From bf2867dec2390e95c0735abaa4d7142fe1eb8d75 Mon Sep 17 00:00:00 2001
From: Gusted <williamzijl7@hotmail.com>
Date: Sat, 26 Feb 2022 16:48:23 +0000
Subject: [PATCH] Don't treat BOM escape sequence as hidden character. (#18909)

* Don't treat BOM escape sequence as hidden character.

- BOM sequence is a common non-harmfull escape sequence, it shouldn't be
shown as hidden character.
- Follows GitHub's behavior.
- Resolves #18837

Co-authored-by: wxiaoguang <wxiaoguang@gmail.com>
---
 modules/charset/escape.go      |  5 +++++
 modules/charset/escape_test.go | 24 ++++++++++++++++++++----
 2 files changed, 25 insertions(+), 4 deletions(-)

diff --git a/modules/charset/escape.go b/modules/charset/escape.go
index d2e8fb0d8..20a4bb2a1 100644
--- a/modules/charset/escape.go
+++ b/modules/charset/escape.go
@@ -63,6 +63,7 @@ func EscapeControlBytes(text []byte) (EscapeStatus, []byte) {
 func EscapeControlReader(text io.Reader, output io.Writer) (escaped EscapeStatus, err error) {
 	buf := make([]byte, 4096)
 	readStart := 0
+	runeCount := 0
 	var n int
 	var writePos int
 
@@ -79,6 +80,8 @@ readingloop:
 
 		for i < len(bs) {
 			r, size := utf8.DecodeRune(bs[i:])
+			runeCount++
+
 			// Now handle the codepoints
 			switch {
 			case r == utf8.RuneError:
@@ -113,6 +116,8 @@ readingloop:
 				lineHasRTLScript = false
 				lineHasLTRScript = false
 
+			case runeCount == 1 && r == 0xFEFF: // UTF BOM
+				// the first BOM is safe
 			case r == '\r' || r == '\t' || r == ' ':
 				// These are acceptable control characters and space characters
 			case unicode.IsSpace(r):
diff --git a/modules/charset/escape_test.go b/modules/charset/escape_test.go
index 180438141..01ccca772 100644
--- a/modules/charset/escape_test.go
+++ b/modules/charset/escape_test.go
@@ -129,6 +129,14 @@ then resh (ר), and finally heh (ה) (which should appear leftmost).`,
 			"\n" + `if access_level != "user<span class="escaped-code-point" data-escaped="[U+202E]"><span class="char">` + "\u202e" + `</span></span> <span class="escaped-code-point" data-escaped="[U+2066]"><span class="char">` + "\u2066" + `</span></span>// Check if admin<span class="escaped-code-point" data-escaped="[U+2069]"><span class="char">` + "\u2069" + `</span></span> <span class="escaped-code-point" data-escaped="[U+2066]"><span class="char">` + "\u2066" + `</span></span>" {` + "\n",
 		status: EscapeStatus{Escaped: true, HasBIDI: true, BadBIDI: true, HasLTRScript: true, HasRTLScript: true},
 	},
+	{
+		// UTF-8/16/32 all use the same codepoint for BOM
+		// Gitea could read UTF-16/32 content and convert into UTF-8 internally then render it, so we only process UTF-8 internally
+		name:   "UTF BOM",
+		text:   "\xef\xbb\xbftest",
+		result: "\xef\xbb\xbftest",
+		status: EscapeStatus{HasLTRScript: true},
+	},
 }
 
 func TestEscapeControlString(t *testing.T) {
@@ -163,10 +171,18 @@ func TestEscapeControlReader(t *testing.T) {
 	// lets add some control characters to the tests
 	tests := make([]escapeControlTest, 0, len(escapeControlTests)*3)
 	copy(tests, escapeControlTests)
+
+	// if there is a BOM, we should keep the BOM
+	addPrefix := func(prefix, s string) string {
+		if strings.HasPrefix(s, "\xef\xbb\xbf") {
+			return s[:3] + prefix + s[3:]
+		}
+		return prefix + s
+	}
 	for _, test := range escapeControlTests {
 		test.name += " (+Control)"
-		test.text = "\u001E" + test.text
-		test.result = `<span class="escaped-code-point" data-escaped="[U+001E]"><span class="char">` + "\u001e" + `</span></span>` + test.result
+		test.text = addPrefix("\u001E", test.text)
+		test.result = addPrefix(`<span class="escaped-code-point" data-escaped="[U+001E]"><span class="char">`+"\u001e"+`</span></span>`, test.result)
 		test.status.Escaped = true
 		test.status.HasControls = true
 		tests = append(tests, test)
@@ -174,8 +190,8 @@ func TestEscapeControlReader(t *testing.T) {
 
 	for _, test := range escapeControlTests {
 		test.name += " (+Mark)"
-		test.text = "\u0300" + test.text
-		test.result = `<span class="escaped-code-point" data-escaped="[U+0300]"><span class="char">` + "\u0300" + `</span></span>` + test.result
+		test.text = addPrefix("\u0300", test.text)
+		test.result = addPrefix(`<span class="escaped-code-point" data-escaped="[U+0300]"><span class="char">`+"\u0300"+`</span></span>`, test.result)
 		test.status.Escaped = true
 		test.status.HasMarks = true
 		tests = append(tests, test)