Add functions for automatic rewriting of country names to brackets at

the end of place names based on lists

scripts/get_wikidata_country_names.php

@@ -0,0 +1,71 @@
+<?PHP
+/**
+ * This file contains tools for fetching data from Wikidata.
+ *
+ * @file
+ * @author Joshua Ramon Enslin <joshua@museum-digital.de>
+ */
+declare(strict_types = 1);
+
+require_once __DIR__ . '/../src/NodaWikidataFetcher.php';
+require_once __DIR__ . '/../../MD_STD/src/MD_STD.php';
+
+/**
+ * Queries wikidata for instances of a Q-ID.
+ *
+ * @param string $lang       Query language.
+ * @param string $instanceOf Q-ID.
+ *
+ * @return array<mixed>
+ */
+function query(string $lang, string $instanceOf):array {
+
+    $sparqlQueryString = 'SELECT ?item ?itemLabel
+    WHERE
+    {
+      ?item wdt:P31/wdt:P279* wd:' . $instanceOf . '.
+      SERVICE wikibase:label { bd:serviceParam wikibase:language "' . $lang . ',[AUTO_LANGUAGE],en". } # Helps get the label in your language, if not, then en language
+    }';
+
+    return NodaWikidataFetcher::sparqlQuery($sparqlQueryString);
+
+}
+
+/**
+ * Returns names from a query.
+ *
+ * @param array<mixed> $data Wikidata output values.
+ *
+ * @return array<string>
+ */
+function getNames(array $data):array {
+
+    $output = [];
+
+    foreach ($data['results']['bindings'] as $entry) {
+        $output[] = $entry['itemLabel']['value'];
+    }
+
+    return $output;
+
+}
+
+// Q6256 => country
+
+$targets = [
+    'Q6256' => 'countries',
+    'Q3024240' => 'historical_countries',
+    'Q10864048' => 'first_lvl_administrative_units',
+];
+
+$langs = ['ar', 'bg', 'bn', 'cs', 'da', 'de', 'el', 'en', 'es', 'fa', 'fi', 'fr', 'ha', 'he', 'hi', 'hu', 'id', 'it', 'ja', 'ka', 'ko', 'nl', 'pl', 'pt', 'ro', 'ru', 'sv', 'sw', 'ta', 'th', 'tl', 'tr', 'uk', 'ur', 'vi', 'zh'];
+
+foreach ($langs as $lang) {
+    foreach ($targets as $qid => $filename) {
+
+        $regionNames = getNames(query($lang, $qid));
+        file_put_contents(__DIR__ . '/../static/' . $filename . '.' . $lang . '.json', json_encode($regionNames));
+        echo "Fetched $lang : $filename ($qid)" . PHP_EOL;
+
+    }
+}

src/NodaConsolidatedNamesForPlaces.php

@@ -64,6 +64,11 @@ final class NodaConsolidatedNamesForPlaces extends NodaConsolidatedNamesAbstract
         'XX' => '20',
     ];
 
+    /**
+     * @var array<string, list<string>>
+     */
+    private static $_placeNameListCaches = [];
+
     /**
      * Rewrites indicators for narrower locations paired with a superordinate location
      * into the format "Narrower (Broader)".
@@ -181,6 +186,75 @@ final class NodaConsolidatedNamesForPlaces extends NodaConsolidatedNamesAbstract
 
     }
 
+    /**
+     * Loads a JSON file, optionally loading it cached through a private static variable
+     * if reuse is expectable (= in the case of CLI usage).
+     *
+     * @param non-empty-string $filename File name to load.
+     *
+     * @return list<string>
+     */
+    private static function _loadJsonList(string $filename):array {
+
+        if (PHP_SAPI === 'cli' && isset(self::$_placeNameListCaches[$filename])) {
+            return self::$_placeNameListCaches[$filename];
+        }
+
+        $output = json_decode(MD_STD::file_get_contents($filename), true);
+        if ($output === false) {
+            throw new Exception("Failed to get list");
+        }
+
+        if (PHP_SAPI === 'cli') {
+            self::$_placeNameListCaches[$filename] = $output;
+        }
+
+        return $output;
+
+    }
+
+    /**
+     * Moves names of regions to brackets using pre-generated lists of countries,
+     * historical country names, etc.
+     *
+     * @param string $lang Instance language.
+     * @param string $name Input string to clean.
+     *
+     * @return string
+     */
+    private static function _move_region_names_to_brackets(string $lang, string $name):string {
+
+        $separators = ['-', ', '];
+
+        foreach ($separators as $separator) {
+
+            if (!str_contains($name, $separator) || substr_count($name, $separator) !== 1) continue;
+
+            // Get parts and trim them
+            $parts = explode($separator, $name);
+            foreach ($parts as $key => $value) {
+                $parts[$key] = trim($value);
+            }
+
+            // Load place names
+            $countryNames = self::_loadJsonList(__DIR__ . "/../static/countries.$lang.json") + self::_loadJsonList(__DIR__ . "/../static/historical_countries.$lang.json");
+
+            $part0IsCountry = in_array($parts[0], $countryNames, true);
+            $part1IsCountry = in_array($parts[1], $countryNames, true);
+
+            if ($part0IsCountry === true && $part1IsCountry === false) {
+                return $parts[1] . ' (' . $parts[0] . ')';
+            }
+            else if ($part0IsCountry === false && $part1IsCountry === true) {
+                return $parts[0] . ' (' . $parts[1] . ')';
+            }
+
+        }
+
+        return $name;
+
+    }
+
     /**
      * Cleans a place name by trimming etc. Also removes uncertainty indicators.
      *
@@ -203,6 +277,8 @@ final class NodaConsolidatedNamesForPlaces extends NodaConsolidatedNamesAbstract
             default => $ort_name,
         };
 
+        $ort_name = self::_move_region_names_to_brackets($lang, $ort_name);
+
         return $ort_name;
 
     }