPUnit\Framework\TestCase;
+use PHPUnit\Framework\Attributes\CoversClass;
+use PHPUnit\Framework\Attributes\Medium;
 
 /**
  * This script contains tests for the automatic translation class for time names.
- *
- * @covers \NodaTimeAutotranslater
  */
+#[Medium]
+#[CoversClass(\NodaIDGetter::class)]
 final class NodaTimeAutotranslaterTest extends TestCase {
     /**
      * Test to check whether the HTML page is correctly generated.
@@ -32,7 +34,7 @@ final class NodaTimeAutotranslaterTest extends TestCase {
             "zeit_zaehlzeit_tag"   => "01",
         ];
         $output = NodaTimeAutotranslater::getTranslations($timeInfo);
-        self::assertEquals($output["de"], "01.05.1920");
+        self::assertEquals("01.05.1920", $output["de"]);
 
     }
 
@@ -671,4 +673,23 @@ final class NodaTimeAutotranslaterTest extends TestCase {
         self::assertEquals($output["de"], "Vor 01.12.1919");
 
     }
+
+    /**
+     * Test to check whether validating works.
+     *
+     * @author Joshua Ramon Enslin <joshua@museum-digital.de>
+     * @group  ValidOutput
+     * @small
+     *
+     * @return void
+     */
+    public function testValidation():void {
+
+        $output = [
+            'de' => '1.12.1920',
+        ];
+
+        self::assertFalse(NodaTimeAutotranslater::validateTranslations("1919", "1919", $output));
+
+    }
 }