Merge pull request #3623 from jruby/truffle-ropes

Truffle ropes
jruby · Jan 27, 2016 · 983325b · 983325b
2 parents 0193bdc + 3d54c55
commit 983325b
Showing 70 changed files with 2,966 additions and 1,121 deletions.
diff --git a/core/src/main/java/org/jruby/util/StringSupport.java b/core/src/main/java/org/jruby/util/StringSupport.java
@@ -319,7 +319,7 @@ public static long strLengthWithCodeRange(Encoding enc, byte[]bytes, int p, int
         }
     }
 
-    private static long strLengthWithCodeRangeAsciiCompatible(Encoding enc, byte[]bytes, int p, int end) {
+    public static long strLengthWithCodeRangeAsciiCompatible(Encoding enc, byte[]bytes, int p, int end) {
         int cr = 0, c = 0;
         while (p < end) {
             if (Encoding.isAscii(bytes[p])) {
@@ -341,7 +341,7 @@ private static long strLengthWithCodeRangeAsciiCompatible(Encoding enc, byte[]by
         return pack(c, cr == 0 ? CR_7BIT : cr);
     }
 
-    private static long strLengthWithCodeRangeNonAsciiCompatible(Encoding enc, byte[]bytes, int p, int end) {
+    public static long strLengthWithCodeRangeNonAsciiCompatible(Encoding enc, byte[]bytes, int p, int end) {
         int cr = 0, c = 0;
         for (c = 0; p < end; c++) {
             int cl = preciseLength(enc, bytes, p, end);
@@ -365,7 +365,7 @@ public static long strLengthWithCodeRange(ByteList bytes, Encoding enc) {
     }
 
     // arg cannot be negative
-    static long pack(int result, int arg) {
+    public static long pack(int result, int arg) {
         return ((long)arg << 31) | result;
     }
 

diff --git a/core/src/main/java/org/jruby/util/io/EncodingUtils.java b/core/src/main/java/org/jruby/util/io/EncodingUtils.java
@@ -1842,12 +1842,12 @@ public static Encoding getEncoding(ByteList str) {
 
     // MRI: get_actual_encoding
     public static Encoding getActualEncoding(Encoding enc, ByteList byteList) {
+        return getActualEncoding(enc, byteList.getUnsafeBytes(), byteList.begin(), byteList.begin() + byteList.realSize());
+    }
+
+    public static Encoding getActualEncoding(Encoding enc, byte[] bytes, int p, int end) {
         if (enc.isDummy() && enc instanceof UnicodeEncoding) {
             // handle dummy UTF-16 and UTF-32 by scanning for BOM, as in MRI
-            byte[] bytes = byteList.unsafeBytes();
-            int p = byteList.begin();
-            int end = p + byteList.getRealSize();
-
             if (enc == UTF16Dummy && end - p >= 2) {
                 int c0 = bytes[p] & 0xff;
                 int c1 = bytes[p + 1] & 0xff;

diff --git a/test/truffle/compiler/pe/core/string_pe.rb b/test/truffle/compiler/pe/core/string_pe.rb
@@ -12,4 +12,25 @@
 example "Truffle::Primitive.create_simple_string.getbyte(0)", simple_string.getbyte(0)
 
 example "'abc'.length", 3
-tagged_example "'abc' == 'abc'", true # seems to fail sometimes
+example "'こにちわ'.length", 4
+
+example "'abc'.bytesize", 3
+example "'こにちわ'.bytesize", 12
+
+example "'abc' == 'abc'", true
+example "x = 'abc'; x == x", true
+example "x = 'abc'; x == x.dup", true
+example "x = 'abc'; 'abc' == x.dup", true
+
+example "'abc'.ascii_only?", true
+example "'こにちわ'.ascii_only?", false
+
+example "''.ascii?", false
+example "'abc'.ascii?", true
+
+example "'abc'.valid_encoding?", true
+example "'こにちわ'.valid_encoding?", true
+
+example "''.empty?", true
+example "'abc'.empty?", false
+example "'こにちわ'.empty?", false
diff --git a/truffle/src/main/java/org/jruby/truffle/format/nodes/read/ReadBase64StringNode.java b/truffle/src/main/java/org/jruby/truffle/format/nodes/read/ReadBase64StringNode.java
@@ -20,6 +20,7 @@
 import org.jruby.truffle.format.nodes.SourceNode;
 import org.jruby.truffle.format.runtime.exceptions.FormatException;
 import org.jruby.truffle.runtime.RubyContext;
+import org.jruby.truffle.runtime.core.StringOperations;
 import org.jruby.truffle.runtime.layouts.Layouts;
 import org.jruby.util.ByteList;
 import org.jruby.util.Pack;
@@ -168,7 +169,7 @@ public Object read(VirtualFrame frame, byte[] source) {
         final ByteList result = new ByteList(lElem, 0, index, encoding, false);
         setSourcePosition(frame, encode.position());
 
-        return Layouts.STRING.createString(getContext().getCoreLibrary().getStringFactory(), result, StringSupport.CR_UNKNOWN, null);
+        return Layouts.STRING.createString(getContext().getCoreLibrary().getStringFactory(), StringOperations.ropeFromByteList(result, StringSupport.CR_UNKNOWN), null);
     }
 
 }
diff --git a/truffle/src/main/java/org/jruby/truffle/format/nodes/read/ReadBinaryStringNode.java b/truffle/src/main/java/org/jruby/truffle/format/nodes/read/ReadBinaryStringNode.java
@@ -18,6 +18,7 @@
 import org.jruby.truffle.format.nodes.SourceNode;
 import org.jruby.truffle.format.runtime.MissingValue;
 import org.jruby.truffle.runtime.RubyContext;
+import org.jruby.truffle.runtime.core.StringOperations;
 import org.jruby.truffle.runtime.layouts.Layouts;
 import org.jruby.util.ByteList;
 import org.jruby.util.StringSupport;
@@ -109,7 +110,7 @@ public Object read(VirtualFrame frame, byte[] source) {
 
         setSourcePosition(frame, start + length);
 
-        return Layouts.STRING.createString(getContext().getCoreLibrary().getStringFactory(), result, StringSupport.CR_UNKNOWN, null);
+        return Layouts.STRING.createString(getContext().getCoreLibrary().getStringFactory(), StringOperations.ropeFromByteList(result, StringSupport.CR_UNKNOWN), null);
     }
 
 }
diff --git a/truffle/src/main/java/org/jruby/truffle/format/nodes/read/ReadBitStringNode.java b/truffle/src/main/java/org/jruby/truffle/format/nodes/read/ReadBitStringNode.java
@@ -17,6 +17,7 @@
 import org.jruby.truffle.format.nodes.PackNode;
 import org.jruby.truffle.format.nodes.SourceNode;
 import org.jruby.truffle.runtime.RubyContext;
+import org.jruby.truffle.runtime.core.StringOperations;
 import org.jruby.truffle.runtime.layouts.Layouts;
 import org.jruby.util.ByteList;
 import org.jruby.util.StringSupport;
@@ -83,7 +84,7 @@ public Object read(VirtualFrame frame, byte[] source) {
         final ByteList result = new ByteList(lElem, encoding, false);
         setSourcePosition(frame, encode.position());
 
-        return Layouts.STRING.createString(getContext().getCoreLibrary().getStringFactory(), result, StringSupport.CR_UNKNOWN, null);
+        return Layouts.STRING.createString(getContext().getCoreLibrary().getStringFactory(), StringOperations.ropeFromByteList(result, StringSupport.CR_UNKNOWN), null);
     }
 
 }
diff --git a/truffle/src/main/java/org/jruby/truffle/format/nodes/read/ReadHexStringNode.java b/truffle/src/main/java/org/jruby/truffle/format/nodes/read/ReadHexStringNode.java
@@ -17,6 +17,7 @@
 import org.jruby.truffle.format.nodes.PackNode;
 import org.jruby.truffle.format.nodes.SourceNode;
 import org.jruby.truffle.runtime.RubyContext;
+import org.jruby.truffle.runtime.core.StringOperations;
 import org.jruby.truffle.runtime.layouts.Layouts;
 import org.jruby.util.ByteList;
 import org.jruby.util.Pack;
@@ -84,7 +85,7 @@ public Object read(VirtualFrame frame, byte[] source) {
         final ByteList result = new ByteList(lElem, encoding, false);
         setSourcePosition(frame, encode.position());
 
-        return Layouts.STRING.createString(getContext().getCoreLibrary().getStringFactory(), result, StringSupport.CR_UNKNOWN, null);
+        return Layouts.STRING.createString(getContext().getCoreLibrary().getStringFactory(), StringOperations.ropeFromByteList(result, StringSupport.CR_UNKNOWN), null);
     }
 
 }
diff --git a/truffle/src/main/java/org/jruby/truffle/format/nodes/read/ReadMIMEStringNode.java b/truffle/src/main/java/org/jruby/truffle/format/nodes/read/ReadMIMEStringNode.java
@@ -18,6 +18,7 @@
 import org.jruby.truffle.format.nodes.PackNode;
 import org.jruby.truffle.format.nodes.SourceNode;
 import org.jruby.truffle.runtime.RubyContext;
+import org.jruby.truffle.runtime.core.StringOperations;
 import org.jruby.truffle.runtime.layouts.Layouts;
 import org.jruby.util.ByteList;
 import org.jruby.util.Pack;
@@ -75,7 +76,7 @@ public Object read(VirtualFrame frame, byte[] source) {
         final ByteList result = new ByteList(lElem, 0, index, encoding, false);
         setSourcePosition(frame, encode.position());
 
-        return Layouts.STRING.createString(getContext().getCoreLibrary().getStringFactory(), result, StringSupport.CR_UNKNOWN, null);
+        return Layouts.STRING.createString(getContext().getCoreLibrary().getStringFactory(), StringOperations.ropeFromByteList(result, StringSupport.CR_UNKNOWN), null);
     }
 
 }
diff --git a/truffle/src/main/java/org/jruby/truffle/format/nodes/read/ReadUUStringNode.java b/truffle/src/main/java/org/jruby/truffle/format/nodes/read/ReadUUStringNode.java
@@ -21,6 +21,7 @@
 import org.jruby.truffle.format.nodes.SourceNode;
 import org.jruby.truffle.format.runtime.exceptions.NoImplicitConversionException;
 import org.jruby.truffle.runtime.RubyContext;
+import org.jruby.truffle.runtime.core.StringOperations;
 import org.jruby.truffle.runtime.layouts.Layouts;
 import org.jruby.util.ByteList;
 import org.jruby.util.Pack;
@@ -156,7 +157,7 @@ else if (encode.hasRemaining()) {
 
         setSourcePosition(frame, encode.position());
 
-        return Layouts.STRING.createString(getContext().getCoreLibrary().getStringFactory(), result, StringSupport.CR_UNKNOWN, null);
+        return Layouts.STRING.createString(getContext().getCoreLibrary().getStringFactory(), StringOperations.ropeFromByteList(result, StringSupport.CR_UNKNOWN), null);
     }
 
     private static int safeGet(ByteBuffer encode) {

diff --git a/truffle/src/main/java/org/jruby/truffle/format/nodes/type/ToStringNode.java b/truffle/src/main/java/org/jruby/truffle/format/nodes/type/ToStringNode.java
@@ -99,7 +99,7 @@ public ByteList toStringString(VirtualFrame frame, DynamicObject string) {
             setTainted(frame);
         }
 
-        return StringOperations.getByteList(string);
+        return StringOperations.getByteListReadOnly(string);
     }
 
     @Specialization(guards = "isRubyArray(array)")
@@ -116,7 +116,7 @@ public ByteList toString(VirtualFrame frame, DynamicObject array) {
                 setTainted(frame);
             }
 
-            return StringOperations.getByteList((DynamicObject) value);
+            return StringOperations.getByteListReadOnly((DynamicObject) value);
         }
 
         CompilerDirectives.transferToInterpreter();
@@ -142,7 +142,7 @@ public ByteList toString(VirtualFrame frame, Object object) {
                 setTainted(frame);
             }
 
-            return StringOperations.getByteList((DynamicObject) value);
+            return StringOperations.getByteListReadOnly((DynamicObject) value);
         }
 
         if (inspectOnConversionFailure) {
@@ -152,7 +152,7 @@ public ByteList toString(VirtualFrame frame, Object object) {
                         getEncapsulatingSourceSection(), new RubyNode[]{null}));
             }
 
-            return StringOperations.getByteList(inspectNode.toS(frame, object));
+            return StringOperations.getByteListReadOnly(inspectNode.toS(frame, object));
         }
 
         CompilerDirectives.transferToInterpreter();

diff --git a/truffle/src/main/java/org/jruby/truffle/nodes/RubyNode.java b/truffle/src/main/java/org/jruby/truffle/nodes/RubyNode.java
@@ -24,6 +24,7 @@
 import org.jruby.truffle.runtime.RubyContext;
 import org.jruby.truffle.runtime.core.StringOperations;
 import org.jruby.truffle.runtime.layouts.Layouts;
+import org.jruby.truffle.runtime.rope.Rope;
 import org.jruby.truffle.runtime.sockets.NativeSockets;
 import org.jruby.util.ByteList;
 
@@ -156,6 +157,10 @@ public DynamicObject getSymbol(ByteList name) {
         return getContext().getSymbol(name);
     }
 
+    public DynamicObject getSymbol(Rope name) {
+        return getContext().getSymbol(name);
+    }
+
     /** Creates a String from the ByteList, with unknown CR */
     protected DynamicObject createString(ByteList bytes) {
         return StringOperations.createString(getContext(), bytes);
@@ -166,6 +171,10 @@ protected DynamicObject create7BitString(ByteList bytes) {
         return StringOperations.create7BitString(getContext(), bytes);
     }
 
+    protected DynamicObject createString(Rope rope) {
+        return StringOperations.createString(getContext(), rope);
+    }
+
     protected POSIX posix() {
         return getContext().getPosix();
     }

diff --git a/truffle/src/main/java/org/jruby/truffle/nodes/StringCachingGuards.java b/truffle/src/main/java/org/jruby/truffle/nodes/StringCachingGuards.java
@@ -12,33 +12,38 @@
 
 import com.oracle.truffle.api.object.DynamicObject;
 import org.jruby.truffle.runtime.core.StringOperations;
+import org.jruby.truffle.runtime.layouts.Layouts;
+import org.jruby.truffle.runtime.rope.Rope;
 import org.jruby.util.ByteList;
 
 public abstract class StringCachingGuards {
 
-    public static ByteList privatizeByteList(DynamicObject string) {
+    public static Rope privatizeRope(DynamicObject string) {
         if (RubyGuards.isRubyString(string)) {
-            return StringOperations.getByteList(string).dup();
+            // TODO (nirvdrum 25-Jan-16) Should we flatten the rope to avoid caching a potentially deep rope tree?
+            return StringOperations.rope(string);
         } else {
             return null;
         }
     }
 
-    public static boolean byteListsEqual(DynamicObject string, ByteList byteList) {
+    public static boolean ropesEqual(DynamicObject string, Rope rope) {
         if (RubyGuards.isRubyString(string)) {
+            final Rope stringRope = StringOperations.rope(string);
+
             // equal below does not check encoding
-            if (StringOperations.getByteList(string).getEncoding() != byteList.getEncoding()) {
+            if (stringRope.getEncoding() != rope.getEncoding()) {
                 return false;
             }
-            // TODO CS 8-Nov-15 this code goes off into the woods - need to break it apart and branch profile it
-            return StringOperations.getByteList(string).equal(byteList);
+
+            return stringRope.equals(rope);
         } else {
             return false;
         }
     }
 
-    public static int byteListLength(ByteList byteList) {
-        return byteList.length();
+    public static int ropeLength(Rope rope) {
+        return rope.byteLength();
     }
 
 }
diff --git a/truffle/src/main/java/org/jruby/truffle/nodes/conversion/ToSymbolNode.java b/truffle/src/main/java/org/jruby/truffle/nodes/conversion/ToSymbolNode.java
@@ -36,7 +36,7 @@ protected DynamicObject toSymbolSymbol(DynamicObject symbol) {
 
     @Specialization(guards = "isRubyString(string)")
     protected DynamicObject toSymbolString(DynamicObject string) {
-        return getSymbol(StringOperations.getByteList(string));
+        return getSymbol(StringOperations.rope(string));
     }
 
     @Specialization

diff --git a/truffle/src/main/java/org/jruby/truffle/nodes/core/BasicObjectNodes.java b/truffle/src/main/java/org/jruby/truffle/nodes/core/BasicObjectNodes.java
@@ -152,7 +152,7 @@ public InstanceEvalNode(RubyContext context, SourceSection sourceSection) {
         @CompilerDirectives.TruffleBoundary
         @Specialization(guards = "isRubyString(string)")
         public Object instanceEval(Object receiver, DynamicObject string, NotProvided block) {
-            return getContext().instanceEval(StringOperations.getByteList(string), receiver, "(eval)", this);
+            return getContext().instanceEval(StringOperations.getByteListReadOnly(string), receiver, "(eval)", this);
         }
 
         @Specialization