Add wrappers to check if an actor, place is usable for new terms

src/NodaBlacklistedTerms.php

@@ -24,6 +24,7 @@ final class NodaBlacklistedTerms {
             'versch.',
             'verschiedenes',
             'unbekannt',
+            '_',
             'weitere',
             'weiteres',
 
@@ -41,6 +42,7 @@ final class NodaBlacklistedTerms {
             'BB',
             'CC',
             'KK',
+            'K.A',
             'S',
             'N) (generell',
             'DD',

src/NodaLogEdit.php

@@ -10,14 +10,14 @@ declare(strict_types = 1);
  * Provides functions for easily logging updates to the main noda DB tables.
  */
 final class NodaLogEdit {
-    const ACTION_WHITELIST = [
+    public const ACTION_WHITELIST = [
         'insert',
         'update',
         'merge',
         'delete',
     ];
 
-    const SECTION_WHITELIST = [
+    public const SECTION_WHITELIST = [
         'base',
         'addition',
         'noda_link',
@@ -32,6 +32,7 @@ final class NodaLogEdit {
         'status',
         'translation',
         'group',
+        'note',
     ];
 
     /**

src/NodaValidationHelper.php

@@ -11,7 +11,7 @@ declare(strict_types = 1);
  */
 final class NodaValidationHelper {
 
-    const ACTOR_DESCRIPTION_REQUIRED_DISTINCT_CHARS = 3;
+    private const ACTOR_DESCRIPTION_REQUIRED_DISTINCT_CHARS = 3;
 
     /**
      * Validates an actor description for completeness. Of course, only an informed
@@ -120,4 +120,93 @@ final class NodaValidationHelper {
         }
 
     }
+
+    /**
+     * Checks if an actor name is valid.
+     * Returns 0 if the actor can be added as a new one / actor name can be used as a new one.
+     * Returns the ID for already known actors.
+     * Throws an exception if the actor is blacklisted.
+     *
+     * @param MDMysqli $mysqli_noda          DB connection to vocabulary.
+     * @param string   $lang                 User language.
+     * @param string   $persinst_name        Actor name.
+     * @param string   $persinst_name_en     Actor name (short).
+     * @param string   $persinst_geburtsjahr Birth year.
+     * @param string   $persinst_sterbejahr  Death year.
+     * @param string   $context_identifier   Context / instance identifier.
+     * @param integer  $institution_id       Institution ID.
+     *
+     * @return integer
+     */
+    public static function checkPersinstNameIsUsable(MDMysqli $mysqli_noda, string $lang, string $persinst_name, string $persinst_name_en, string $persinst_geburtsjahr, string $persinst_sterbejahr, string $context_identifier, int $institution_id):int {
+
+        if ($storedType = NodaDistinctlyTypedStrings::lookup($mysqli_noda, $lang, $persinst_name_en)) {
+
+            if ($storedType !== 'persinst') {
+                throw new MDpageParameterMissingException("This term is marked to be distinctly a " . $storedType);
+            }
+
+        }
+
+        // Special case: ? or "unbekannt" as a given name.
+        if (str_starts_with($persinst_name_en, '? ') || str_starts_with(strtolower($persinst_name_en), 'unbekannt ') ) {
+            throw new MDBlacklistedInputException("Term is blacklisted");
+        }
+
+        // Check if the place name is in the current language's blacklist.
+        if (NodaBlacklistedTerms::checkPersinstBlacklistedInDb($mysqli_noda, $lang, $persinst_name) === true) {
+            throw new MDBlacklistedInputException("Term is blacklisted");
+        }
+        if (NodaBlacklistedTerms::checkPersinstBlacklistedInDb($mysqli_noda, $lang, $persinst_name_en) === true) {
+            throw new MDBlacklistedInputException("Term is blacklisted");
+        }
+
+        if ($existingPersinstId = NodaIDGetter::getPersinstIDByNamesAndRewrites($mysqli_noda, $lang, $persinst_name_en, $persinst_geburtsjahr, $persinst_sterbejahr, $context_identifier, $institution_id)) {
+            return $existingPersinstId;
+        }
+
+        if ($existingPersinstId = NodaIDGetter::getPersinstIDByNamesAndRewrites($mysqli_noda, $lang, $persinst_name, $persinst_geburtsjahr, $persinst_sterbejahr, $context_identifier, $institution_id)) {
+            return $existingPersinstId;
+        }
+
+        return 0;
+
+    }
+
+    /**
+     * Checks if an place name is valid.
+     * Returns 0 if the place can be added as a new one / place name can be used as a new one.
+     * Returns the ID for already known places.
+     * Throws an exception if the place is blacklisted.
+     *
+     * @param MDMysqli $mysqli_noda        DB connection to vocabulary.
+     * @param string   $lang               User language.
+     * @param string   $place_name         Place name.
+     * @param string   $context_identifier Context / instance identifier.
+     * @param integer  $institution_id     Institution ID.
+     *
+     * @return integer
+     */
+    public static function checkPlaceNameIsUsable(MDMysqli $mysqli_noda, string $lang, string $place_name, string $context_identifier, int $institution_id):int {
+
+        if ($storedType = NodaDistinctlyTypedStrings::lookup($mysqli_noda, $lang, $place_name)) {
+
+            if ($storedType !== 'orte') {
+                throw new MDpageParameterMissingException("This term is marked to be distinctly a " . $storedType);
+            }
+
+        }
+
+        // Check if the place name is in the current language's blacklist.
+        if (NodaBlacklistedTerms::checkPlaceBlacklistedInDb($mysqli_noda, $lang, $place_name) === true) {
+            throw new MDBlacklistedInputException("Term is blacklisted");
+        }
+
+        if ($existingPlaceId = NodaIDGetter::getPlaceIDByNamesAndRewrites($mysqli_noda, $lang, $place_name, $context_identifier, $institution_id)) {
+            return $existingPlaceId;
+        }
+
+        return 0;
+
+    }
 }

src/NodaWikidataFetcher.php

@@ -138,7 +138,7 @@ final class NodaWikidataFetcher {
             return '';
         }
         $firstPageId = array_keys($json_decoded['query']['pages'])[0];
-        $datafromwiki = strval($json_decoded['query']['pages'][$firstPageId]['extract']);
+        $datafromwiki = strval($json_decoded['query']['pages'][$firstPageId]['extract'] ?? "");
 
         return self::_cleanInputSimple($datafromwiki);