From ed0fc0ec46466eb25e77cfefbf7f6d05f6f902b8 Mon Sep 17 00:00:00 2001
From: Giteabot <teabot@gitea.io>
Date: Mon, 17 Jun 2024 15:41:47 +0800
Subject: [PATCH] Fix natural sort (#31384) (#31394)

Backport #31384 by wxiaoguang

Fix #31374

Co-authored-by: wxiaoguang <wxiaoguang@gmail.com>
---
 modules/base/natural_sort.go      | 57 ++++++++++++++++++++++++++++++-
 modules/base/natural_sort_test.go | 43 +++++++++++++++--------
 2 files changed, 85 insertions(+), 15 deletions(-)

diff --git a/modules/base/natural_sort.go b/modules/base/natural_sort.go
index 0f90ec70ce..acb9002276 100644
--- a/modules/base/natural_sort.go
+++ b/modules/base/natural_sort.go
@@ -4,12 +4,67 @@
 package base
 
 import (
+	"unicode/utf8"
+
 	"golang.org/x/text/collate"
 	"golang.org/x/text/language"
 )
 
+func naturalSortGetRune(str string, pos int) (r rune, size int, has bool) {
+	if pos >= len(str) {
+		return 0, 0, false
+	}
+	r, size = utf8.DecodeRuneInString(str[pos:])
+	if r == utf8.RuneError {
+		r, size = rune(str[pos]), 1 // if invalid input, treat it as a single byte ascii
+	}
+	return r, size, true
+}
+
+func naturalSortAdvance(str string, pos int) (end int, isNumber bool) {
+	end = pos
+	for {
+		r, size, has := naturalSortGetRune(str, end)
+		if !has {
+			break
+		}
+		isCurRuneNum := '0' <= r && r <= '9'
+		if end == pos {
+			isNumber = isCurRuneNum
+			end += size
+		} else if isCurRuneNum == isNumber {
+			end += size
+		} else {
+			break
+		}
+	}
+	return end, isNumber
+}
+
 // NaturalSortLess compares two strings so that they could be sorted in natural order
 func NaturalSortLess(s1, s2 string) bool {
+	// There is a bug in Golang's collate package: https://github.com/golang/go/issues/67997
+	// text/collate: CompareString(collate.Numeric) returns wrong result for "0.0" vs "1.0" #67997
+	// So we need to handle the number parts by ourselves
 	c := collate.New(language.English, collate.Numeric)
-	return c.CompareString(s1, s2) < 0
+	pos1, pos2 := 0, 0
+	for pos1 < len(s1) && pos2 < len(s2) {
+		end1, isNum1 := naturalSortAdvance(s1, pos1)
+		end2, isNum2 := naturalSortAdvance(s2, pos2)
+		part1, part2 := s1[pos1:end1], s2[pos2:end2]
+		if isNum1 && isNum2 {
+			if part1 != part2 {
+				if len(part1) != len(part2) {
+					return len(part1) < len(part2)
+				}
+				return part1 < part2
+			}
+		} else {
+			if cmp := c.CompareString(part1, part2); cmp != 0 {
+				return cmp < 0
+			}
+		}
+		pos1, pos2 = end1, end2
+	}
+	return len(s1) < len(s2)
 }
diff --git a/modules/base/natural_sort_test.go b/modules/base/natural_sort_test.go
index f27a4eb53a..b001bc4ac9 100644
--- a/modules/base/natural_sort_test.go
+++ b/modules/base/natural_sort_test.go
@@ -10,21 +10,36 @@ import (
 )
 
 func TestNaturalSortLess(t *testing.T) {
-	test := func(s1, s2 string, less bool) {
-		assert.Equal(t, less, NaturalSortLess(s1, s2), "s1=%q, s2=%q", s1, s2)
+	testLess := func(s1, s2 string) {
+		assert.True(t, NaturalSortLess(s1, s2), "s1<s2 should be true: s1=%q, s2=%q", s1, s2)
+		assert.False(t, NaturalSortLess(s2, s1), "s2<s1 should be false: s1=%q, s2=%q", s1, s2)
+	}
+	testEqual := func(s1, s2 string) {
+		assert.False(t, NaturalSortLess(s1, s2), "s1<s2 should be false: s1=%q, s2=%q", s1, s2)
+		assert.False(t, NaturalSortLess(s2, s1), "s2<s1 should be false: s1=%q, s2=%q", s1, s2)
 	}
-	test("v1.20.0", "v1.2.0", false)
-	test("v1.20.0", "v1.29.0", true)
-	test("v1.20.0", "v1.20.0", false)
-	test("abc", "bcd", true)
-	test("a-1-a", "a-1-b", true)
-	test("2", "12", true)
-	test("a", "ab", true)
 
-	test("A", "b", true)
-	test("a", "B", true)
+	testEqual("", "")
+	testLess("", "a")
+	testLess("", "1")
 
-	test("cafe", "café", true)
-	test("café", "cafe", false)
-	test("caff", "café", false)
+	testLess("v1.2", "v1.2.0")
+	testLess("v1.2.0", "v1.10.0")
+	testLess("v1.20.0", "v1.29.0")
+	testEqual("v1.20.0", "v1.20.0")
+
+	testLess("a", "A")
+	testLess("a", "B")
+	testLess("A", "b")
+	testLess("A", "ab")
+
+	testLess("abc", "bcd")
+	testLess("a-1-a", "a-1-b")
+	testLess("2", "12")
+
+	testLess("cafe", "café")
+	testLess("café", "caff")
+
+	testLess("A-2", "A-11")
+	testLess("0.txt", "1.txt")
 }