From 8a4a8f7ed8489f68eb6871210746d3e84d539d0c Mon Sep 17 00:00:00 2001
From: Joshua Ramon Enslin <joshua@museum-digital.de>
Date: Sun, 4 Oct 2020 23:20:58 +0200
Subject: [PATCH] Split more variations of dots in dates, century ranges

---
 src/NodaTimeSplitter.php | 35 ++++++++++++++++++++++++-----------
 1 file changed, 24 insertions(+), 11 deletions(-)

diff --git a/src/NodaTimeSplitter.php b/src/NodaTimeSplitter.php
index e9f2aeb..a87bd57 100644
--- a/src/NodaTimeSplitter.php
+++ b/src/NodaTimeSplitter.php
@@ -657,13 +657,13 @@ final class NodaTimeSplitter {
 
         $datum = self::clean_input($datum);
 
-        if (preg_match("/^[0-9]{4}\.[0-9]{2}\.[0-9]{2}\-$/", $datum)) { // Hungarian Y-m
+        if (preg_match("/^[0-9]{4}\.[0-9]{2}\.[0-9]{2}(\.|)\-$/", $datum)) { // Hungarian Y-m
             $start = substr($datum, 0, 4);
             $month = substr($datum, 5, 2);
             $day = substr($datum, 8, 2);
             return [$start, "?", $month, $day, "+", ""];
         }
-        if (preg_match("/^[0-9]{4}\.[0-9]{2}\-$/", $datum)) { // Hungarian Y-m
+        if (preg_match("/^[0-9]{4}\.[0-9]{2}(\.|)\-$/", $datum)) { // Hungarian Y-m
             $start = substr($datum, 0, 4);
             $month = substr($datum, 5, 2);
             return [$start, "?", $month, "00", "+", ""];
@@ -707,7 +707,7 @@ final class NodaTimeSplitter {
             }
         }
 
-        if (preg_match("/^[0-9]{4}\ (\(nach)/", $datum)) {
+        if (preg_match("/\ (\(nach|nach)$/", $datum)) {
             if (($spacePos = strpos($datum, " ")) === false) {
                 return [];
             }
@@ -738,6 +738,19 @@ final class NodaTimeSplitter {
             }
         }
 
+        if (preg_match("/\ (\(vor|Vor)$/", $datum)) {
+            if (($spacePos = strpos($datum, " ")) === false) {
+                return [];
+            }
+            if ($output = self::attempt_splitting(substr($datum, 0, $spacePos))) {
+
+                $output[0] = "?";
+                if (empty(trim($output[2], "0 .,"))) $output[1] = strval((intval($output[1]) - 1));
+                $output[5] = "Vor";
+                return $output;
+            }
+        }
+
         if (preg_match("/^(Ab|Seit|seit)\ /", $datum)) {
             if (($spacePos = strpos($datum, " ")) === false) {
                 return [];
@@ -843,12 +856,10 @@ final class NodaTimeSplitter {
         }
 
         // 17.-18. Jahrhundert
-        if (preg_match("/^[0-9]{2}\.\-[0-9]{2}\.\ (Jh\.|Jahrhundert|sz|század)$/", $datum)) {
-            return self::negotiate_century_span_bce_ce(substr($datum, 0, 2), substr($datum, 4, 2));
-        }
-        // 17-18. Jahrhundert
-        if (preg_match("/^[0-9]{2}\-[0-9]{2}\.\ (Jh\.|Jahrhundert|sz|század)$/", $datum)) {
-            return self::negotiate_century_span_bce_ce(substr($datum, 0, 2), substr($datum, 3, 2));
+        if (preg_match("/^[0-9]{2}(\.|)(|\ Jh\.||\ Jahrhundert||\ sz||\ század)\-[0-9]{2}\.\ (Jh\.|Jahrhundert|sz|század)$/", $datum)) {
+            if (($dashPos = strpos($datum, "-")) !== false) {
+                return self::negotiate_century_span_bce_ce(substr($datum, 0, 2), substr($datum, $dashPos + 1, 2));
+            }
         }
 
         // 1.-12. Jahrhundert
@@ -860,8 +871,10 @@ final class NodaTimeSplitter {
             return self::negotiate_century_span_bce_ce(substr($datum, 0, 1), substr($datum, 3, 1));
         }
         // 1-2. Jahrhundert
-        if (preg_match("/^[0-9]\-[0-9]\.\ (Jh\.|Jahrhundert|sz|század)$/", $datum)) {
-            return self::negotiate_century_span_bce_ce(substr($datum, 0, 1), substr($datum, 2, 1));
+        if (preg_match("/^[0-9](\.|)(|\ Jh\.||\ Jahrhundert||\ sz||\ század)\-[0-9]\.\ (Jh\.|Jahrhundert|sz|század)$/", $datum)) {
+            if (($dashPos = strpos($datum, "-")) !== false) {
+                return self::negotiate_century_span_bce_ce(substr($datum, 0, 1), substr($datum, $dashPos + 1, 1));
+            }
         }
 
         return [];