caching string conversions in the scalar scanner

3 files changed, 23 insertions, 17 deletions

diff --git a/lib/psych/scalar_scanner.rb b/lib/psych/scalar_scanner.rb
index aa00d37..3d10276 100644
--- a/lib/psych/scalar_scanner.rb
+++ b/lib/psych/scalar_scanner.rb
@@ -5,24 +5,26 @@ module Psych
   # Scan scalars for built in types
   class ScalarScanner
     # Taken from http://yaml.org/type/timestamp.html
-    TIME = /\d{4}-\d{1,2}-\d{1,2}([Tt]|\s+)\d{1,2}:\d\d:\d\d(\.\d*)?(\s*Z|[-+]\d{1,2}(:\d\d)?)?/
+    TIME = /^\d{4}-\d{1,2}-\d{1,2}([Tt]|\s+)\d{1,2}:\d\d:\d\d(\.\d*)?(\s*Z|[-+]\d{1,2}(:\d\d)?)?/
 
-    def initialize string
-      @string = string
+    def initialize
+      @string_cache = {}
     end
 
-    def tokenize
-      self.class.tokenize @string
-    end
-
-    def self.tokenize string
+    def tokenize string
       return nil if string.empty?
+      return string if @string_cache.key?(string)
 
       case string
       when /^[A-Za-z~]/
-        return string if string.length > 5
+        if string.length > 5
+          @string_cache[string] = true
+          return string
+        end
+
         case string
         when /^[^ytonf~]/i
+          @string_cache[string] = true
           string
         when '~', /^null$/i
           nil
@@ -31,6 +33,7 @@ module Psych
         when /^(no|false|off)$/i
           false
         else
+          @string_cache[string] = true
           string
         end
       when TIME
@@ -77,6 +80,7 @@ module Psych
       else
         return Integer(string.gsub(/[,_]/, '')) rescue ArgumentError
         return Float(string.gsub(/[,_]/, '')) rescue ArgumentError
+        @string_cache[string] = true
         string
       end
     end
diff --git a/lib/psych/visitors/to_ruby.rb b/lib/psych/visitors/to_ruby.rb
index 9217ccc..6d6dae0 100644
--- a/lib/psych/visitors/to_ruby.rb
+++ b/lib/psych/visitors/to_ruby.rb
@@ -8,12 +8,13 @@ module Psych
       def initialize
         super
         @st = {}
+        @ss = ScalarScanner.new
       end
 
       def accept target
         result = super
-        return result if Psych.domain_types.empty?
         return result unless target.tag
+        return result if Psych.domain_types.empty?
 
         short_name = target.tag.sub(/^!/, '').split('/', 2).last
         if Psych.domain_types.key? short_name
@@ -28,7 +29,7 @@ module Psych
         @st[o.anchor] = o.value if o.anchor
 
         return o.value if o.quoted
-        return ScalarScanner.tokenize(o.value) unless o.tag
+        return @ss.tokenize(o.value) unless o.tag
 
         case o.tag
         when '!binary', 'tag:yaml.org,2002:binary'
@@ -40,7 +41,7 @@ module Psych
         when "!ruby/object:Rational"
           Rational(o.value)
         when "tag:yaml.org,2002:float", "!float"
-          Float(ScalarScanner.tokenize(o.value))
+          Float(@ss.tokenize(o.value))
         when "!ruby/regexp"
           o.value =~ /^\/(.*)\/([mix]*)$/
           source  = $1
@@ -62,7 +63,7 @@ module Psych
           args.push(args.delete_at(1) == '...')
           Range.new(*args)
         else
-          ScalarScanner.tokenize o.value
+          @ss.tokenize o.value
         end
       end
 
@@ -89,7 +90,7 @@ module Psych
           members = Hash[*o.children.map { |c| accept c }]
           string = members.delete 'str'
           init_with(string, members.map { |k,v| [k.to_s.sub(/^@/, ''),v] })
-        when /!ruby\/struct:?(.*)?$/
+        when /^!ruby\/struct:?(.*)?$/
           klass   = resolve_class($1)
 
           if klass
@@ -118,7 +119,7 @@ module Psych
           h = Hash[*o.children.map { |c| accept c }]
           Range.new(h['begin'], h['end'], h['excl'])
 
-        when /!ruby\/exception:?(.*)?$/
+        when /^!ruby\/exception:?(.*)?$/
           h = Hash[*o.children.map { |c| accept c }]
 
           e = build_exception((resolve_class($1) || Exception),
@@ -142,7 +143,7 @@ module Psych
           h = Hash[*o.children.map { |c| accept c }]
           Rational(h['numerator'], h['denominator'])
 
-        when /!ruby\/object:?(.*)?$/
+        when /^!ruby\/object:?(.*)?$/
           name = $1 || 'Object'
           h = Hash[*o.children.map { |c| accept c }]
           s = name.split('::').inject(Object) { |k,sub|
diff --git a/lib/psych/visitors/yaml_tree.rb b/lib/psych/visitors/yaml_tree.rb
index 6b97799..8732200 100644
--- a/lib/psych/visitors/yaml_tree.rb
+++ b/lib/psych/visitors/yaml_tree.rb
@@ -9,6 +9,7 @@ module Psych
         @tree.children << Nodes::Document.new
         @stack = @tree.children.dup
         @st = {}
+        @ss = ScalarScanner.new
 
         @dispatch_cache = Hash.new do |h,klass|
           method = "visit_#{(klass.name || '').split('::').join('_')}"
@@ -149,7 +150,7 @@ module Psych
         else
           str   = o
           tag   = nil
-          quote = !(String === ScalarScanner.new(o).tokenize)
+          quote = !(String === @ss.tokenize(o))
           plain = !quote
         end