Enable the creation of dummy encodings from C extensions.

PullRequest: truffleruby/747

Author: eregon

lib/truffle/truffle/cext.rb

@@ -733,6 +733,11 @@ def rb_enc_to_index(enc)
     Truffle.invoke_primitive :encoding_get_encoding_index, enc
   end
 
+  def rb_define_dummy_encoding(name)
+    _, idx = Truffle::EncodingOperations.dummy_encoding(name)
+    idx
+  end
+
   def rb_str_new_frozen(value)
     if value.frozen?
       value
@@ -899,7 +904,7 @@ def rb_str_encode(str, to, ecflags, ecopts)
   end
 
   def rb_str_conv_enc_opts(str, from, to, ecflags, ecopts)
-    if (to.ascii_compatible? && str.ascii_only?) || to == Encoding::ASCII_8BIT
+    if (to.ascii_compatible? && str.ascii_only?) || to == Encoding::ASCII_8BIT || to.dummy?
       if str.encoding != to
         str = str.dup
         str.force_encoding(to)

src/main/c/cext/ruby.c

@@ -3091,7 +3091,7 @@ int rb_enc_replicate(const char *name, rb_encoding *encoding) {
 }
 
 int rb_define_dummy_encoding(const char *name) {
-  rb_tr_error("rb_define_dummy_encoding not implemented");
+  return polyglot_as_i32(RUBY_CEXT_INVOKE_NO_WRAP("rb_define_dummy_encoding", rb_str_new_cstr(name)));
 }
 
 #undef rb_enc_str_new_cstr

src/main/java/org/truffleruby/core/CoreLibrary.java

@@ -1424,6 +1424,7 @@ public DynamicObject getWarningModule() {
             "/core/truffle/boot.rb",
             "/core/truffle/debug.rb",
             "/core/truffle/warnings.rb",
+            "/core/truffle/encoding_operations.rb",
             "/core/truffle/exception_operations.rb",
             "/core/truffle/hash_operations.rb",
             "/core/truffle/numeric_operations.rb",

src/main/java/org/truffleruby/core/encoding/EncodingLayout.java

@@ -23,8 +23,7 @@ DynamicObjectFactory createEncodingShape(DynamicObject logicalClass,
 
     DynamicObject createEncoding(DynamicObjectFactory factory,
                                  Encoding encoding,
-                                 DynamicObject name,
-                                 boolean dummy);
+                                 DynamicObject name);
 
     boolean isEncoding(DynamicObject object);
     boolean isEncoding(Object object);
@@ -33,6 +32,4 @@ DynamicObject createEncoding(DynamicObjectFactory factory,
 
     DynamicObject getName(DynamicObject object);
 
-    boolean getDummy(DynamicObject object);
-
 }

src/main/java/org/truffleruby/core/encoding/EncodingManager.java

@@ -170,15 +170,15 @@ private void initializeLocaleEncoding(TruffleNFIPlatform nfi, NativeConfiguratio
     }
 
     @TruffleBoundary
-    private static DynamicObject newRubyEncoding(RubyContext context, Encoding encoding, byte[] name, int p, int end, boolean dummy) {
+    private static DynamicObject newRubyEncoding(RubyContext context, Encoding encoding, byte[] name, int p, int end) {
         assert p == 0 : "Ropes can't be created with non-zero offset: " + p;
         assert end == name.length : "Ropes must have the same exact length as the name array (len = " + end + "; name.length = " + name.length + ")";
 
         final Rope rope = RopeOperations.create(name, USASCIIEncoding.INSTANCE, CodeRange.CR_7BIT);
         final Rope cachedRope = context.getRopeCache().getRope(rope.getBytes(), rope.getEncoding(), rope.getCodeRange());
         final DynamicObject string = StringOperations.createFrozenString(context, cachedRope);
 
-        return Layouts.ENCODING.createEncoding(context.getCoreLibrary().getEncodingFactory(), encoding, string, dummy);
+        return Layouts.ENCODING.createEncoding(context.getCoreLibrary().getEncodingFactory(), encoding, string);
     }
 
     public static Encoding getEncoding(String name) {
@@ -240,7 +240,7 @@ public DynamicObject getRubyEncoding(Encoding encoding) {
     public synchronized DynamicObject defineEncoding(EncodingDB.Entry encodingEntry, byte[] name, int p, int end) {
         final Encoding encoding = encodingEntry.getEncoding();
         final int encodingIndex = encoding.getIndex();
-        final DynamicObject rubyEncoding = newRubyEncoding(context, encoding, name, p, end, encodingEntry.isDummy());
+        final DynamicObject rubyEncoding = newRubyEncoding(context, encoding, name, p, end);
 
         assert encodingIndex >= ENCODING_LIST_BY_ENCODING_INDEX.size() || ENCODING_LIST_BY_ENCODING_INDEX.get(encodingIndex) == null;
 
@@ -260,6 +260,18 @@ public DynamicObject defineAlias(Encoding encoding, String name) {
         return rubyEncoding;
     }
 
+    @TruffleBoundary
+    public synchronized DynamicObject createDummyEncoding(String name) {
+        if (getRubyEncoding(name) != null) {
+            return null;
+        }
+
+        byte[] nameBytes = name.getBytes();
+        EncodingDB.dummy(nameBytes);
+        final Entry entry = EncodingDB.getEncodings().get(nameBytes);
+        return defineEncoding(entry, nameBytes, 0, nameBytes.length);
+    }
+
     @TruffleBoundary
     public synchronized DynamicObject replicateEncoding(Encoding encoding, String name) {
         if (getRubyEncoding(name) != null) {

src/main/java/org/truffleruby/core/encoding/EncodingNodes.java

@@ -421,7 +421,7 @@ protected int getCacheLimit() {
         protected static boolean isDummy(DynamicObject encoding) {
             assert RubyGuards.isRubyEncoding(encoding);
 
-            return Layouts.ENCODING.getDummy(encoding);
+            return Layouts.ENCODING.getEncoding(encoding).isDummy();
         }
     }
 
@@ -652,21 +652,29 @@ public DynamicObject encodingGetObjectEncodingNil(DynamicObject object) {
 
     }
 
+    public static abstract class EncodingCreationNode extends PrimitiveArrayArgumentsNode {
+
+        public DynamicObject setIndexOrRaiseError(String name, DynamicObject newEncoding) {
+            if (newEncoding == null) {
+                throw new RaiseException(getContext(), coreExceptions().argumentErrorEncodingAlreadyRegistered(name, this));
+            }
+
+            final int index = Layouts.ENCODING.getEncoding(newEncoding).getIndex();
+            return createArray(new Object[]{ newEncoding, index }, 2);
+        }
+
+    }
+
     @Primitive(name = "encoding_replicate")
-    public static abstract class EncodingReplicateNode extends PrimitiveArrayArgumentsNode {
+    public static abstract class EncodingReplicateNode extends EncodingCreationNode {
 
         @Specialization(guards = "isRubyString(nameObject)")
         public DynamicObject encodingReplicate(DynamicObject self, DynamicObject nameObject) {
             final String name = StringOperations.getString(nameObject);
             final Encoding encoding = EncodingOperations.getEncoding(self);
 
             final DynamicObject newEncoding = replicate(name, encoding);
-            if (newEncoding == null) {
-                throw new RaiseException(getContext(), coreExceptions().argumentErrorEncodingAlreadyRegistered(name, this));
-            }
-
-            final int index = Layouts.ENCODING.getEncoding(newEncoding).getIndex();
-            return createArray(new Object[]{ newEncoding, index }, 2);
+            return setIndexOrRaiseError(name, newEncoding);
         }
 
         @TruffleBoundary
@@ -676,6 +684,24 @@ private DynamicObject replicate(String name, Encoding encoding) {
 
     }
 
+    @Primitive(name = "encoding_create_dummy", needsSelf = false)
+    public static abstract class DummyEncodingeNode extends EncodingCreationNode {
+
+        @Specialization(guards = "isRubyString(nameObject)")
+        public DynamicObject createDummyEncoding(DynamicObject nameObject) {
+            final String name = StringOperations.getString(nameObject);
+
+            final DynamicObject newEncoding = createDummy(name);
+            return setIndexOrRaiseError(name, newEncoding);
+        }
+
+        @TruffleBoundary
+        private DynamicObject createDummy(String name) {
+            return getContext().getEncodingManager().createDummyEncoding(name);
+        }
+
+    }
+
     @Primitive(name = "encoding_get_encoding_by_index", needsSelf = false, lowerFixnum = 1)
     public static abstract class GetEncodingObjectByIndexNode extends PrimitiveArrayArgumentsNode {
 

src/main/ruby/core/truffle/encoding_operations.rb

@@ -0,0 +1,19 @@
+# frozen_string_literal: true
+
+# Copyright (c) 2019 Oracle and/or its affiliates. All rights reserved. This
+# code is released under a tri EPL/GPL/LGPL license. You can use it,
+# redistribute it and/or modify it under the terms of the:
+#
+# Eclipse Public License version 1.0, or
+# GNU General Public License version 2, or
+# GNU Lesser General Public License version 2.1.
+
+module Truffle
+  module EncodingOperations
+    def self.dummy_encoding(name)
+      new_encoding, index = Truffle.invoke_primitive :encoding_create_dummy, name
+      ::Encoding::EncodingMap[name.upcase.to_sym] = [nil, index]
+      [new_encoding, index]
+    end
+  end
+end