From f3831965a3c4f3a8478fd037b3b6700227b647e1 Mon Sep 17 00:00:00 2001
From: Joshua Ramon Enslin <joshua@museum-digital.de>
Date: Sat, 9 Dec 2023 23:54:45 +0100
Subject: [PATCH] Extend NodaTimeSplitter

---
 src/NodaTimeSplitter.php       | 27 ++++++++++++++++---------
 tests/NodaTimeSplitterTest.php | 36 ++++++++++++++++++++++++++++++++++
 2 files changed, 54 insertions(+), 9 deletions(-)

diff --git a/src/NodaTimeSplitter.php b/src/NodaTimeSplitter.php
index 09628a4..d48b8ed 100644
--- a/src/NodaTimeSplitter.php
+++ b/src/NodaTimeSplitter.php
@@ -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.
@@ -555,10 +559,6 @@ final class NodaTimeSplitter {
             }
         }
 
-        if ($datum === '1978. július 7 elött') {
-            throw new Exception(var_export($monat, true));
-        }
-
         if (!empty($monat) && empty($day) && preg_match('~[0-9]+~', substr($datum, -3))) {
             return [];
         }
@@ -864,14 +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;
             }
@@ -951,6 +954,12 @@ final class NodaTimeSplitter {
                 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 [];
 
@@ -1054,14 +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", $bcBceIndicator, ""];
diff --git a/tests/NodaTimeSplitterTest.php b/tests/NodaTimeSplitterTest.php
index 1ae5577..a9b4583 100644
--- a/tests/NodaTimeSplitterTest.php
+++ b/tests/NodaTimeSplitterTest.php
@@ -500,6 +500,18 @@ final class NodaTimeSplitterTest extends TestCase {
         self::assertEquals(NodaTimeSplitter::timePartsToTimeName($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",
@@ -609,6 +621,18 @@ final class NodaTimeSplitterTest extends TestCase {
         self::assertEquals(NodaTimeSplitter::timePartsToTimeName($output), "1920-1929");
         self::assertEquals(NodaTimeSplitter::timePartsToCountingYear($output), 1925);
 
+        $output = NodaTimeSplitter::attempt_splitting("1920-es évekig");
+        self::assertEquals($output, [
+            0 => "?",
+            1 => "1929",
+            2 => "00",
+            3 => "00",
+            4 => "+",
+            5 => "",
+        ]);
+        self::assertEquals(NodaTimeSplitter::timePartsToTimeName($output), "Bis 1929");
+        self::assertEquals(NodaTimeSplitter::timePartsToCountingYear($output), 1929);
+
         $output = NodaTimeSplitter::attempt_splitting("2020. Januar");
         self::assertEquals([
             0 => "2020",
@@ -768,6 +792,18 @@ final class NodaTimeSplitterTest extends TestCase {
         self::assertEquals(NodaTimeSplitter::timePartsToTimeName($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);
+
     }
 
     /**