Add rewriting for Ukrainian place names based on specified hierarchies

src/NodaConsolidatedNamesForPlaces.php

@@ -144,7 +144,7 @@ final class NodaConsolidatedNamesForPlaces extends NodaConsolidatedNamesAbstract
     /**
      * Cleans and consolidates name parts appearing regularly in German place names.
      *
-     * @param string $name Name of an actor.
+     * @param string $name Name of a place.
      *
      * @return string
      */
@@ -175,7 +175,7 @@ final class NodaConsolidatedNamesForPlaces extends NodaConsolidatedNamesAbstract
     /**
      * Cleans and consolidates name parts appearing regularly in Hungarian place names.
      *
-     * @param string $name Name of an actor.
+     * @param string $name Name of a place.
      *
      * @return string
      */
@@ -218,6 +218,131 @@ final class NodaConsolidatedNamesForPlaces extends NodaConsolidatedNamesAbstract
 
     }
 
+    /**
+     * Rewrites a Ukrainian language name based on abbreviations explaining the
+     * hierarchy of named places.
+     *
+     * @param string $name Input name to rewrite.
+     *
+     * @return string
+     */
+    private static function _rewrite_ukrainian_names_by_hierarchy($name):string {
+
+        $identifiersByLevel = [
+            'state' => [' РСР', 'РСР ', ' губернія', 'губернія '],
+            'oblast' => ['обл.'],
+            'region' => ['р-н'],
+            'city' => ['м.'],
+            'district' => [],   // Is also р-н; which it is is determined based on position
+            'street' => ['вул. '],
+        ];
+
+        $levels = [
+            'country' => '',
+            'state' => '',
+            'oblast' => '',
+            'region' => '',
+            'city' => '',
+            'district' => '',
+            'street' => '',
+        ];
+
+        $parts = explode(',', $name);
+        foreach ($parts as $part) {
+            $part = trim($part);
+            foreach ($identifiersByLevel as $level => $identifiers) {
+                foreach ($identifiers as $identifier) {
+
+                    if (str_starts_with($part, $identifier) || str_ends_with($part, $identifier)) {
+
+                        // Special case: Region can both be rajon or a district within a city
+                        // If both oblast and city are already known, the region will be a
+                        // district within the city.
+                        // Otherwise, it is to be assumed that it is a super-city region.
+                        if ($level === 'region' && !empty($levels['oblast']) && !empty($levels['city'])) {
+                            $level = 'district';
+                        }
+
+                        if (!empty($levels[$level])) {
+                            throw new Exception("Used the same level (" . $level . ") twice");
+                        }
+                        $levels[$level] = $part;
+                        continue 3;
+                    }
+
+                }
+            }
+
+            // Unspecified part level: Attempt identifying country
+            if (!isset($countryNames)) {
+                $countryNames = self::_loadJsonList(__DIR__ . "/../static/countries.uk.json") + self::_loadJsonList(__DIR__ . "/../static/historical_countries.uk.json");
+                $countryNames[] = 'СРСР';
+                $countryNames[] = 'УНР';
+                $countryNames[] = 'Російська імперія';
+                $countryNames[] = 'Рос.імперія';
+                $countryNames[] = 'Рос.имперія';
+            }
+            if (in_array($part, $countryNames, true)) {
+                $levels['country'] = $part;
+                continue;
+            }
+
+            // Unspecified level; return
+            throw new Exception("Unknown " . $part);
+            return $name;
+        }
+
+        $main_name = '';
+        $specifiers = [];
+
+        foreach (array_reverse($levels) as $level => $partname) {
+            if (empty($partname)) continue;
+
+            if ($level === 'city') {
+                $strtr = [];
+                foreach ($identifiersByLevel[$level] as $identifier) $strtr[$identifier] = '';
+                $partname = trim(strtr($partname, $strtr));
+            }
+
+            if (empty($main_name)) {
+                $main_name = $partname;
+            }
+            else {
+                $specifiers[] = $partname;
+            }
+        }
+
+        $output = $main_name;
+        if (!empty($specifiers)) $output .= ' (' . implode(', ', $specifiers) . ')';
+
+        return $output;
+
+    }
+
+
+    /**
+     * Cleans and consolidates name parts appearing regularly in Ukrainian place names.
+     *
+     * @param string $name Name of an place.
+     *
+     * @return string
+     */
+    private static function _clean_ukrainian_abbreviations(string $name):string {
+
+        /*
+        if (str_contains($name, " krt. ") && \preg_match("/\ krt\.\ [0-9]/", $name)) {
+            $name = str_replace(" krt. ", " körut ", $name);
+        }
+         */
+
+        if (str_contains($name, ',')) {
+            $name = self::_rewrite_ukrainian_names_by_hierarchy($name);
+        }
+
+        return $name;
+
+    }
+
     /**
      * Loads a JSON file, optionally loading it cached through a private static variable
      * if reuse is expectable (= in the case of CLI usage).
@@ -361,6 +486,7 @@ final class NodaConsolidatedNamesForPlaces extends NodaConsolidatedNamesAbstract
         $ort_name = match ($lang) {
             'de' => self::_clean_german_abbreviations($ort_name),
             'hu' => self::_clean_hungarian_abbreviations($ort_name),
+            'uk' => self::_clean_ukrainian_abbreviations($ort_name),
             default => $ort_name,
         };
 

tests/NodaConsolidatedNamesForPlacesTest.php

@@ -71,5 +71,24 @@ final class NodaConsolidatedNamesForPlacesTest extends TestCase {
 
         self::assertEquals("Västerdås, Schweden", NodaConsolidatedNamesForPlaces::consolidate_name("hu", "Västerdås, Schweden, Schweden"));
 
+        //
+        // Ukrainian: Rewrite by hierarchy
+        //
+        # self::assertEquals("Приморський (Україна, Запорізька обл., м. Запоріжжя,  р-н", NodaConsolidatedNamesForPlaces::consolidate_name("uk", "Україна, Запорізька обл., м. Запоріжжя, Приморський р-н"));
+        self::assertEquals("Рига (Латвійська РСР)", NodaConsolidatedNamesForPlaces::consolidate_name("uk", "Латвійська РСР, м.Рига"));
+        self::assertEquals("Латвійська РСР, м.Рига", NodaConsolidatedNamesForPlaces::consolidate_name("de", "Латвійська РСР, м.Рига"));
+        self::assertEquals("Рига (Латвійська РСР, СРСР)", NodaConsolidatedNamesForPlaces::consolidate_name("uk", "СРСР, Латвійська РСР, м.Рига"));
+        self::assertEquals("СРСР, Латвійська РСР, м.Рига", NodaConsolidatedNamesForPlaces::consolidate_name("de", "СРСР, Латвійська РСР, м.Рига"));
+
+        // Steet
+        self::assertEquals("вул. Шевченка (Рівне)", NodaConsolidatedNamesForPlaces::consolidate_name("uk", "м. Рівне, вул. Шевченка"));
+        self::assertEquals("м. Рівне, вул. Шевченка", NodaConsolidatedNamesForPlaces::consolidate_name("de", "м. Рівне, вул. Шевченка"));
+
+        // Special case: Region can both be rajon or a district within a city
+        // If both oblast and city are already known, the region will be a
+        // district within the city.
+        // Otherwise, it is to be assumed that it is a super-city region.
+        self::assertEquals("Приморський р-н (Запоріжжя, Запорізька обл., Україна)", NodaConsolidatedNamesForPlaces::consolidate_name("uk", "Україна, Запорізька обл., м. Запоріжжя, Приморський р-н"));
+
     }
 }