Enable NodaTimeSplitter to split dates with uncertain end or start

(seit, bis)

src/NodaTimeSplitter.php

@@ -175,6 +175,16 @@ final class NodaTimeSplitter {
      */
     public static function timePartsToTimeName(array $moda):string {
 
+        if ($moda[0] === "?") {
+            $prefix = "Bis ";
+            $moda[0] = $moda[1];
+        }
+        else if ($moda[1] === "?") {
+            $prefix = "Seit ";
+            $moda[1] = $moda[0];
+        }
+        else $prefix = "";
+
         $moda[0] = intval($moda[0]);
         $moda[1] = intval($moda[1]);
 
@@ -190,17 +200,17 @@ final class NodaTimeSplitter {
         $moda[1] = abs($moda[1]);
 
         if ($moda[0] !== $moda[1]) {
-            return "{$moda[0]}-{$moda[1]}{$suffix}";
+            return "{$prefix}{$moda[0]}-{$moda[1]}{$suffix}";
         }
         else if (intval($moda[2]) !== 0 and intval($moda[3]) !== 0) {
-            return "{$moda[3]}.{$moda[2]}.{$moda[0]}{$suffix}";
+            return "{$prefix}{$moda[3]}.{$moda[2]}.{$moda[0]}{$suffix}";
         }
         else if ($moda[0] === $moda[1] && trim((string)$moda[2], " 0") === "" && trim((string)$moda[3], " 0") === "") {
-            return "{$moda[0]}{$suffix}";
+            return "{$prefix}{$moda[0]}{$suffix}";
         }
         else if ($moda[0] === $moda[1] && trim((string)$moda[2], " 0") !== "" && trim((string)$moda[3], " 0") === "") {
             setlocale(LC_TIME, NodaTimeAutotranslater::LANGS_TO_LOCALES['de']);
-            return strftime(getMonthFormatByLang("de"), MD_STD::strtotime("{$moda[0]}-{$moda[2]}-01 01:01:01")) . $suffix;
+            return $prefix . strftime(getMonthFormatByLang("de"), MD_STD::strtotime("{$moda[0]}-{$moda[2]}-01 01:01:01")) . $suffix;
         }
 
         return "";
@@ -216,6 +226,14 @@ final class NodaTimeSplitter {
      */
     public static function timePartsToCountingYear(array $moda):int {
 
+        if ($moda[0] === "?") {
+            return intval($moda[1]);
+        }
+
+        if ($moda[1] === "?") {
+            return intval($moda[0]);
+        }
+
         return abs((int)ceil(intval($moda[1]) - ((intval($moda[1]) - intval($moda[0])) / 2)));
 
     }
@@ -292,22 +310,22 @@ final class NodaTimeSplitter {
         if (preg_match("/^[0-9][0-9][0-9][0-9](\-|\/)[0-9][0-9][0-9][0-9] v\. Chr\.$/", $datum)) {
             $start = "-" . substr($datum, 0, 4);
             $end = "-" . substr($datum, 5, 4);
-            return [$start, $end, "00", "00", "-"];
+            return [$start, $end, "00", "00", "-", ""];
         }
         if (preg_match("/^[0-9][0-9][0-9][0-9](\-|\/)[0-9][0-9][0-9] v\. Chr\.$/", $datum)) {
             $start = "-" . substr($datum, 0, 4);
             $end = "-" . substr($datum, 5, 3);
-            return [$start, $end, "00", "00", "-"];
+            return [$start, $end, "00", "00", "-", ""];
         }
         if (preg_match("/^[0-9][0-9][0-9](\-|\/)[0-9][0-9][0-9] v\. Chr\.$/", $datum)) {
             $start = "-" . substr($datum, 0, 3);
             $end = "-" . substr($datum, 4, 3);
-            return [$start, $end, "00", "00", "-"];
+            return [$start, $end, "00", "00", "-", ""];
         }
         if (preg_match("/^[0-9][0-9](\-|\/)[0-9][0-9] v\. Chr\.$/", $datum)) {
             $start = "-00" . substr($datum, 0, 2);
             $end = "-00" . substr($datum, 3, 2);
-            return [$start, $end, "00", "00", "-"];
+            return [$start, $end, "00", "00", "-", ""];
         }
 
         if (self::stri_occurs($datum, self::STOP_STRINGS_GERMAN)) {
@@ -333,10 +351,10 @@ final class NodaTimeSplitter {
         if (self::is_numeric((string)substr($datum, -4))) $year = substr($datum, -4);
 
         if (!empty($year) and !empty($monat) and !empty($day) and $use_day) {
-            return [$year, $year, $monat, $day, '+'];
+            return [$year, $year, $monat, $day, '+', ""];
         }
         else if (!empty($year) and !empty($monat)) {
-            return [$year, $year, $monat, "00", '+'];
+            return [$year, $year, $monat, "00", '+', ""];
         }
         return [];
 
@@ -390,10 +408,10 @@ final class NodaTimeSplitter {
         if (self::is_numeric((string)substr($datum, 0, 4))) $year = substr($datum, 0, 4);
 
         if (!empty($year) and !empty($monat) and !empty($day)) {
-            return [$year, $year, $monat, $day, '+'];
+            return [$year, $year, $monat, $day, '+', ""];
         }
         else if (!empty($year) and !empty($monat)) {
-            return [$year, $year, $monat, "00", '+'];
+            return [$year, $year, $monat, "00", '+', ""];
         }
         return [];
 
@@ -414,7 +432,7 @@ final class NodaTimeSplitter {
             return [];
         }
 
-        return [date("Y", $timeInt), date("m", $timeInt), date("d", $timeInt), '+'];
+        return [date("Y", $timeInt), date("m", $timeInt), date("d", $timeInt), '+', ""];
 
     }
 
@@ -432,72 +450,105 @@ final class NodaTimeSplitter {
         if (preg_match("/^[0-9][0-9][0-9][0-9](\-|\/)[0-9][0-9][0-9][0-9]$/", $datum)) {
             $start = substr($datum, 0, 4);
             $end = substr($datum, -4);
-            return [$start, $end, "00", "00", "+"];
+            return [$start, $end, "00", "00", "+", ""];
         }
         if (preg_match("/^[0-9][0-9]\.[0-9]\.[0-9][0-9][0-9][0-9]$/", $datum)) { // German T.MM.JJJJ
             $start = substr($datum, 5, 4);
             $month = "0" . substr($datum, 3, 1);
             $day   = substr($datum, 0, 2);
-            return [$start, $start, $month, $day, "+"];
+            return [$start, $start, $month, $day, "+", ""];
         }
         if (preg_match("/^[0-9]\.[0-9][0-9]\.[0-9][0-9][0-9][0-9]$/", $datum)) { // German T.MM.JJJJ
             $start = substr($datum, 5, 4);
             $month = substr($datum, 2, 2);
             $day   = "0" . substr($datum, 0, 1);
-            return [$start, $start, $month, $day, "+"];
+            return [$start, $start, $month, $day, "+", ""];
         }
         if (preg_match("/^[0-9]\.[0-9]\.[0-9][0-9][0-9][0-9]$/", $datum)) { // German T.M.JJJJ
             $start = substr($datum, 4, 4);
             $month = "0" . substr($datum, 2, 1);
             $day   = "0" . substr($datum, 0, 1);
-            return [$start, $start, $month, $day, "+"];
+            return [$start, $start, $month, $day, "+", ""];
         }
         if (preg_match("/^[0-9][0-9]\.[0-9][0-9][0-9][0-9]$/", $datum)) { // German Y-m
             $start = substr($datum, 3, 4);
             $month = substr($datum, 0, 2);
-            return [$start, $start, $month, "00", "+"];
+            return [$start, $start, $month, "00", "+", ""];
         }
         if (preg_match("/^[0-9]\.[0-9][0-9][0-9][0-9]$/", $datum)) { // German Y-m
             $start = substr($datum, 2, 4);
             $month = "0" . substr($datum, 0, 1);
-            return [$start, $start, $month, "00", "+"];
+            return [$start, $start, $month, "00", "+", ""];
         }
         if (preg_match("/^[0-9][0-9][0-9][0-9]\.[0-9][0-9]$/", $datum)) { // Hungarian Y-m
             $start = substr($datum, 0, 4);
             $month = substr($datum, 5, 2);
-            return [$start, $start, $month, "00", "+"];
+            return [$start, $start, $month, "00", "+", ""];
         }
-        if (preg_match("/^[0-9][0-9][0-9][0-9]\-[0-9][0-9]$/", $datum)) { // Hungarian Y-m
+        if (preg_match("/^[0-9][0-9][0-9][0-9]\-[0-9][0-9]$/", $datum)) { // Time spans: 1945-46
             $start = substr($datum, 0, 4);
             $endDigits = substr($datum, 5, 2);
-            if (intval($endDigits) > 12) return [$start, substr($datum, 0, 2) . $endDigits, "00", "00", "+"];
+            if (intval($endDigits) > 12) return [$start, substr($datum, 0, 2) . $endDigits, "00", "00", "+", ""];
         }
         if (preg_match("/^01\.01\.[0-9][0-9][0-9][0-9]\-31\.12\.[0-9][0-9][0-9][0-9]$/", $datum)) { // Hungarian Y-m
             $start = substr($datum, 6, 4);
             $end = substr($datum, -4);
-            return [$start, $end, "00", "00", "+"];
+            return [$start, $end, "00", "00", "+", ""];
         }
         if (preg_match("/^[0-9][0-9][0-9]\-[0-9][0-9][0-9]$/", $datum)) { // Hungarian Y-m
             $start = substr($datum, 0, 3);
             $end = substr($datum, -3);
-            if ($end > $start) return ["0" . $start, "0" . $end, "00", "00", "+"];
+            if ($end > $start) return ["0" . $start, "0" . $end, "00", "00", "+", ""];
         }
-        if (preg_match("/^[0-9][0-9]\-[0-9][0-9]$/", $datum)) { // Hungarian Y-m
+        if (preg_match("/^[0-9][0-9]\-[0-9][0-9]$/", $datum)) { // 20-40 (n. Chr.)
             $start = substr($datum, 0, 2);
             $end = substr($datum, -2);
-            if ($end > $start) return ["00" . $start, "00" . $end, "00", "00", "+"];
+            if ($end > $start) return ["00" . $start, "00" . $end, "00", "00", "+", ""];
         }
 
         if (preg_match("/^[0-9][0-9][0-9][0-9]$/", $datum)) {
 
             $start = substr($datum, 0, 4);
-            return [$start, $start, "00", "00", "+"];
+            return [$start, $start, "00", "00", "+", ""];
 
         }
         return [];
 
     }
 
+    /**
+     * Checks if an input date is an incomplete date: Before 1920, after 1930.
+     *
+     * @param string $datum Input date.
+     *
+     * @return array<string>
+     */
+    public static function is_incomplete_date(string $datum):array {
+
+        if (preg_match("/^(Ab|Seit|seit)\ /", $datum)) {
+            if (($spacePos = strpos($datum, " ")) === false) {
+                return [];
+            }
+            if ($output = self::attempt_splitting(substr($datum, $spacePos))) {
+                $output[1] = "?";
+                return $output;
+            }
+        }
+
+        if (preg_match("/^(Bis|bis)\ /", $datum)) {
+            if (($spacePos = strpos($datum, " ")) === false) {
+                return [];
+            }
+            if ($output = self::attempt_splitting(substr($datum, $spacePos))) {
+                $output[0] = "?";
+                return $output;
+            }
+        }
+
+        return [];
+
+    }
+
     /**
      * Wrapper to check if any splitting command works.
      *
@@ -508,6 +559,7 @@ final class NodaTimeSplitter {
     public static function attempt_splitting(string $datum):array {
 
         $moda = NodaTimeSplitter::is_timespan($datum);
+        if (!$moda) $moda = NodaTimeSplitter::is_incomplete_date($datum);
         if (!$moda) $moda = NodaTimeSplitter::is_valid_date($datum);
         if (!$moda) $moda = NodaTimeSplitter::is_valid_date_hungarian($datum);