Use RopeOperations.decodeAscii() instead of new String() without a Charset

eregon · eregon · commit 7b8a3eafba9e · 2020-09-15T17:36:12.000+02:00
diff --git a/mx.truffleruby/spotbugs-filters.xml b/mx.truffleruby/spotbugs-filters.xml
@@ -10,6 +10,8 @@
         <Bug pattern="MS_MUTABLE_COLLECTION" />
         <Bug pattern="SF_SWITCH_NO_DEFAULT" />
         <Bug pattern="RI_REDUNDANT_INTERFACES" />
+        <Bug pattern="EQ_COMPARETO_USE_OBJECT_EQUALS" />
+        <Bug pattern="SE_BAD_FIELD" />
         <!-- Intentional -->
         <Bug pattern="BC_UNCONFIRMED_CAST" />
         <Bug pattern="BC_UNCONFIRMED_CAST_OF_RETURN_VALUE" />
@@ -18,6 +20,7 @@
         <Bug pattern="EI_EXPOSE_REP" />
         <Bug pattern="EI_EXPOSE_REP2" />
         <Bug pattern="FE_FLOATING_POINT_EQUALITY" />
+        <Bug pattern="NM_CLASS_NOT_EXCEPTION" />
       </Or>
     </Not>
   </Match>
diff --git a/src/main/java/org/truffleruby/core/encoding/EncodingConverterNodes.java b/src/main/java/org/truffleruby/core/encoding/EncodingConverterNodes.java
@@ -117,7 +117,7 @@ protected Object initialize(
 
                 final byte[] segmentSource = transcoder.getSource();
                 ret[retIndex++] = getSymbol(
-                        RopeOperations.decodeAscii(segmentSource, 0, segmentSource.length).toUpperCase());
+                        RopeOperations.decodeAscii(segmentSource).toUpperCase());
             }
 
             final int retSize = retIndex + 1;
@@ -129,7 +129,7 @@ protected Object initialize(
 
             final byte[] destinationName = destinationEncoding.getName();
             ret[retIndex] = getSymbol(
-                    RopeOperations.decodeAscii(destinationName, 0, destinationName.length).toUpperCase());
+                    RopeOperations.decodeAscii(destinationName).toUpperCase());
 
             return createArray(ret);
         }
diff --git a/src/main/java/org/truffleruby/core/encoding/TranscodingManager.java b/src/main/java/org/truffleruby/core/encoding/TranscodingManager.java
@@ -45,6 +45,8 @@
 import org.jcodings.util.Hash;
 
 import com.oracle.truffle.api.TruffleOptions;
+import org.truffleruby.core.rope.RopeOperations;
+import org.truffleruby.core.string.StringUtils;
 
 public class TranscodingManager {
 
@@ -56,8 +58,8 @@ public class TranscodingManager {
             for (Hash.HashEntry<TranscoderDB.Entry> destinationEntry : sourceEntry.entryIterator()) {
                 final TranscoderDB.Entry e = destinationEntry.value;
 
-                final String sourceName = new String(e.getSource()).toUpperCase();
-                final String destinationName = new String(e.getDestination()).toUpperCase();
+                final String sourceName = StringUtils.toUpperCase(RopeOperations.decodeAscii(e.getSource()));
+                final String destinationName = StringUtils.toUpperCase(RopeOperations.decodeAscii(e.getDestination()));
 
                 if (TruffleOptions.AOT) {
                     // Load the classes eagerly
diff --git a/src/main/java/org/truffleruby/core/format/format/FormatCharacterNode.java b/src/main/java/org/truffleruby/core/format/format/FormatCharacterNode.java
@@ -18,6 +18,7 @@
 import org.truffleruby.core.format.convert.ToStringNodeGen;
 import org.truffleruby.core.format.exceptions.NoImplicitConversionException;
 import org.truffleruby.core.format.write.bytes.WriteByteNodeGen;
+import org.truffleruby.core.rope.RopeOperations;
 import org.truffleruby.core.string.StringUtils;
 import org.truffleruby.language.RubyGuards;
 import org.truffleruby.language.control.RaiseException;
@@ -83,7 +84,8 @@ protected String getCharString(VirtualFrame frame, Object value) {
             // TODO BJF check char length is > 0
             charString = Character.toString((char) charValue);
         } else {
-            final String resultString = StringUtils.create((byte[]) toStrResult);
+            byte[] bytes = (byte[]) toStrResult;
+            final String resultString = RopeOperations.decodeAscii(bytes);
             final int size = resultString.length();
             if (size > 1) {
                 throw new RaiseException(
diff --git a/src/main/java/org/truffleruby/core/rope/RopeOperations.java b/src/main/java/org/truffleruby/core/rope/RopeOperations.java
@@ -152,11 +152,17 @@ public static byte[] encodeAsciiBytes(String value) {
         return bytes;
     }
 
+    public static String decodeAscii(byte[] bytes) {
+        return decodeAscii(bytes, 0, bytes.length);
+    }
+
     public static String decodeAscii(byte[] bytes, int byteOffset, int byteLength) {
         final char[] buffer = new char[byteLength];
 
         for (int i = 0; i < byteLength; i++) {
-            buffer[i] = (char) bytes[byteOffset + i];
+            byte b = bytes[byteOffset + i];
+            assert b >= 0;
+            buffer[i] = (char) b;
         }
 
         return newString(buffer);
diff --git a/src/main/java/org/truffleruby/core/string/EncodingUtils.java b/src/main/java/org/truffleruby/core/string/EncodingUtils.java
@@ -33,6 +33,7 @@
 import org.jcodings.ascii.AsciiTables;
 import org.jcodings.specific.ASCIIEncoding;
 import org.truffleruby.core.rope.CodeRange;
+import org.truffleruby.core.rope.RopeOperations;
 
 public class EncodingUtils {
 
@@ -80,7 +81,7 @@ public static List<String> encodingNames(byte[] name, int p, int end) {
         boolean isValid = false;
         if (s >= end) {
             isValid = true;
-            names.add(new String(name, p, end));
+            names.add(RopeOperations.decodeAscii(name, p, end));
         }
 
         if (!isValid || hasLower) {
@@ -111,7 +112,7 @@ public static List<String> encodingNames(byte[] name, int p, int end) {
                     }
                 }
                 if (hasUpper) {
-                    names.add(new String(constName, 0, constName.length));
+                    names.add(RopeOperations.decodeAscii(constName));
                 }
             }
             if (hasLower) {
@@ -121,7 +122,7 @@ public static List<String> encodingNames(byte[] name, int p, int end) {
                         constName[s] = AsciiTables.ToUpperCaseTable[code];
                     }
                 }
-                names.add(new String(constName, 0, constName.length));
+                names.add(RopeOperations.decodeAscii(constName));
             }
         }
 
diff --git a/src/main/java/org/truffleruby/core/string/StringUtils.java b/src/main/java/org/truffleruby/core/string/StringUtils.java
@@ -11,18 +11,14 @@
 
 import java.util.Locale;
 
-import com.oracle.truffle.api.CompilerDirectives.TruffleBoundary;
 import org.truffleruby.core.rope.RopeOperations;
 
+import com.oracle.truffle.api.CompilerDirectives.TruffleBoundary;
+
 public abstract class StringUtils {
 
     public static final String[] EMPTY_STRING_ARRAY = new String[0];
 
-    @TruffleBoundary
-    public static String create(byte[] bytes) {
-        return new String(bytes);
-    }
-
     @TruffleBoundary
     public static String toString(Object value) {
         return value.toString();
@@ -57,4 +53,9 @@ public static String toLowerCase(String string) {
         return string.toLowerCase(Locale.ENGLISH);
     }
 
+    @TruffleBoundary
+    public static String toUpperCase(String string) {
+        return string.toUpperCase(Locale.ENGLISH);
+    }
+
 }
diff --git a/src/main/java/org/truffleruby/core/time/RubyTimeOutputFormatter.java b/src/main/java/org/truffleruby/core/time/RubyTimeOutputFormatter.java
@@ -26,6 +26,7 @@
  ***** END LICENSE BLOCK *****/
 package org.truffleruby.core.time;
 
+import org.truffleruby.core.string.StringUtils;
 import org.truffleruby.core.time.RubyDateFormatter.FieldType;
 
 /** Support for GNU-C output formatters, see:
@@ -84,9 +85,9 @@ public String format(String sequence, long value, FieldType type) {
                 case '#': // change case
                     char last = sequence.charAt(sequence.length() - 1);
                     if (Character.isLowerCase(last)) {
-                        sequence = sequence.toUpperCase();
+                        sequence = StringUtils.toUpperCase(sequence);
                     } else {
-                        sequence = sequence.toLowerCase();
+                        sequence = StringUtils.toLowerCase(sequence);
                     }
                     break;
             }
diff --git a/src/main/java/org/truffleruby/interop/ToJavaStringNode.java b/src/main/java/org/truffleruby/interop/ToJavaStringNode.java
@@ -61,7 +61,7 @@ protected String stringUncached(RubyString value,
         final byte[] bytes = bytesNode.execute(rope);
 
         if (asciiOnlyProfile.profile(asciiOnlyNode.execute(rope))) {
-            return RopeOperations.decodeAscii(bytes, 0, bytes.length);
+            return RopeOperations.decodeAscii(bytes);
         } else {
             return RopeOperations.decodeNonAscii(rope.getEncoding(), bytes, 0, bytes.length);
         }

Original file line number	Diff line number	Diff line change
`@@ -117,7 +117,7 @@ protected Object initialize(`
`117`	`117`
`118`	`118`	`final byte[] segmentSource = transcoder.getSource();`
`119`	`119`	`ret[retIndex++] = getSymbol(`
`120`		`- RopeOperations.decodeAscii(segmentSource, 0, segmentSource.length).toUpperCase());`
	`120`	`+ RopeOperations.decodeAscii(segmentSource).toUpperCase());`
`121`	`121`	`}`
`122`	`122`
`123`	`123`	`final int retSize = retIndex + 1;`
`@@ -129,7 +129,7 @@ protected Object initialize(`
`129`	`129`
`130`	`130`	`final byte[] destinationName = destinationEncoding.getName();`
`131`	`131`	`ret[retIndex] = getSymbol(`
`132`		`- RopeOperations.decodeAscii(destinationName, 0, destinationName.length).toUpperCase());`
	`132`	`+ RopeOperations.decodeAscii(destinationName).toUpperCase());`
`133`	`133`
`134`	`134`	`return createArray(ret);`
`135`	`135`	`}`
Original file line number	Diff line number	Diff line change
`@@ -33,6 +33,7 @@`
`33`	`33`	`import org.jcodings.ascii.AsciiTables;`
`34`	`34`	`import org.jcodings.specific.ASCIIEncoding;`
`35`	`35`	`import org.truffleruby.core.rope.CodeRange;`
	`36`	`+import org.truffleruby.core.rope.RopeOperations;`
`36`	`37`
`37`	`38`	`public class EncodingUtils {`
`38`	`39`
`@@ -80,7 +81,7 @@ public static List<String> encodingNames(byte[] name, int p, int end) {`
`80`	`81`	`boolean isValid = false;`
`81`	`82`	`if (s >= end) {`
`82`	`83`	`isValid = true;`
`83`		`- names.add(new String(name, p, end));`
	`84`	`+ names.add(RopeOperations.decodeAscii(name, p, end));`
`84`	`85`	`}`
`85`	`86`
`86`	`87`	`if (!isValid \|\| hasLower) {`
`@@ -111,7 +112,7 @@ public static List<String> encodingNames(byte[] name, int p, int end) {`
`111`	`112`	`}`
`112`	`113`	`}`
`113`	`114`	`if (hasUpper) {`
`114`		`- names.add(new String(constName, 0, constName.length));`
	`115`	`+ names.add(RopeOperations.decodeAscii(constName));`
`115`	`116`	`}`
`116`	`117`	`}`
`117`	`118`	`if (hasLower) {`
`@@ -121,7 +122,7 @@ public static List<String> encodingNames(byte[] name, int p, int end) {`
`121`	`122`	`constName[s] = AsciiTables.ToUpperCaseTable[code];`
`122`	`123`	`}`
`123`	`124`	`}`
`124`		`- names.add(new String(constName, 0, constName.length));`
	`125`	`+ names.add(RopeOperations.decodeAscii(constName));`
`125`	`126`	`}`
`126`	`127`	`}`
`127`	`128`
Original file line number	Diff line number	Diff line change
`@@ -61,7 +61,7 @@ protected String stringUncached(RubyString value,`
`61`	`61`	`final byte[] bytes = bytesNode.execute(rope);`
`62`	`62`
`63`	`63`	`if (asciiOnlyProfile.profile(asciiOnlyNode.execute(rope))) {`
`64`		`- return RopeOperations.decodeAscii(bytes, 0, bytes.length);`
	`64`	`+ return RopeOperations.decodeAscii(bytes);`
`65`	`65`	`} else {`
`66`	`66`	`return RopeOperations.decodeNonAscii(rope.getEncoding(), bytes, 0, bytes.length);`
`67`	`67`	`}`