[GR-19220] Add String#bytesplice

PullRequest: truffleruby/3828

Author: andrykonchin

CHANGELOG.md

@@ -17,6 +17,7 @@ Compatibility:
 * Fix `Range#size` and return `nil` for beginningless Range when end isn't Numeric (#3039, @rwstauner).
 * Alias `String#-@` to `String#dedup` (#3039, @itarato).
 * Fix `Pathname#relative_path_from` to convert string arguments to Pathname objects (@rwstauner).
+* Add `String#bytesplice` (#3039, @itarato).
 
 Performance:
 

spec/ruby/core/string/bytesplice_spec.rb

@@ -0,0 +1,133 @@
+# -*- encoding: utf-8 -*-
+require_relative '../../spec_helper'
+
+describe "String#bytesplice" do
+  ruby_version_is "3.2" do
+    it "raises IndexError when index is less than -bytesize" do
+      -> { "hello".bytesplice(-6, 0, "xxx") }.should raise_error(IndexError, "index -6 out of string")
+    end
+
+    it "raises IndexError when index is greater than bytesize" do
+      -> { "hello".bytesplice(6, 0, "xxx") }.should raise_error(IndexError, "index 6 out of string")
+    end
+
+    it "raises IndexError for negative length" do
+      -> { "abc".bytesplice(0, -2, "") }.should raise_error(IndexError, "negative length -2")
+    end
+
+    it "replaces with integer indices" do
+      "hello".bytesplice(-5, 0, "xxx").should == "xxxhello"
+      "hello".bytesplice(0, 0, "xxx").should == "xxxhello"
+      "hello".bytesplice(0, 1, "xxx").should == "xxxello"
+      "hello".bytesplice(0, 5, "xxx").should == "xxx"
+      "hello".bytesplice(0, 6, "xxx").should == "xxx"
+    end
+
+    it "raises RangeError when range left boundary is less than -bytesize" do
+      -> { "hello".bytesplice(-6...-6, "xxx") }.should raise_error(RangeError, "-6...-6 out of range")
+    end
+
+    it "replaces with ranges" do
+      "hello".bytesplice(-5...-5, "xxx").should == "xxxhello"
+      "hello".bytesplice(0...0, "xxx").should == "xxxhello"
+      "hello".bytesplice(0..0, "xxx").should == "xxxello"
+      "hello".bytesplice(0...1, "xxx").should == "xxxello"
+      "hello".bytesplice(0..1, "xxx").should == "xxxllo"
+      "hello".bytesplice(0..-1, "xxx").should == "xxx"
+      "hello".bytesplice(0...5, "xxx").should == "xxx"
+      "hello".bytesplice(0...6, "xxx").should == "xxx"
+    end
+
+    it "raises TypeError when integer index is provided without length argument" do
+      -> { "hello".bytesplice(0, "xxx") }.should raise_error(TypeError, "wrong argument type Integer (expected Range)")
+    end
+
+    it "replaces on an empty string" do
+      "".bytesplice(0, 0, "").should == ""
+      "".bytesplice(0, 0, "xxx").should == "xxx"
+    end
+
+    it "mutates self" do
+      s = "hello"
+      s.bytesplice(2, 1, "xxx").should.equal?(s)
+    end
+
+    it "raises when string is frozen" do
+      s = "hello".freeze
+      -> { s.bytesplice(2, 1, "xxx") }.should raise_error(FrozenError, "can't modify frozen String: \"hello\"")
+    end
+  end
+end
+
+describe "String#bytesplice with multibyte characters" do
+  ruby_version_is "3.2" do
+    it "raises IndexError when index is out of byte size boundary" do
+      -> { "こんにちは".bytesplice(-16, 0, "xxx") }.should raise_error(IndexError, "index -16 out of string")
+    end
+
+    it "raises IndexError when index is not on a codepoint boundary" do
+      -> { "こんにちは".bytesplice(1, 0, "xxx") }.should raise_error(IndexError, "offset 1 does not land on character boundary")
+    end
+
+    it "raises IndexError when length is not matching the codepoint boundary" do
+      -> { "こんにちは".bytesplice(0, 1, "xxx") }.should raise_error(IndexError, "offset 1 does not land on character boundary")
+      -> { "こんにちは".bytesplice(0, 2, "xxx") }.should raise_error(IndexError, "offset 2 does not land on character boundary")
+    end
+
+    it "replaces with integer indices" do
+      "こんにちは".bytesplice(-15, 0, "xxx").should == "xxxこんにちは"
+      "こんにちは".bytesplice(0, 0, "xxx").should == "xxxこんにちは"
+      "こんにちは".bytesplice(0, 3, "xxx").should == "xxxんにちは"
+      "こんにちは".bytesplice(3, 3, "はは").should == "こははにちは"
+      "こんにちは".bytesplice(15, 0, "xxx").should == "こんにちはxxx"
+    end
+
+    it "replaces with range" do
+      "こんにちは".bytesplice(-15...-16, "xxx").should == "xxxこんにちは"
+      "こんにちは".bytesplice(0...0, "xxx").should == "xxxこんにちは"
+      "こんにちは".bytesplice(0..2, "xxx").should == "xxxんにちは"
+      "こんにちは".bytesplice(0...3, "xxx").should == "xxxんにちは"
+      "こんにちは".bytesplice(0..5, "xxx").should == "xxxにちは"
+      "こんにちは".bytesplice(0..-1, "xxx").should == "xxx"
+      "こんにちは".bytesplice(0...15, "xxx").should == "xxx"
+      "こんにちは".bytesplice(0...18, "xxx").should == "xxx"
+    end
+
+    it "treats negative length for range as 0" do
+      "こんにちは".bytesplice(0...-100, "xxx").should == "xxxこんにちは"
+      "こんにちは".bytesplice(3...-100, "xxx").should == "こxxxんにちは"
+      "こんにちは".bytesplice(-15...-100, "xxx").should == "xxxこんにちは"
+    end
+
+    it "raises when ranges not match codepoint boundaries" do
+      -> { "こんにちは".bytesplice(0..0, "x") }.should raise_error(IndexError, "offset 1 does not land on character boundary")
+      -> { "こんにちは".bytesplice(0..1, "x") }.should raise_error(IndexError, "offset 2 does not land on character boundary")
+      # Begin is incorrect
+      -> { "こんにちは".bytesplice(-4..-1, "x") }.should raise_error(IndexError, "offset 11 does not land on character boundary")
+      -> { "こんにちは".bytesplice(-5..-1, "x") }.should raise_error(IndexError, "offset 10 does not land on character boundary")
+      # End is incorrect
+      -> { "こんにちは".bytesplice(-3..-2, "x") }.should raise_error(IndexError, "offset 14 does not land on character boundary")
+      -> { "こんにちは".bytesplice(-3..-3, "x") }.should raise_error(IndexError, "offset 13 does not land on character boundary")
+    end
+
+    it "deals with a different encoded argument" do
+      s = "こんにちは"
+      s.encoding.should == Encoding::UTF_8
+      sub = "xxxxxx"
+      sub.force_encoding(Encoding::US_ASCII)
+
+      result = s.bytesplice(0, 3, sub)
+      result.should == "xxxxxxんにちは"
+      result.encoding.should == Encoding::UTF_8
+
+      s = "xxxxxx"
+      s.force_encoding(Encoding::US_ASCII)
+      sub = "こんにちは"
+      sub.encoding.should == Encoding::UTF_8
+
+      result = s.bytesplice(0, 3, sub)
+      result.should == "こんにちはxxx"
+      result.encoding.should == Encoding::UTF_8
+    end
+  end
+end

spec/tags/truffle/methods_tags.txt

@@ -113,3 +113,4 @@ fails:Public methods on Thread should include native_thread_id
 fails:Public methods on UnboundMethod should include private?
 fails:Public methods on UnboundMethod should include protected?
 fails:Public methods on UnboundMethod should include public?
+fails:Public methods on String should not include bytesplice

spec/truffleruby.next-specs

@@ -16,3 +16,4 @@ spec/ruby/core/hash/shift_spec.rb
 spec/ruby/core/range/size_spec.rb
 
 spec/ruby/core/string/dedup_spec.rb
+spec/ruby/core/string/bytesplice_spec.rb

src/main/java/org/truffleruby/core/string/StringNodes.java

@@ -4459,4 +4459,16 @@ private String formatTooLongError(int count, RubyString string) {
 
     }
 
+    @Primitive(name = "string_is_character_head?", lowerFixnum = 2)
+    public abstract static class IsCharacterHeadPrimitiveNode extends PrimitiveArrayArgumentsNode {
+
+        @Specialization
+        protected boolean isCharacterHead(RubyEncoding enc, Object string, int byteOffset,
+                @Cached RubyStringLibrary libString,
+                @Cached IsCharacterHeadNode isCharacterHeadNode) {
+            var tstring = libString.getTString(string);
+            return isCharacterHeadNode.execute(enc, tstring, byteOffset);
+        }
+    }
+
 }

src/main/ruby/truffleruby/core/string.rb

@@ -70,6 +70,49 @@ def byteslice(index_or_range, length = undefined)
     byteslice index, length
   end
 
+  def bytesplice(index_or_range, length = undefined, str)
+    is_range = Primitive.is_a?(index_or_range, Range)
+
+    if Primitive.undefined?(length)
+      raise TypeError, "wrong argument type #{Primitive.class(index_or_range)} (expected Range)" unless is_range
+
+      start, len = Primitive.range_normalized_start_length(index_or_range, bytesize)
+      len = 0 if len < 0
+    else
+      start = Primitive.rb_to_int(index_or_range)
+      start += bytesize if start < 0
+      len = Primitive.rb_to_int(length)
+    end
+
+    str = StringValue(str)
+
+    if len < 0
+      raise IndexError, "negative length #{len}"
+    end
+
+    if bytesize < start || start < 0
+      if is_range
+        raise RangeError, "#{index_or_range} out of range"
+      else
+        raise IndexError, "index #{index_or_range} out of string"
+      end
+    end
+
+    len = bytesize - start if len > bytesize - start
+    finish = start + len
+
+    if start < bytesize && !Primitive.string_is_character_head?(encoding, self, start)
+      raise IndexError, "offset #{start} does not land on character boundary"
+    end
+    if finish < bytesize && !Primitive.string_is_character_head?(encoding, self, finish)
+      raise IndexError, "offset #{finish} does not land on character boundary"
+    end
+
+    Primitive.check_mutable_string(self)
+    enc = Primitive.encoding_ensure_compatible_str(self, str)
+    Primitive.string_splice(self, str, start, len, enc)
+  end
+
   def self.try_convert(obj)
     Truffle::Type.try_convert obj, String, :to_str
   end

src/main/ruby/truffleruby/core/truffle/polyglot_methods.rb

@@ -80,6 +80,10 @@ def byteslice(...)
       to_s.byteslice(...)
     end
 
+    def bytesplice(...)
+      to_s.bytesplice(...)
+    end
+
     def capitalize(...)
       to_s.capitalize(...)
     end