From c8d0292ca88fc2bc1124188812b822f8148f991a Mon Sep 17 00:00:00 2001
From: Joshua Ramon Enslin <joshua@museum-digital.de>
Date: Thu, 8 Sep 2022 16:28:27 +0200
Subject: [PATCH] Fix bug in German dates like "1 November 1921"

---
 src/NodaTimeSplitter.php       | 39 ++++++++++++++++++++++++----------
 tests/NodaTimeSplitterTest.php | 15 +++++++++++++
 2 files changed, 43 insertions(+), 11 deletions(-)

diff --git a/src/NodaTimeSplitter.php b/src/NodaTimeSplitter.php
index 03487b7..3ef2e97 100644
--- a/src/NodaTimeSplitter.php
+++ b/src/NodaTimeSplitter.php
@@ -428,6 +428,10 @@ final class NodaTimeSplitter {
         if (strlen($datum) <= 9) $use_day = false;
         else $use_day = true;
 
+        if (self::is_numeric((string)\substr($datum, -4))) $year = \substr($datum, -4);
+        // Further code requires a year to be present, skip if none is set
+        if (empty($year)) return [];
+
         foreach (self::MONTH_NAMES_ENGLISH as $monthVal => $monthValidNames) {
             if (self::stri_occurs($datum, $monthValidNames)) {
                 if (!empty($monat)) break;
@@ -444,10 +448,12 @@ final class NodaTimeSplitter {
 
         if (empty($monat) and self::is_numeric((string)\substr($datum, 3, 2))) $monat = \substr($datum, 3, 2);
 
-        if (self::is_numeric((string)\substr($datum, 0, 2))) $day = \substr($datum, 0, 2);
-        else if (\substr($datum, 1, 1) === "." and self::is_numeric((string)\substr($datum, 0, 1))) $day = "0" . \substr($datum, 0, 1);
-
-        if (self::is_numeric((string)\substr($datum, -4))) $year = \substr($datum, -4);
+        if (self::is_numeric((string)\substr($datum, 0, 2))) {
+            $day = \substr($datum, 0, 2);
+        }
+        else if (\in_array(\substr($datum, 1, 1), [".", " "]) && self::is_numeric((string)\substr($datum, 0, 1))) {
+            $day = "0" . \substr($datum, 0, 1);
+        }
 
         if (!empty($year) and !empty($monat) and !empty($day) and $use_day) {
             return [$year, $year, $monat, $day, '+', ""];
@@ -493,8 +499,19 @@ final class NodaTimeSplitter {
             return [];
         }
 
+        //
+        // Rest: Only those entries, where there are spelled out months
+        //
         if (strlen($datum) <= 9) return [];
 
+        // The year is only parse-able if it is a four digit year at the start
+        if (self::is_numeric((string)\substr($datum, 0, 4))) {
+            $year = \substr($datum, 0, 4);
+        }
+
+        // Further code requires a year to be present, skip if none is set
+        if (empty($year)) return [];
+
         foreach (self::MONTH_NAMES_HUNGARIAN as $monthVal => $monthValidNames) {
             if (self::stri_occurs($datum, $monthValidNames)) {
                 if (!empty($monat)) return [];
@@ -511,14 +528,14 @@ final class NodaTimeSplitter {
         if (empty($day)) $day = self::validateDateSubstr($datum, -5, 2);
         if (empty($day)) $day = self::validateDateSubstr($datum, -6, 2);
 
-        if (\substr($datum, -2, 1) === " " and self::is_numeric((string)\substr($datum, -1, 1))) {
-            $day = "0" . \substr($datum, -1, 1);
+        if (empty($day)) {
+            if (\substr($datum, -2, 1) === " " and self::is_numeric((string)\substr($datum, -1, 1))) {
+                $day = "0" . \substr($datum, -1, 1);
+            }
+            else if (\substr($datum, -3, 1) === " " and self::is_numeric((string)\substr($datum, -2, 1))) {
+                $day = "0" . \substr($datum, -2, 1);
+            }
         }
-        else if (empty($day) and \substr($datum, -3, 1) === " " and self::is_numeric((string)\substr($datum, -2, 1))) {
-            $day = "0" . \substr($datum, -2, 1);
-        }
-
-        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, $day, '+', ""];
diff --git a/tests/NodaTimeSplitterTest.php b/tests/NodaTimeSplitterTest.php
index 826658c..e2d1a8a 100644
--- a/tests/NodaTimeSplitterTest.php
+++ b/tests/NodaTimeSplitterTest.php
@@ -374,6 +374,18 @@ final class NodaTimeSplitterTest extends TestCase {
         self::assertEquals(NodaTimeSplitter::timePartsToTimeName($output), "Bis 1801 v. Chr.");
         self::assertEquals(NodaTimeSplitter::timePartsToCountingYear($output), 1801);
 
+        $output = NodaTimeSplitter::attempt_splitting("6 November 1978");
+        self::assertEquals($output, [
+            0 => "1978",
+            1 => "1978",
+            2 => "11",
+            3 => "06",
+            4 => "+",
+            5 => "",
+        ]);
+        self::assertEquals(NodaTimeSplitter::timePartsToTimeName($output), "06.11.1978");
+        self::assertEquals(NodaTimeSplitter::timePartsToCountingYear($output), 1978);
+
     }
 
     /**
@@ -603,6 +615,9 @@ final class NodaTimeSplitterTest extends TestCase {
         $output = NodaTimeSplitter::attempt_splitting("2022-10-40");
         self::assertEquals($output, []);
 
+        $output = NodaTimeSplitter::attempt_splitting("6;November 1978");
+        self::assertEquals($output, []);
+
         # $output = NodaTimeSplitter::attempt_splitting("Nach 1944-1964");
         # self::assertEquals($output, []);