[Truffle] Pulled in String#index from Rubinius.

Author: nirvdrum

spec/truffle/tags/core/string/index_tags.txt

@@ -1,9 +1,5 @@
-fails:String#index raises a TypeError if passed nil
-fails:String#index raises a TypeError if passed a boolean
-fails:String#index raises a TypeError if passed a Symbol
 fails:String#index calls #to_str to convert the first argument
 fails:String#index calls #to_int to convert the second argument
-fails:String#index raises a TypeError if passed a Fixnum
 fails:String#index with String behaves the same as String#index(char) for one-character strings
 fails:String#index with String returns the index of the first occurrence of the given substring
 fails:String#index with String doesn't set $~
@@ -14,17 +10,9 @@ fails:String#index with String returns nil if the substring isn't found
 fails:String#index with String returns the character index of a multibyte character
 fails:String#index with String returns the character index after offset
 fails:String#index with String returns the character index after a partial first match
-fails:String#index with String raises an Encoding::CompatibilityError if the encodings are incompatible
 fails:String#index with Regexp behaves the same as String#index(string) for escaped string regexps
-fails:String#index with Regexp returns the index of the first match of regexp
-fails:String#index with Regexp sets $~ to MatchData of match and nil when there's none
-fails:String#index with Regexp starts the search at the given offset
 fails:String#index with Regexp starts the search at offset + self.length if offset is negative
-fails:String#index with Regexp returns nil if the substring isn't found
-fails:String#index with Regexp returns nil if the Regexp matches the empty string and the offset is out of range
 fails:String#index with Regexp supports \G which matches at the given start offset
-fails:String#index with Regexp converts start_offset to an integer via to_int
 fails:String#index with Regexp returns the character index of a multibyte character
 fails:String#index with Regexp returns the character index after offset
 fails:String#index with Regexp treats the offset as a character index
-fails:String#index with Regexp raises an Encoding::CompatibilityError if the encodings are incompatible

truffle/src/main/java/org/jruby/truffle/nodes/rubinius/StringPrimitiveNodes.java

@@ -363,11 +363,22 @@ public StringCharacterByteIndexPrimitiveNode(StringCharacterByteIndexPrimitiveNo
             super(prev);
         }
 
-        @Specialization
-        public Object stringCharacterByteIndex(RubyString string, Object index, Object start) {
-            throw new UnsupportedOperationException("string_character_byte_index");
+        @Specialization(guards = "isSingleByteOptimizable")
+        public int stringCharacterByteIndex(RubyString string, int index, int start) {
+            return start + index;
+        }
+
+        @Specialization(guards = "!isSingleByteOptimizable")
+        public int stringCharacterByteIndexMultiByteEncoding(RubyString string, int index, int start) {
+            final ByteList bytes = string.getBytes();
+
+            return StringSupport.nth(bytes.getEncoding(), bytes.getUnsafeBytes(), bytes.getBegin(),
+                    bytes.getBegin() + bytes.getRealSize(), start + index);
         }
 
+        public static boolean isSingleByteOptimizable(RubyString string) {
+            return StringSupport.isSingleByteOptimizable(string, string.getBytes().getEncoding());
+        }
     }
 
     @RubiniusPrimitive(name = "string_byte_character_index", needsSelf = false)

truffle/src/main/ruby/core/rubinius/common/string.rb

@@ -699,6 +699,41 @@ def rjust(width, padding=" ")
     str.force_encoding enc
   end
 
+  def index(str, start=undefined)
+    if undefined.equal?(start)
+      start = 0
+    else
+      start = Rubinius::Type.coerce_to start, Fixnum, :to_int
+
+      start += size if start < 0
+      return if start < 0 or start > size
+    end
+
+    if str.kind_of? Regexp
+      Rubinius::Type.compatible_encoding self, str
+
+      m = Rubinius::Mirror.reflect self
+      start = m.character_to_byte_index start
+      if match = str.match_from(self, start)
+        Regexp.last_match = match
+        return match.begin(0)
+      else
+        Regexp.last_match = nil
+        return
+      end
+    end
+
+    str = StringValue(str)
+    return start if str == ""
+
+    Rubinius::Type.compatible_encoding self, str
+
+    return if str.size > size
+
+    m = Rubinius::Mirror.reflect self
+    m.character_index str, start
+  end
+
   def upto(stop, exclusive=false)
     return to_enum :upto, stop, exclusive unless block_given?
     stop = StringValue(stop)