From 04422d716cf226b5e8a6c11306e07adbde26fd67 Mon Sep 17 00:00:00 2001
From: Evgeny Bobkin <evgen.ibqn@gmail.com>
Date: Wed, 17 Jul 2013 17:16:06 +0200
Subject: [PATCH] Add support for the \uXXXX escape sequence

An additional checks to validate the escape sequences \xYY and \uYYYY
were added, where Y represents a hex digit.
https://bugzilla.gnome.org/show_bug.cgi?id=704709
---
 tests/Makefile.am                   |  1 +
 tests/basic-types/escape-chars.vala | 17 ++++++++
 vala/valascanner.vala               | 63 ++++++++++++++++++++++++++---
 3 files changed, 75 insertions(+), 6 deletions(-)
 create mode 100644 tests/basic-types/escape-chars.vala

diff --git a/tests/Makefile.am b/tests/Makefile.am
index 8a0df29c3..b84ee0411 100644
--- a/tests/Makefile.am
+++ b/tests/Makefile.am
@@ -16,6 +16,7 @@ TESTS_ENVIRONMENT = EXEEXT=$(EXEEXT) CFLAGS='$(CFLAGS)' CPPFLAGS='$(CPPFLAGS)' L
 
 TESTS = \
 	basic-types/integers.vala \
+	basic-types/escape-chars.vala \
 	basic-types/floats.vala \
 	basic-types/strings.vala \
 	basic-types/arrays.vala \
diff --git a/tests/basic-types/escape-chars.vala b/tests/basic-types/escape-chars.vala
new file mode 100644
index 000000000..6ff1662a6
--- /dev/null
+++ b/tests/basic-types/escape-chars.vala
@@ -0,0 +1,17 @@
+void test_x_escape_chars () {
+	string s = "Copyright \xc2\xa9";
+
+	assert (s == "Copyright ©");
+}
+
+void test_u_escape_chars () {
+	string s = "Copyright \u00a9";
+
+	assert (s == "Copyright ©");
+}
+
+void main () {
+	// Test case for bug report 704709
+	test_x_escape_chars ();
+	test_u_escape_chars ();
+}
diff --git a/vala/valascanner.vala b/vala/valascanner.vala
index ba4736966..ab1628cc0 100644
--- a/vala/valascanner.vala
+++ b/vala/valascanner.vala
@@ -212,14 +212,31 @@ public class Vala.Scanner {
 							current++;
 							token_length_in_chars++;
 							break;
+						case 'u':
+							// u escape character has four hex digits
+							current++;
+							token_length_in_chars++;
+							int digit_length;
+							for (digit_length = 0; digit_length < 4 && current < end && current[0].isxdigit (); digit_length++) {
+								current++;
+								token_length_in_chars++;
+							}
+							if (digit_length != 4) {
+								Report.error (get_source_reference (token_length_in_chars), "\\u requires four hex digits");
+							}
+							break;
 						case 'x':
-							// hexadecimal escape character
+							// hexadecimal escape character requires two hex digits
 							current++;
 							token_length_in_chars++;
-							while (current < end && current[0].isxdigit ()) {
+							int digit_length;
+							for (digit_length = 0; digit_length < 2 && current < end && current[0].isxdigit (); digit_length++) {
 								current++;
 								token_length_in_chars++;
 							}
+							if (digit_length != 2) {
+								Report.error (get_source_reference (token_length_in_chars), "\\x requires two hex digits");
+							}
 							break;
 						default:
 							Report.error (get_source_reference (token_length_in_chars), "invalid escape sequence");
@@ -693,14 +710,31 @@ public class Vala.Scanner {
 							current++;
 							token_length_in_chars++;
 							break;
+						case 'u':
+							// u escape character has four hex digits
+							current++;
+							token_length_in_chars++;
+							int digit_length;
+							for (digit_length = 0; digit_length < 4 && current < end && current[0].isxdigit (); digit_length++) {
+								current++;
+								token_length_in_chars++;
+							}
+							if (digit_length != 4) {
+								Report.error (get_source_reference (token_length_in_chars), "\\u requires four hex digits");
+							}
+							break;
 						case 'x':
-							// hexadecimal escape character
+							// hexadecimal escape character requires two hex digits
 							current++;
 							token_length_in_chars++;
-							while (current < end && current[0].isxdigit ()) {
+							int digit_length;
+							for (digit_length = 0; digit_length < 2 && current < end && current[0].isxdigit (); digit_length++) {
 								current++;
 								token_length_in_chars++;
 							}
+							if (digit_length != 2) {
+								Report.error (get_source_reference (token_length_in_chars), "\\x requires two hex digits");
+							}
 							break;
 						default:
 							Report.error (get_source_reference (token_length_in_chars), "invalid escape sequence");
@@ -1110,14 +1144,31 @@ public class Vala.Scanner {
 							current++;
 							token_length_in_chars++;
 							break;
+						case 'u':
+							// u escape character has four hex digits
+							current++;
+							token_length_in_chars++;
+							int digit_length;
+							for (digit_length = 0; digit_length < 4 && current < end && current[0].isxdigit (); digit_length++) {
+								current++;
+								token_length_in_chars++;
+							}
+							if (digit_length != 4) {
+								Report.error (get_source_reference (token_length_in_chars), "\\u requires four hex digits");
+							}
+							break;
 						case 'x':
-							// hexadecimal escape character
+							// hexadecimal escape character requires two hex digits
 							current++;
 							token_length_in_chars++;
-							while (current < end && current[0].isxdigit ()) {
+							int digit_length;
+							for (digit_length = 0; digit_length < 2 && current < end && current[0].isxdigit (); digit_length++) {
 								current++;
 								token_length_in_chars++;
 							}
+							if (digit_length != 2) {
+								Report.error (get_source_reference (token_length_in_chars), "\\x requires two hex digits");
+							}
 							break;
 						default:
 							Report.error (get_source_reference (token_length_in_chars), "invalid escape sequence");
-- 
2.47.2