Extend NodaTimeSplitter

src/NodaTimeSplitter.php

@@ -42,7 +42,7 @@ final class NodaTimeSplitter {
     ];
 
     const MONTH_NAMES_HUNGARIAN = [
-        "01" => ['január', 'jan'],
+        "01" => ['január', 'januar', 'jan'],
         "02" => ['február', 'feb'],
         "03" => ['március', 'mar.', 'már.'],
         "04" => ['április', 'apr.', 'ápr.'],
@@ -58,6 +58,8 @@ final class NodaTimeSplitter {
 
     const STRINGS_TO_CLEAN = [
         "között" => "",
+        " рр." => "",
+        " рр" => "",
         "nach Christus" => "",
         "n. Christus" => "",
         "nach Chr." => "",
@@ -69,6 +71,7 @@ final class NodaTimeSplitter {
         "BCE" => "v. Chr.",
         "CE" => "",
         "vor Christus" => "v. Chr.",
+        " до н. е." => "v. Chr.",
     ];
 
     const STRINGS_TO_CLEAN_START = [
@@ -133,6 +136,7 @@ final class NodaTimeSplitter {
     ];
 
     private const REGEX_CENTURIES = '(\ |)(Jh\.|Jhd(|\.)|Jhdt(|\.)|Jahrhundert|sz|század|th century|ст|ст\.)';
+    private const REGEX_DECADES = '(s|er|er\ Jahre|(\-|\ )es\ évek|(\-|\ )as\ \évek|\ évek|\-es\ években|\-ті)';
 
     /**
      * Cleans input strings by trimming obsolete stuff.
@@ -521,20 +525,25 @@ final class NodaTimeSplitter {
         // Skip, if dates are too long and do not contain spaces (= no translatable names)
         if (str_contains($datum, " ") === false && strlen($datum) > 12) return [];
 
+        $unparsed = trim(strtolower(str_replace($year, '', $datum)), ' ,.');
         foreach (self::MONTH_NAMES_HUNGARIAN as $monthVal => $monthValidNames) {
             if (self::stri_occurs($datum, $monthValidNames)) {
                 $monat = (string)$monthVal;
+                foreach ($monthValidNames as $name) {
+                    $unparsed = str_replace($name, '', $unparsed);
+                }
                 break;
             }
         }
 
+        if (strlen($unparsed) > 5) {
+            return [];
+        }
+
         if (empty($monat) and self::is_numeric((string)\substr($datum, 5, 2))) $monat = \substr($datum, 5, 2);
         else if (empty($monat) and self::is_numeric((string)\substr($datum, 6, 2))) $monat = \substr($datum, 6, 2);
 
         // Last four characters must contain at least one space or one dot
-        $lastChars = substr($datum, -4);
-        if (str_contains($lastChars, '.') === false && str_contains($lastChars, ' ') === false)  return [];
-
         $day = self::validateDateSubstr($datum, -2);
         if (empty($day)) $day = self::validateDateSubstr($datum, -3, 2);
         if (empty($day)) $day = self::validateDateSubstr($datum, -4, 2);
@@ -550,6 +559,10 @@ final class NodaTimeSplitter {
             }
         }
 
+        if (!empty($monat) && empty($day) && preg_match('~[0-9]+~', substr($datum, -3))) {
+            return [];
+        }
+
         if (!empty($monat) and !empty($day)) {
             return [$year, $year, $monat, $day, '+', ""];
         }
@@ -851,14 +864,17 @@ final class NodaTimeSplitter {
             }
         }
 
-        if (\preg_match("/\ (\(vor|\(Vor|vor)$/", $datum)) {
-            if (($spacePos = \strpos($datum, " ")) === false) {
+        if (\preg_match("/\ (\(vor|\(Vor|vor|előtt)$/", $datum)) {
+            if (($spacePos = \strrpos($datum, " ")) === false) {
                 return [];
             }
             if ($output = self::attempt_splitting(\substr($datum, 0, $spacePos))) {
 
                 $output[0] = "?";
-                if (empty(trim($output[2], "0 .,"))) $output[1] = \strval(\intval($output[1]) - 1);
+                // If month and day are unknown, lower end year by one
+                if (empty(trim($output[2], "0 .,"))) {
+                    $output[1] = \strval(\intval($output[1]) - 1);
+                }
                 $output[5] = "Vor";
                 return $output;
             }
@@ -932,6 +948,18 @@ final class NodaTimeSplitter {
                 return $output;
             }
         }
+        if (!empty(\preg_match("/^[0-9]{4}ig$/", $datum))) {
+            if ($output = self::attempt_splitting(\substr($datum, 0, 4))) {
+                $output[0] = "?";
+                return $output;
+            }
+        }
+        if (str_ends_with($datum, '-as évekig') || str_ends_with($datum, '-es évekig')) {
+            if ($output = self::attempt_splitting(\substr($datum, 0, -2))) {
+                $output[0] = "?";
+                return $output;
+            }
+        }
 
         return [];
 
@@ -1035,14 +1063,14 @@ final class NodaTimeSplitter {
         $bcBceIndicator = '+';
 
         // 20er Jahre
-        if (\preg_match("/^[0-9]0(er|er\ Jahre|\-es\ évek|\-as\ \évek)$/", $datum)) {
+        if (\preg_match("/^[0-9]0" . self::REGEX_DECADES . "$/", $datum)) {
             $start = "19" . \substr($datum, 0, 2);
             $ende = (string)(\intval($start) + 9);
             return [$start, $ende, "00", "00", $bcBceIndicator, ""];
         }
 
         // 1920er Jahre
-        if (\preg_match("/^[0-9]{3}0(s|er|er\ Jahre|(\-|\ )es\ évek|(\-|\ )as\ \évek)$/", $datum)) {
+        if (\preg_match("/^[0-9]{3}0" . self::REGEX_DECADES . "$/", $datum)) {
             $start = \substr($datum, 0, 4);
             $ende = (string)(\intval($start) + 9);
             return [$start, $ende, "00", "00", $bcBceIndicator, ""];
@@ -1102,7 +1130,7 @@ final class NodaTimeSplitter {
 
         // Hungarian year and month until month
         // 2005.01.-02. => 2005.01.-2005.02.
-        if ($inputLength === 12 && (\preg_match("/^[0-9]{4}\.[0-1][0-9]\.\-[0-1][0-9]\.$/", $datum)) !== false) {
+        if ($inputLength === 12 && (\preg_match("/^[0-9]{4}\.[0-1][0-9]\.\-[0-1][0-9]\.$/", $datum))) {
             $reconstituted = substr($datum, 0, 8) . '-';
             $reconstituted .= substr($datum, 0, 4) . '.' . substr($datum, -3);
             return $reconstituted;
@@ -1111,7 +1139,7 @@ final class NodaTimeSplitter {
         // Hungarian year and month until month without a dot after the first YYYY-MM
         // 2005.01-02. => 2005.01.-2005.02.
 
-        if (in_array($inputLength, [10, 11], true) && (\preg_match("/^[0-9]{4}\.[0-1][0-9]\-[0-1][0-9](\.|)$/", $datum)) !== false) {
+        if (in_array($inputLength, [10, 11], true) && (\preg_match("/^[0-9]{4}\.[0-1][0-9]\-[0-1][0-9](\.|)$/", $datum))) {
             $reconstituted = substr($datum, 0, 7) . '.-';
             $reconstituted .= substr($datum, 0, 4) . '.' . substr(rtrim($datum, '.'), -2) . '.';
             return $reconstituted;
@@ -1120,7 +1148,7 @@ final class NodaTimeSplitter {
         // Hungarian year and month until month
         // 2005.01.01.-02.02. => 2005.01.01-2005.02.02.
         // 2005.01.01-02.02 => 2005.01.01-2005.02.02.
-        if ($inputLength >= 16 && $inputLength <= 18 && (\preg_match("/^[0-9]{4}\.[0-1][0-9]\.[0-3][0-9](\.|)\-[0-1][0-9]\.[0-3][0-9](\.|)$/", $datum)) !== false) {
+        if ($inputLength >= 16 && $inputLength <= 18 && (\preg_match("/^[0-9]{4}\.[0-1][0-9]\.[0-3][0-9](\.|)\-[0-1][0-9]\.[0-3][0-9](\.|)$/", $datum))) {
             $parts = explode('-', $datum);
             if (count($parts) !== 2) return '';
             $reconstituted = substr($datum, 0, 10) . '.-';
@@ -1129,7 +1157,7 @@ final class NodaTimeSplitter {
         }
 
         // Hungarian; without trailing dots: YYYY.MM.DD-DD
-        if ($inputLength >= 13 && $inputLength <= 15 && (\preg_match("/^[0-9]{4}\.[0-1][0-9]\.[0-3][0-9](\.|)\-[0-3][0-9](\.|)$/", $datum)) !== false) {
+        if ($inputLength >= 13 && $inputLength <= 15 && \preg_match("/^[0-9]{4}\.[0-1][0-9]\.[0-3][0-9](\.|)\-[0-3][0-9](\.|)$/", $datum)) {
             $parts = explode('-', $datum);
             if (count($parts) !== 2) return '';
             $reconstituted = substr($datum, 0, 10) . '.-';
@@ -1154,8 +1182,8 @@ final class NodaTimeSplitter {
 
         // If es évek / as évek is contained in the string (e.g. 1880-1990-es évek), there
         // will be more than one hyphens
-        if (MD_STD::stri_contains_any($datum, ['-as évek', '-es évek'])) {
-            return strtr($datum, ['-as évek' => ' as évek', '-es évek' => ' es évek']);
+        if (MD_STD::stri_contains_any($datum, ['-as évek', '-es-évek', '-es évek'])) {
+            return strtr($datum, ['-as évek' => ' as évek', '-es-évek' => ' es évek', '-es évek' => ' es évek']);
         }
 
         // 1981. július-augusztus > 1981.07-08

tests/NodaTimeSplitterTest.php

@@ -500,6 +500,18 @@ final class NodaTimeSplitterTest extends TestCase {
         self::assertEquals(NodaTimeSplitter::timePartsToTimeName($output), "02.01.2020");
         self::assertEquals(NodaTimeSplitter::timePartsToCountingYear($output), 2020);
 
+        $output = NodaTimeSplitter::attempt_splitting("2020. januar. 2.");
+        self::assertEquals($output, [
+            0 => "2020",
+            1 => "2020",
+            2 => "01",
+            3 => "02",
+            4 => "+",
+            5 => "",
+        ]);
+        self::assertEquals(NodaTimeSplitter::timePartsToTimeName($output), "02.01.2020");
+        self::assertEquals(NodaTimeSplitter::timePartsToCountingYear($output), 2020);
+
         $output = NodaTimeSplitter::attempt_splitting("2020.01.2.");
         self::assertEquals($output, [
             0 => "2020",
@@ -609,15 +621,27 @@ final class NodaTimeSplitterTest extends TestCase {
         self::assertEquals(NodaTimeSplitter::timePartsToTimeName($output), "1920-1929");
         self::assertEquals(NodaTimeSplitter::timePartsToCountingYear($output), 1925);
 
-        $output = NodaTimeSplitter::attempt_splitting("2020. Januar");
+        $output = NodaTimeSplitter::attempt_splitting("1920-es évekig");
         self::assertEquals($output, [
+            0 => "?",
+            1 => "1929",
+            2 => "00",
+            3 => "00",
+            4 => "+",
+            5 => "",
+        ]);
+        self::assertEquals(NodaTimeSplitter::timePartsToTimeName($output), "Bis 1929");
+        self::assertEquals(NodaTimeSplitter::timePartsToCountingYear($output), 1929);
+
+        $output = NodaTimeSplitter::attempt_splitting("2020. Januar");
+        self::assertEquals([
             0 => "2020",
             1 => "2020",
             2 => "01",
             3 => "00",
             4 => "+",
             5 => "",
-        ]);
+        ], $output);
         self::assertEquals(NodaTimeSplitter::timePartsToTimeName($output), "Januar 2020");
         self::assertEquals(NodaTimeSplitter::timePartsToCountingYear($output), 2020);
 
@@ -729,6 +753,57 @@ final class NodaTimeSplitterTest extends TestCase {
         self::assertEquals(NodaTimeSplitter::timePartsToTimeName($output), "Bis 1801 v. Chr.");
         self::assertEquals(NodaTimeSplitter::timePartsToCountingYear($output), 1801);
 
+        /*
+         * TODO
+        $output = NodaTimeSplitter::attempt_splitting("1900 előtt");
+        self::assertEquals($output, [
+            0 => "?",
+            1 => "1899",
+            2 => "00",
+            3 => "00",
+            4 => "+",
+            5 => "",
+        ]);
+        self::assertEquals(NodaTimeSplitter::timePartsToTimeName($output), "Vor 1900");
+        self::assertEquals(NodaTimeSplitter::timePartsToCountingYear($output), 1899);
+         */
+
+        $output = NodaTimeSplitter::attempt_splitting("1900-ig");
+        self::assertEquals($output, [
+            0 => "?",
+            1 => "1900",
+            2 => "00",
+            3 => "00",
+            4 => "+",
+            5 => "",
+        ]);
+        self::assertEquals(NodaTimeSplitter::timePartsToTimeName($output), "Bis 1900");
+        self::assertEquals(NodaTimeSplitter::timePartsToCountingYear($output), 1900);
+
+        $output = NodaTimeSplitter::attempt_splitting("1900ig");
+        self::assertEquals($output, [
+            0 => "?",
+            1 => "1900",
+            2 => "00",
+            3 => "00",
+            4 => "+",
+            5 => "",
+        ]);
+        self::assertEquals(NodaTimeSplitter::timePartsToTimeName($output), "Bis 1900");
+        self::assertEquals(NodaTimeSplitter::timePartsToCountingYear($output), 1900);
+
+        $output = NodaTimeSplitter::attempt_splitting("2020. december előtt");
+        self::assertEquals($output, [
+            0 => "?",
+            1 => "2020",
+            2 => "12",
+            3 => "00",
+            4 => "+",
+            5 => "Vor",
+        ]);
+        self::assertEquals(NodaTimeSplitter::timePartsToTimeName($output), "Vor Dezember 2020");
+        self::assertEquals(NodaTimeSplitter::timePartsToCountingYear($output), 2020);
+
     }
 
     /**
@@ -881,6 +956,9 @@ final class NodaTimeSplitterTest extends TestCase {
         $output = NodaTimeSplitter::attempt_splitting("1978. július7");
         self::assertEmpty($output);
 
+        $output = NodaTimeSplitter::attempt_splitting("1978. július 7 elött");
+        self::assertEmpty($output);
+
         $output = NodaTimeSplitter::attempt_splitting("Anfang September 1903");
         self::assertEmpty($output);