[Truffle] Last of the encoding specs.

jruby · Dec 24, 2014 · 7da6fd7 · 7da6fd7
1 parent 53610d2
commit 7da6fd7
Show file tree

Hide file tree

Showing 3 changed files with 31 additions and 8 deletions.
diff --git a/core/src/main/java/org/jruby/truffle/nodes/core/InteroplatedRegexpNode.java b/core/src/main/java/org/jruby/truffle/nodes/core/InteroplatedRegexpNode.java
@@ -14,8 +14,10 @@
 import org.jruby.truffle.nodes.RubyNode;
 import org.jruby.truffle.nodes.dispatch.DispatchHeadNode;
 import org.jruby.truffle.runtime.RubyContext;
+import org.jruby.truffle.runtime.core.RubyEncoding;
 import org.jruby.truffle.runtime.core.RubyRegexp;
 import org.jruby.truffle.runtime.core.RubyString;
+import org.jruby.truffle.translator.BodyTranslator;
 import org.jruby.util.RegexpOptions;
 
 public class InteroplatedRegexpNode extends RubyNode {
@@ -43,7 +45,20 @@ public RubyRegexp executeRubyRegexp(VirtualFrame frame) {
         }
 
         final org.jruby.RubyString preprocessed = org.jruby.RubyRegexp.preprocessDRegexp(getContext().getRuntime(), strings, options);
-        return new RubyRegexp(this, getContext().getCoreLibrary().getRegexpClass(), preprocessed.getByteList(), options.toOptions());
+
+        final RubyRegexp regexp = new RubyRegexp(this, getContext().getCoreLibrary().getRegexpClass(), preprocessed.getByteList(), options.toOptions());
+
+        if (options.isEncodingNone()) {
+            // This isn't quite right - we shouldn't be looking up by name, we need a real reference to this constants
+
+            if (!BodyTranslator.all7Bit(preprocessed.getByteList().bytes())) {
+                regexp.forceEncoding((RubyEncoding) getContext().getCoreLibrary().getEncodingClass().getConstants().get("ASCII_8BIT").getValue());
+            } else {
+                regexp.forceEncoding((RubyEncoding) getContext().getCoreLibrary().getEncodingClass().getConstants().get("US_ASCII").getValue());
+            }
+        }
+
+        return regexp;
     }
 
     @Override

diff --git a/core/src/main/java/org/jruby/truffle/translator/BodyTranslator.java b/core/src/main/java/org/jruby/truffle/translator/BodyTranslator.java
@@ -2026,10 +2026,10 @@ public RubyNode visitRegexpNode(org.jruby.ast.RegexpNode node) {
         if (node.getOptions().isEncodingNone()) {
             // This isn't quite right - we shouldn't be looking up by name, we need a real reference to this constants
 
-            if (all7Bit(node.getValue().bytes())) {
-                regexp.forceEncoding((RubyEncoding) context.getCoreLibrary().getEncodingClass().getConstants().get("US_ASCII").getValue());
+            if (!all7Bit(node.getValue().bytes())) {
+                regexp.forceEncoding((RubyEncoding) context.getCoreLibrary().getEncodingClass().getConstants().get("ASCII_8BIT").getValue());
             } else {
-                regexp.forceEncoding((RubyEncoding) context.getCoreLibrary().getEncodingClass().getConstants().get("ASCII-8BIT").getValue());
+                regexp.forceEncoding((RubyEncoding) context.getCoreLibrary().getEncodingClass().getConstants().get("US_ASCII").getValue());
             }
         } else if (node.getOptions().getKCode().getKCode().equals("SJIS")) {
             regexp.forceEncoding((RubyEncoding) context.getCoreLibrary().getEncodingClass().getConstants().get("Windows_31J").getValue());
@@ -2048,11 +2048,21 @@ public RubyNode visitRegexpNode(org.jruby.ast.RegexpNode node) {
         return literalNode;
     }
 
-    private static boolean all7Bit(byte[] bytes) {
+    public static boolean all7Bit(byte[] bytes) {
         for (int n = 0; n < bytes.length; n++) {
-            if (bytes[n] < 0 || bytes[n] > 0x7F) {
+            if (bytes[n] < 0 || bytes[n] > Byte.MAX_VALUE) {
                 return false;
             }
+
+            if (bytes[n] == '\\' && n + 1 < bytes.length && bytes[n + 1] == 'x') {
+                int b = Integer.parseInt(new String(Arrays.copyOfRange(bytes, n + 2, n + 4)), 16);
+
+                if (b > 0x7F) {
+                    return false;
+                }
+
+                n += 3;
+            }
         }
 
         return true;

diff --git a/spec/truffle/tags/language/regexp/encoding_tags.txt b/spec/truffle/tags/language/regexp/encoding_tags.txt