Do not allocate a string to check if a scalar is an integer

2 files changed, 24 insertions, 9 deletions

diff --git a/ext/psych/lib/psych/scalar_scanner.rb b/ext/psych/lib/psych/scalar_scanner.rb
index e92fe27fd7..cea2a453dd 100644
--- a/ext/psych/lib/psych/scalar_scanner.rb
+++ b/ext/psych/lib/psych/scalar_scanner.rb
@@ -14,10 +14,10 @@ module Psych
               |\.(nan|NaN|NAN)(?# not a number))$/x
 
     # Taken from http://yaml.org/type/int.html
-    INTEGER = /^(?:[-+]?0b[0-1_]+          (?# base 2)
-                  |[-+]?0[0-7_]+           (?# base 8)
-                  |[-+]?(?:0|[1-9][0-9_]*) (?# base 10)
-                  |[-+]?0x[0-9a-fA-F_]+    (?# base 16))$/x
+    INTEGER = /^(?:[-+]?0b[0-1_,]+          (?# base 2)
+                  |[-+]?0[0-7_,]+           (?# base 8)
+                  |[-+]?(?:0|[1-9][0-9_,]*) (?# base 10)
+                  |[-+]?0x[0-9a-fA-F_,]+    (?# base 16))$/x
 
     attr_reader :class_loader
 
@@ -91,10 +91,9 @@ module Psych
         else
           Float(string.gsub(/[,_]|\.([Ee]|$)/, '\1'))
         end
+      elsif string.match?(INTEGER)
+        parse_int string
       else
-        int = parse_int string.gsub(/[,_]/, '')
-        return int if int
-
         string
       end
     end
@@ -102,8 +101,7 @@ module Psych
     ###
     # Parse and return an int from +string+
     def parse_int string
-      return unless INTEGER === string
-      Integer(string)
+      Integer(string.gsub(/[,]/, ''))
     end
 
     ###
diff --git a/test/psych/test_scalar_scanner.rb b/test/psych/test_scalar_scanner.rb
index ebe8daf672..d12a905330 100644
--- a/test/psych/test_scalar_scanner.rb
+++ b/test/psych/test_scalar_scanner.rb
@@ -113,5 +113,22 @@ module Psych
     def test_scan_strings_starting_with_underscores
       assert_equal "_100", ss.tokenize('_100')
     end
+
+    def test_scan_int_commas_and_underscores
+      # NB: This test is to ensure backward compatibility with prior Psych versions,
+      # not to test against any actual YAML specification.
+      assert_equal 123_456_789, ss.tokenize('123_456_789')
+      assert_equal 123_456_789, ss.tokenize('123,456,789')
+      assert_equal 123_456_789, ss.tokenize('1_2,3,4_5,6_789')
+
+      assert_equal 0b010101010, ss.tokenize('0b010101010')
+      assert_equal 0b010101010, ss.tokenize('0b0,1_0,1_,0,1_01,0')
+
+      assert_equal 01234567, ss.tokenize('01234567')
+      assert_equal 01234567, ss.tokenize('0_,,,1_2,_34567')
+
+      assert_equal 0x123456789abcdef, ss.tokenize('0x123456789abcdef')
+      assert_equal 0x123456789abcdef, ss.tokenize('0x12_,34,_56,_789abcdef')
+    end
   end
 end