Improve splitting and translating of times BC

src/NodaTimeAutotranslater.php

@@ -208,13 +208,13 @@ final class NodaTimeAutotranslater {
         'fa' => 'قرن %s',
         'fr' => '%sème siècle',
         'hu' => '%s. század',
-        'id' => 'abad ke-%s',
+        'id' => 'Abad ke-%s',
         'it' => '%sesimo secolo',
         'ka' => 'მე -%s საუკუნე',
         'ko' => '%s 세기',
         'pl' => '%s wiek',
-        'pt' => 'século %s',
+        'pt' => 'Século %s',
-        'ro' => 'secolul al %s-lea',
+        'ro' => 'Secolul al %s-lea',
         'ru' => '%s век',
         'ta' => '%s ஆம் நூற்றாண்டு',
         'tl' => 'Ika-%s na siglo',
@@ -231,13 +231,13 @@ final class NodaTimeAutotranslater {
         'fa' => 'قرن %s-%s',
         'fr' => '%s-%sème siècle',
         'hu' => '%s.-%s. század',
-        'id' => 'abad ke-%s-%s',
+        'id' => 'Abad ke-%s-%s',
         'it' => '%s-%sesimo secolo',
         'ka' => 'მე -%s-%s საუკუნე',
         'ko' => '%s-%s 세기',
         'pl' => '%s-%s wiek',
-        'pt' => 'século %s-%s',
+        'pt' => 'Século %s-%s',
-        'ro' => 'secolul al %s-%s-lea',
+        'ro' => 'Secolul al %s-%s-lea',
         'ru' => '%s-%s век',
         'ta' => '%s-%s ஆம் நூற்றாண்டு',
         'tl' => 'Ika-%s hanggang ika-%s na siglo',
@@ -332,7 +332,9 @@ final class NodaTimeAutotranslater {
             return self::TRANSLATABLE_NOT;
         }
 
-        if (intval($zeit_ende) % 100 === 0 and (intval($zeit_beginn) - 1) % 100 === 0) {
+        if (($zeit_ende > 0 && intval($zeit_ende) % 100 === 0 and (intval($zeit_beginn) - 1) % 100 === 0)
+            || (intval($zeit_beginn) < 0 && intval($zeit_beginn) % 100 === 0 and (intval($zeit_ende) + 1) % 100 === 0)
+        ) {
             return self::TRANSLATABLE_CENTURY;
         }
 
@@ -481,10 +483,18 @@ final class NodaTimeAutotranslater {
      */
     public static function translateYearsAsCentury(array $timeInfo):array {
 
-        // Beginn: 1500. (1501 - 1) / 100 + 1 = 16. 16th century is the time.
+        if (intval($timeInfo['zeit_beginn']) > 0) {
-        $start_cen = ((intval($timeInfo['zeit_beginn']) - 1) / 100) + 1;
+            // Beginn: 1500. (1501 - 1) / 100 + 1 = 16. 16th century is the time.
-        // End: 1600. 16th century is the time.
+            $start_cen = ((intval($timeInfo['zeit_beginn']) - 1) / 100) + 1;
-        $end_cen = (intval($timeInfo['zeit_ende']) / 100);
+            // End: 1600. 16th century is the time.
+            $end_cen = (intval($timeInfo['zeit_ende']) / 100);
+        }
+        else {
+            // End: -1600. 16th century is the time.
+            $start_cen = (intval($timeInfo['zeit_beginn']) / 100);
+            // End: -1500. (1501 - 1) / 100 + 1 = 16. 16th century is the time.
+            $end_cen = ((intval($timeInfo['zeit_ende']) + 1) / 100) - 1;
+        }
 
         $suffixMode = self::getSuffixModeForYearsWSuffix((intval($timeInfo['zeit_beginn']) - 1), intval($timeInfo['zeit_ende']));
 
@@ -492,13 +502,13 @@ final class NodaTimeAutotranslater {
 
         if ($start_cen === $end_cen) {
             foreach (self::LANGS_CENTURY_FORMAT as $tLang => $format) {
-                $tLangValue = sprintf($format, (string)$start_cen);
+                $tLangValue = sprintf($format, (string)abs($start_cen));
                 $output[$tLang] = self::applyBcBceFormat($tLang, $tLangValue, $suffixMode);
             }
         }
         else {
             foreach (self::LANGS_CENTURIES_FORMAT as $tLang => $format) {
-                $tLangValue = sprintf($format, (string)$start_cen, (string)$end_cen);
+                $tLangValue = sprintf($format, (string)abs($start_cen), (string)abs($end_cen));
                 $output[$tLang] = self::applyBcBceFormat($tLang, $tLangValue, $suffixMode);
             }
         }

src/NodaTimeSplitter.php

@@ -378,9 +378,18 @@ final class NodaTimeSplitter {
 
         $datum = self::clean_input($datum);
 
-        if (preg_match("/^Kr\.\ e\.\ [0-9][0-9][0-9][0-9]$/", $datum)) {
+        if (preg_match("/^Kr\.\ e\.\ /", $datum)) {
-            $start = "-" . substr($datum, 7, 4);
+            if ($output = self::attempt_splitting(substr($datum, 7))) {
-            return [$start, $start, "00", "00", "-"];
+                $start = strval(-1 * intval($output[1]));
+                $end = strval(-1 * intval($output[0]));
+                return [$start, $end, $output[2], $output[3], '-'];
+            }
+        }
+
+        if (preg_match("/^[0-9][0-9][0-9][0-9]\-től(\ |\-)[0-9][0-9][0-9][0-9]\-ig$/", $datum)) {
+            $start = substr($datum, 0, 4);
+            $end = substr($datum, -7, 4);
+            return [$start, $end, "00", "00", "+", ""];
         }
 
         if (self::stri_occurs($datum, self::STOP_STRINGS_HUNGARIAN)) {
@@ -502,23 +511,33 @@ final class NodaTimeSplitter {
             $end = substr($datum, -4);
             return [$start, $end, "00", "00", "+", ""];
         }
+
+        // 303-305 (n. Chr.)
         if (preg_match("/^[0-9][0-9][0-9]\-[0-9][0-9][0-9]$/", $datum)) { // Hungarian Y-m
             $start = substr($datum, 0, 3);
             $end = substr($datum, -3);
             if ($end > $start) return ["0" . $start, "0" . $end, "00", "00", "+", ""];
         }
+
+        // 20-30 (n. Chr.)
         if (preg_match("/^[0-9][0-9]\-[0-9][0-9]$/", $datum)) { // 20-40 (n. Chr.)
             $start = substr($datum, 0, 2);
             $end = substr($datum, -2);
             if ($end > $start) return ["00" . $start, "00" . $end, "00", "00", "+", ""];
         }
 
+        // 1920
         if (preg_match("/^[0-9][0-9][0-9][0-9]$/", $datum)) {
-
             $start = substr($datum, 0, 4);
             return [$start, $start, "00", "00", "+", ""];
-
         }
+
+        // 1920
+        if (preg_match("/^[0-9][0-9][0-9]$/", $datum)) {
+            $start = "0" . substr($datum, 0, 3);
+            return [$start, $start, "00", "00", "+", ""];
+        }
+
         return [];
 
     }