[stringio] fix stringio codepoint enumerator off by one error

ruby · Aug 27, 2020 · b3c1c76 · b3c1c76
1 parent 96d701f
commit b3c1c76
Show file tree

Hide file tree

Showing 3 changed files with 24 additions and 1 deletion.
diff --git a/ext/stringio/stringio.c b/ext/stringio/stringio.c
@@ -1095,8 +1095,8 @@ strio_each_codepoint(VALUE self)
 
 	c = rb_enc_codepoint_len(RSTRING_PTR(ptr->string)+ptr->pos,
 				 RSTRING_END(ptr->string), &n, enc);
-	rb_yield(UINT2NUM(c));
 	ptr->pos += n;
+	rb_yield(UINT2NUM(c));
     }
     return self;
 }

diff --git a/test/ruby/test_io.rb b/test/ruby/test_io.rb
@@ -405,6 +405,19 @@ def test_each_codepoint
     }
   end
 
+  def test_each_codepoint_enumerator
+    make_tempfile {|t|
+      a = ""
+      b = ""
+      File.open(t, 'rt') {|f|
+        a = f.each_codepoint.take(4).pack('U*')
+        b = f.read(8)
+      }
+      assert_equal("foo\n", a)
+      assert_equal("bar\nbaz\n", b)
+    }
+  end
+
   def test_codepoints
     make_tempfile {|t|
       bug2959 = '[ruby-core:28650]'

diff --git a/test/stringio/test_stringio.rb b/test/stringio/test_stringio.rb
@@ -525,6 +525,16 @@ def test_each_codepoint
     assert_equal([49, 50, 51, 52], f.each_codepoint.to_a)
   end
 
+  def test_each_codepoint_enumerator
+    io = StringIO.new('你好построить')
+
+    chinese_part = io.each_codepoint.take(2).pack('U*')
+    russian_part = io.read(40).force_encoding('UTF-8')
+
+    assert_equal("你好", chinese_part)
+    assert_equal("построить", russian_part)
+  end
+
   def test_gets2
     f = StringIO.new("foo\nbar\nbaz\n")
     assert_equal("fo", f.gets(2))