end-data-patch

diff --git a/kernel/common/compiled_method.rb b/kernel/common/compiled_method.rb
index df80cc9..338c18a 100644
--- a/kernel/common/compiled_method.rb
+++ b/kernel/common/compiled_method.rb
@@ -198,6 +198,8 @@ module Rubinius
 
     class Script
       attr_accessor :path
+      def root_script=(value); @root_script = value; end
+      def root_script?; @root_script; end
     end
 
     def as_script(script=nil)
diff --git a/kernel/common/data.rb b/kernel/common/data.rb
new file mode 100644
index 0000000..56901c5
--- /dev/null
+++ b/kernel/common/data.rb
@@ -0,0 +1,6 @@
+module Rubinius
+  def self.set_data(str)
+    require 'stringio'
+    Object.const_set :DATA, StringIO.new(str)
+  end
+end
diff --git a/kernel/common/load_order.txt b/kernel/common/load_order.txt
index cf063c3..f521835 100644
--- a/kernel/common/load_order.txt
+++ b/kernel/common/load_order.txt
@@ -37,6 +37,7 @@ compiled_method.rbc
 continuation.rbc
 delegated_method.rbc
 fixnum.rbc
+data.rbc
 dir.rbc
 dir_186.rbc
 dir_glob.rbc
diff --git a/kernel/common/static_scope.rb b/kernel/common/static_scope.rb
index e4003b1..299c20e 100644
--- a/kernel/common/static_scope.rb
+++ b/kernel/common/static_scope.rb
@@ -74,12 +74,8 @@ module Rubinius
     end
 
     def active_path
-      scope = self
-      while scope and !scope.script
-        scope = scope.parent
-      end
-
-      if scope and script = scope.script
+      script = current_script
+      if script
         if path = script.path
           return path.dup
         end
@@ -88,6 +84,19 @@ module Rubinius
       return "__unknown__.rb"
     end
 
+    def root_script?
+      script = current_script
+      return script && script.root_script?
+    end
+
+    def current_script
+      scope = self
+      while scope and !scope.script
+        scope = scope.parent
+      end
+      return scope && scope.script
+    end
+
     def const_defined?(name)
       scope = self
       while scope and scope.module != Object
diff --git a/kernel/compiler/bytecode.rb b/kernel/compiler/bytecode.rb
index 13e40fb..143ad55 100644
--- a/kernel/compiler/bytecode.rb
+++ b/kernel/compiler/bytecode.rb
@@ -1052,6 +1052,27 @@ class Compiler
       end
     end
 
+    class EndData
+      def bytecode(g)
+        # if (scope.root_script?)
+        #   Rubinius.set_data(...)
+        # end
+        not_root = g.new_label
+        done = g.new_label
+
+        g.push_scope
+        g.send :root_script?, 0
+        g.gif not_root
+        g.push_const(:Rubinius)
+        g.push_literal(data)
+        g.send :set_data, 1
+        g.goto done
+        not_root.set!
+        g.push_nil
+        done.set!
+      end
+    end
+
     class Ensure
       def bytecode(g)
         pos(g)
diff --git a/kernel/compiler/compile.rb b/kernel/compiler/compile.rb
index a9b47f2..e5ffb86 100644
--- a/kernel/compiler/compile.rb
+++ b/kernel/compiler/compile.rb
@@ -243,6 +243,7 @@ class Compiler
           end
           cm.as_script do |script|
             script.path = rb_path
+            script.root_script = options[:root_script]
           end
         rescue Exception => e
           $LOADED_FEATURES.delete(rb) if requiring
diff --git a/kernel/compiler/nodes.rb b/kernel/compiler/nodes.rb
index 97839e1..805a3ed 100644
--- a/kernel/compiler/nodes.rb
+++ b/kernel/compiler/nodes.rb
@@ -989,6 +989,16 @@ raise "no"
       kind :zarray
     end
 
+    class EndData < Node
+      kind :end_data
+
+      def args(data)
+        @data = data
+      end
+
+      attr_accessor :data
+    end
+
     class Ensure < Node
       kind :ensure
 
diff --git a/kernel/loader.rb b/kernel/loader.rb
index ab74f73..d45b303 100644
--- a/kernel/loader.rb
+++ b/kernel/loader.rb
@@ -317,7 +317,8 @@ module Rubinius
       if File.exist?(@script)
         $0 = @script
         Compiler::Utils.debug_script! if @debugging
-        Compiler::Utils.load_from_extension @script, :no_rbc => @no_rbc
+        Compiler::Utils.load_from_extension @script,
+          :no_rbc => @no_rbc, :root_script => true
       else
         if @script.suffix?(".rb")
           puts "Unable to find '#{@script}'"
diff --git a/vm/codegen/node_types.rb b/vm/codegen/node_types.rb
new file mode 100644
index 0000000..5022267
--- /dev/null
+++ b/vm/codegen/node_types.rb
@@ -0,0 +1,191 @@
+node_types = %w{
+  method
+  fbody
+  cfunc
+  scope
+  block
+  if
+  case
+  when
+  opt_n
+  while
+  until
+  iter
+  for
+  break
+  next
+  redo
+  retry
+  begin
+  rescue
+  resbody
+  ensure
+  and
+  or
+  not
+  masgn
+  lasgn
+  dasgn
+  dasgn_curr
+  gasgn
+  iasgn
+  cdecl
+  cvasgn
+  cvdecl
+  op_asgn1
+  op_asgn2
+  op_asgn_and
+  op_asgn_or
+  call
+  fcall
+  vcall
+  super
+  zsuper
+  array
+  zarray
+  hash
+  return
+  yield
+  lvar
+  dvar
+  gvar
+  ivar
+  const
+  cvar
+  nth_ref
+  back_ref
+  match
+  match2
+  match3
+  lit
+  str
+  dstr
+  xstr
+  dxstr
+  evstr
+  dregx
+  dregx_once
+  args
+  argscat
+  argspush
+  splat
+  to_ary
+  svalue
+  block_arg
+  block_pass
+  defn
+  defs
+  alias
+  valias
+  undef
+  class
+  module
+  sclass
+  colon2
+  colon3
+  cref
+  dot2
+  dot3
+  flip2
+  flip3
+  attrset
+  self
+  nil
+  true
+  false
+  defined
+  newline
+  postexe
+  dmethod
+  bmethod
+  memo
+  ifunc
+  dsym
+  attrasgn
+  regex
+  fixnum
+  number
+  hexnum
+  binnum
+  octnum
+  float
+  negate
+  last
+  file
+  end_data
+}
+
+File.open("vm/parser/node_types.cpp", "w") do |f|
+  f.puts <<EOF
+/* This file is generated by node_types.rb. Do not edit. */
+
+#include "node_types.hpp"
+
+namespace rubinius {
+  namespace parser {
+
+    static const char node_types[] = {
+EOF
+
+  node_types.each do |type|
+    f.puts("      \"#{type}\\0\"")
+  end
+  
+  f.puts("    };")
+  f.puts
+  
+  f.puts("    static const unsigned short node_types_offsets[] = {")
+  offset = 0
+  
+  node_types.each_with_index do |type, index|
+    f.puts(",") if index > 0
+    f.write("      #{offset}")
+    offset += node_types[index].length + 1
+  end
+  
+  f.puts <<EOF
+
+    };
+
+    const char *get_node_type_string(enum node_type nt) {
+      return node_types + node_types_offsets[nt];
+    }
+
+  };  // namespace parser
+};  // namespace rubinius
+EOF
+end
+
+File.open("vm/parser/node_types.hpp", "w") do |f|
+  f.puts <<EOF
+/* This file is generated by node_types.rb. Do not edit. */
+
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+namespace rubinius {
+  namespace parser {
+
+    enum node_type {
+EOF
+
+  node_types.each_with_index do |type, index|
+    f.puts(",") if index > 0
+    f.write("      NODE_#{type.upcase}")
+  end
+
+  f.puts <<EOF
+
+    };
+
+    const char *get_node_type_string(enum node_type nt);
+
+  };  // namespace parser
+};  // namespace rubinius
+
+#ifdef __cplusplus
+}
+#endif
+EOF
+end
diff --git a/vm/parser/grammar.y b/vm/parser/grammar.y
index c746b1e..455052d 100644
--- a/vm/parser/grammar.y
+++ b/vm/parser/grammar.y
@@ -337,16 +337,17 @@ static NODE *extract_block_vars(rb_parse_state *parse_state, NODE* node, var_tab
         klEND
         k__LINE__
         k__FILE__
+        k__END__
 
 %token <id>   tIDENTIFIER tFID tGVAR tIVAR tCONSTANT tCVAR tXSTRING_BEG
-%token <node> tINTEGER tFLOAT tSTRING_CONTENT
+%token <node> tINTEGER tFLOAT tSTRING_CONTENT tEND_DATA
 %token <node> tNTH_REF tBACK_REF
 %token <num>  tREGEXP_END
 %type <node> singleton strings string string1 xstring regexp
 %type <node> string_contents xstring_contents string_content
 %type <node> words qwords word_list qword_list word
 %type <node> literal numeric dsym cpath
-%type <node> bodystmt compstmt stmts stmt expr arg primary command command_call method_call
+%type <node> bodystmt compstmt opt_end_data stmts stmt expr arg primary command command_call method_call
 %type <node> expr_value arg_value primary_value
 %type <node> if_tail opt_else case_body cases opt_rescue exc_list exc_var opt_ensure
 %type <node> args when_args call_args call_args2 open_args paren_args opt_paren_args
@@ -425,7 +426,7 @@ program         :  {
                         vps->variables = new LocalState(0);
                         class_nest = 0;
                     }
-                  compstmt
+                  compstmt opt_end_data
                     {
                         if ($2 && !compile_for_eval) {
                             /* last expression should not be void */
@@ -439,6 +440,8 @@ program         :  {
                             }
                         }
                         vps->top = block_append(vps, vps->top, $2);
+                        if ($3)
+                            vps->top = block_append(vps, $3, vps->top);
                         class_nest = 0;
                     }
                 ;
@@ -470,6 +473,12 @@ compstmt        : stmts opt_terms
                     }
                 ;
 
+opt_end_data    : none
+                | k__END__ tEND_DATA
+                  {
+                    $$ = $2;
+                  }
+                
 stmts           : none
                 | stmt
                     {
@@ -1033,7 +1042,7 @@ op              : '|'           { $$ = '|'; }
                 | '`'           { $$ = '`'; }
                 ;
 
-reswords        : k__LINE__ | k__FILE__  | klBEGIN | klEND
+reswords        : k__LINE__ | k__FILE__  | klBEGIN | klEND | k__END__
                 | kALIAS | kAND | kBEGIN | kBREAK | kCASE | kCLASS | kDEF
                 | kDEFINED | kDO | kELSE | kELSIF | kEND | kENSURE | kFALSE
                 | kFOR | kIN | kMODULE | kNEXT | kNIL | kNOT
@@ -2569,6 +2578,7 @@ yycompile(rb_parse_state *parse_state, char *f, int line)
     /* Setup an initial empty scope. */
     heredoc_end = 0;
     lex_strterm = 0;
+    parse_state->end_seen = 0;
     ruby_sourcefile = f;
     n = yyparse(parse_state);
     ruby_debug_lines = 0;
@@ -3475,6 +3485,18 @@ yylex(void *yylval_v, void *vstate)
         }
         return token;
     }
+    
+    if (parse_state->end_seen) {  /* After __END__ */
+      newtok(parse_state);
+      while ((c = nextc()) != -1) {
+        tokadd(c, parse_state);
+      }
+      tokfix();
+      pslval->node = NEW_END_DATA(string_new(tok(), toklen()));
+      parse_state->end_seen = 0;
+      return tEND_DATA;
+    }
+
     cmd_state = command_start;
     command_start = FALSE;
   retry:
@@ -4493,7 +4515,9 @@ yylex(void *yylval_v, void *vstate)
       case '_':
         if (was_bol() && whole_match_p("__END__", 7, 0, parse_state)) {
             parse_state->lex_lastline = 0;
-            return -1;
+            parse_state->end_seen = 1;
+            parse_state->lex_p = parse_state->lex_pend;
+            return k__END__;
         }
         newtok(parse_state);
         break;
diff --git a/vm/parser/grammar_node.hpp b/vm/parser/grammar_node.hpp
index c1f7f9b..3c25b0b 100644
--- a/vm/parser/grammar_node.hpp
+++ b/vm/parser/grammar_node.hpp
@@ -248,6 +248,7 @@ namespace rubinius {
 #define NEW_NEGATE(l) NEW_NODE(NODE_NEGATE,l,0,0)
 #define NEW_REGEX(l,o) NEW_NODE(NODE_REGEX,l,0,o)
 #define NEW_FILE() NEW_NODE(NODE_FILE,0,0,0)
+#define NEW_END_DATA(s) NEW_NODE(NODE_END_DATA,s,0,0)
 #define NOEX_PUBLIC    0
 #define NOEX_NOSUPER   1
 #define NOEX_PRIVATE   2
diff --git a/vm/parser/grammar_runtime.cpp b/vm/parser/grammar_runtime.cpp
index 2659cc6..1872618 100644
--- a/vm/parser/grammar_runtime.cpp
+++ b/vm/parser/grammar_runtime.cpp
@@ -813,6 +813,7 @@ namespace rubinius {
 
       case NODE_XSTR:             /* u1    (%x{ls}) */
       case NODE_STR:              /* u1 */
+      case NODE_END_DATA:
         array_push(state, current, string_newfrombstr(state, node->nd_str));
         bdestroy(node->nd_str);
         break;
diff --git a/vm/parser/node_types.cpp b/vm/parser/node_types.cpp
index 5bf7819..79e3d02 100644
--- a/vm/parser/node_types.cpp
+++ b/vm/parser/node_types.cpp
@@ -119,6 +119,7 @@ namespace rubinius {
       "negate\0"
       "last\0"
       "file\0"
+      "end_data\0"
     };
 
     static const unsigned short node_types_offsets[] = {
@@ -234,7 +235,8 @@ namespace rubinius {
       697,
       703,
       710,
-      715
+      715,
+      720
     };
 
     const char *get_node_type_string(enum node_type nt) {
diff --git a/vm/parser/node_types.hpp b/vm/parser/node_types.hpp
index f00eaf2..513c4de 100644
--- a/vm/parser/node_types.hpp
+++ b/vm/parser/node_types.hpp
@@ -120,7 +120,8 @@ namespace rubinius {
       NODE_FLOAT,
       NODE_NEGATE,
       NODE_LAST,
-      NODE_FILE
+      NODE_FILE,
+      NODE_END_DATA
     };
 
     const char *get_node_type_string(enum node_type nt);