[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

3 files changed, 19 insertions, 11 deletions

diff --git a/ext/psych/lib/psych/nodes/node.rb b/ext/psych/lib/psych/nodes/node.rb
index 1f841625ca..f44fce5f05 100644
--- a/ext/psych/lib/psych/nodes/node.rb
+++ b/ext/psych/lib/psych/nodes/node.rb
@@ -46,8 +46,8 @@ module Psych
       # Convert this node to Ruby.
       #
       # See also Psych::Visitors::ToRuby
-      def to_ruby(symbolize_names: false, freeze: false)
-        Visitors::ToRuby.create(symbolize_names: symbolize_names, freeze: freeze).accept(self)
+      def to_ruby(symbolize_names: false, freeze: false, strict_integer: false)
+        Visitors::ToRuby.create(symbolize_names: symbolize_names, freeze: freeze, strict_integer: strict_integer).accept(self)
       end
       alias :transform :to_ruby
 
diff --git a/ext/psych/lib/psych/scalar_scanner.rb b/ext/psych/lib/psych/scalar_scanner.rb
index 604cb3e399..58deea3baa 100644
--- a/ext/psych/lib/psych/scalar_scanner.rb
+++ b/ext/psych/lib/psych/scalar_scanner.rb
@@ -12,24 +12,32 @@ module Psych
     FLOAT = /^(?:[-+]?([0-9][0-9_,]*)?\.[0-9]*([eE][-+][0-9]+)?(?# base 10))$/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]|,[0-9]|_[0-9])*) (?# base 10)
-                  |[-+]?0x[0-9a-fA-F_,]+                  (?# base 16))$/x
+    INTEGER_STRICT = /^(?:[-+]?0b[0-1_]+                  (?# base 2)
+                         |[-+]?0[0-7_]+                   (?# base 8)
+                         |[-+]?(0|[1-9][0-9_]*)           (?# base 10)
+                         |[-+]?0x[0-9a-fA-F_]+            (?# base 16))$/x
+
+    # Same as above, but allows commas.
+    # Not to YML spec, but kept for backwards compatibility
+    INTEGER_LEGACY = /^(?:[-+]?0b[0-1_,]+                        (?# base 2)
+                         |[-+]?0[0-7_,]+                         (?# base 8)
+                         |[-+]?(?:0|[1-9](?:[0-9]|,[0-9]|_[0-9])*) (?# base 10)
+                         |[-+]?0x[0-9a-fA-F_,]+                  (?# base 16))$/x
 
     attr_reader :class_loader
 
     # Create a new scanner
-    def initialize class_loader
+    def initialize class_loader, strict_integer: false
       @symbol_cache = {}
       @class_loader = class_loader
+      @strict_integer = strict_integer
     end
 
     # Tokenize +string+ returning the Ruby object
     def tokenize string
       return nil if string.empty?
       return @symbol_cache[string] if @symbol_cache.key?(string)
-
+      integer_regex = @strict_integer ? INTEGER_STRICT : INTEGER_LEGACY
       # Check for a String type, being careful not to get caught by hash keys, hex values, and
       # special floats (e.g., -.inf).
       if string.match?(%r{^[^\d.:-]?[[:alpha:]_\s!@#$%\^&*(){}<>|/\\~;=]+}) || string.match?(/\n/)
@@ -89,7 +97,7 @@ module Psych
         else
           Float(string.gsub(/[,_]|\.([Ee]|$)/, '\1'))
         end
-      elsif string.match?(INTEGER)
+      elsif string.match?(integer_regex)
         parse_int string
       else
         string
diff --git a/ext/psych/lib/psych/visitors/to_ruby.rb b/ext/psych/lib/psych/visitors/to_ruby.rb
index 4c1f561070..935bc74f21 100644
--- a/ext/psych/lib/psych/visitors/to_ruby.rb
+++ b/ext/psych/lib/psych/visitors/to_ruby.rb
@@ -12,9 +12,9 @@ module Psych
     ###
     # This class walks a YAML AST, converting each node to Ruby
     class ToRuby < Psych::Visitors::Visitor
-      def self.create(symbolize_names: false, freeze: false)
+      def self.create(symbolize_names: false, freeze: false, strict_integer: false)
         class_loader = ClassLoader.new
-        scanner      = ScalarScanner.new class_loader
+        scanner      = ScalarScanner.new class_loader, strict_integer: strict_integer
         new(scanner, class_loader, symbolize_names: symbolize_names, freeze: freeze)
       end