_rl_adjust_point: performance fix with large point

_rl_adjust_point() takes a long time when inserting text. If point
argument is a large value, performance is poor. In this case, counting
byte size from tail of the string will improve performance. Resetting
encoding instead of String#dup will give a small improvement.

Tested-by: Mark Somerville <mark@scottishclimbs.com> [F16, MRI 1.8.7, MRI 1.9.3]
Signed-off-by: Mark Somerville <mark@scottishclimbs.com>

bench/_rl_adjust_point.rb

@@ -0,0 +1,26 @@
+$LOAD_PATH.unshift "#{File.dirname(__FILE__)}/../lib/"
+require 'rbreadline'
+require 'benchmark'
+
+N = 100_000
+
+Benchmark.bmbm do |x|
+  x.report do
+    N.times { RbReadline._rl_adjust_point("a", 0) }
+  end
+  x.report do
+    N.times { RbReadline._rl_adjust_point("a", 1) }
+  end
+  x.report do
+    N.times { RbReadline._rl_adjust_point("aaaaaaaaaaaaaaaaaaaaa", 0) }
+  end
+  x.report do
+    N.times { RbReadline._rl_adjust_point("aaaaaaaaaaaaaaaaaaaaa", 40) }
+  end
+  x.report do
+    N.times { RbReadline._rl_adjust_point("aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa", 0) }
+  end
+  x.report do
+    N.times { RbReadline._rl_adjust_point("aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa", 40) }
+  end
+end

lib/rbreadline.rb

@@ -8524,12 +8524,27 @@ def _rl_adjust_point(string, point)
         i += 1
       end
     when 'X'
-      str = string.dup.force_encoding(@encoding_name)
-      i, len = 0, str.length
-      while (pos < point && i < len)
-        pos += str[i].bytesize
-        i += 1
+      enc = string.encoding
+      str = string.force_encoding(@encoding_name)
+      len = str.length
+      if point <= length / 2
+        # count byte size from head
+        i = 0
+        while (pos < point && i < len)
+          pos += str[i].bytesize
+          i += 1
+        end
+      else
+        # count byte size from tail
+        pos = str.bytesize
+        i = len - 1
+        while (pos > point && i >= 0)
+          pos -= str[i].bytesize
+          i -= 1
+        end
+        pos += str[i + 1].bytesize if pos < point
       end
+      string.force_encoding(enc)
     else
       pos = point
     end

test/test_rbreadline.rb

@@ -6,4 +6,18 @@ def test_versions
     assert_equal('5.2', RbReadline::RL_LIBRARY_VERSION)
     assert_equal(0x0502, RbReadline::RL_READLINE_VERSION)
   end
+
+  def test_rl_adjust_point
+    encoding_name = RbReadline.instance_variable_get(:@encoding_name)
+    RbReadline.instance_variable_set(:@encoding_name, Encoding.find('UTF-8'))
+
+    assert_equal(0, RbReadline._rl_adjust_point("a".force_encoding('ASCII-8BIT'), 0))
+    assert_equal(0, RbReadline._rl_adjust_point("a".force_encoding('ASCII-8BIT'), 1))
+    assert_equal(0, RbReadline._rl_adjust_point(("a" * 40).force_encoding('ASCII-8BIT'), 0))
+    assert_equal(0, RbReadline._rl_adjust_point(("a" * 40).force_encoding('ASCII-8BIT'), 40))
+    assert_equal(2, RbReadline._rl_adjust_point(("\u3042" * 10).force_encoding('ASCII-8BIT'), 1))
+    assert_equal(1, RbReadline._rl_adjust_point(("\u3042" * 15).force_encoding('ASCII-8BIT'), 38))
+  ensure
+    RbReadline.instance_variable_set(:@encoding_name, encoding_name)
+  end if defined?(Encoding)
 end