[ruby/json] Fix a regression in parsing of unicode surogate pairs

Fix: https://github.com/ruby/json/issues/912

In the case of surogate pairs we consume two backslashes, so
`json_next_backslash` need to ensure it's not sending us back in the
stream.

https://github.com/ruby/json/commit/0fce370c41

2 files changed, 10 insertions, 1 deletions

diff --git a/ext/json/parser/parser.c b/ext/json/parser/parser.c
index 5b7cd835cd..c84c7ed660 100644
--- a/ext/json/parser/parser.c
+++ b/ext/json/parser/parser.c
@@ -651,7 +651,9 @@ static inline const char *json_next_backslash(const char *pe, const char *string
         positions->size--;
         const char *next_position = positions->positions[0];
         positions->positions++;
-        return next_position;
+        if (next_position >= pe) {
+            return next_position;
+        }
     }
 
     if (positions->has_more) {
diff --git a/test/json/json_parser_test.rb b/test/json/json_parser_test.rb
index 3e662bda32..257e4f1736 100644
--- a/test/json/json_parser_test.rb
+++ b/test/json/json_parser_test.rb
@@ -325,6 +325,13 @@ class JSONParserTest < Test::Unit::TestCase
     assert_raise(JSON::ParserError) { parse('"\u111___"') }
   end
 
+  def test_unicode_followed_by_newline
+    # Ref: https://github.com/ruby/json/issues/912
+    assert_equal "🌌\n".bytes, JSON.parse('"\ud83c\udf0c\n"').bytes
+    assert_equal "🌌\n", JSON.parse('"\ud83c\udf0c\n"')
+    assert_predicate JSON.parse('"\ud83c\udf0c\n"'), :valid_encoding?
+  end
+
   def test_invalid_surogates
     assert_raise(JSON::ParserError) { parse('"\\uD800"') }
     assert_raise(JSON::ParserError) { parse('"\\uD800_________________"') }