[ruby/psych] Add strict_integer option to parse numbers with commas as strings

Authored-by: Seth Boyles <sethboyles@gmail.com>

https://github.com/ruby/psych/commit/75bebb37b8

2 files changed, 36 insertions, 0 deletions

diff --git a/test/psych/test_numeric.rb b/test/psych/test_numeric.rb
index 8c3dcd173c..9c75c016cd 100644
--- a/test/psych/test_numeric.rb
+++ b/test/psych/test_numeric.rb
@@ -43,5 +43,16 @@ module Psych
       str = Psych.load('--- 1.1.1')
       assert_equal '1.1.1', str
     end
+
+    # This behavior is not to YML spec, but is kept for backwards compatibility
+    def test_string_with_commas
+      number = Psych.load('--- 12,34,56')
+      assert_equal 123456, number
+    end
+
+    def test_string_with_commas_with_strict_integer
+      str = Psych.load('--- 12,34,56', strict_integer: true)
+      assert_equal '12,34,56', str
+    end
   end
 end
diff --git a/test/psych/test_scalar_scanner.rb b/test/psych/test_scalar_scanner.rb
index ebc9fbdcd2..145db58fd9 100644
--- a/test/psych/test_scalar_scanner.rb
+++ b/test/psych/test_scalar_scanner.rb
@@ -149,6 +149,31 @@ module Psych
       assert_equal 0x123456789abcdef, ss.tokenize('0x12_,34,_56,_789abcdef__')
     end
 
+    def test_scan_strict_int_commas_and_underscores
+      # this test is to ensure adherance to YML spec using the 'strict_integer' option
+      scanner = Psych::ScalarScanner.new ClassLoader.new, strict_integer: true
+      assert_equal 123_456_789, scanner.tokenize('123_456_789')
+      assert_equal '123,456,789', scanner.tokenize('123,456,789')
+      assert_equal '1_2,3,4_5,6_789', scanner.tokenize('1_2,3,4_5,6_789')
+
+      assert_equal 1, scanner.tokenize('1')
+      assert_equal 1, scanner.tokenize('+1')
+      assert_equal(-1, scanner.tokenize('-1'))
+
+      assert_equal 0b010101010, scanner.tokenize('0b010101010')
+      assert_equal 0b010101010, scanner.tokenize('0b01_01_01_010')
+      assert_equal '0b0,1_0,1_,0,1_01,0', scanner.tokenize('0b0,1_0,1_,0,1_01,0')
+
+      assert_equal 01234567, scanner.tokenize('01234567')
+      assert_equal '0_,,,1_2,_34567', scanner.tokenize('0_,,,1_2,_34567')
+
+      assert_equal 0x123456789abcdef, scanner.tokenize('0x123456789abcdef')
+      assert_equal 0x123456789abcdef, scanner.tokenize('0x12_34_56_789abcdef')
+      assert_equal '0x12_,34,_56,_789abcdef', scanner.tokenize('0x12_,34,_56,_789abcdef')
+      assert_equal '0x_12_,34,_56,_789abcdef', scanner.tokenize('0x_12_,34,_56,_789abcdef')
+      assert_equal '0x12_,34,_56,_789abcdef__', scanner.tokenize('0x12_,34,_56,_789abcdef__')
+    end
+
     def test_scan_dot
       assert_equal '.', ss.tokenize('.')
     end