From 08ff84627cfbcce47d9a053f8ed8cc919053a114 Mon Sep 17 00:00:00 2001
From: shove <shove@163.com>
Date: Fri, 29 Dec 2023 20:25:11 +0800
Subject: [PATCH] scanner: fix escape character handling in character/rune
 literals (fix #20301) (#20304)

---
 vlib/v/scanner/scanner.v                      | 26 ++++++++++++++-----
 .../tests/invalid_character_literal_err_1.out |  7 +++++
 .../tests/invalid_character_literal_err_1.vv  |  1 +
 .../tests/invalid_character_literal_err_2.out |  7 +++++
 .../tests/invalid_character_literal_err_2.vv  |  1 +
 5 files changed, 36 insertions(+), 6 deletions(-)
 create mode 100644 vlib/v/scanner/tests/invalid_character_literal_err_1.out
 create mode 100644 vlib/v/scanner/tests/invalid_character_literal_err_1.vv
 create mode 100644 vlib/v/scanner/tests/invalid_character_literal_err_2.out
 create mode 100644 vlib/v/scanner/tests/invalid_character_literal_err_2.vv

diff --git a/vlib/v/scanner/scanner.v b/vlib/v/scanner/scanner.v
index 9402ba73f..e8b97526d 100644
--- a/vlib/v/scanner/scanner.v
+++ b/vlib/v/scanner/scanner.v
@@ -1537,11 +1537,14 @@ pub fn (mut s Scanner) ident_char() string {
 	mut len := 0
 
 	// set flags for advanced escapes first
-	escaped_hex := s.expect('\\x', start + 1)
-	escaped_unicode_16 := s.expect('\\u', start + 1)
-	escaped_unicode_32 := s.expect('\\U', start + 1)
+	escaped_hex := s.expect('\\x', start + 1) && s.text.len > start + 3
+		&& s.text[start + 3].is_hex_digit()
+	escaped_unicode_16 := s.expect('\\u', start + 1) && s.text.len > start + 3
+		&& s.text[start + 3].is_hex_digit()
+	escaped_unicode_32 := s.expect('\\U', start + 1) && s.text.len > start + 3
+		&& s.text[start + 3].is_hex_digit()
 	escaped_octal := !escaped_hex && !escaped_unicode_16 && !escaped_unicode_32
-		&& s.expect('\\', start + 1)
+		&& s.expect('\\', start + 1) && s.text.len > start + 2 && s.text[start + 2].is_oct_digit()
 
 	// walk the string to get characters up to the next backtick
 	for {
@@ -1599,8 +1602,19 @@ pub fn (mut s Scanner) ident_char() string {
 
 		u := c.runes()
 		if u.len != 1 {
+			mut err_info := []string{cap: u.len}
+			mut i := 0
+			for i < u.len {
+				if u[i] != `\\` || i == u.len - 1 {
+					err_info << '`${u[i]}`'
+					i++
+					continue
+				}
+				err_info << '`\\${u[i + 1]}`'
+				i += 2
+			}
 			if escaped_hex || escaped_unicode_16 || escaped_unicode_32 {
-				s.error_with_pos('invalid character literal `${orig}` => `${c}` (${u}) (escape sequence did not refer to a singular rune)',
+				s.error_with_pos('invalid character literal `${orig}` => `${c}` ([${err_info.join(', ')}]) (escape sequence did not refer to a singular rune)',
 					lspos)
 			} else if u.len == 0 {
 				s.add_error_detail_with_pos('use quotes for strings, backticks for characters',
@@ -1609,7 +1623,7 @@ pub fn (mut s Scanner) ident_char() string {
 			} else {
 				s.add_error_detail_with_pos('use quotes for strings, backticks for characters',
 					lspos)
-				s.error_with_pos('invalid character literal `${orig}` => `${c}` (${u}) (more than one character)',
+				s.error_with_pos('invalid character literal `${orig}` => `${c}` ([${err_info.join(', ')}]) (more than one character)',
 					lspos)
 			}
 		}
diff --git a/vlib/v/scanner/tests/invalid_character_literal_err_1.out b/vlib/v/scanner/tests/invalid_character_literal_err_1.out
new file mode 100644
index 000000000..5d2fe747b
--- /dev/null
+++ b/vlib/v/scanner/tests/invalid_character_literal_err_1.out
@@ -0,0 +1,7 @@
+vlib/v/scanner/tests/invalid_character_literal_err_1.vv:1:6: error: invalid character literal `\n\t` => `\n\t` ([`\n`, `\t`]) (more than one character)
+    1 | a := `\n\t`
+      |      ^
+Details: 
+vlib/v/scanner/tests/invalid_character_literal_err_1.vv:1:6: details: use quotes for strings, backticks for characters
+    1 | a := `\n\t`
+      |      ^
diff --git a/vlib/v/scanner/tests/invalid_character_literal_err_1.vv b/vlib/v/scanner/tests/invalid_character_literal_err_1.vv
new file mode 100644
index 000000000..28f6b6aa4
--- /dev/null
+++ b/vlib/v/scanner/tests/invalid_character_literal_err_1.vv
@@ -0,0 +1 @@
+a := `\n\t`
diff --git a/vlib/v/scanner/tests/invalid_character_literal_err_2.out b/vlib/v/scanner/tests/invalid_character_literal_err_2.out
new file mode 100644
index 000000000..c89fda852
--- /dev/null
+++ b/vlib/v/scanner/tests/invalid_character_literal_err_2.out
@@ -0,0 +1,7 @@
+vlib/v/scanner/tests/invalid_character_literal_err_2.vv:1:6: error: invalid character literal `\nb` => `\nb` ([`\n`, `b`]) (more than one character)
+    1 | a := `\nb`
+      |      ^
+Details: 
+vlib/v/scanner/tests/invalid_character_literal_err_2.vv:1:6: details: use quotes for strings, backticks for characters
+    1 | a := `\nb`
+      |      ^
diff --git a/vlib/v/scanner/tests/invalid_character_literal_err_2.vv b/vlib/v/scanner/tests/invalid_character_literal_err_2.vv
new file mode 100644
index 000000000..56abba72b
--- /dev/null
+++ b/vlib/v/scanner/tests/invalid_character_literal_err_2.vv
@@ -0,0 +1 @@
+a := `\nb`
-- 
2.39.5