A number of fixes for Regexp found running rails3

@source must be dup'd before returning.
Made our named capture API mirror the API in 1.9.

kernel/bootstrap/regexp.rb

@@ -8,12 +8,14 @@ def self.allocate
   ##
   # See Regexp.new. This may be overridden by subclasses.
 
-  def initialize(pattern, opts, lang)
+  def compile(pattern, opts, lang)
     Ruby.primitive :regexp_initialize
     raise PrimitiveFailure,
           "regexp_new(#{str.inspect}, #{opts}, #{lang.inspect}) primitive failed"
   end
 
+  private :compile
+
   def search_region(str, start, finish, forward) # equiv to MRI's re_search
     Ruby.primitive :regexp_search_region
     raise PrimitiveFailure, "Regexp#search_region primitive failed"

kernel/common/regexp.rb

@@ -1,7 +1,4 @@
 class Regexp
-  attr_reader :source
-  attr_reader :names
-
   ValidKcode    = [?n,?e,?s,?u]
   KcodeValue    = [16,32,48,64]
 
@@ -64,7 +61,7 @@ class Regexp
   # You may also explicitly pass in 'n', 'N' or 'none' to disable multibyte
   # support. Any other values are ignored.
 
-  def self.new(pattern, opts = nil, lang = nil)
+  def initialize(pattern, opts=nil, lang=nil)
     if pattern.is_a?(Regexp)
       opts = pattern.options
       pattern  = pattern.source
@@ -82,9 +79,7 @@ def self.new(pattern, opts = nil, lang = nil)
       opts |= KcodeValue[idx] if idx
     end
 
-    r = allocate
-    r.send :initialize, pattern, opts, lang
-    r
+    compile pattern, opts, lang
   end
 
   def self.escape(str)
@@ -140,6 +135,10 @@ def self.union(*patterns)
     Regexp.new(patterns.join('|'))
   end
 
+  def source
+    @source.dup
+  end
+
   def ~
     line = $_
     if !line.is_a?(String)
@@ -466,6 +465,19 @@ def option_to_string(option)
     string << 'x' if (option & EXTENDED) > 0
     string
   end
+
+  def name_table
+    @names
+  end
+
+  def named_captures
+    hash = {}
+    @names.each do |k,v|
+      hash[k.to_s] = [v + 1] # we only have one location currently for a key
+    end
+
+    return hash
+  end
 end
 
 class MatchData
@@ -544,7 +556,7 @@ def [](idx, len = nil)
     if len
       return to_a[idx, len]
     elsif idx.is_a?(Symbol)
-      num = @regexp.names[idx]
+      num = @regexp.name_table[idx]
       raise ArgumentError, "Unknown named group '#{idx}'" unless num
       return get_capture(num)
     elsif !idx.is_a?(Integer) or idx < 0

vm/builtin/regexp.cpp

@@ -228,6 +228,7 @@ namespace rubinius {
   // 'self' is passed in automatically by the primitive glue
   Regexp* Regexp::allocate(STATE, Object* self) {
     Regexp* re = Regexp::create(state);
+    re->onig_data = 0;
     re->klass(state, (Class*)self);
     return re;
   }
@@ -367,6 +368,8 @@ namespace rubinius {
     Regexp* reg_o = force_as<Regexp>(obj);
     regex_t* reg = reg_o->onig_data;
 
+    if(!reg) return;
+
     ByteArray* reg_ba = ByteArray::from_body(reg);
 
     if(ByteArray* reg_tmp = force_as<ByteArray>(mark.call(reg_ba))) {
@@ -436,6 +439,8 @@ namespace rubinius {
     Regexp* reg_o = force_as<Regexp>(obj);
     regex_t* reg = reg_o->onig_data;
 
+    if(!reg) return;
+
     ByteArray* reg_ba = ByteArray::from_body(reg);
     visit.call(reg_ba);