From 395c5fcb098376f5a6267283b9e87e4f85170471 Mon Sep 17 00:00:00 2001
From: nova <email@nova.gay>
Date: Sun, 21 Sep 2025 21:30:15 +0200
Subject: [PATCH] improvements to sort_natural

---
 sorting.c | 112 ++++++++++++++++++++++++++++--------------------------
 1 file changed, 58 insertions(+), 54 deletions(-)

diff --git a/sorting.c b/sorting.c
index 32f9fc8..e7aae3e 100644
--- a/sorting.c
+++ b/sorting.c
@@ -24,75 +24,79 @@ int skip_dot(const struct dirent *entry){
 	return 1;
 }
 
-int sort_natural(const void *file0_, const void *file1_){
-	file *file0 = (file*)file0_;
-	file *file1 = (file*)file1_;
+int sort_natural(const void *file0, const void *file1){
 
-	unsigned char *a = (unsigned char*)file0->file_name;
-	unsigned char *b = (unsigned char*)file1->file_name;
-
-	if (file0->file_type & (FILE_TYPE_SYMLINK | FILE_TYPE_DIR) && !(file1->file_type & (FILE_TYPE_SYMLINK | FILE_TYPE_DIR))) {
+	if (((file*)file0)->file_type & (FILE_TYPE_SYMLINK | FILE_TYPE_DIR) && !(((file*)file1)->file_type & (FILE_TYPE_SYMLINK | FILE_TYPE_DIR))) {
 			return -1;
 	}
-	if (!(file0->file_type & (FILE_TYPE_SYMLINK | FILE_TYPE_DIR)) && file1->file_type & (FILE_TYPE_SYMLINK | FILE_TYPE_DIR)) {
+	if (!(((file*)file0)->file_type & (FILE_TYPE_SYMLINK | FILE_TYPE_DIR)) && ((file*)file1)->file_type & (FILE_TYPE_SYMLINK | FILE_TYPE_DIR)) {
 			return 1;
 	}
-	unsigned long num0 = 0;
-	unsigned long num1 = 0;
-	/* bitwise OR a with ' ' turns turns caps into small letters
-	 * while doing this on all chars may cause unexpected behaviour on the extended ascii set,
-	 * for now i dont care */
-	while ((*a | ' ') == (*b | ' ') && *a != '\0') {
+	const unsigned char *a = (unsigned char*)((file*)file0)->file_name;
+	const unsigned char *b = (unsigned char*)((file*)file1)->file_name;
+
+	long parsed_number0 = 0;
+	long parsed_number1 = 0;
+	char is_num = 0;
+	char result = 0;
+
+	do {
+		is_num = 0;
+
 		if ((*a >= '0') && (*a <= '9')) {
+			parsed_number0 = 0;
 			while((*a >= '0') && (*a <= '9')) {
-				num0 = (num0 * 10) + (*a - '0');
+				parsed_number0 = (parsed_number0 * 10) + (*a - '0');
 				a++;
 			}
+			is_num |= 1;
+		}
+		if ((*b >= '0') && (*b <= '9')) {
+			parsed_number1 = 0;
 			while((*b >= '0') && (*b <= '9')) {
-				num1 = (num1 * 10) + (*b - '0');
+				parsed_number1 = (parsed_number1 * 10) + (*b - '0');
 				b++;
 			}
-			if (num0 != num1) {
-				break;
-			}
-		} else {
-			a++;
-			b++;
-		}
-	}
-	if (num0 == num1) {
-		if (*a == '\0') {
-			a--;
-		}
-		if (*b == '\0') {
-			b--;
+			is_num |= 2;
 		}
 
-		unsigned char c0;
-		unsigned char c1;
-		/* in this case we actually check for a through z as otherwise unicode characters get ordered wrongly */
-		if (*a >= 'A' && *a <= 'Z') {
-			c0 = (*a | ' ');
-		} else {
-			c0 = *a;
+		if (is_num) {
+			if (is_num == 1) {
+				if (*b < '0') {
+					result = 1;
+				} else {
+					result = -1;
+				}
+				break;
+			} else if (is_num == 2) {
+				if (*a < '0') {
+					result = -1;
+				} else {
+					result = 1;
+				}
+				break;
+			} else {
+				if (parsed_number0 > parsed_number1) {
+					result = 1;
+					break;
+				} else if (parsed_number0 < parsed_number1) {
+					result = -1;
+					break;
+				}
+			}
+			/* those breaks are not set here, due to the possibillity that both numbers are equal
+			 * in which case the comparison should continue */
 		}
-		if (*b >= 'A' && *b <= 'Z') {
-			c1 = (*b | ' ');
-		} else {
-			c1 = *b;
-		}
-		if (c0 > c1) {
-			return 1;
-		} else if (c0 < c1) {
-			return -1;
-		} else {
-			return 0;
-		}
-	} else if (num0 > num1) {
-		return 1;
-	} else {
-		return -1;
-	}
+
+		unsigned char aa = ((*a >= 'A') && (*a <= 'Z')) ? (*a | ' ') : *a;
+		unsigned char bb = ((*b >= 'A') && (*b <= 'Z')) ? (*b | ' ') : *b;
+		/*using a simple aa - bb would occasionaly cause underflows with wide chars*/
+		result = ((aa == bb) ? 0 : ((aa > bb) ? 1 : -1 ));
+		a++;
+		b++;
+	} while (result == 0);
+
+	return result;
 }
 int sort_alpha(const void *file0, const void *file1){
 	char *file_name0 = ((file*)file0)->file_name;